Overcoming Controllable and Uncontrollable Work
Difficulties: Change Environment or Self?
Elise Maher, B.Sc. (Hons)
Submitted in fulfillment of the requirements for the degree of
Doctor of Philosophy
Deakin University, December 2002
ii
DEAKIN UNIVERSITY
CANDIDATE DECLARATION
I certify that the thesis entitled “Overcoming Controllable and Uncontrollable Work
Difficulties: Change Environment or Self?”
submitted for the degree of Doctor of Philosophy is the result of my own work and
that where reference is made to the work of others, due acknowledgment is given.
I also certify that any material in the thesis which has been accepted for a degree or
diploma by any other university or institution is identified in the text.
Full Name
Elise Catherine Maher
Signed ..................................................................................……………….
Date......................................................................................……………….
iii
Acknowledgements
First and foremost, I would like to acknowledge my supervisor, Professor
Robert Cummins. Bob offered endless support, guidance, and encouragement
throughout my studies. His dedication to research and his tremendous work ethic
motivated and inspired me. I am most grateful that he challenged me and allowed
me the freedom and respect to develop my own ideas and theories.
Second, I must thank all the organisations that allowed me to enter their
workplaces, and all the participants involved in the studies. I especially want to
thank staff at Australian Unity, members of the Australian Centre of Quality of Life,
and the hundreds of supermarket workers, teachers and academics that assisted me.
These people invested time and energy into completing my survey purely for the
benefit of helping others. I am so appreciative of their efforts and I am determined to
share the knowledge that I have gained from them.
Third, I would like to acknowledge the academic and administrative staff, and
fellow students at Deakin University. I especially want to acknowledge Rose-Anne
and Helen (my surrogate parents) for taking me under their wings. Their delightful,
vibrant personalities made work times pleasurable. Also, I would like to
acknowledge Carolyn and Catherine for always being there to listen and share.
Finally, I would like to acknowledge my family and friends. My parents,
John and Frances, and my brothers have provided endless support throughout the last
seven years. I also want to acknowledge Lauren, Taylah, Tyson and Buffy for
iv
always lighting up my life. My partner, Tim Davis, has been my tower of strength,
and I am forever grateful for his love and support.
Elise
v
Table of Contents
ACKNOWLEDGEMENTS ...................................................................................................................... III
LIST OF TABLES ...............................................................................................................................XIII
LIST OF FIGURES ............................................................................................................................. XVI
LIST OF APPENDICES...................................................................................................................... XVII
ABSTRACT .................................................................................................................................... XVIII
CHAPTER 1 - LITERATURE REVIEW .......................................................................................... 1
1.1
ABSTRACT ............................................................................................................................ 2
1.2
JOB SATISFACTION ............................................................................................................... 3
1.2.1
1.3
Theories of Job Satisfaction: Environmental and Dispositional Predictors ................... 3
MASLOW’S (1954, 1970) HIERARCHY OF NEEDS.................................................................. 5
1.3.1
Need Hierarchy Theory .................................................................................................. 5
1.3.2
Applying Maslow’s (1954, 1970) Theory to Organisations ............................................ 6
1.3.3
Criticisms of Maslow’s (1954, 1970) Need Hierarchy Theory ....................................... 7
1.3.4
Conclusion .................................................................................................................... 20
1.4
HERZBERG, MAUSNER AND SNYDERMAN’S (1959, 1993) TWO-FACTOR THEORY OF JOB
SATISFACTION .................................................................................................................................. 22
1.4.1
How the Two-Factor Theory has Contributed to our Understanding of Job
Satisfaction .................................................................................................................................. 22
1.4.2
Development of the Two-Factor Theory ....................................................................... 22
1.4.3
Criticisms of Herzberg et al’s., (1959) Theory ............................................................. 24
1.4.4
Conclusion .................................................................................................................... 33
1.5
VROOM’S (1964) EXPECTANCY THEORY OF JOB SATISFACTION ........................................ 34
1.5.1
How Expectancy Theory has Contributed to our Knowledge of Job Satisfaction ........ 34
1.5.2
Description of Expectancy Theory ................................................................................ 34
1.5.3
Applications of the Valence Model ............................................................................... 36
1.5.4
Studies of the Valence Model ........................................................................................ 36
vi
1.5.5
Methodological Limitations .......................................................................................... 37
1.5.6
Conclusion .................................................................................................................... 43
1.6
DISCREPANCY THEORIES .................................................................................................... 44
1.6.1
How Discrepancy Theories have Contributed to our Knowledge of Job Satisfaction .. 44
1.6.2
Description of Discrepancy Theories ........................................................................... 44
1.6.3
Empirical Studies Investigating Discrepancy Theories ................................................ 44
1.6.4
Theoretical Problems with Discrepancy Theories ........................................................ 45
1.6.5
Conclusion .................................................................................................................... 46
1.7
1.7.1
JOB CHARACTERISTICS MODEL (JCM; HACKMAN & OLDHAM, 1976) ............................... 47
How the Job Characteristics Model has Contributed to our Knowledge of Job
Satisfaction .................................................................................................................................. 47
1.7.2
Description of the Job Characteristics Model .............................................................. 47
1.7.3
Empirical Studies of the Model ..................................................................................... 50
1.7.4
Conclusion .................................................................................................................... 56
1.8
1.8.1
JOB DEMAND-CONTROL MODEL (KARASEK, 1979; KARASEK & THEORELL, 1990) .......... 57
How the Job Demand-Control Model Contributes to our Understanding of Job
Satisfaction .................................................................................................................................. 57
1.8.2
Description of the Job Demand-Control Model ........................................................... 57
1.8.3
Empirical Studies of the Job Demand-Control Model .................................................. 59
1.8.4
Conclusion .................................................................................................................... 62
1.8.5
Extensions on the Job Demand-Control Model ............................................................ 63
1.8.6
Addressing the “Gaps” in the Job Demand-Control Model ......................................... 63
1.9
DEVELOPMENT OF A NEW EXPLANATION FOR THE RELATIONSHIP BETWEEN JOB
AUTONOMY AND JOB SATISFACTION: INFLUENCING EMPLOYEES’ RESPONSES TO WORK
DIFFICULTIES.................................................................................................................................... 65
1.9.1
a) Primary Control Strategies and Secondary Control Strategies .............................. 66
1.9.2
b) Amounts of Primary Control and Secondary Control .............................................. 68
1.9.3
c) Which Control Strategies are more Adaptive for Employees? ................................. 69
vii
1.9.4
1.10
Summary ....................................................................................................................... 75
EXPLAINING THE RELATIONSHIP BETWEEN JOB AUTONOMY AND JOB SATISFACTION: HOW
JOB AUTONOMY INFLUENCES PRIMARY AND SECONDARY CONTROL............................................... 75
1.10.1
1) Use of Primary and Secondary Control............................................................... 76
1.10.2
2) Adaptiveness of Primary and Secondary Control ............................................... 77
1.10.3
Summary .................................................................................................................. 86
1.11
OTHER MAJOR PREDICTORS OF JOB SATISFACTION ........................................................... 87
1.11.1
Personality ............................................................................................................... 87
1.11.2
Life Satisfaction ....................................................................................................... 90
1.12
MODEL OF JOB SATISFACTION ........................................................................................... 97
2 CHAPTER 2 - STUDY ONE ........................................................................................................ 100
2.1
ABSTRACT ........................................................................................................................ 101
2.2
PROPOSAL FOR STUDY ONE.............................................................................................. 102
2.2.1
Identifying Employees with Low/High Job Autonomy ................................................ 102
2.3
AIMS AND HYPOTHESES ................................................................................................... 104
2.4
METHOD........................................................................................................................... 108
2.4.1
Participants ................................................................................................................ 108
2.4.2
Materials ..................................................................................................................... 108
2.4.3
Procedure ................................................................................................................... 114
2.5
RESULTS........................................................................................................................... 116
2.5.1
Data Screening and Checking of Assumptions ........................................................... 116
2.5.2
Descriptive Statistics and Inter-Correlations ............................................................. 118
2.5.3
Factor Analyses .......................................................................................................... 120
2.5.4
Factor Analysis of the Job Descriptive Index ............................................................. 120
2.5.5
Factor Analysis of the Primary and Secondary Control Scale ................................... 122
2.5.6
Factor Analysis of the Job Autonomy Scale................................................................ 125
2.6
HYPOTHESIS TESTING ...................................................................................................... 127
viii
2.6.1
Hypothesis One- Assumption Testing ......................................................................... 127
2.6.2
Hypothesis Two- Occupational Differences in the Use of the Control Strategies ..... 128
2.6.3
Hypothesis Three- Examining how Job Autonomy Relates to the Control Strategies. 130
2.6.4
Hypothesis Four- Examining how Job Autonomy Influences the Adaptiveness of the
Control Strategies ..................................................................................................................... 131
2.6.5
Hypothesis Five- Does Job Autonomy Moderate the Relationship Between the Control
Strategies and Job Satisfaction? ............................................................................................... 132
2.6.6
Hypothesis Six- Do the Control Strategies Mediate the Relationship Between Job
Autonomy and Job Satisfaction? ............................................................................................... 137
2.6.7
Hypothesis Seven- Occupational Differences in Job and Life Satisfaction ................ 142
2.6.8
Hypothesis Eight- Predictors of Job Satisfaction ....................................................... 146
2.6.9
Conclusion .................................................................................................................. 148
2.7
DISCUSSION...................................................................................................................... 150
2.7.1
Assumption Testing ..................................................................................................... 150
2.7.2
Does Job Autonomy Influence the Use of the Control Strategies? ............................. 152
2.7.3
Does Job Autonomy Influence the Relationship Between the Control Strategies and Job
Satisfaction? .............................................................................................................................. 154
2.7.4
Do the Control Strategies Mediate the Relationship Between Job Autonomy and Job
Satisfaction? .............................................................................................................................. 156
2.7.5
Examining Occupational Differences in Job Satisfaction .......................................... 158
2.7.6
Examining Occupational Differences in Life Satisfaction .......................................... 163
2.7.7
Predicting Job Satisfaction from Job Autonomy, Control Strategies, Personality, and
Life Satisfaction ........................................................................................................................ 164
2.7.8
Conclusion .................................................................................................................. 166
3 CHAPTER 3 - STUDY TWO ....................................................................................................... 167
3.1
ABSTRACT ........................................................................................................................ 168
3.2
PROPOSAL FOR STUDY TWO ............................................................................................. 169
ix
3.2.1
a) The Primary and Secondary Control Scale ............................................................ 169
3.2.2
b) Job Autonomy Scale ............................................................................................... 181
3.2.3
c) Occupational Groups.............................................................................................. 182
3.2.4
d) Need for Job Autonomy .......................................................................................... 184
3.2.5
e) Addition of Social Support ...................................................................................... 186
3.3
MODEL OF JOB SATISFACTION ......................................................................................... 189
3.4
AIMS AND HYPOTHESES ................................................................................................... 191
3.5
METHOD........................................................................................................................... 195
3.5.1
Participants ................................................................................................................ 195
3.5.2
Materials ..................................................................................................................... 196
3.5.3
Procedure ................................................................................................................... 203
3.6
RESULTS........................................................................................................................... 204
3.6.1
Data Screening and Checking of Assumptions ........................................................... 204
3.6.2
Descriptive Statistics and Inter-Correlations ............................................................. 205
3.6.3
Preliminary Examination of the Primary Control and Secondary Control Scale....... 206
3.7
HYPOTHESIS TESTING ...................................................................................................... 211
3.7.1
Hypothesis One: Levels of Job Autonomy and Job Satisfaction ................................. 211
3.7.2
Hypotheses Two and Three: Examining how Job Autonomy Influences the Amount of
Primary and Secondary Control Strategies .............................................................................. 212
3.7.3
Hypotheses Four and Five: Examining how Job Autonomy Influences the Relationship
Between the Control Strategies and Job Satisfaction ............................................................... 213
3.7.4
Hypothesis Six: Examining the Proposed Explanation for the Relationship Between Job
Autonomy and Job Satisfaction ................................................................................................. 216
3.7.5
Hypothesis Seven: Occupational Differences in Job Satisfaction and Life
Satisfaction ................................................................................................................................ 218
3.7.6
Hypothesis Eight: Examining how Social Support at Work Moderates the Relationship
between Difficulties at Work and Job Satisfaction .................................................................... 222
x
3.7.7
Hypothesis Nine: The Moderating Role of Need for Autonomy on the Relationship
Between Job Autonomy and Job Satisfaction ............................................................................ 227
3.7.8
Hypothesis Ten: Major Predictors of Job Satisfaction ............................................... 228
3.7.9
Conclusion .................................................................................................................. 231
3.8
DISCUSSION...................................................................................................................... 232
3.8.1
Assumption- The Academics Represent a High Job Autonomy Group and the Teachers
Represent a Low Job Autonomy Group..................................................................................... 232
3.8.2
Hypothesis Testing ...................................................................................................... 235
3.8.3
Job Autonomy Influences the Amount of the Control Strategies ................................. 235
3.8.4
Job Autonomy Influences the Relationship Between the Control Strategies and Job
Satisfaction ................................................................................................................................ 238
3.8.5
Limitations in the Hypotheses Examining Job Autonomy and Control Strategies ...... 239
3.8.6
Other Predictors of Job Satisfaction........................................................................... 241
3.8.7
Occupational Differences in Job Satisfaction and Life Satisfaction ........................... 241
3.8.8
The Influence of Social Support at Work on the Relationship Between Work Difficulties
and Job Satisfaction .................................................................................................................. 245
3.8.9
The Influence that Need for Job Autonomy has on the Relationship Between Job
Autonomy and Job Satisfaction ................................................................................................. 246
3.8.10
Major predictors of Job Satisfaction...................................................................... 247
3.8.11
Conclusion ............................................................................................................. 248
4 CHAPTER 4 - STUDY THREE ................................................................................................... 249
4.1
ABSTRACT ........................................................................................................................ 250
4.2
PROPOSAL FOR STUDY THREE .......................................................................................... 251
4.2.1
Specificity of Hypotheses Testing the Proposal that Job Autonomy Influences the
Control Strategies ..................................................................................................................... 251
4.2.2
Examining how the Controllability of a Difficulty Influences the Use of the Control
Strategies................................................................................................................................... 252
xi
4.2.3
Empirical Studies Examining if the Controllability of a Situation Influences the Use of
Control Strategies ..................................................................................................................... 253
4.2.4
Examining how Controllability Influences the Adaptiveness of the Control
Strategies................................................................................................................................... 257
4.2.5
Developing a Situation Specific Primary and Secondary Control Scale .................... 261
4.2.6
Examining the Moderating Role of Primary and Secondary Control Strategies ........ 265
4.2.7
Examining the Moderating Role of Social Support at Work ....................................... 267
4.3
REVISED MODEL OF JOB SATISFACTION ........................................................................... 269
4.4
HYPOTHESES .................................................................................................................... 272
4.5
METHOD........................................................................................................................... 274
4.5.1
Participants ................................................................................................................ 274
4.5.2
Materials ..................................................................................................................... 275
4.5.3
Procedure ................................................................................................................... 280
4.6
RESULTS........................................................................................................................... 281
4.6.1
Data Screening and Checking of Assumptions ........................................................... 281
4.6.2
Descriptive Statistics and Inter-Correlations ............................................................. 282
4.6.3
Factor Analyses .......................................................................................................... 284
4.6.4
Primary and Secondary Control Scale ....................................................................... 284
4.6.5
Social Support at Work ............................................................................................... 289
ITEMS WITH LOADINGS LESS THAN 0.30 ARE NOT SHOWN. .............................................................. 290
4.7
4.7.1
HYPOTHESIS TESTING ...................................................................................................... 291
Hypothesis One- Use of Control Strategies for Controllable and Uncontrollable
Difficulties ................................................................................................................................. 291
4.7.2
Hypothesis Two- Adaptiveness of the Control Strategies for Controllable and
Uncontrollable Difficulties........................................................................................................ 295
4.7.3
Hypothesis Three- The Moderating Role of Primary and Secondary Control ............ 298
4.7.4
Hypothesis Four - Moderating Role of Instrumental Support .................................... 303
4.7.5
Hypothesis Five- Moderating Role of Emotional Support .......................................... 307
xii
4.7.6
Hypothesis Six- Major Predictors of Job Satisfaction ................................................ 310
4.7.7
Conclusion .................................................................................................................. 311
4.8
DISCUSSION...................................................................................................................... 313
4.9
HYPOTHESES TESTING ..................................................................................................... 313
4.9.1
Primary Control, Self-Protective Secondary Control, and Self-Affirmative
SecondaryControl ..................................................................................................................... 314
4.9.2
Proposal One: The Controllability of the Difficulty Influences the Amount and
Adaptiveness of the Control Strategies Used to Manage that Difficulty ................................... 317
4.9.3
Proposal Two: Moderators of Controllable and Uncontrollable Difficulties on Job
Satisfaction ................................................................................................................................ 329
4.9.4
Proposal Three: Predictors of Job Satisfaction .......................................................... 334
4.9.5
Conclusion .................................................................................................................. 336
5 CHAPTER 5 - FINAL DISCUSSION .......................................................................................... 338
5.1
ABSTRACT ........................................................................................................................ 339
5.2
THE DEVELOPMENT OF A NEW MODEL OF JOB SATISFACTION ........................................ 340
1)
Primary and Secondary Control Strategies Mediate the Relationship Between Job
Autonomy and Job Satisfaction ................................................................................................. 342
5.2.2
Conclusion: Do the Control Strategies Mediate the Relationship Between Job
Autonomy and Job Satisfaction? ............................................................................................... 350
5.2.3
2) Social Support at Work and Life Satisfaction Directly Predict Job Satisfaction .... 351
5.2.4
3) The Control Strategies and Social Support at Work Moderate the Relationship
Between Work Difficulties and Job Satisfaction ....................................................................... 354
5.3
REVISED MODEL OF JOB SATISFACTION ........................................................................... 357
5.4
CONCLUSION .................................................................................................................... 360
5.5
FINAL WORD .................................................................................................................... 361
5.6
REFERENCES .................................................................................................................... 362
5.7
APPENDICES ..................................................................................................................... 399
xiii
List of Tables
TABLE 1- SOURCES OF GOOD/BAD TIMES FOR ACCOUNTANTS AND ENGINEERS (N=200) .................... 23
TABLE 2- SECONDARY CONTROL STRATEGIES ...................................................................................... 74
TABLE 3- MEANS AND STANDARD DEVIATIONS OF MAJOR VARIABLES FOR ACADEMICS AND
SUPERMARKET WORKERS.......................................................................................................... 118
TABLE 4- INTER-CORRELATIONS FOR THE ACADEMICS AND THE SUPERMARKET WORKERS .............. 119
TABLE 5- FACTOR ANALYSIS OF JOB SATISFACTION SCALE................................................................ 121
TABLE 6- TOTAL VARIANCE EXPLAINED BY A FIVE-FACTOR SOLUTION............................................. 123
TABLE 7- FACTOR ANALYSIS OF PRIMARY AND SECONDARY CONTROL SCALE .................................. 124
TABLE 8- FACTOR ANALYSIS OF JOB AUTONOMY SCALE.................................................................... 126
TABLE 9- MEANS AND STANDARD DEVIATIONS OF CONTROL MEASURES FOR ACADEMICS AND
SUPERMARKET WORKERS.......................................................................................................... 130
TABLE 10- MULTIPLE REGRESSION OF PRIMARY AND SECONDARY CONTROL ON JOB SATISFACTION FOR
ACADEMICS AND SUPERMARKET WORKERS .............................................................................. 132
TABLE 11- MODERATING ROLE OF PRIMARY AND SECONDARY CONTROL ON THE RELATIONSHIP
BETWEEN JOB AUTONOMY AND JOB SATISFACTION .................................................................. 137
TABLE 12 -HIERARCHICAL MULTIPLE REGRESSION TESTING THE MEDIATING ROLE OF THE CONTROL
STRATEGIES ............................................................................................................................... 140
TABLE 13- MEANS AND STANDARD DEVIATIONS OF JOB SATISFACTION SCALE FOR ACADEMICS AND
SUPERMARKET WORKERS.......................................................................................................... 144
TABLE 14- MEANS AND STANDARD DEVIATIONS OF LIFE SATISFACTION FOR ACADEMICS AND
SUPERMARKET WORKERS.......................................................................................................... 146
TABLE 15- MULTIPLE REGRESSION OF JOB AUTONOMY, CONTROL STRATEGIES, PERSONALITY, AND
LIFE SATISFACTION FOR ACADEMICS AND SUPERMARKET WORKERS ....................................... 148
TABLE 16- FACTOR ANALYSIS OF PRIMARY AND SECONDARY CONTROL SCALE ................................ 171
xiv
TABLE 17- ORIGINAL AND REVISED PRIMARY CONTROL ITEMS ......................................................... 175
TABLE 18- ORIGINAL AND REVISED SECONDARY CONTROL ITEMS .................................................... 179
TABLE 19- FUNCTIONS OF THE SECONDARY CONTROL STRATEGIES ................................................... 179
TABLE 20- DEMOGRAPHICS OF THE ACADEMICS AND TEACHERS ....................................................... 196
TABLE 21- FACTOR ANALYSIS OF THE NEED FOR JOB AUTONOMY SCALE .......................................... 199
TABLE 22- MEANS AND STANDARD DEVIATIONS OF MAJOR VARIABLES FOR ACADEMICS AND
TEACHERS ................................................................................................................................. 205
TABLE 23- INTER-CORRELATIONS FOR THE ACADEMICS AND TEACHERS ........................................... 206
TABLE 24- FREQUENCY OF PRIMARY AND SECONDARY CONTROL...................................................... 207
TABLE 25- FACTOR ANALYSIS OF THE REVISED PRIMARY AND SECONDARY CONTROL SCALE .......... 209
TABLE 26- MULTIPLE REGRESSION OF PRIMARY AND SECONDARY CONTROL ON JOB SATISFACTION FOR
ACADEMICS AND TEACHERS ...................................................................................................... 214
TABLE 27- HIERARCHICAL MULTIPLE REGRESSION TESTING THE MODERATING ROLE OF THE CONTROL
STRATEGIES ON THE RELATIONSHIP BETWEEN JOB AUTONOMY AND JOB SATISFACTION ......... 216
TABLE 28- HIERARCHICAL MULTIPLE REGRESSION TESTING THE MEDIATING ROLE OF THE CONTROL
STRATEGIES ............................................................................................................................... 217
TABLE 29- MEANS AND STANDARD DEVIATIONS OF THE INTRINSIC AND EXTRINSIC JOB SATISFACTION
ITEMS FOR ACADEMICS AND TEACHERS .................................................................................... 220
TABLE 30- MEANS AND STANDARD DEVIATIONS OF THE DOMAINS OF LIFE SATISFACTION FOR
ACADEMICS AND TEACHERS ...................................................................................................... 221
TABLE 31- HIERARCHICAL MULTIPLE REGRESSION ANALYSIS EXAMINING IF SUPERVISOR SUPPORT
MODERATES THE RELATIONSHIP BETWEEN WORK DIFFICULTIES AND JOB SATISFACTION ....... 223
TABLE 32- HIERARCHICAL REGRESSION ANALYSES EXAMINING WHETHER CO-WORKER SUPPORT
MODERATES THE RELATIONSHIP BETWEEN WORK DIFFICULTIES AND JOB SATISFACTION. ...... 226
TABLE 33- HIERARCHICAL REGRESSION ANALYSES EXAMINING WHETHER NEED FOR JOB AUTONOMY
MODERATES THE RELATIONSHIP BETWEEN JOB AUTONOMY AND JOB SATISFACTION. ............. 228
TABLE 34- STANDARD MULTIPLE REGRESSION PREDICTING JOB SATISFACTION FOR EMPLOYEES WITH
LOW AUTONOMY AND EMPLOYEES WITH HIGH AUTONOMY ..................................................... 230
xv
TABLE 35- NORMATIVE DATA FOR HACKMAN AND OLDHAM’S (1980) AUTONOMY SCALE ............... 234
TABLE 36- DEMOGRAPHIC CHARACTERISTICS OF THE SAMPLE........................................................... 274
TABLE 37- MEANS AND STANDARD DEVIATIONS OF THE MAJOR VARIABLES .................................... 283
TABLE 38- INTER-CORRELATIONS AMONG MAJOR VARIABLES .......................................................... 283
TABLE 39- FACTOR ANALYSIS OF PRIMARY AND SECONDARY CONTROL ITEM IN CONTROLLABLE
SITUATIONS ............................................................................................................................... 286
TABLE 40- FACTOR ANALYSIS OF PRIMARY AND SECONDARY CONTROL ITEMS IN UNCONTROLLABLE
SITUATION ................................................................................................................................. 288
TABLE 41-SECONDARY CONTROL ITEMS INCLUDED IN ANALYSES ..................................................... 289
TABLE 42- FACTOR ANALYSIS OF THE SOCIAL SUPPORT AT WORK SCALE ......................................... 290
TABLE 43 - CONTROLLABLE AND UNCONTROLLABLE DIFFICULTIES REPORTED BY EMPLOYEES........ 292
TABLE 44- EMPLOYEES USE OF PRIMARY AND SECONDARY CONTROL IN CONTROLLABLE AND
UNCONTROLLABLE SITUATIONS ................................................................................................ 293
TABLE 45- MEANS AND STANDARD DEVIATIONS OF INDIVIDUAL CONTROL STRATEGIES .................. 295
TABLE 46- STANDARD MULTIPLE REGRESSION ANALYSIS PREDICTING JOB SATISFACTION FROM
PRIMARY AND SECONDARY CONTROL ....................................................................................... 297
TABLE 47- CORRELATIONS BETWEEN INDIVIDUAL CONTROL STRATEGIES AND JOB SATISFACTION FOR
CONTROLLABLE AND UNCONTROLLABLE DIFFICULTIES ........................................................... 298
TABLE 48- HIERARCHICAL MULTIPLE REGRESSION TESTING THE MODERATING ROLE OF CONTROL
STRATEGIES ON THE RELATIONSHIP BETWEEN WORK DIFFICULTIES AND JOB SATISFACTION ... 302
TABLE 49- HIERARCHICAL REGRESSION ANALYSES TESTING THE MODERATING ROLE OF
INSTRUMENTAL SUPPORT .......................................................................................................... 307
TABLE 50- HIERARCHICAL REGRESSION ANALYSES TESTING THE MODERATING ROLE OF EMOTIONAL
SUPPORT .................................................................................................................................... 309
TABLE 51- STANDARD MULTIPLE REGRESSION PREDICTING JOB SATISFACTION ................................ 311
xvi
List of Figures
FIGURE 1-JOB CHARACTERISTICS MODEL ............................................................................................. 49
FIGURE 2- MODEL OF JOB SATISFACTION .............................................................................................. 99
FIGURE 3- EXPECTED MODERATED EFFECT OF JOB AUTONOMY ON A) PRIMARY CONTROL AND JOB
SATISFACTION AND B) SECONDARY CONTROL AND JOB SATISFACTION .................................... 133
FIGURE 4- JOB AUTONOMY MODERATES THE RELATIONSHIP BETWEEN A) PRIMARY CONTROL AND B)
SECONDARY CONTROL, AND JOB SATISFACTION. ...................................................................... 135
FIGURE 5- MEDIATING ROLE OF CONTROL STRATEGIES ON THE RELATIONSHIP BETWEEN JOB
AUTONOMY AND JOB SATISFACTION ......................................................................................... 138
FIGURE 6 -REVISED MODEL OF JOB SATISFACTION FOR STUDY 2 ....................................................... 190
FIGURE 7 - RELATIONSHIP BETWEEN WORK DIFFICULTIES AND JOB SATISFACTION FOR EMPLOYEES
WITH LOW/HIGH SUPERVISOR SUPPORT .................................................................................... 225
FIGURE 8- REVISED MODEL OF JOB SATISFACTION ............................................................................. 271
FIGURE 9 – PRIMARY AND SECONDARY CONTROL MODERATE THE RELATIONSHIP BETWEEN WORK
DIFFICULTIES AND JOB SATISFACTION....................................................................................... 300
FIGURE 10 - REGRESSION OF CONTROLLABLE WORK DIFFICULTIES ON JOB SATISFACTION FOR
EMPLOYEES WITH LOW INSTRUMENTAL CO-WORKER SUPPORT AND EMPLOYEES WITH HIGH
INSTRUMENTAL CO-WORKER SUPPORT ..................................................................................... 305
FIGURE 11- REVISED MODEL OF JOB SATISFACTION ........................................................................... 359
xvii
List of Appendices
APPENDIX A- PLAIN LANGUAGE STATEMENT FOR STUDY ONE .......................................................... 400
APPENDIX B- JOB AUTONOMY SCALE USED IN STUDY ONE (REVISION OF GANSTER, 1989, CITED IN
DWYER & GANSTER, 1991) ....................................................................................................... 401
APPENDIX C- PRIMARY AND SECONDARY CONTROL SCALE USED IN STUDY ONE (REVISION OF HEEPS
ET AL., 2000) ............................................................................................................................. 403
APPENDIX D- JOB SATISFACTION SCALE USED IN STUDY ONE (REVISION OF ROZNOWSKI, 1989) ...... 406
APPENDIX E- LIFE SATISFACTION SCALE USED IN STUDY ONE (CUMMINS, 1997) .............................. 408
APPENDIX F- PERSONALITY SCALE USED IN STUDY ONE (COSTA & MCCRAE, 1992) ......................... 409
APPENDIX G-LEVELS OF JOB SATISFACTION REPORTED BY VARIOUS OCCUPATIONAL GROUPS ........... 412
APPENDIX H- PRIMARY AND SECONDARY CONTROL SCALE FOR STUDY TWO (MAHER ET AL., 2001) 415
APPENDIX I- PLAIN LANGUAGE STATEMENT FOR STUDY TWO ........................................................... 419
APPENDIX J- JOB AUTONOMY SCALE FOR STUDY TWO (HACKMAN & OLDHAM, 1975) ..................... 420
APPENDIX K- NEED FOR AUTONOMY SCALE FOR STUDY TWO (DE RIJK ET AL., 1998) ....................... 421
APPENDIX L- JOB SATISFACTION SCALE FOR STUDY TWO (WEISS ET AL., 1967) ................................ 422
APPENDIX M- SOCIAL SUPPORT SCALE FOR STUDY TWO (REVISION OF KARASEK & THEORELL, 1990)
.................................................................................................................................................. 425
APPENDIX N-PLAIN LANGUAGE STATEMENT USED IN STUDY THREE ................................................. 426
APPENDIX O- PRIMARY AND SECONDARY CONTROL SCALE FOR STUDY THREE (MAHER & CUMMINS,
2002) ......................................................................................................................................... 427
APPENDIX P- LIFE SATISFACTION SCALE FOR STUDY 3 (CUMMINS ET AL., 2001) ............................... 433
APPENDIX Q- SOCIAL SUPPORT SCALE FOR STUDY 3 (DUCHARME & MARTIN, 2000) ........................ 434
xviii
Abstract
Although theories of job satisfaction have been extensively studied, researchers are
yet to agree on the major predictors of job satisfaction. One theory, which is
particularly appealing to the workplace, is Karasek and Theorell’s (1990) job
demand-control model. Essentially, this model proposes that job autonomy can
reduce the effects of job demands on job satisfaction by allowing workers to redirect
the physiological arousal produced from job demands into an appropriate response.
This explanation is criticised however for being tautological, and a new explanation
is developed which incorporates the life span theory of control (Heckhausen &
Schulz, 1995) and the discrimination model (Thompson et al., 1998). Specifically it
is proposed that job autonomy influences the use, and the adaptiveness of primary
and secondary control strategies. This proposal is developed into a model of job
satisfaction that includes job autonomy, primary and secondary control, life
satisfaction, personality, and social support at work. This model of job satisfaction is
tested over three studies using university academic staff, secondary school teachers,
supermarket workers and general employees. Overall, the results demonstrated that
job autonomy did not influence the use or adaptiveness of the control strategies.
These results suggest that employees have trait control strategies, and they also
challenge the assumptions about primary control failure. The proposed model of job
satisfaction was revised to include job autonomy, primary and secondary control
strategies and their successfulness, life satisfaction, work difficulties, and
personality.
1
Chapter 1 - Literature Review
2
1.1
Abstract
Although theories of job satisfaction have been extensively researched in the
organisational psychology literature, researchers are yet to agree on the major
predictors of job satisfaction. Several predictors have been investigated such as
needs, values, expectations and specific job characteristics such as job autonomy and
job demands. This chapter reviews such theories, focussing on the ones that have
made the greatest contribution to the understanding of job satisfaction. Although
these theories are well cited, many of them have theoretical and empirical problems
as well as having limited applicability to the workplace. One theory, which is less
problematic, and particularly appealing to the workplace, is the job demand-control
model. This model proposes that job autonomy can reduce the effect of job demands
on job satisfaction, and that the most satisfied workers are those with high job
demands and high job autonomy. According to the model, job autonomy influences
job satisfaction because it allows workers to redirect the physiological arousal
produced from job demands into an appropriate response. This explanation is
criticised however for being non-specific and tautological. A new explanation is
developed, where it is proposed that job autonomy influences how employees
respond to work difficulties. This explanation forms the basis of a model of job
satisfaction, which includes the following predictors: job autonomy; primary control
and secondary control; personality; and life satisfaction.
3
1.2
Job Satisfaction
Job satisfaction, the extent to which employees like their job and its
components (Spector, 1997), is one of the most extensively researched topic in the
industrial and organisational psychology literature (Highhouse & Becker, 1993).
The number of articles and books investigating this construct has increased from
over 3000 in 1976 (Locke, 1976), to over 5000 in 1992 (Harwood & Rice, 1992).
Today, a review of psychology and business databases demonstrates that over 10,000
publications on job satisfaction are available. Although this increasing interest in job
satisfaction is no doubt beneficial to the field of industrial and organisational
psychology, the amount of research has become overwhelming to both researchers
and practitioners. Nowhere is this more clearly evident than in the theories of job
satisfaction.
1.2.1
Theories of Job Satisfaction: Environmental and Dispositional
Predictors
Theories of job satisfaction include dispositional and environmental
predictors. The dispositional predictors of job satisfaction refer to characteristics of
the employee, such as needs, values, and expectations. The environmental predictors
refer to job characteristics, such as job control, workload, feedback, role ambiguity,
and role conflict. Some theorists focus on the dispositional predictors, whilst others
focus on the environmental predictors. More recent theorists recognise the
importance of both types of predictors.
4
Dispositional and environmental theories of job satisfaction have been
extensively researched, however researchers have still not reached consensus as to
the major predictors of job satisfaction. As a result, researchers continue to rely on
theories that have theoretical and empirical problems, or have limited applicability to
the workplace. In order to determine which theories are valid and useful, this review
will examine the theories that have made the greatest contribution to a shift in focus
of the determinants of job satisfaction. These include Maslow’s (1970) need
hierarchy theory, Herzberg, Mausner and Snyderman’s (1959) two-factor theory of
job satisfaction, Vroom’s (1964) expectancy theory, discrepancy theories, Hackman
and Oldham’s (1976) job characteristics model, and Karasek’s (1979) job demandcontrol model.
5
1.3
Maslow’s (1954, 1970) Hierarchy of Needs
The need hierarchy theory was one of the first theories to focus on the
dispositional predictors of job satisfaction. It proposed that employees’ needs
determine their level of job satisfaction.
1.3.1
Need Hierarchy Theory
The need hierarchy theory (Maslow, 1954, 1970) posits that individuals are
born with a set of needs. There are five needs: physiological, safety, belongingness,
esteem, and self-actualisation. These are arranged in a hierarchy of relative
prepotency, meaning that lower-order needs are satisfied before higher-order needs
are activated.
The lowest need, physiological, refers to basic biological drives, such as
hunger, thirst and sex. These physiological needs are the most prepotent of all, as an
individual deprived of all needs would seek to gratify these needs first. They would
not be concerned with safety, belongingness, esteem, or self-actualisation. Once they
have gratified the physiological needs however, the strength of that need decreases,
and the next highest need, safety, becomes important.
The safety need refers to security, stability, dependency, protection, and need
for structure, order, law and limits. To gratify the safety need, an individual requires
a safe, orderly, predictable, lawful world. Once the safety need is gratified, its need
strength is reduced, and the strength of the belongingness need increases. The
individual will begin to hunger for affectionate relationships with people, and for a
6
place in their group or family. Once these belongingness needs are gratified, the
strength of the esteem need increases, and the individual will desire a high evaluation
of themselves, and others. Once an individual has gratified these four needs,
collectively known as deficiency needs (D-needs), they may begin to feel restless.
This restlessness is indicative of the need for self-actualisation.
The need for self-actualisation refers to the need for the individual to become
everything they are capable of becoming. When the strength of this need increases,
the individual strives for self-fulfilment. This fifth need is referred to as a being need
(B-need) because it sustains an individual’s interest without being driven by feelings
of deprivation. Unlike the previous four needs, when the need for self-actualisation is
gratified, it increases in need strength (Maslow, 1962). Growth is a continued
upward development, where the more that one gets, the more that one wants. This
growth is “endless, and can never be attained or satisfied” (Maslow, 1962, p. 31).
1.3.2
Applying Maslow’s (1954, 1970) Theory to Organisations
In terms of applying this theory to organisations, the theory proposes that the
lower-order needs must be gratified before the higher-order needs are activated. As
such, employers must ensure that their employees’ physiological, safety,
belongingness and esteem needs are satisfied. The employer can help the employee
to gratify each need. For example, to help them gratify their physiological and safety
needs, employers can increase their employees’ pay. Once these needs are satisfied,
the relationship between the employee and their supervisors and co-workers takes on
increased strength. The employer can help the employee to gratify this need through
7
increasing the amount of social interaction among employees. This process needs to
be continued until the employees have gratified all of the lower-order needs, and are
reaching for self-actualisation, should the nature of the job permit this level to be
attained.
1.3.3
Criticisms of Maslow’s (1954, 1970) Need Hierarchy Theory
Almost every aspect of Maslow’s (1954, 1970) work has been disputed on
both theoretical and empirical grounds (Neher, 1991; Wahba & Bridwell, 1976).
Five fundamental propositions of Maslow’s (1954, 1970) theory have been
questioned, including: 1) the higher the deprivation of a need, the higher its need
strength (i.e., deprivation/domination paradigm); 2) the higher the satisfaction with a
need, the higher the need strength of the need at the next level (i.e.,
gratification/activation paradigm); 3) the measurement of self-actualisation; 4) the
ability to achieve self-actualisation; and 5) the applicability of the theory to
organisations. Each of these will now be considered.
1.3.3.1
Criticism One: Deprivation/Domination Paradigm
The deprivation/domination paradigm postulates that the higher the
deprivation of a need, the higher its need strength. An early review concluded that
the deprivation/domination paradigm was only partially supported for selfactualisation, and not supported for safety, belongingness and esteem needs (Wahba
& Bridwell, 1976). On the basis of this review, many researchers have assumed that
the proposition is not supported (Wicker, Brown, Wiehe, Hagen & Reed, 1993).
8
This assumption may be inaccurate however, as many of the studies included in the
review have methodological limitations. These limitations concern: a) the
operationalisation of need strength; and b) establishing causality.
1.3.3.2
a) Operationalising Need Strength
One of the main limitations in studies examining the deprivation/domination
paradigm concerns the operationalisation of need strength. Some researchers
measure need strength through desire, others through important or intention. Two
studies have measured need strength through desire. In Alderfer’s (1969) study,
subjects were asked to rate how much more of the following factors they would like
to have in their jobs; pay, fringe benefits, love, status, and growth. Similarly, in
Graham and Balloun’s (1973) study, subjects were asked how much improvement
they wanted in their physiological, security, social and self-actualisation needs.
These measures of need strength were then correlated with corresponding measures
of satisfaction.
Both studies provided some support for Maslow’s (1954) theory suggesting
that as satisfaction with a need increases, the strength of that need decreases. For
example, in Graham and Balloun’s (1973) study, the correlations between need
strength and satisfaction ranged from r = –0.42 to r = –0.72. Furthermore, in
Alderfer’s (1969) study, satisfaction and need strength were negatively correlated for
the relatedness need, which was composed of a respect from co-workers’ need, and a
respect from supervisors’ need. For the respect from co-workers’ need, the
correlations were all significant, ranging from r = -0.21 to r = -0.38. For the respect
9
from supervisors’ need, the correlations ranged from r = -0.06 to r = -0.49. Although
the correlations in Alderfer’s (1969) study were in the expected direction, they were
often small, and the correlations between satisfaction and need strength for the
belongingness need were insignificant (r = 0.02 to r = 0.07).
These two studies appear to provide some support for Maslow's (1954)
theory. Both of these studies assessed need strength ratings by desire, where
participants were asked how much more they wanted of a need. It must be
questioned however, if wanting or desiring more of a need is a measure of the
strength of the need. Wanting more of a need may actually be another way of
demonstrating dissatisfaction with the area covered by that need.
Other researchers have overcome this limitation by assessing need strength
using importance ratings, which may be less likely to measure satisfaction. For
example, Hall and Nougaim (1968) conducted a longitudinal study on managers,
interviewing them annually for five years. The participants rated the importance of,
and satisfaction with a number of needs including safety, affiliation, achievement and
esteem, and self-actualisation. Inconsistent with Maslow’s (1954) theory, the
correlations between the satisfaction of needs and the importance of needs were
positive. For safety, importance and satisfaction correlated r = 0.26, for affiliation
r = 0.16, for achievement and esteem r = 0.54, and for self-actualisation r = 0.29.
In addition, Hall and Nougaim (1968) also examined the longitudinal changes
in satisfaction and importance for each need. According to Maslow’s (1954) theory,
it would be expected that if satisfaction of a need increased from one year to the next,
importance of that need would decrease. However, they found that the importance of
10
a need in a given year was positively correlated with its own satisfaction in the
previous year. These correlations were moderate for safety (r = 0.25), affiliation
(r = 0.21), achievement and esteem (r = 0.53) and self-actualisation (r = 0.28).
Although Hall and Nougaim (1968) failed to discuss these correlations in detail, they
clearly contradict Maslow’s (1954) theory. Importance was positively related to
need satisfaction, suggesting that a satisfied need is an important need. This finding
does not support Maslow’s (1970, p. 393) proposal that “a satisfied need is not a
motivator.”
Although Hall and Nougaim’s (1968) findings are inconsistent with
Maslow’s (1954) theory, their validity has been questioned. Specifically, the study
relied on a small sample, and the interview was not designed to produce data relevant
to Maslow’s (1954) theory (Lawler & Suttle, 1972). Furthermore, the inter-rater
reliability of the coding of interviews was low (0.55 to 0.59).
A study designed to overcome the limitations identified in Hall and
Nougaim’s (1968) study was conducted by Lawler and Suttle (1972). They
employed a reasonably large sample of employees from government agencies and
retail stores. Their questionnaire, developed by Porter (1963), was designed to
measure Maslow’s (1954) needs. According to Maslow’s (1954) theory, the
importance of a need should be negatively correlated with satisfaction of that need.
Hence, as satisfaction with a need increases, the importance of that need decreases.
Lawler and Suttle’s (1972) results did not support this proposal for either the
government or retail organisations respectively, for social (r = -0.09, r = 0.07),
esteem (r = 0.06, r = -0.04), autonomy (r = 0.07, r =0.01), and self-actualisation
11
needs (r = 0.01, r = -0.10). There was however, some support for the security needs
(r = -0.34, r = -0.12).
As their study was longitudinal they also conducted change analyses. They
correlated the change in need importance with the change in need satisfaction. It was
expected that these correlations would be negative, indicating that increases in the
satisfaction of a need were associated with decreases in its importance. However,
these correlations were also positive ranging from r = 0.07 to r = 0.24. Hence, the
direction of the correlations were inconsistent with Maslow’s (1954) theory.
In summary, Hall and Nougaim’s (1968) and Lawler and Suttle’s (1972)
findings are inconsistent with those of Alderfer (1969) and Graham and Balloun
(1972). The major difference between these studies is that the latter two measured
need strength with desire or improvement, while the former two relied on measures
of importance. Although the desire and improvement measures were criticised
earlier for being too similar to measures of satisfaction, the use of importance as an
indicator of need strength has also been criticised (Wicker et al., 1993).
Although Maslow (1970) postulates that a need is important because of
deprivation, it has been suggested that a person may report that a need is important
because they have attained it and value it (Wicker et al., 1993). Indeed, Maslow
(1954, p. 148) proposed that “greater value is usually placed on higher-order needs
by persons who have gratified both kinds (i.e., lower and higher-order needs).”
Hence, people who are self-actualising may report that all the higher needs are
important because they value them. A person may thus report that a higher-order
need is important because they are deprived of it, or because they have attained it and
12
value it. If individuals report that a higher-order need is important because they have
attained it, it would be positively related to satisfaction (Wicker et al., 1993).
Although importance may be an ambiguous construct, the early studies
conducted by Hall and Nougaim (1968) and Lawler and Suttle (1972) should still be
valid. The majority of participants in these studies would not have gratified both
lower-order and higher-order needs. As such, they would only be expected to report
that a need was important if they were deprived of the need. Hence, although the
early studies tested the deprivation/domination paradigm using importance ratings,
this is not expected to reduce the validity of the findings, which are inconsistent with
Maslow’s (1954) theory.
More recent researchers have found some support for Maslow’s (1954)
theory using a different measure of need strength, namely intention. Wicker et al.,
(1993) examined how need strength relates to satisfaction when need strength is
operationalised in a number of different ways. They used, among others, ratings of
importance (i.e., “To what extent is it an important goal”) and ratings of intention
(i.e., “How much do you want to pursue it”). They correlated these variables with
attainment as a measure of deprivation (i.e., “To what extent do you already have
it”). According to Maslow’s (1970) theory, it would be expected that as attainment
of a need decreased, the intention of that need would increase. However, they found
the correlations of past attainment (deprivation) and intention were positive, ranging
from r = 0.39 to r = 0.96. This suggests that as attainment of a need increases, the
intention to pursue the need also increases.
13
Although Wicker et al’s., (1993) findings are inconsistent with Maslow’s
(1970) theory, they suggest that the correlations may have been inflated by
halo-effects or carryover rating bias. They postulate that the ratings may be affected
by a general motivation factor, and by earlier ratings. To control for such effects,
deviation scores were computed and correlated. Deviation scores are calculated by
subtracting the grand mean over all scales for a need from the mean of that need on
each particular scale. This removed a need-means factor from the data, “reducing
any biasing effect on correlations resulting from mean differences among needs”
(Wicker et al., 1993, p. 126). Using these deviation scores, the direction of the
correlations were reversed. For importance, two of the four correlations were in the
expected negative direction, however they were very small (r = -0.13 and r = –0.07).
For intention however, all four of the correlations were strong and negative, ranging
from r = –0.62 to r = –0.74. This suggests that if need strength is measured through
intention, and deviation scores are used, then it is negatively related to attainment.
On this basis, Wicker et al., (1993) postulate that it is too early to discard the
deprivation/domination paradigm. They propose that participants in earlier studies
(e.g., Hall & Nougaim, 1968; Lawler & Suttle, 1972) may have reported that a lower
order need was important because they had attained it and they valued it (high
satisfaction), or because they were deprived of it (low satisfaction). As a result, the
correlations between importance and satisfaction could be positive or negative,
depending on how need strength was operationalised. Despite this, it remains
concerning that the deprivation/domination paradigm is only supported when need
strength is operationalised as intention.
14
1.3.3.3
b) Establishing Causality
A second methodological problem, which may reduce the validity of the
studies examining the deprivation/domination paradigm is that although the
deprivation/domination paradigm is causal, the relationship is assessed through
correlational analyses (e.g., Alderfer, 1969; Graham & Balloun, 1972; Hall &
Nougaim, 1968; Lawler & Suttle, 1972). Only one study has attempted to establish
causality through experimentally manipulating deprivation and measuring
subsequent need strength. Wicker and Wiehe (1999) divided forty students into two
groups, where one group wrote about a past event where they felt especially close to
another person and the other group wrote about a time when they tried to get close to
someone, but felt unsuccessful. Both groups then rated their needs on each level of
the hierarchy on prior attainment (i.e., “To what extent to do you already have it”),
intention (i.e., “How much do you intend to pursue it”), and importance (i.e., “To
what extent is it an important goal”).
The interpersonal scenario was expected to affect their belongingness
responses, where the unsuccessful group would report lower attainment, and higher
need strength for the belongingness need. Inconsistently however, the two groups
did not report different levels of attainment on the belongingness need. The two
groups did report different levels of esteem attainment where the unsuccessful group
reported less past attainment of esteem needs than the successful group. The
unsuccessful group also reported higher intention on all levels of the hierarchy than
the successful group. The two groups did not however differ on importance ratings.
15
These data were interpreted as supporting Maslow’s (1970) theory, as when
the past attainment of esteem needs were low, intentions were higher. The results
must be interpreted with caution however as there were methodological limitations in
the study. Aside from each group having a small sample size (N = 20), need strength
was not assessed prior to the intervention. Hence, the differences in their intentions
may have been a pre-existing difference. Furthermore, although the groups were
asked to report a story relating to belongingness needs, the two groups did not report
different level of past attainment on belongingness needs. Hence, the belongingness
manipulation was not successful. In summary, although Wicker and Wiehe (1999)
present their study as supporting Maslow’s (1970) theory, the findings should be
viewed with caution.
1.3.3.4
Summary: Deprivation/Domination Paradigm
The deprivation/domination paradigm was rejected after several early studies
failed to find supportive correlations. Wicker et al., (1993) re-introduced the
proposition into the literature, attributing the inconsistent findings to the
operationalisation of need strength. They demonstrated that positive correlations
between attainment and need strength could be reversed if deviation scores were
used, and need strength was measured by intentions rather than importance. The
validity of these findings continues to be questioned however, as the relationship
between need strength and satisfaction, although causal has been assessed through
correlational analyses. In summary, the majority of research demonstrates that as
deprivation increases, need strength does not necessarily increase.
16
1.3.3.5
Criticism Two: Gratification/Activation Paradigm
The gratification/activation paradigm postulates that the higher the
satisfaction with a need, the higher the need strength of the need at the next level of
the hierarchy. The gratification/activation paradigm is different from the
deprivation/domination paradigm as the former examines the correlation between the
satisfaction of a need at one level with the importance of the need at the next level,
whereas the latter examines the correlation between satisfaction and need strength of
a need on the same level.
Two longitudinal studies have been conducted to evaluate the
gratification/activation paradigm. As previously mentioned, Hall and Nougaim
(1968) interviewed managers annually throughout a five-year period, coding their
responses on need strength and satisfaction. For each year, they correlated the
changes in need satisfaction from one year to the next with changes in need strength
at the next highest level during the same period of time. According to Maslow’s
(1954) theory, it was expected that high correlations would exist between the change
in satisfaction of a given need level and the change in strength of the next highest
level. The pooled correlations were low however, ranging from r = 0.05 to r = 0.22.
Hence, there was little evidence to suggest that the increasing satisfaction of a need
results in the increasing need strength of the next highest need. It must be noted
however that this study relied on a small sample size, and the interview used in the
study was not designed to produce data relevant to Maslow’s (1954) theory. These
limitations were addressed in Lawler and Suttle’s (1972) study.
17
As previously mentioned, Lawler and Suttle (1972) relied on Porter’s (1963)
questionnaire, which was specifically designed to measure Maslow’s (1954) needs.
According to Maslow’s (1954) theory, it was expected that the satisfaction of a need
would be positively correlated with the need strength of the need in the next highest
level. Lawler and Suttle’s (1972) results demonstrated that one correlation between
security satisfaction, and social importance was significant for the retail group
(r = 0.21), however the rest were all low ranging from r = -0.01 to r = 0.10. These
findings, as with Hall and Nougaim’s (1968) findings clearly raise questions
concerning the validity of the gratification/activation paradigm.
In summary, the gratification/activation paradigm proposes that as
satisfaction with a need increases, the need strength of the next highest need
increases. Studies investigating this paradigm generally demonstrate that the
correlations between need satisfaction and need strength of the next highest need are
low.
1.3.3.6
Criticism Three: Measurement of Self-Actualisation
There is a poor level of concordance between the definition of the need for
self-actualisation, and the measurement of the need for self-actualisation. Selfactualisation is defined as “the full use of one’s talents, capacities, potentialities”
(Maslow, 1970, p. 150). It is the need for the individual to become everything they
are capable of becoming. Self-actualisers have a more efficient perception of reality,
accept others, are autonomous, do not need others, are less concerned with
themselves, and have deeper interpersonal relationships (Maslow, 1970). These
18
characteristics must be regarded with caution however as they were based on a social
discussion with a sample of 22 people whom Maslow (1954) believed to be selfactualisers. These people were selected as they seemed to be fulfilling themselves,
and doing the best they were capable of. Perhaps as a consequence of this vague
definition, operational definitions of the need for self-actualisation vary extensively.
Several early studies measured self-actualisation using Porter’s (1963) need
scale (i.e., Lawler & Suttle, 1972; Roberts, Walter & Miles, 1971). This scale
includes three items which assess the opportunity for personal growth and
development in the job, the feelings of self-fulfilment a person gets from being in the
job, and the feelings of worthwhile accomplishment in the job. One problem with
these items however, is that they appear to assess how the person feels about their
work rather than whether they feel they are have reached their potential.
Although more recent scales tend to be more comprehensive, their validity is
still questioned. For example, Shoura and Singh (1999) assessed self-actualisation
through items measuring meaningfulness, self-sufficiency, effortlessness, creativity,
professional creativeness, self-understanding, independence, and harmony with the
universe. Examples of these items are “do you think you have enough talents and
capabilities to perform the job”, “does your work come as second nature to you” and
“do you feel your job is in harmony with the universe.” These items are criticised for
being vague, and it is questioned whether they measure if a person has become all
that they are capable of. Furthermore, these items only refer to self-actualisation on
the job, and in some cases, self-actualisation may occur off the job. In summary,
19
there seems to be a great deal of discrepancy between the definition and
measurement of self-actualisation.
1.3.3.7
Criticism Four: Ability to Achieve Self-Actualisation
The need for self-actualisation is the need for the individual to become
everything that they are capable of becoming. This suggests that anyone performing
their job to the best of their abilities is self-actualising. However, Maslow (1970)
screened 3000 college students and concluded that only one student was
self-actualising. Following this study, Maslow (1970) proposed that
self-actualisation of the sort he had found in older adults was not possible for
younger developing people. He proposed that young people lack many of the
experiences needed for self-actualisation such as identity, autonomy, and romantic
relationships. The proposal that younger people do not self-actualise has not
received empirical support. A study conducted on engineers demonstrated that the
junior engineers reported higher scores on self-actualisation than the senior engineers
(Shoura & Singh, 1999). Furthermore, in a study of academics, ranging in age from
30 to 68 years, age and self-actualisation were not related (Hawkins, Hawkins &
Ryan, 1989). It must be noted however that, as previously mentioned, these studies
relied on questionable measures of self-actualisation.
20
1.3.3.8
Criticism Five: Applicability of Maslow’s Hierarchy of Needs to
Organisations
Although some of the propositions in the need hierarchy theory have not
received empirical support, the theory has been extensively accepted in the
management literature (Roberts, 1982). Moreover, the general idea that the concepts
of love, safety, self-esteem, and growth contribute to motivation and satisfaction are
acceptable to both psychologists and management scientists (Shoura & Singh, 1999).
The fundamental problem in applying Maslow’s (1970) theory to work
organisations is that little is known about how to reach the ultimate goal of selfactualisation. Maslow’s (1970, p.46) definition of self-actualisation as “what a man
can be, he must be” is extremely vague, and there is no agreed upon way of
operationalising the construct, or facilitating it in employees.
1.3.4
Conclusion
The need hierarchy theory proposes that individuals strive to gratify five needs,
namely physiological, safety, belongingness, esteem and self-actualisation needs.
The theory proposes that the higher the deprivation of a need, the higher its need
strength, and the higher the satisfaction with a need, the higher the need strength of
the next highest need. Although early studies tended to reject these propositions,
more supportive results were found when need strength was operationalised as
intentions rather than importance or desire. Even with some supportive findings, the
21
validity of the theory is still questioned as very little is known about the ultimate goal
for humans, the need for self-actualisation.
22
1.4
Herzberg, Mausner and Snyderman’s (1959, 1993) Two-Factor Theory of
Job Satisfaction
1.4.1
How the Two-Factor Theory has Contributed to our Understanding of
Job Satisfaction
The two-factor theory (Herzberg et al., 1959) questioned the assumption that
job satisfaction and job dissatisfaction lie on a single continuum. Rather, the theory
proposed that job satisfaction and job dissatisfaction are separate continua, and that
the factors which affect job satisfaction are different from the factors which affect
job dissatisfaction.
1.4.2
Development of the Two-Factor Theory
The two-factor theory is based on a study of accountants and engineers.
Through an interview, employees recalled experiences about times when they felt
especially good or bad about their jobs, and then rated how seriously their feelings
(good/bad) about their jobs had been affected by what happened. Using content
analysis, their responses were coded into 14 categories.
As demonstrated in Table 1, employees reporting the sources of good times
tended to recall events related to achievement, recognition, work itself,
responsibility, and advancement. These sources of satisfaction were termed
motivator factors. Employees reporting the sources of bad times tended to recall
events related to company policy and administration, supervision-technical, salary,
23
recognition, and interpersonal relations with supervisor. These sources of
dissatisfaction were termed hygiene factors. An obvious exception to this
classification is for the factor salary. Salary was reported a similar number of times
for employees reporting the source of good events and for those reporting the source
of bad events.
On the basis of these findings, Herzberg et al., (1959) proposed that paying
attention to motivator factors will increase job satisfaction, but will not affect job
dissatisfaction. Alternatively, paying attention to hygiene factors will decrease job
dissatisfaction but will not increase job satisfaction. For example, increasing status
is expected to reduce job dissatisfaction, but not increase job satisfaction.
Table 1- Sources of Good/Bad Times for Accountants and Engineers (N=200)
Factor
Achievement
Recognition
Work Itself
Responsibility
Advancement
Salary
Possibility of Growth
IR-subordinate
Status
IR-Supervisor
IR-Peer
Supervision-technical
Company policy and
administration
Working conditions
Personal life
Job Security
Time felt especially good
41**
33**
26**
23**
20**
15
6
6
4
4
3
3
3
Time felt especially bad
7
18
14
6
11
17
8
3
4
15**
8**
20**
31**
1
1
1
11**
6**
1
**p<0.01; Motivator factors are bolded
From Herzberg, F., Mausner, B., & Snyderman, B. (1959). The Motivation to Work.
(p.60, 72). New York: Wiley.
24
1.4.3
Criticisms of Herzberg et al’s., (1959) Theory
The two-factor theory is criticised for deducing conclusions from a study that:
a) failed to test the main propositions; and b) was methodologically flawed. In
regards to the first criticism, there is insufficient evidence to demonstrate how
motivator and hygiene factors relate to job satisfaction. Although the study
demonstrated that employees recalling good times tended to recall motivator factors,
and employees recalling bad times tended to recall hygiene factors, there is no
empirical evidence for the proposal that motivator factors can only contribute to job
satisfaction and that hygiene factors can only contribute to job dissatisfaction. The
study did not measure job satisfaction, and as such, there is no basis for assuming
that the factors described in the incidents caused, or were even related to job
satisfaction (Ewen, 1964).
In regards to the second criticism of Herzberg et al’s., (1959) theory,
concerning the methodology of the study, several problems have been identified.
These include: 1) some of the findings contradict the theory; 2) the findings differ
depending on the method of data collection; and 3) the hypotheses and criterion
measures are ambiguous. These limitations will now be discussed more extensively.
1.4.3.1
Criticism One: Evaluation of Results
The results from Herzberg et al’s., (1959) study did not completely support
the theory. As can be seen in Table 1, employees often report motivator factors, such
as recognition when they are recalling a time when they felt bad. Although they
25
reported recognition significantly less for bad times than good times, recognition was
still the third highest source of a bad time. Furthermore, some of the hygiene factors
were reported only slightly more for bad events than good events (i.e., salary, status
and job security). Hence, some of the findings are not supportive of the two-factor
theory.
1.4.3.2
Criticism Two: The Interview Method
Replications of Herzberg et al’s., (1959) study have produced mixed results.
Some researchers have found support for the theory (i.e., Schmidt, 1976), whilst
others have contradicted the theory (e.g., Armstrong, 1971; Brenner, Carmack &
Weinstein, 1971; Hill, 1986; King 1970; Waters & Waters, 1969). A commonality
among the studies that have contradicted the theory is that they have departed from
the traditional interview method (Gardner, 1977; Salancik & Pfeffer, 1977). The
interview method is criticised for being retrospective, and selective (Gardner, 1977).
The employees are expected to more readily recall positive events which reflect upon
themselves, and negative events which can be attributed to external conditions
(Vroom, 1964). As a result, many researchers have tested Herzberg et al’s., (1959)
theory with rating scales.
1.4.3.3
Rating Scales
One example of such a study is Waters and Waters (1969) study of office
employees. Rather than using Herzberg et al’s., (1959) critical incidents interview,
employees completed a job satisfaction scale, a job dissatisfaction scale (as these are
26
proposed to be two separate dimensions), and a scale examining satisfaction with
specific facets of work. They correlated facet satisfaction with overall satisfaction
and overall dissatisfaction.
According to Herzberg et al’s., (1959) theory, it was expected that the
motivator factors (i.e., responsibility, work, sense of achievement etc.) would
correlate with overall satisfaction more than overall dissatisfaction. This finding was
not supported as the pattern of relationships with satisfaction and dissatisfaction were
similar (i.e., responsibility of job correlated with satisfaction r = 0.41 and with
dissatisfaction, r = -0.37). Similar results were obtained for the hygiene factor,
where for example, competent supervision correlated r = 0.44 with satisfaction and
r = -0.40 with dissatisfaction, and salary correlated r = 0.43 with satisfaction and
r = –0.28 with dissatisfaction. As motivator and hygiene factors acted as both
satisfiers and dissatisfiers, this study did not provide support for the two-factor
theory.
Other researchers who have relied on rating scales have also found that their
results fail to support the theory. For example, Brenner et al., (1971) conducted a
study on accountants, assessing “how much is there now” for each motivator and
hygiene factor. They correlated each of the items with a measure of overall job
satisfaction. Consistent with the two-factor theory, the motivator factors were
positively related to measures of job satisfaction, with the correlations ranging from
r = 0.39 to r = 0.62. Inconsistently however, the hygiene factors were also positively
related to job satisfaction, with the correlations ranging from r = 0.41 to r = 0.59.
27
These findings, fail to conform with Herzberg et al’s., (1959) theory, and suggest that
as motivator and hygiene factors increase, job satisfaction increases.
Although Waters and Waters (1969) and Brenner et al’s., (1971) studies
failed to support the two-factor theory using a rating scale, Hill’s (1986) study claims
to offer more support. Hill (1986) developed a 45-item questionnaire to measure
intrinsic and extrinsic factors of work in academia. The intrinsic factors
(i.e., teaching, convenience, recognition-support) were similar to motivator factors,
whilst the extrinsic factors (i.e., economic, administration, and collegial) were similar
to the hygiene factors. It was expected that the intrinsic factors would lead to job
satisfaction and that the extrinsic factors would lead to job dissatisfaction. To test
this proposal, Hill (1986) compared the mean level of satisfaction with each
dimension. The employees were more satisfied with the intrinsic dimension
(M = 4.43) than the extrinsic dimension (M = 4.18). Specifically, the following
means were observed where one is very dissatisfied and six is very satisfied: teaching
(M = 4.82), convenience (M = 4.52), recognition-support (M = 3.96), economic
(M = 4.24), administration (M = 4.00), and collegial (M = 4.23). From these results,
Hill (1986) concluded that the academics’ dissatisfaction with their work came from
extrinsic factors (i.e., hygiene factors), whilst their satisfaction came from intrinsic
factors (i.e., motivator factors).
The validity of this conclusion is questioned however, as the mean level of
satisfaction for the intrinsic and extrinsic factors were very similar. The difference
was significant, however this may be due, in part, to the large sample size
28
(N = 1000). More importantly however, it must be questioned whether Hill’s (1986)
study is even testing Herzberg et al’s., (1959) theory. The two-factor theory did not
propose that employees are more satisfied with the motivator factors than the
hygiene factors, but rather that the motivators serve to bring about job satisfaction,
and hygiene factors prevent job dissatisfaction. As such, although Hill’s (1986)
study claims to support the two-factor theory using a rating scale, the validity of the
findings are questioned.
In summary, it appears that studies testing the two-factor theory using rating
scales tend to be inconsistent with those using the interview method. The rating
scale may be superior to the interview method, however it is still problematic
(Herzberg, 1966; Silver, 1987; Whitsett & Winslow, 1967). Researchers propose
that the rating scales may induce respondents to indicate an attitude towards every
item, even on items that they have never thought about before (Herzberg, 1966).
Furthermore, there is pressure for the respondents to appear rational when they report
their satisfaction with the job facets and overall satisfaction, where they may attempt
to keep their responses consistent. As a result of these limitations, some researchers
have opted for free response scales (e.g., Silver, 1987).
1.4.3.4
Free Response Scales
Studies attempting to overcome the limits of both interview and ratings scales
have relied on free response scales. These scales are not retrospective, and allow the
employee to develop their own answers. For example, Friesen, Holdaway and Rice’s
(1983) study of school Principals relied on two questions including “which two
29
factors contribute most to your overall satisfaction with the principalship” and
“which two factors contribute most to your overall dissatisfaction with the
principalship.” They then calculated how often the Principals mentioned motivator
factors and hygiene factors when they referred to sources of their satisfaction and
dissatisfaction. These were converted into ratios, which included the number of
times each factor was mentioned as a satisfier, and the number of times each factor
was mentioned as a dissatisfier (satisfier: dissatisfier). For example, sense of
achievement was reported as a source of satisfaction 85 times, and a source of
dissatisfaction 5 times (i.e., 85: 5). Other factors that were reported as satisfiers
more than dissatisfiers included interpersonal relationship (77: 0), importance of the
work (24: 0), and relationship with central office (11: 0). These findings were
generally consistent with the two-factor theory, the exception being factors involving
relationships (i.e., interpersonal relationships and relationships with central office).
Relationship factors are hygiene factors, and as such, are expected to be reported as
dissatisfiers more than satisfiers.
The factors that were mentioned more as dissatisfiers than satisfiers include
amount of work (0: 68), overall constraints (0: 56), attitudes of society (0: 49), stress
(0: 21) and impact on home life (0: 14). These were also generally consistent with
the two-factor theory.
It must be noted however that many other factors were identified as sources
of both satisfaction and dissatisfaction, such as relationship with teachers (94: 42),
responsibility (81: 20), autonomy (70: 19), student attitudes (51: 25), challenge of
work (41: 36), relationships with parents (22: 51) and salary (6: 7). In fact, only
30
eight of the 20 factors occurred uniquely as either satisfiers or dissatisfiers and two
of these were in the wrong direction (i.e., interpersonal relationships, relationships
with central office). Hence, although researchers have proposed that this study
“represents a major step in resolving the controversy in favour of Herzberg’s
assertion” (Silver, 1987, p. 5), it provides at best, only partial support.
A similar study was conducted on educators by Silver (1987). The
participants were required to think of a time when they felt especially good/bad about
their jobs, and write a paragraph describing what happened. It was hypothesised that
the employees would cite motivator factors more often than hygiene factors when
describing positive events, and cite hygiene factors more often than motivator factors
when describing negative events. As hypothesised, the employees mentioned more
motivator factors (85) than hygiene factors (6), when recalling a positive event.
Inconsistently however, the employees reported more motivator factors (48) than
hygiene factors (40), when recalling a negative event. As such, Silver’s (1987) study
provides only partial support for the two-factor theory.
Silver (1987) conducted a second study using a questionnaire developed by
Wernimont (1966). The questionnaire contained two lists of statements, one positive
and one negative, each referring to one of Herzberg et al’s., (1959) 16 categories.
The participants were required to indicate whether an event had occurred, and then to
indicate whether it was a positive or negative event. For example, for the pay facet,
on the negative list was “the pay increase I got was insufficient for putting some
aside for the future” and on the positive list was “I received a substantial increase in
pay.” It was hypothesised that on the positive-feelings list, respondents would check
31
more motivator than hygiene items, and on the negative-feelings list, respondents
would check more hygiene than motivator items. On the positive list, the employees
checked 322 motivator factors and 259 hygiene factors, whilst on the negative list,
they checked 255 hygiene and 178 motivators factors. These results are assumed to
be supportive of the two-factor theory as respondents checked more motivator than
hygiene factors on the positive list and more hygiene than motivator factors on the
negative list. However, it is concerning that motivator and hygiene factors were
reported for both positive and negative events.
1.4.3.5
Summary
Studies that contradict the two-factor theory tend to depart from the
traditional interview method. These studies, relying on rating scales or free response
scales, claim to provide some support for the theory. Closer examination of the
results however, demonstrates that these studies provide at best, partial support of the
theory.
1.4.3.6
Criticism Three: Ambiguous Hypotheses and Criterion Measures
Researchers testing the two-factor theory have been criticised for employing
several different hypotheses and criterion measures (King, 1970). First, in regard to
the hypotheses, King (1970) cites several different ways that researchers test the
main propositions of the theory. Some researchers propose that all motivator factors
combined together should contribute more to job satisfaction than job dissatisfaction,
and that all hygiene factors combined should contribute more to job dissatisfaction
32
than job satisfaction. Other researchers examine each factor separately, proposing
that each motivator factor should contribute more to job satisfaction than job
dissatisfaction, and each hygiene factor should contribute more to job dissatisfaction
than job satisfaction. A more precise version of the theory proposes that only
motivators determine job satisfaction, and that only hygienes determine job
dissatisfaction. These examples serve to demonstrate that one researcher using a
broad hypothesis may report that their findings support the theory, whilst another
researcher using a specific hypothesis may report that their results are inconsistent
with the two-factor theory.
In regard to the criterion measures, researchers tend to evaluate their findings
differently (King, 1970). For example Sergiovanni (1967) conducted a study on
teachers using the critical incident technique. The results indicated that teachers
reported achievement as a source of a positive event (30) more than a source of a
negative event (9). Some researchers, including Sergiovanni (1967) propose that this
ratio is supportive of the two-factor theory as it is reported more in positive
experiences than negative experiences. However, other researchers (e.g., Friesen et
al., 1983) propose that it is not supportive as achievement was reported for some
negative experiences. Most researchers opt for the former, proposing that if one part
of the ratio is greater than the other part, the results are supportive of the two-factor
theory (i.e., Silver, 1987). Even so, these different criterion measures certainly
create confusion.
It must also be questioned whether a study can provide support for the twofactor theory when some of the ratios are in the wrong direction (i.e., salary 20: 12).
33
Herzberg et al., (1993) did not comment on the issue, however they accepted results
that were not in the proposed direction in their study. King (1970) attempted to
specify some guidelines, proposing that failure to conform one item would not
contradict the whole theory unless that one item had a significant negative difference.
However, it still remains unclear how many items would need to be inconsistent for
the theory to be refuted.
1.4.4
Conclusion
The two-factor theory was notable for proposing that job satisfaction and job
dissatisfaction are separate continua, and that the factors which affect job satisfaction
are different to the factors which affect job dissatisfaction. The original study from
which the theory developed was methodologically flawed, and as such, it is not
surprising that empirical studies evaluating the two-factor theory often demonstrate
that motivator and hygiene factors affect both job satisfaction and job dissatisfaction.
Researchers that report supportive findings often rely on less stringent hypotheses
and criterion measures. In conclusion, the two-factor theory of job satisfaction has
received little empirical or theoretical support.
34
1.5
1.5.1
Vroom’s (1964) Expectancy Theory of Job Satisfaction
How Expectancy Theory has Contributed to our Knowledge of Job
Satisfaction
Expectancy theory (Vroom, 1964) was one of the first theories to focus on the
cognitive processes that underlie job satisfaction. It has received considerable
theoretical and empirical attention for over 30 years (Van Eerde & Thierry, 1996).
The number of studies examining expectancy theory has decreased recently however,
with only ten studies being conducted since the 1990’s (Ambrose & Kulik, 1999).
As such, this review will mainly be based on the earlier studies.
1.5.2
Description of Expectancy Theory
Expectancy theory describes its major constructs and propositions using its
own jargon. It refers to three major constructs, namely expectancy, valence, and
instrumentality. Expectancy refers to how much a person perceives that an action
will result in a certain outcome. For example, how much a person believes that if
they work harder, they will get a pay rise. Valence refers to the degree of anticipated
satisfaction or desirability of an outcome. Hence, in the previous example, the
valence would be a measure of how much the person desires a pay rise.
Instrumentality refers to the degree to which the person sees the outcome in question
as leading to the attainment of other outcomes. Hence, in our example,
35
instrumentality would be how much a person believes that a pay rise will result in
other outcomes, such as buying a house.
The way these constructs are combined depends on the variable that is being
predicted. Three dependent variables have been examined, namely job effort, job
performance and job satisfaction. This review will only examine the model
predicting job satisfaction, referred to as the valence model. This incorporates two of
the above-mentioned constructs, namely valence and instrumentality. It proposes
that job satisfaction can be predicted by multiplying the valence of an outcome by its
instrumentality. Hence, to predict job satisfaction, we would need to determine how
much a person likes or values an outcome of their job (i.e., being promoted) and
multiply this measure by how much they believe that this outcome will lead to other
outcomes (i.e., being offered a partnership in a business).
There is a great deal of ambiguity surrounding the measurement of the major
constructs in the expectancy theory (Van Eerde & Thierry, 1996). The
instrumentality construct has proved to be the most troublesome for researchers
(Wahba & House, 1978). Vroom (1964) referred to instrumentality as the
probability that an outcome will result in other outcomes (i.e., outcome-outcome
relationship), and expectancy as the probability that an action will result in an
outcome (i.e., action-outcome relationship). Researchers have confused these
variables however, and have measured instrumentality through examining the
probability that an action will result in an outcome (eg., Constantinople, 1967;
Pulakos & Schmitt, 1983; Reinharth & Wahba, 1976). These different
36
conceptualisations of instrumentality influence the application of the valence model
to the workplace.
1.5.3
Applications of the Valence Model
According to the valence model as defined by Vroom (1964), an employer
can increase their employees’ levels of job satisfaction through ensuring that
employees value the outcomes of their job (i.e., gaining admiration from other
workers, being promoted, feeling a sense of accomplishment, pay rise), and believe
that these outcomes will lead to other outcomes.
According to researchers who operationalise instrumentality as expectancy,
employers should ensure that their employees value the outcomes of their jobs, and
believe that their work will help them achieve those outcomes.
1.5.4
Studies of the Valence Model
Several early studies examined the relationship between job satisfaction and
the valence model (e.g., Constantinople, 1967; Ferris, 1977; Pulakos & Schmitt,
1983; Reinharth & Wahba, 1976; Sobel, 1971, Teas, 1981). A review of such studies
demonstrates that correlations between the valence model (valence x instrumentality)
and job satisfaction are generally positive, ranging from r = 0.03 to r = 0.57
(Mitchell, 1974). This demonstrates that together, valence and instrumentality
predict job satisfaction.
An example of a typical study conducted to assess how the valence model
influences satisfaction, is that conducted by Constantinople (1967). This study
37
examined how valence and instrumentality contributed to satisfaction in university
students. The students were given a list of 14 outcomes of university (e.g., learning
how to learn from books and teachers). Each outcome was rated in terms of its
importance (i.e., valence) and on the degree to which the university was helping the
students to achieve the outcome (i.e., instrumentality). The product of these two
ratings (i.e., instrumentality and valence) was obtained for each outcome, and the
products were summed across all 14 outcomes. This measure was then correlated
with a measure of satisfaction with college. According to the valence model, it was
expected that the valence times instrumentality interaction would be positively
related to satisfaction. The results were generally supportive of the model with the
correlations ranging from r = 0.34 to r = 0.49. It must be noted however that
Constantinople (1967) did not examine how much each component of the model
contributed to satisfaction.
1.5.5
Methodological Limitations
Although many studies testing Vroom’s (1964) valence model claim to
provide moderate support for Vroom’s (1964) expectancy theory (e.g., Ferris, 1977;
Pulakos & Schmitt, 1983; Reinharth & Wahba, 1976; Sobel, 1971, Teas, 1981), these
studies have some methodological limitations. Three such limitations have been
identified and will be discussed below as: 1) the finding that the components of the
valence model account for more of the variance in satisfaction on their own than
when combined; 2) violations of the assumptions of the multiplicative composite;
and 3) inflated correlations due to common method variance.
38
1.5.5.1
1) The Finding that the Components of the Valence Model Account for
more of the Variance in Satisfaction on their own than when Combined.
The valence model proposes that job satisfaction can be predicted by the
product of valence and instrumentality. However, many studies have demonstrated
that the components of expectancy theory account for more of the variance in
satisfaction on their own than when included in the expectancy model (e.g., Pulakos
& Schmitt, 1983; Reinharth & Wahba, 1976; Teas, 1981; Van Eerde & Thierry,
1996). In these studies, one of the components, either valence or instrumentality, has
predicted job satisfaction as well, or better than, the valence times instrumentality
interaction.
An example of such a study is that conducted by Reinharth and Wahba
(1976). They measured valence and expectancy in a sample of sales force
employees. Although instrumentality should have been included in the model, their
measure of expectancy was similar to a measure of instrumentality. They measured
expectancy by assessing the extent of agreement with the following items; “The
harder I work, the more I produce”, “there are no rewards for working hard in this
company” and “poor job performance may get me fired.” Their results demonstrated
that expectancy was as strongly correlated to job satisfaction (r = 0.43) as the
expectancy times valence interaction (r = 0.40).
Similar findings were reported in Pulakos and Schmitt’s (1983) study of
graduating students. Valence of work outcomes was assessed through rating the
importance of job facets (e.g., good pay, cooperative workers, opportunities for
39
personal growth), and instrumentality was assessed through rating the likelihood of
each facet. They correlated these measures with internal job satisfaction and external
job satisfaction. The results demonstrated that valence and instrumentality
considered separately correlated with job satisfaction as well or better than the
valence times instrumentality interaction. For example, in regard to the co-workers
facet, the correlations between the valence times instrumentality interaction (r = 0.04,
internal, r = 0.11, external) were lower than the correlation for instrumentality
considered on its own, (r = 0.11, internal, r = 0.12, external). Hence, in this example,
the valence model was not more strongly related to job satisfaction than the
components considered separately.
A recent meta-analysis of studies using the valence model to predict
occupational choice reached similar conclusions (Van Eerde & Thierry, 1996). The
results demonstrated that valence (r = 0.27) and instrumentality (r = 0.27) considered
separately correlated as well with choice as the valence times instrumentality model
(r = 0.28)
In conclusion, these studies suggest that the components of the valence model
often account for more of the variance in job satisfaction when considered separately
rather than when combined into the valence model. These results not only question
the usefulness of the two components of the valence model, but also how these
components are combined.
40
1.5.5.2
2) Violations of the Assumptions of the Multiplicative Composite
Although the valence model proposes that valence should be multiplied by
instrumentality, many assumptions underlying the multiplicative process may not be
met. First, although it is assumed that for a multiplication to be valid, the two
constructs are independent (Campbell & Pritchard, 1970), instrumentality and
valence are related to each other (e.g., r = 0.47; Pritchard & Sanders, 1973). Second,
although it is assumed that multiplicative composites are based on a ratio scales with
a true zero point (Evans, 1991), most researchers rely on interval scales (Mitchell,
1974). Some researchers have attempted to establish a zero-point on a likert scale by
having a scale that ranges from 0 to 10 (i.e., Pritchard & Sanders, 1973). This scale
does not have a true zero point, and rather, to establish a true zero point, a complex
and time-consuming process needs to be undertaken, that requires the scaling of
pairs, as well as individual outcomes or objects (Thurstone & Jones, 1957).
In summary, although the valence model proposes that the components of the
model should be multiplied, two major assumptions underlying multiplicative
composites may not be met.
1.5.5.3
3) Inflated Correlations due to Common Method Variance
Although the assumptions of the multiplicative composites are often ignored,
the correlations between the components, considered either on their own or in the
valence model, with job satisfaction, are still moderate. Critics suggest that these
41
moderate correlations occur as the measures of instrumentality, valence, and
satisfaction are all based on self-report (Schwab, Olian-Gottlieb & Heneman, 1979).
It has been proposed that when both the independent variables and dependent
variables are measured through self-report, they correlate higher than if one of the
variables is observed (Mitchell, 1974; Schwab et al., 1979). The problem with this
reasoning however is that self-report measures are expected to differ from objective
measures. Objective life satisfaction, for example, is poorly correlated with
subjective life satisfaction (r = 0.12; Cummins, 2000a). Thus, the subjective
measures cannot be verified through objective measures. Furthermore, it is the
subjective measures, which are important to the individuals’ levels of satisfaction.
As long as the employee perceives that by working hard, they will receive a pay rise
(instrumentality), and value a pay rise (valence), their satisfaction will be influenced.
As such, there is no evidence for the proposal that the correlations among variables
in the valence model are inappropriately inflated through common method variance.
Rather, the correlations used to make such claims are based on invalid comparisons
between objective and subjective variables.
Although common method variance is not deemed to be a problem in this
regard, researchers have tested the valence model using measures other than
self-report. Sobel (1971) conducted a study with students, experimentally
manipulating instrumentality. Two groups were formed; one with high
instrumentality and one with low instrumentality. Both groups were told that they
were required to complete a task of mental agility. Before completing this task, they
rated the valence of this task to themselves. They then completed the task, and their
42
score was calculated. They were given a table of norm probabilities which indicated
how likely it was that they would perform well on the next task. One group was
given a table of norms, which contained high probabilities (i.e., high instrumentality
group), whilst the other group was given a table of norms which contained low
probabilities (i.e., low instrumentality group). Both groups were then asked to rate
their satisfaction whilst considering the importance of the task, and the probability
that they would do well in the next task.
According to the valence model, it was expected that people with higher
instrumentality and higher valence would report higher satisfaction. In regard to
instrumentality, the high instrumentality group consistently reported higher
satisfaction (M = 19.5) than the low instrumentality group (M = 13.5), thus
supporting the model. In regard to the proposed interaction effect, it was expected
that for the high or low instrumentality group, people who reported high valence
would also report higher satisfaction than people who reported low valence.
Inconsistently however, the results demonstrated that for the high instrumentality
group, there was no difference in the level of satisfaction reported by the high
valence group (M = 20.3) and the low valence group (M = 18.7). Furthermore, for
the low instrumentality group, the low valence group reported significantly higher
satisfaction (M = 15.0) than the high valence group (M = 12.1).
Although these results are generally inconsistent with the valence model,
there was a major limitation in the study. The researchers failed to measure
instrumentality after the subjects had completed the intervention. As such, they
failed to demonstrate that their intervention altered levels of instrumentality.
43
In summary, researchers have suggested that instrumentality and valence
correlate well with job satisfaction because they are measured by self-report.
Although there is no evidence for this proposition, Sobel’s (1971) study suggests that
when the variables are experimentally manipulated, the results are inconsistent with
the valence model.
1.5.6
Conclusion
Expectancy theory proposes that job satisfaction can be predicted by
multiplying the valence of an outcome by its instrumentality. Reviews conducted on
the valence model have demonstrated that the two major components of the model
correlate well with job satisfaction. However, the individual components of the
model often account for more of the variance in job satisfaction than the
multiplicative composite. This has led researchers to not only question the validity
of the individual components of the model, but also the validity of the multiplicative
composite. In conclusion, while the valence model appears to be simple, it combines
a set of complex variables in a problematic manner.
44
1.6
1.6.1
Discrepancy Theories
How Discrepancy Theories have Contributed to our Knowledge of Job
Satisfaction
Discrepancy theories of job satisfaction focus on the cognitive processes that
underlie job satisfaction. These theories are particularly notable for proposing that
employees’ levels of job satisfaction are dependent on this source of comparison.
1.6.2
Description of Discrepancy Theories
Discrepancy theories propose that job satisfaction is a result of a comparison
between the perception of the current situation and some standard of comparison
(Lawler & Suttle, 1973; Locke, 1969; Michalos, 1985; Porter, 1961). Researchers
have defined this standard of comparison in various ways, including what they want,
what they feel they are entitled to, what they see others as getting, what they had in
the past, or what they expected to have (Harwood & Rice, 1992; Michalos, 1985). In
all of these theories however, the larger the difference between the perceptions of the
current situation and the standard of comparison, the lower the level of job
satisfaction.
1.6.3
Empirical Studies Investigating Discrepancy Theories
Although only a few empirical studies have examined the relationship
between discrepancy and job satisfaction, they have generally been supportive. For
example, Rice, McFarlin and Bennett (1989) measured how much employees have,
45
and want, thirteen job facets. They then calculated the amount of discrepancy
between what the employee has and what they want. They found that the perceived
have-want discrepancies were moderately negatively correlated with facet
satisfaction, where r = -0.48. Hence, as the have-want discrepancy increases,
satisfaction decreases.
Although Rice et al’s., (1989) study only examined have-want discrepancies,
similar results have also been found for other discrepancies. For example, Harwood
and Rice (1992) examined comparisons with: a) co-workers; b) what the person
believed that they should have; c) what they expected; and d) what they currently
expect. The correlations between these discrepancies and satisfaction, although all in
the predicted direction, varied depending on the comparison. The have-want
discrepancy was most highly correlated with satisfaction, where the average
correlation was r = -0.51. For the have-should have, r = -0.42, have-expected,
r = -0.33, have-expect, r = -0.25, and have-co-workers discrepancy, r = -0.22. In
summary, studies examining the discrepancies theories are generally supportive.
1.6.4
Theoretical Problems with Discrepancy Theories
Although the discrepancy between what a person has and some standard of
comparison correlates well with job satisfaction, there are difficulties in using
discrepancies to explain satisfaction (Cummins & Nistico, in press). When the
discrepancy theory is used to explain job satisfaction in the workplace, the
explanation becomes tautological. For example, the theory would propose that an
46
employee has a low level of job satisfaction because they want more from their job.
As such, these discrepancies may define job satisfaction rather than explain it.
1.6.5
Conclusion
Discrepancy theories propose that job satisfaction can be determined by
cognitive comparative processes. Empirical studies have demonstrated that the
discrepancy between what an employee has and some standard of comparison is
moderately correlated with job satisfaction. However, when discrepancies are used
as an explanation of job satisfaction, the explanation becomes tautological.
47
1.7
1.7.1
Job Characteristics Model (JCM; Hackman & Oldham, 1976)
How the Job Characteristics Model has Contributed to our Knowledge
of Job Satisfaction
The job characteristics model (Hackman & Oldham, 1976) was one of the
first theories to focus on the environmental determinants of job satisfaction.
1.7.2
Description of the Job Characteristics Model
The job characteristics model proposes that complex jobs are associated with
increased job satisfaction, motivation and performance. It postulates that five core
job characteristics are associated with positive outcomes (refer to Figure 1). These
include skill variety, task identity, task significance, autonomy, and feedback.
Skill variety is the degree to which the job requires employees to use different
skills. Task identity is the degree to which the job requires completion of a whole
piece of work. Task significance is the degree to which the job has an effect on other
peoples’ lives, and autonomy is the degree to which the job provides freedom.
Finally, feedback is the degree to which the job provides clear information about the
effectiveness of the employees’ performance.
These five variables do not directly relate to job satisfaction, rather the
relationship is mediated by three critical psychological states, including experienced
meaningfulness of the work, responsibility for outcomes, and knowledge of results.
Scores on these critical psychological states are determined by the five job
48
characteristics. Experienced meaningfulness of the work refers to the degree to
which the individual experiences the job as being meaningful and worthwhile. It is
determined by skill variety, task identity, and task significance. Experienced
responsibility, determined by autonomy, is the degree to which the individual feels
accountable for their work. Knowledge of results, determined by feedback, refers to
the degree to which the individual is aware of how they are performing the work.
These critical psychological states are expected to predict a number of
personal and work outcome measures including work motivation, work performance,
work satisfaction, absenteeism and turnover. However, the relationship between the
critical psychological states and outcomes is mediated by growth need strength.
Growth need strength is the need for personal growth and development. It is
proposed that individuals with high growth need strength will respond more
positively to their critical psychological states than those with low growth need
strength.
49
Figure 1-Job Characteristics Model
Core job
dimensions
Skill Variety
Task Identity
Task Significance
Critical
Psychological States
Experienced meaningfulness
of work
Autonomy
Experienced responsibility
for outcomes of the work
Feedback
Knowledge of the actual results
of work activities
Personal and
work outcomes
High internal
work motivation
High quality
work performance
High satisfaction
with work
Low absenteeism
and turnover
Employee Growth
Need Strength
Source: Hackman, J.R., & Oldham, G.R. (1975). Development of the Job Diagnostic
Survey. Journal of Applied Psychology, 60, p. 161.
50
1.7.3
Empirical Studies of the Model
The job characteristics model has been extensively researched in over 200
studies (Renn & Vandenberg, 1995), and at least three reviews (Fried & Ferris, 1987;
Loher, Noe, Moeller & Fitzgerald, 1985; Roberts & Glick, 1981). These studies
have examined four major propositions of the model. These include that: 1) the five
core job characteristics contribute to the three critical psychological states; 2) the
critical psychological states will mediate the relationship between the job
characteristics and the outcome variables; 3) the model is moderated by growth need
strength; and 4) the model can be applied to the workplace. These propositions will
be examined for only one outcome variable, general satisfaction. General
satisfaction is an overall measure of the degree to which the employee is satisfied
and happy with their job (Hackman & Oldham, 1975).
1.7.3.1
Proposal One: The Five Core Job Characteristics Contribute to the
Three Critical Psychological States
As demonstrated in Figure 1, each job characteristic contributes to one
critical psychological state. The first three job characteristics (skill variety, task
identity and task significance) contribute to experienced meaningfulness. Autonomy
contributes to experienced responsibility, and feedback contributes to knowledge. It
is proposed that each job characteristic should only correlate with its designated
critical psychological state. However, studies examining the relationships among the
51
core job characteristics and the critical psychological states have provided only
moderate support for this proposal.
Generally, the job characteristics correlate with their designated critical
psychological state, but they also correlate with other critical psychological states.
For example, autonomy, as expected, correlates with experienced responsibility
(r = 0.40, Fox & Feldman, 1988; r = 0.41, Hackman & Oldham, 1976; r = 0.45, Wall,
Clegg & Jackson, 1978). However, autonomy also correlates with experienced
meaningfulness (r = 0.46, Hackman & Oldham, 1976; r = 0.37, Wall et al., 1978),
and knowledge of results (r = 0.25, Fox & Feldman, 1988; r = 0.26, Hackman &
Oldham, 1976; r = 0.32, Wall et al., 1978).
Another example of a job characteristic that correlates with more than one
critical psychological state is skill variety. Skill variety correlates with experienced
meaningfulness (r = 0.46, Fox & Feldman, 1988; r = 0.51, Hackman & Oldham,
1976; r = 0.30, Wall et al., 1978) and with experienced responsibility (r = 0.35, Fox
& Feldman, 1988; r = 0.40, Hackman & Oldham, 1976; r = 0.22, Wall et al., 1978).
These results suggest that, inconsistent with the job characteristics model, the core
job characteristics may predict several critical psychological states.
1.7.3.2
Proposal Two: The Degree to which the Critical Psychological States
Mediate the Relationship Between the Job Characteristics and the Outcome
Variables
Although the three critical psychological states are proposed to be the “causal
core of the model” (Hackman & Oldham, 1976, p. 8), only a few researchers have
52
examined the mediation hypothesis (e.g., Arnold & House, 1980; Fox & Feldman,
1988; Hackman & Oldham, 1976; Renn & Vandenberg, 1995; Wall et al., 1978).
This hypothesis proposes that the relationship between the five core job
characteristics and outcome variables is mediated by the three critical psychological
states.
This hypothesis has been tested by examining the correlations between the
job characteristics and the outcome variables before, and after, the relevant critical
psychological states have been controlled for. These results have provided some
support for the mediation hypothesis. For example, Hackman and Oldham (1976)
found that the correlations between the job characteristics and the outcome variables
were lower after controlling for the critical psychological state for skill variety and
task significance. However, for autonomy and feedback, the correlations remained
moderate (r = 0.29, r = 0.23). These results suggest that the critical psychological
states may be partial mediators for only some of the job characteristics.
The mediation hypothesis has also been tested using multiple regression
analyses. To support the mediation hypothesis, these analyses should demonstrate
that the critical psychological states account for sizeable proportions of the variance
in each of the dependent variable, and that the core job dimensions add little to this
when considered in the same analysis (Hackman & Oldham, 1976). Results have
demonstrated that the critical psychological states have accounted for sizeable
amounts of the variance in job satisfaction, where R = 0.68 (Hackman & Oldham,
1976), and R = 0.54 (Wall et al., 1978). When the five core job dimensions were
added to these analyses, the value of R increased by 0.01 in Hackman and Oldham’s
53
(1976) study and by 0.10 in Wall et al’s., (1978) study. This increase was significant
in Hackman and Oldham’s (1976) study, suggesting that the variance in the five core
job characteristics is explained by the three critical psychological states. It must be
noted however that the increase in R was small, and the significance may have
reflected that they employed a large sample size (N=658).
Although the above studies examined the five job characteristics together,
Renn and Vandenberg (1995) examined the job characteristics separately. They
examined the effects of the job characteristics before and after the relevant critical
psychological state were controlled for. They demonstrated that the effects of the job
characteristics were lower in magnitude when the critical psychological states were
controlled for, than when considered on their own. For example, when predicting
general satisfaction, the partial regression coefficient of task identity on its own was
0.27, and when meaningfulness of work was controlled for, the partial regression
coefficient was 0.20. However, the partial regression coefficient representing task
identity effects on general satisfaction after meaningfulness was controlled for was
still significant (0.20). This was the case for three of the five job characteristics.
Specifically, after the relevant critical psychological state was controlled for, the
partial regression coefficients for skill variety was 0.08, task identity was 0.20, task
significance was 0.15, autonomy was 0.53, and feedback was 0.11. These results
concur with the earlier studies that the critical psychological states are only partial
mediators of the relationship between job characteristics and outcomes.
In summary, although only a few studies have tested the mediation
hypothesis, they generally suggest that the critical psychological states are, at best,
54
only partial mediators of the relationship between the core job characteristics and
general satisfaction.
1.7.3.3
Proposal Three: The Degree to which the Model is Moderated by
Growth Need Strength.
Growth need strength is a need for personal growth and development. It is
postulated that people who have a high need for personal growth will respond more
positively to the critical psychological states than people who have a low need for
personal growth. Although an early study conducted by Hackman and Oldham
(1976) demonstrated that the relationship between the critical psychological states
and general satisfaction was significantly higher for employees with high growth
need strength than for those with low growth need strength, later studies have been
less supportive (Champoux, 1980; Fried & Ferris, 1987; Tiegs, Tetrick, & Fried,
1992).
For example, Tiegs et al., (1992) tested the moderating role of growth need
strength with over 6,000 subjects. Using univariate and multivariate hierarchical
moderated regression analyses, they demonstrated that growth need strength did not
moderate the relationships among job characteristics, critical psychological states,
and motivation and affective outcomes. In summary, more recent studies have
questioned the moderating role of growth need strength.
55
1.7.3.4
Proposal Four: Applying the Job Characteristics Model to Work
Organisations
According to the job characteristics model, an employer can increase job
satisfaction through increasing the five job characteristics (e.g., skill variety, task
identity etc). Through increasing these job characteristics, the employees’ critical
psychological states will increase, and job satisfaction will subsequently increase.
However, as tests of the theory have not examined changes in job characteristics, and
the theory does not specify how to make changes to the job characteristics, it may be
problematic to apply the job characteristics model to the workplace.
Tests of the theory tend focus on naturally occurring variations rather than
examining changes in job characteristics. However, the effects of changing the job
characteristics for an employee through job re-design may have different effects than
if the person was recruited into the already re-designed job (Kelly, 1992). This is
important because if the model were applied to a workplace, the five job
characteristics would be changed in an attempt to increase job satisfaction.
As researchers have not examined the effects of changing the job
characteristics, there is little research specifying how to change the job
characteristics (Roberts & Glick, 1981). Researchers have attempted to change them
using their own techniques, however these have not been particularly successful.
Kelly (1992) reviewed such studies, demonstrating that job re-design led to
improvements in job satisfaction in 17 out of 30 cases, a distribution that was not
significantly different from chance. This suggests that job re-design did not
56
consistently lead to increased job satisfaction. It must be noted however, that in
many of the studies, the employees did not alter their perceptions of the job after job
re-design. When this finding was taken into account, perceptions of job content and
job satisfaction were associated. The important finding from this review however is
that job re-design may not change employees’ perceptions of their jobs. This finding
has serious implications for employers intending to implement job re-design. It may
be costly and time-consuming to change the job characteristics, particularly if only a
few employees recognise and benefit from the changes.
1.7.4
Conclusion
The job characteristics model focuses on the environmental determinants of
job satisfaction. The model proposes that five job characteristics relate to job
satisfaction through influencing three critical psychological states. Empirical tests of
the model have provided partial support for the main propositions, however these
tests have also demonstrated that many of the relationships that exist between
variables were excluded from the model. Even if these relationships were added to
the model, practical difficulties in applying the findings to the workplace reduce the
usefulness of the theory.
57
1.8
1.8.1
Job Demand-Control Model (Karasek, 1979; Karasek & Theorell, 1990)
How the Job Demand-Control Model Contributes to our Understanding
of Job Satisfaction
The job demand-control model, developed by Karasek (1979) is one of the
most well known models in the occupational job stress literature (Fletcher & Jones,
1993). Like the job characteristics model (Hackman & Oldham, 1976), it focuses on
the characteristics of the job rather than the person. Unlike the job characteristics
model however, it proposes that job satisfaction can be increased without altering
work demands.
1.8.2
Description of the Job Demand-Control Model
The job demand-control model proposes that job satisfaction is a function of
the job demands placed on the worker (job demands), and the discretion permitted in
deciding how to address those demands (job decision latitude; Karasek & Theorell,
1990). Job demands are the psychological stressors in the work environment (i.e.,
high pressure of time, high working pace, difficult and mentally exacting work). Job
decision latitude is the worker’s potential control over his/her tasks and conduct
during the working day. Using the job demand and job decision latitude dimensions,
the job demand-control model predicts four outcomes. Two of these outcomes occur
when job demands are high (i.e., active model, high-strain model), whilst the other
two occur when job demands are low (i.e., low-strain model, and passive model).
58
The most positive outcomes, including learning and growth, result from
active jobs, where both job demands and job decision latitude are high. Although
high job demands increase physiological arousal (i.e., increase heart rate or
adrenaline), high job decision latitude allows this arousal to be reduced. Workers
with high job decision latitude redirect the arousal into an appropriate response.
They can choose how they deal with their demands. Through dealing with demands
in their own way, they can reduce the arousal.
A high strain job is one in which job demands are high and job decision
latitude is low. This type of job results in the most adverse reactions of
psychological strain (i.e., fatigue, anxiety, depression, physical illness). This is
because the arousal from the high job demands cannot be redirected. As the
employees have low job decision latitude, they cannot choose how to handle their
work demands. As a result, their arousal increases, producing a larger physiological
reaction.
The two other models are the low-strain model and the passive model. Low
strain jobs are those in which job demands are low and job decision latitude is high.
These low-strain jobs, although clearly not common, may characterise some selfemployed workers, who only have the occasional customer. Employees in these jobs
have a low risk of job strain as they have few demands that produce arousal. Even
when they do have the occasional demand, they can redirect the arousal into an
appropriate response. Finally, a passive job is one in which both job demands and
job decision latitude are low. Employees in these jobs face few challenges and are
59
unable to test ideas for improving the work environment. As a result, they often
suffer from reduced work motivation.
More recently, in addition to job demands and job decision latitude, Karasek
and Theorell (1990) added social support at work. Social support at work is defined
as the “overall levels of helpful social interaction available on the job from both coworkers and supervisors” (Karasek & Theorell, 1990, p. 69). It is proposed that
social support at work is positively related to job satisfaction, and that job demands,
job decision latitude and social support at work interact to predict job satisfaction.
1.8.3
Empirical Studies of the Job Demand-Control Model
Initial tests of the job demand-control model demonstrated that both job
demands and job decision latitude predicted a number of dependent variables,
including exhaustion, depression, job dissatisfaction, life satisfaction, pill
consumption and sick days (Karasek, 1979). Job demands were positively related to
these variables, whilst job decision latitude was negatively related to these variables.
Replications of Karasek’s (1979) study have demonstrated that job demands and job
decision latitude separately predict the dependent variables (Dwyer & Ganster, 1991;
Fletcher & Jones, 1993; Payne & Fletcher, 1983; Spector, 1987; Warr, 1990).
Although these results are supportive of the model, the central proposition of
the job-demand control model is that job demands and job decision latitude interact
to predict job strain. This interaction effect was tested through regression analyses
where the interaction term was added (Karasek, 1979). These analyses demonstrated
that job demands and job decision latitude interacted to predict exhaustion, job
60
dissatisfaction, and life dissatisfaction. The following beta values were observed for
exhaustion, (decision latitude = -0.004, demands = 0.07, interaction = 0.11), job
dissatisfaction, (decision latitude = -0.22, demands = 0.001, interaction = 0.12), and
life dissatisfaction, (decision latitude = -0.13, demands = -0.03, interaction = 0.11).
Although these interaction terms were significant, the method of analysis was
subsequently criticised (Fletcher & Jones, 1993; Ganster & Fusilier, 1989).
Researchers propose that Karasek (1979) rejected the traditional tests of interaction
based on partialed product terms in regression analyses, and rather relied on variables
that reflected differences between demands and control (Fletcher & Jones, 1993;
Ganster & Fusilier, 1989).
When researchers have replicated these analyses using an appropriate test of
the interaction effect specified by Cohen and Cohen (1983), the interaction effect
tends to be insignificant (Fletcher & Jones, 1993; Payne & Fletcher, 1983; Warr,
1990). For example, Payne and Fletcher (1983) tested the job demand-control model
on secondary school teachers. Using multiple regression they demonstrated that the
interaction term did not predict the dependent variables, including depression,
anxiety, obsession, somatic symptoms, and cognitive failures.
Although these studies suggest that job demands and job decision latitude do
not interact to predict job satisfaction, a major problem has been identified in the
measurement of job decision latitude. Job decision latitude is defined as “the
working individual’s potential control over his tasks and his conduct during the
working day” (Karasek, 1979, p. 289-290). However, the most recent measure of job
decision latitude, developed by Karasek and Theorell (1990), includes items
61
reflecting decision latitude and decision authority. Decision latitude refers to
whether the job involves learning new things, and developing skills. Decision
authority refers to whether the person has the freedom to make their own decisions
and if they can choose how they perform their work. Although the decision authority
items are consistent with the definition of job decision latitude, the decision latitude
items have been criticised for measuring skill level, skill variety, and job complexity
(Ganster, 1989). Factor analyses of this scale have confirmed that two factors
emerge (Smith, Tisak, Hahn, & Schmeider, 1997), namely decision latitude and
decision authority. However, only the decision authority items are consistent with
the definition.
Many researchers have proposed that the definition of job decision latitude is
similar to the definition of job autonomy (de Jonge, Breukelen, Landeweerd &
Nijhuis, 1999; Ganster & Fusilier, 1989; Spector, 1986). Indeed, Ganster and
Fusilier (1989, p. 256) propose that the “definition of job decision latitude mirrors
job autonomy.” Job decision latitude is “the working individuals potential control
over his tasks and his conduct during the working day” (Karasek, 1979, p. 289-290),
whilst job autonomy is “the degree to which the job provides substantial freedom,
independence, and discretion to the individual in scheduling the work and in
determining the procedures to be used in carrying it out” (Hackman & Oldham,
1976, p.258). As a result of this similarity, researchers have tested the job demandcontrol model using measures of job autonomy (de Jonge, Mulder & Nijhuis, 1999;
Dwyer & Ganster, 1991). It must be noted that these researchers may refer to their
62
scales as measuring job control, however job control and job autonomy appear to be
interchangeable (de Jonge et al., 1999b).
For example, the interaction effect of the job demand-control model was
tested using Ganster’s (1989, cited in Dwyer & Ganster, 1991) multidimensional
control scale. This scale examines the amount of choice employees have in several
areas of their work, such as their work tasks, work pacing, work scheduling, physical
environment, decision making, interaction and mobility.
Using regression analyses, Dwyer and Ganster (1991) demonstrated that the
interaction term predicted absenteeism, satisfaction with work, tardiness and sick
days. Specifically, the interaction term contributed an additional 15% to explaining
the variance in absences, 4% in satisfaction with work, 26% in tardiness, and 4% in
sick days. These findings suggest that further research on the job demand control
model is required using Ganster’s (1989, cited in Dwyer & Ganster, 1991) scale.
1.8.4
Conclusion
The job demand-control model is intuitively appealing, proposing that job
decision latitude can ameliorate job demands. This theory has received partial
support as job demands and job decision latitude have separately predicted the
dependent variables. Whether these two variables interact to predict job satisfaction
continues to be debated.
63
1.8.5
Extensions on the Job Demand-Control Model
Although the evidence for the job demand-control model has been equivocal,
there are two main reasons why this theory, over the other reviewed theories,
deserves further attention. First, the proposition that job autonomy can somehow
ameliorate job demands is certainly appealing to employers (Ganster & Fusilier,
1989). It suggests that employers can increase job satisfaction without altering work
demands. Second, although few researchers are continuing to investigate the other
theories, the job demand-control model continues to be the subject of many papers
(e.g., de Jonge et al., 1999a; Dollard, Winefield, Winefield, & de Jonge, 2000;
Hallqvist, Diderichsen, Theorell, Reuterwall, & Ahlbom, 1998; Lu, 1999; Parker &
Sprigg, 1999).
1.8.6
Addressing the “Gaps” in the Job Demand-Control Model
Although the job demand-control model deserves further attention, it must be
recognised that, in addition to the operationalisation of job decision latitude, there is
a major gap in the theory. This involves the explanation of how job decision latitude
results in positive outcomes.
1.8.6.1
The Explanation of how Job Decision Latitude Results in Positive
Outcomes
The model proposes that job decision latitude increases job satisfaction by
allowing employees to redirect the physiological arousal produced from job
64
demands. Specifically, Karasek and Theorell (1990) propose that employees with
high job decision latitude can translate the physiological arousal produced from job
demands into action through effective problem solving. They propose that workers
with high job autonomy are “given the freedom to decide what is the most effective
course of action in response to a stressor” (Karasek & Theorell, 1990, p. 36). Job
decision latitude gives employees the “freedom of action in accomplishing the formal
work task…and the freedom to engage in the informal rituals” (Karasek & Theorell,
1990, p. 34).
A major problem with this explanation however is that it is tautological. This
explanation proposes that job decision latitude, or the ability to choose at work, is
beneficial because it allows people to choose how they deal with their work
demands. Furthermore, the model is non-specific, failing to discuss how the
physiological arousal produced from job demands is redirected, and failing to define
the most effective course of action. As such, it is unknown how a person with low
job decision latitude handles a job demand, and how this is different from a person
with high job decision latitude. In response to this criticism, a new explanation for
the relationship between job decision latitude and job satisfaction is developed. This
explanation specifies how employees with low job autonomy differ from employees
with high job autonomy.
65
1.9
Development of a new Explanation for the Relationship Between Job
Autonomy and Job Satisfaction: Influencing Employees’ Responses to
Work Difficulties
The job demand-control model proposes that workers with higher job
autonomy have higher job satisfaction because they can channel the arousal produced
from job demands into an appropriate response. A new explanation is developed
which proposes that employees with low job autonomy respond differently to work
difficulties than employees with high job autonomy.
It must be noted that this explanation focuses on work difficulties rather than
job demands. Job demands are the psychological stressors in the work environment
(i.e., high pressure of time, high working pace, difficult and mentally exacting work;
Karasek & Theorell, 1990). It is expected that job autonomy will influence
employees’ responses to these job demands, but that the hypothesis can be extended
to any type of work difficulty. Thus, job autonomy is expected to influence
employees’ responses to their supervisors, co-workers, pay, opportunities for
promotion, and so forth.
In response to a work difficulty, employees can change the situation to suit
themselves, or they can change themselves to suit the situation (Heckhausen &
Schulz, 1995; Rothbaum, Weisz & Snyder, 1982). These two strategies are referred
to as primary control and secondary control strategies respectively. Before
discussing how job autonomy influences the control strategies that employees use,
the two strategies will firstly be examined. Specifically, the nature of the strategies
66
will be examined, followed by a discussion of the strategies that people generally use
and the most adaptive strategies.
1.9.1
a) Primary Control Strategies and Secondary Control Strategies
Two strategies implemented by employees when they face difficult situations
are primary control strategies and secondary control strategies (Rothbaum et al.,
1982). Primary control involves changing the work environment to suit one’s needs,
whilst secondary control strategies involve changing oneself to suit the work
environment. For example, if an employee felt they were being underpaid, they
could use a primary control strategy, such as confronting their employer, or they
could use a secondary control strategy and compare themselves to others who are
worse off.
This conceptualisation of primary and secondary control is similar to Lazarus
and Folkman’s (1984) conceptualisation of problem-focussed coping and emotionfocussed coping. In this case, coping refers to the “constantly changing cognitive
and behavioral efforts to manage specific external and/ or internal demands that are
appraised as taxing or exceeding the resources of the person” (Lazarus & Folkman,
1984, p. 141). Coping strategies are employed to manage the problem causing the
distress (i.e., problem-focussed coping) and to regulate the accompanying emotions
(i.e., emotion-focussed coping; Folkman & Lazarus, 1980).
The theory underlying problem-focussed and emotion-focussed coping and
the questionnaire designed to assess these strategies (i.e., Ways of Coping
Questionnaire; WCQ; Folkman & Larazus, 1980) is shrouded in methodological
67
limitations (Edwards & O’Neill, 1998). First, the definition of coping focuses on
managing demands that tax or exceed personal resources. Thus, coping strategies
should manage or reduce demands and enhance personal resources to meet demands.
The Ways of Coping Questionnaire (Folkman & Larazus, 1980) examines how an
individual can cope with a situation by changing the environment or the self,
however it is not specified how these strategies manage demands or enhance
personal resources (Edwards & O’Neill, 1998).
Second, there is often a great deal of overlap among the coping dimensions,
where some problem-focussed coping strategies resemble emotion-focussed coping
strategies (Edwards & O’Neill, 1998). Problem-focussed coping is aimed at problem
solving, or doing something to alter the situation, however it also includes strategies
that alter the self. For example, the problem-focussed coping strategy of “shifting
one’s aspiration level” involves the person attempting to move one’s goals to be
more in line with the current situation (Lazarus & Folkman, 1984). Furthermore,
“reducing ego involvement” involves the person attempting to reduce the overall
significance of the situation to oneself (Lazarus & Folkman, 1984). These strategies
alter the self and are more consistent with emotion-focussed coping strategies, which
aim to reduce the emotional distress associated with the problem.
Third, and perhaps most concerning is that factor analyses of the Ways of
Coping Questionnaire are highly inconsistent. Researchers have found support for
three factors (Parkes, 1984), six factors (Vitaliano, Russo, Carr, Maiuro & Becker,
1985) and eight factors (Aldwin & Revenson, 1987; Folkman et al., 1986; Lazarus &
Folkman, 1984). Edwards and O’Neill (1998) used confirmatory factor analysis to
68
evaluate these alternative factor structures, concluding that these models yielded poor
fit.
The conceptualisation of primary control and secondary control is superior to
problem-focussed and emotion-focussed coping because it maintains the distinction
between changing the environment (i.e., primary control), and changing the self (i.e.,
secondary control). The control model not only addresses responses to threat and
negative events, but also behaviour directed at growth and potential (Schulz &
Heckhausen, 1996). Furthermore, the items on the scale are consistent with the
definitions of the control strategies.
1.9.2
b) Amounts of Primary Control and Secondary Control
The life span theory of control, developed by Heckhausen and Schulz (1995)
specifies which kind of strategies people rely on throughout their life. They propose
that adults implement both primary and secondary control strategies, however in
Western samples, primary control strategies tend to be implemented first, and are
generally preferred over secondary control strategies.
Research examining the frequency of primary control and secondary controltype strategies in the work environment is generally supportive of Heckhausen and
Schulz's (1995) propositions (e.g., Boey, 1998; Koeske, Kirk & Koeske, 1993). For
example, Boey’s (1998) study on nurses demonstrated that, using a scale ranging
from 0 to 4, on average, problem-focussed strategies (M = 2.47) were reported more
than emotion-focussed strategies (M = 1.63). In general, theoretical and empirical
research suggests that people tend to rely on primary control more than secondary
69
control. The next step is to determine whether primary control strategies are also the
most adaptive strategies.
1.9.3
c) Which Control Strategies are more Adaptive for Employees?
To determine which control strategies are more adaptive for employees,
theoretical and empirical research is examined. The theoretical propositions are
based on the life span theory of control (Heckhausen & Schulz, 1995), and the
empirical studies specify the correlations between the control strategies and job
satisfaction.
1.9.3.1
Theoretical Propositions: The Life span Theory of Control
The life span theory of control (Heckhausen & Schulz, 1995) proposes that
primary control is more adaptive than secondary control as it allows individuals to
meet their own needs. If a person successfully changes their environment using
primary control, they overcome their difficulty and also enhance their general
perceptions of control.
Secondary control strategies are less adaptive than primary control strategies,
however they have two main benefits (Heckhausen & Schulz, 1995). They
compensate for primary control failure and they assist individuals to focus on goals
that expand primary control (Heckhausen & Schulz, 1995).
Secondary control compensates for primary control failure, which may
threaten self-esteem, self-efficacy, and general perceptions of control (Heckhausen,
Schulz & Wrosch, 1997). If an individual experiences repeated primary control
70
failure, they may become vulnerable to experiencing learned helplessness. However,
if they implement secondary control after primary control failure, they can protect
their self-esteem, and reduce the likelihood of experiencing repeated primary control
failure.
For example, an individual may face a difficulty at work, where a co-worker
is working at a slow pace. To handle this difficulty, they could use a primary control
strategy or a secondary control strategy. It is expected that they would firstly
implement a primary control strategy, where they may confront their co-worker.
They may discuss the problem with them, and the co-worker may agree to put in
more effort. If this primary control strategy is successful, they overcome their
difficulty. If the strategy fails however and the co-worker continues to work at the
same pace, the employee is likely to experience a loss in their general perception of
primary control. To avoid repeating this situation, they could implement a secondary
control strategy, such as wisdom control, where they think, “I can’t always get what I
want.” Through implementing this strategy, they avoid risking repeated primary
control failure.
Secondary control strategies are also beneficial in assisting individuals to
focus on goals that expand primary control. An individual may continue to persist to
solve a difficulty if they implement secondary control strategies such as focussing on
past success. Through such a focus, the individual may feel more confident in their
ability to overcome the problem. In summary, theoretically, primary control is more
adaptive than secondary control.
71
1.9.3.2
ii) Empirical Studies Examining the Adaptiveness of Primary and
Secondary Control
Empirical studies examining the relationship between primary control and
secondary control-type strategies and job satisfaction/job stress have provided some
support for the life span theory of control (Boey, 1998; Burke & Greenglass, 2000;
Koeske et al., 1993; Kohn, Hay & Legere, 1994; Norman, Collins, Conner, Martin &
Rance, 1995).
These studies generally demonstrate that primary control strategies are more
positively related to job satisfaction than secondary control strategies. For example,
Norman et al’s., (1995) study of teleworkers demonstrated that problem-focussed
coping was positively correlated with job satisfaction (r = 0.33) and
emotion-focussed coping was negatively related to job satisfaction (r = -0.22). In
Burke and Greenglass’s (2000) study of nurses, control coping was also positively
related to job satisfaction (r = 0.14) and escape coping was negatively related to job
satisfaction (r = -0.12). Furthermore, in Kohn et al’s., (1994) study of teachers,
task-oriented coping was negatively related to perceived stress (r = -0.38) and
emotion-oriented coping was positively related to stress (r = 0.63).
Although these results suggest that, consistent with the life span theory of
control, primary control is more adaptive than secondary control, it is important to
note however that these studies have conceptualised primary control and secondarytype strategies using several different constructs and scales.
72
For primary control, many of the scales are poorly designed, including items
that do not appear to measure primary control-type strategies. For example, Burke
and Greenglass (1995) relied on Latack’s (1986) measure of control coping. Control
coping refers to actions and cognitive reappraisals that are proactive. Many of these
items refer to secondary control-type strategies that make the person feel better about
the problem. For example, the item “try to see the situations as an opportunity to
learn and develop new skills” is measuring a secondary control strategy known as
positive re-interpretation. Furthermore, the items “try to think of myself as a winneras someone who always comes through” and “tell myself that I can probably work
things out to my advantage” refers to another secondary control strategy known as
illusory optimism.
Another scale which confounds primary control strategies with secondary
control type strategies is the control coping scale implemented in Koeske et al’s.,
(1993) study. Many of the items included in this scale appear to measure secondary
control strategies. For example, “talked with spouse or other relative about the
problem”, “tried to see the positive side of the situation”, “got busy with other things
to keep my mind off the problem”, “told myself things that helped me feel better”,
“let my feelings out somehow”, and “exercising more.” These strategies attempt to
make the person feel better, rather than change a situation.
The secondary control scales have also been criticised for confounding
secondary control strategies with primary control strategies. For example, Burke and
Greenglass (2000) relied on Latack’s (1986) measure of escape coping. This scale
included primary control-type strategies, such as “delegate work to others” and “set
73
my own priorities based on what I like to do.” Many other scales focused on
negative responses, such as avoidance and denial. For example, Norman et al's.,
(1995) revised version of the COPE scale relied on only five emotion-focussed
coping items, such as “I use alcohol or drugs to make me feel better” and “I give up
the attempt to get what I want.” Furthermore, Boey (1998) measured avoidance
coping through items involving suppression of feelings, blaming others, and getting
mad at people (i.e., taking more tranquillising drugs, drinking more, avoided being
with people in general).
Although avoidance and denial are two types of secondary control, there are
many other ways that people can change the self to fit in with the environment.
Fourteen secondary control strategies have actually been identified in the Primary
and Secondary Scale (Heeps, Croft & Cummins, 2000). These strategies, displayed
in Table 2, concur with Rothbaum et al’s., (1982) and Heckhausen and Schulz's
(1995) definition of secondary control.
In summary, the empirical studies suggest that secondary control-type
strategies are negatively related to job satisfaction. However, these findings may not
be generalised to secondary control as conceptualised by Heckhausen and Schulz
(1995), since these authors recognise that there are many positive secondary control
strategies.
74
Table 2- Secondary Control Strategies
Item
I can see that something good will come of it.
I remember you can't always get what you want.
I know things will work out OK in the end.
I remember I am better off than many other people.
I remember I have already accomplished a lot in
life.
I remember the success of my family and friends.
I think nice thoughts to take my mind off it.
I tell myself it doesn't matter.
I don't feel disappointed because I knew it might
Happen.
I can see it was not my fault.
I ignore it by thinking about other things.
I realise I didn't need to control it anyway.
I think about my success in other areas.
Secondary control strategy
Positive re-interpretation
Wisdom
Illusory-optimism
Downward social
comparison
Past success
Vicarious
Positive approach
Goal disengagement
Predictive-negative
Attribution
Active avoidance
Sour grapes
Present success
Source: Heeps, L., Croft, C., & Cummins, R.A. (2000). Primary control and
Secondary Control Scale (2nded.). Melbourne: Deakin University.
1.9.3.3
Comparing the Life Span Theory of Control and Empirical Studies
Examining the most Adaptive Control Strategy
The life span theory of control (Heckhausen & Schulz, 1995) proposes that
primary control strategies are more adaptive than secondary control strategies as they
allow individuals to meet their own needs, and they facilitate general perceptions of
control. Secondary control strategies are still useful however in compensating for
primary control failure and assisting individuals to focus on goals that expand
primary control. The empirical studies partly concur with these propositions,
demonstrating that primary control-type strategies are positively related to job
satisfaction. Inconsistently however, several studies demonstrate that secondary
75
control type strategies are negatively related to job satisfaction (Boey, 1998; Burke &
Greenglass, 2000; Friedman et al., 1992; Koeske et al., 1993; Kohn et al., 1994;
Norman et al., 1995). It must be noted however, that these empirical studies have
relied on many different scales, some of which are methodologically flawed.
1.9.4
Summary
Employees implement primary control and secondary control strategies when
they face a difficulty at work, however they tend to rely on primary control more
than secondary control. Primary control strategies allow individuals to meet their
own needs, and are positively related to job satisfaction. Secondary control
strategies are assumed to compensate for primary control failure, and assist
individuals to focus on goals that expand primary control. Although they have been
negatively related to job satisfaction in previous studies, the scales have been
criticised for focussing on negative strategies. It is expected that secondary control
strategies, as assessed through the Primary Control and Secondary Control Scale
(Heeps et al., 2000), will be beneficial for people after they have experienced
primary control failure.
1.10 Explaining the Relationship Between Job Autonomy and Job Satisfaction:
How Job Autonomy Influences Primary and Secondary Control
The explanation for the relationship between job autonomy and job
satisfaction proposes that job autonomy influences the way employees respond to
76
their work difficulties. It is expected that job autonomy will influence the use and
adaptiveness of primary and secondary control strategies.
Past research has tended to confuse job autonomy and primary control
(Thompson, Collins, Newcomb & Hunt, 1996) and as such, these two will be
differentiated. Job autonomy refers to whether employees perceive that they can
control aspects of their work environment, whereas primary control is a strategy that
employees use to change their work environment. An employee who has high job
autonomy perceives that they can choose, or control many aspects of their work. An
employee who has high primary control perceives that they change their environment
when they face a difficulty at work.
Although job autonomy and primary control are different, they are expected
to be related to each another. Specifically, job autonomy should influence: 1), which
control strategies employees rely on; and 2) the adaptiveness of the control strategies
(i.e., the relationship between the control strategies and job satisfaction).
1.10.1
1) Use of Primary and Secondary Control
It is expected that all individuals, with either low or high job autonomy, will
implement both primary control and secondary control strategies. Both groups will
implement primary control strategies first, and if they experience primary control
failure, they will then implement secondary control strategies. The difference
between the two groups lies in the amount of primary control failure that they
experience. Employees with low job autonomy have little influence over their work
environment are expected to experience more primary control failure than employees
77
with high job autonomy. As they need to compensate for this failure, it is expected
that these employees (i.e., low job autonomy) will implement more secondary
control than employees with high job autonomy. Hence, it is proposed that the
ability to choose is inversely related to the probability of primary control failure,
which in turn, influences the use of secondary control strategies.
These propositions are based on Heckhausen and Schulz’s (1995) life span
theory of control. This theory proposes that as people age, they experience reduced
autonomy, and they begin to experience primary control failure more often. To
compensate for this primary control failure, they need to increase their reliance on
secondary control strategies. For example, an older individual with restricted
mobility may experience primary control failure when working hard to maintain their
garden. To reduce the amount of primary control failure that they experience, they
can rely on secondary control strategies such as downward comparison (e.g. “I am
better off than others my age”). The proposal that older people rely on more
secondary control strategies than younger people has been confirmed in several
studies (i.e., Chipperfield, Perry & Menec, 1999; Maher & Cummins, 2001;
McConatha & Huba, 1999).
1.10.2
2) Adaptiveness of Primary and Secondary Control
In addition to influencing the relative use of primary and secondary control
strategies, job autonomy may also influence the adaptiveness of such strategies.
Although it was previously demonstrated that primary control-type strategies were
positively related to job satisfaction, and secondary control-type strategies were
78
negatively related to job satisfaction, it has been suggested that these relationships
may change if the person perceives that the situation is uncontrollable (Thompson et
al., 1996; Thompson, Nanni & Levine, 1994; Thompson, Sobolew-Shubin,
Galbraith, Schwankovsky & Cruzen, 1993; Thompson et al., 1998).
Two models have been developed to explain this relationship, namely the
discrimination model and the primacy/back-up model (Thompson et al., 1998). The
discrimination model proposes that primary control is more adaptive than secondary
control when the situation is controllable, and that secondary control is more
adaptive than primary control when the situation is uncontrollable. This model
underlies the philosophy of the serenity prayer; “Grant me the strength to change
what I can, the patience to accept what I cannot, and the wisdom to know the
difference” (Thompson et al., 1998, p. 587). In regard to job autonomy, this model
suggests that primary control strategies are more adaptive for employees with high
job autonomy, and secondary control strategies are more adaptive for employees
with low job autonomy.
The primacy/back-up model, on the other hand, proposes that primary control
is more adaptive than secondary control in controllable and relatively uncontrollable
situations. The role of secondary control is only to “compensate for low primary
control, and help increase feelings of overall control” (Thompson et al., 1998,
p. 587). Thus secondary control is only beneficial when primary control is low. In
regard to job autonomy, the primacy/back-up model proposes that primary control
strategies are the most adaptive strategy for employees with high job autonomy and
for employees with low autonomy, unless primary control is low.
79
It is difficult to differentiate the primacy/back-up model from the
discrimination model. The primacy/back-up model proposes that primary control is
more adaptive than secondary control unless primary control is low. If a person has
low primary control, they believe that they cannot change the environment using an
active strategy, such as working hard. However, this means that they perceive the
situation to be uncontrollable. Hence, the primacy/back-up model is proposing that
secondary control is only useful when primary control is low, however primary
control is low when the situation is perceived as being uncontrollable. This is indeed
similar to the discrimination model, which proposes that secondary control is best in
uncontrollable situations. As such, it appears that there may be some overlap in the
models.
In order to reduce the overlap in the models, the primacy/back-up model
should be revised to propose that primary control is the most adaptive strategy in
controllable and uncontrollable situations. The proposal that secondary control is
beneficial when primary control is low needs to be excluded as it overlaps with the
discrimination model. Researchers who have tested the primacy/back-up model
generally focus on the proposal that primary control is adaptive in low-control and
high-control situations.
1.10.2.1
Empirical Studies Examining the Discrimination Model and the
Primacy/Back-Up Model
Only a few studies have examined the most adaptive control strategies in
low-control situations (Thompson et al., 1996; 1994; 1993; 1998). In a review of
80
these studies, Thompson et al., (1998) concluded that they generally supported the
primacy/back-up model. As several serious methodological problems have been
identified in these studies, they will be reviewed.
1.10.2.2
Primacy/Back-Up Model
The first study that Thompson et al., (1998) cites as supporting the
primacy/back-up model is Thompson et al’s., (1993) study. According to Thompson
et al., (1998), this study demonstrated that cancer patients with higher levels of
primary control were less depressed than those with lower levels of primary control.
Control was negatively related to maladjustment (r = -0.46) and positively related to
physical functioning (r = 0.39) and marital satisfaction (r = 0.24). As such, it was
concluded that this study supported the primacy/back-up model (Thompson et al.,
1998).
However, this conclusion is inaccurate as the study did not measure primary
control, rather it measured perceived control. The participants were firstly asked
about how much control they had over various facets of their lives (i.e., perceived
control over emotions, physical symptoms, relationship with family). They were
then asked what type of things they have done to control their feelings over each
facet, and how effective these were. The items measuring amount of perceived
control were then added to the effectiveness item for each facet. The resulting scale
assessed perceived control, and the effectiveness of the control strategies, but clearly
failed to measure primary control.
81
When the findings are reinterpreted using perceived control rather than
primary control, they are intuitive. It is not surprising that it is beneficial for cancer
patients to believe that they can control areas of their lives. Indeed, a fundamental
belief about human nature is that we have a need to control events, people and
situations (DeCharms, 1968; White, 1959). However, perceived control is not the
same as primary control. Whereas autonomy or control refers to whether a person
perceives that can change the environment, primary control refers to the specific
strategies people use to change the environment to suit their needs. As such,
Thompson et al’s., (1993) study does not adequately test the primacy/back-up model.
Another study which claims to support the primacy/back-up model is
Thompson et al’s., (1996) study on HIV-positive men in prison. They examined the
relationship between primary and secondary control and distress. Regression
analyses demonstrated that primary control was negatively related to distress and
secondary control was positively related to distress. Although these findings suggest
that people in a low-control environment should rely on primary control, the
measurement of primary and secondary control in the study is questionable.
To measure primary control, the participants were asked how much control
they had over a variety of outcomes, such as their feelings, day-to-day activities,
nutrition, and HIV-related symptoms. This measure is criticised however, as the
items do not refer to primary control strategies, but rather refer to levels of perceived
control. As mentioned previously, primary control is not the same as perceived
control. As such, a person may report that they can control their relationship with
their cellmates and how their correctional officers treat them, however this does not
82
indicate that they use primary control strategies when they have a difficulty with
their cellmates or correctional officers.
A further problem with Thompson et al’s., (1996) study concerns the
measurement of secondary control. Secondary control was measured by the
following item; “How much do you feel okay about (an outcome) because you just
accept it and don’t try to change it?” Although secondary control generally involves
acceptance of the situation, this item is criticised as it fails to make respondents
aware of the different ways they can accept a situation. For example, they can
believe that it will work out okay in the end (i.e., illusory optimism), or they can
think that they can't always get what they want (i.e., wisdom). The respondents in
Thompson et al’s., (1996) study were not made aware of these different strategies,
and as such may have underestimated their use of secondary control. A further
problem with this measure of secondary control is that it does not just ask if the
person uses acceptance, rather it confounds acceptance with feeling okay.
One final study which claims to support the primacy/back-up model is
Thompson et al’s., (1998). They examined whether adults (young, middle, and
older) use primary or secondary control to handle their appearance-related changes
due to aging. The youngest group was expected to have the most perceived control
over age-related changes, whilst the oldest group was expected to have the least.
Averaging over all age groups, primary control (r = 0.46) and secondary
control (r = 0.42) were positively related to satisfaction with physical appearance,
and primary control (r = -0.20) and secondary control (r = -0.24) were negatively
related to emotional distress. Although these correlations suggest that secondary
83
control is adaptive, multiple regression analyses indicated that secondary control was
only beneficial when primary control was low. There was no relationship between
secondary control and distress for those with high primary control, but for those with
low primary control, secondary control was negatively related to distress.
Although these results appear to provide support for the primacy/back-up
model, the items measuring primary and secondary control were poorly constructed.
Primary control was measured by the following five items rated on a scale from
strongly agree to strongly disagree: 1) “I feel that I have some control over the
effects of aging on my appearance”; 2) “I dread the thought of aging, but there is not
much I can do about it” (reversed coded); 3) “As long as I put the effort in I can keep
looking attractive”; 4) “I can stay attractive and youthful as long as possible if I just
work at it"; and 5) “I get depressed when I think about what’s coming as I get older”
(reverse coded).
These items are criticised as some of them are based on the assumption that
aging is a negative process (items two and five). For example, although Thompson
et al., (1998) proposed that a person who disagreed with the item “I dread the thought
of aging, but there is not much I can do about it” has high primary control, it may be
that they do not dread the thought of aging. Furthermore, the item “I get depressed
when I think about what’s coming as I get older” does not refer to a secondary
control strategy, and simply refers to the persons attitude towards aging. Other items
are based on the assumption that people perceive themselves as being attractive
(items three and four). For example, a person may disagree with the item “I can stay
84
attractive and youthful as long as possible if I just work at it”, not because they have
low primary control, but because they do not believe that they are attractive.
The items in the secondary control scale are also criticised for being based on
the assumption that aging is a negative process. The scale includes items such as “as
long as I know what’s coming, it doesn’t bother me too much to get older” and “I am
not worried about getting older, because I trust that God will take care of me.” These
items confound the perceptions of aging with the secondary control strategy. As
such, it is impossible to tell if the person is referring to the part of the question
referring to aging or the part referring to the strategy. For example, a respondent
may report that they strongly agree they are "not worried about getting older, because
they trust that God will take care of them” because they are not worried about getting
older, or because they trust that God will take care of them. In summary, as with the
other reviewed studies, several measurement issues limit the validity of Thompson et
al’s., (1998) findings.
1.10.2.3
Discrimination Model
One study conducted by Thompson et al., (1994) supported the
discrimination model. This study examined the relationship between primary and
secondary control and depression for men with a diagnosis of HIV. Both strategies
appeared to be adaptive for people who presumably were in a low-control situation.
Primary control (r = -0.36) and secondary control (r = -0.41) were negatively related
to depression. Furthermore, for the group that was low in primary control, secondary
control was negatively related to depression. For those high in primary control,
85
secondary control was weakly related to depression. The values of these correlations
cannot be discussed however, as the authors only reported them in graphical form.
This study provided some support for the discrimination model, however the
measurement of primary control and secondary control strategies was once again
limited. The items measuring the control strategies were the same as Thompson et
al’s., (1996) study, where primary control strategies were measured by perceived
control and secondary control strategies were measured by acceptance.
1.10.2.4
Conclusion: Does Research Support the Primacy/Back-Up Model or
the Discrimination Model?
Most of the studies reviewed thus far have concluded that their findings
support the primacy/back-up model. However as these studies have often failed to
validly measure primary control and secondary control strategies, further research is
required to determine whether primary control is adaptive in low-control situations.
This research must rely on a measure of primary and secondary control that concurs
with Heckhausen and Schulz's (1995) conceptualisation of control.
It is expected that employees with low job autonomy will rely on less primary
control and more secondary control than employees with high job autonomy.
Employees with low job autonomy are expected to have a higher probability of
failing when implementing primary control. To compensate for this primary control
failure, they can rely on secondary control.
86
1.10.3
Summary
Job autonomy refers to the perceived ability to exert choice in the work
environment. It may influence employees use of primary and secondary control, and
the adaptiveness of the control strategies. In regard to the use of the control
strategies, it is expected that the ability to choose facilitates the probability of
primary control failure, which in turn, influences the use of secondary control
strategies. In regard to the adaptiveness, empirical studies suggest that primary
control strategies are more positively related to job satisfaction than secondary
control strategies in low-control situations. These studies are criticised however for
their measurement of primary and secondary control, and it is expected that when
they are measured validly, primary control is more adaptive in controllable
situations, and secondary control is more adaptive in uncontrollable situations.
87
1.11 Other Major Predictors of Job Satisfaction
In addition to job autonomy and the control strategies, two other major
predictors of job satisfaction are examined. They are personality and life
satisfaction.
1.11.1
Personality
Researchers have recently paid considerable attention to the role of
personality in predicting job satisfaction. The most common taxonomy of
personality, the five-factor model (Costa & McCrae, 1985) includes neuroticism,
extroversion, conscientiousness, agreeableness and openness to experience.
Researchers have examined how some of these personality variables, namely
neuroticism and extroversion, influence levels of job satisfaction.
1.11.1.1
Personality and Job Satisfaction
Personality may directly influence job satisfaction. As evidence for this
proposal, researchers have demonstrated that job satisfaction is consistent over time
and across situations. For example, Staw and Ross (1985) demonstrated that job
attitudes remained consistent over time, even if the person changed employer, and/or
occupation. They conducted a longitudinal survey, administering a one-item
measure of job satisfaction to over 5000 men in 1966, 1969 and 1971. They
correlated the scores on this measure of job satisfaction over time. The correlation
between satisfaction scores when the employer and occupation were the same, were
88
moderate (r = 0.37 to r = 0.48). When the employer or the occupation had changed,
the correlations were only slightly lower (r = 0.19 to r = 0.34). These correlations
provide support for the stability of job satisfaction, however more supportive results
were provided by the regression analyses.
In the regression analyses, the authors used the 1966 and 1969 job
satisfaction scores to predict the 1971 job satisfaction scores. The prior job
satisfaction scores (i.e., 1966, 1969 data) were almost always a better predictor of the
1971 job satisfaction scores than situational variables, such as changes in pay and job
status. This was even the case when the sample had changed employers but had the
same occupation, and when the sample had changed occupation but still had the
same employer. Situational variables, including change in pay was a significant
predictor of the 1971 job satisfaction score only when the employer and occupation
had changed. However, the strength of the relationship was considerably less than
prior job attitudes. Hence, Staw and Ross’s (1985) study demonstrated that job
satisfaction scores could be predicted five years later by earlier job satisfaction
scores, even if the individual had changed their employers or changed their
occupation.
Although Staw and Ross’s (1985) study demonstrated that job satisfaction
remained stable, the authors did not specifically examine the relationship between
personality and job satisfaction. However, Staw, Bell and Clausen (1986) used
measures of childhood personality to predict adulthood levels of job satisfaction.
They combined three longitudinal surveys, and compared the subjects at early
adolescence (12-14 years), late adolescence (15-18 years) and adulthood. They
89
correlated childhood measures of personality with facet job satisfaction, and an
overall one-item measure of career satisfaction. The correlations were all positive,
ranging from r = 0.04 to r = 0.45. Hence, these results suggest that childhood
personality is related to job satisfaction in adulthood.
To add support to the proposal that job satisfaction is influenced by
dispositional variables such as personality, researchers have more recently tested
whether there is a genetic component to job satisfaction. Arvey, Bouchard, Segal
and Abraham’s (1989) studied monozygotic twins who were reared apart. They
completed the Minnesota Satisfaction Questionnaire (Weiss, Dawis, England &
Lofquist, 1967), which consists of an intrinsic satisfaction scale, an extrinsic
satisfaction scale, and a general satisfaction scale. Intraclass correlations, adjusted
for age and sex, were significant for intrinsic satisfaction (r = 0.32) and for general
satisfaction (r = 0.31). Similar findings were found by Arvey, McCall, Bouchard,
Taubman and Cavanaugh (1994) where r = 0.27, and Lykken and Tellegen (1995)
where r = 0.44 to r = 0.52.
In summary, these findings suggest that job attitudes are consistent over time,
that personality measured in adolescence predicts job satisfaction in adulthood, and
that there is a genetic component to job satisfaction. The next step is to examine the
relationship between specific personality characteristics and levels of job
satisfaction.
90
1.11.1.2
The Relationship Between Neuroticism and Extroversion and Job
Satisfaction
Neuroticism tends to be negatively related to job satisfaction, where r = -0.29
(Judge, Bono & Locke, 2000), r = -0.18 (Tokar & Subich, 1997), r = -0.25 (Terry,
Nielsen & Perchard, 1993), r = -0.21 (Smith, Organ & Near, 1983), and r = -0.40,
r = -0.26, r = -0.34 (Hart, 1999). The relationship between extroversion and job
satisfaction tends to be much weaker than that of neuroticism. The following
correlations between extroversion and job satisfaction have been reported; r = 0.25,
r = 0.08, r = 0.18 (Hart, 1999), and r = 0.16 (Tokar & Subich, 1997). In summary,
people reporting higher extroversion and lower neuroticism tend to report higher job
satisfaction.
1.11.1.3
Summary
Personality appears to be an important predictor of job satisfaction. Research
has demonstrated that job attitudes are consistent over time, and that personality
measured in adolescence predicts job satisfaction in adulthood. People high on
extroversion and low on neuroticism tend to report higher job satisfaction.
1.11.2
Life Satisfaction
Researchers have long been interested in the relationship between life
satisfaction and job satisfaction (Judge & Watanabe, 1994). Although varying
definitions and theories of life satisfaction have been proposed, theoretical and
91
empirical support has been provided for seven domains of life satisfaction. These
include material well-being, emotional well-being, productivity, health, intimacy,
safety, and community (Cummins, 1996; Felce & Perry, 1995). Before the
relationship between life satisfaction and job satisfaction is examined, it will be
demonstrated that life satisfaction, like job satisfaction, is influenced by personality.
1.11.2.1
Personality and Life Satisfaction
In addition to job satisfaction, neuroticism and extroversion also predict life
satisfaction (DeNeve, 1999). Neuroticism is negatively related to life satisfaction
where r = -0.29 to r = -0.37 (McCrae & Costa, 1991), r = -0.42 (Costa & McCrae,
1989), and r = -0.46 (Judge et al., 2000). Extroversion is positively related to life
satisfaction, (e.g., r = 0.19 to r = 0.22 (McCrae & Costa, 1991), and r = 0.17 to
r = 0.20 (Costa & McCrae, 1989). On the basis of these low to moderate
correlations, extroversion and neuroticism have been proposed as the key to the
relationship between personality and life satisfaction (DeNeve, 1999; Diener, Suh,
Lucas & Smith, 1999).
1.11.2.2
Life Satisfaction and Job Satisfaction
Job satisfaction is expected to be related to life satisfaction, as work is a
significant and central aspect of many peoples’ lives. Two models have been
developed to explain the linkage between job satisfaction and life satisfaction,
namely the spillover model, and the compensatory model (Wilensky, 1960). The
spillover model assumes that satisfaction in one domain of an individual’s life
92
extends into other areas. Life satisfaction may spillover into job satisfaction or job
satisfaction may spillover into life satisfaction. Either way, life satisfaction and job
satisfaction would be positively related. Alternatively, the compensatory model
proposes that job satisfaction and life satisfaction would be negatively related. An
employee with low job satisfaction would be expected to compensate for this by
engaging in satisfying non-work activities.
A meta-analysis of 34 studies examining the relationship between job
satisfaction and life satisfaction demonstrated that the two variables were positively
correlated, with an average correlation of r = 0.44 (Tait, Padgett & Baldwin, 1989).
Several more recent studies found correlations of similar magnitudes. For example,
Iverson and Maguire (2000) found a correlation of r = 0.23, and Beutell and
Wittig-Berman (1999) reported a correlation of r = 0.39. Judge, Locke, Durham and
Kluger (1998) found that r = 0.68 and r = 0.42, and Landry (2000) found that
r = 0.44. These findings demonstrate that job satisfaction and life satisfaction are
moderately related, and as such, support the spillover model.
Researchers have also examined how life satisfaction relates to the specific
facets of job satisfaction (Wright, Bennett & Dun, 1999; Judge & Locke, 1993).
Judge and Locke’s (1993) study of clerical workers demonstrated that life
satisfaction was positively related to all facets of job satisfaction, including nature of
work (r = 0.39), co-workers (r = 0.17), supervision (r = 0.26), pay (r = 0.35), and
promotion (r = 0.24). In Wright et al’s., (1999) study of professional card dealers,
only satisfaction with pay (r = 0.33) and satisfaction with the work itself (r = 0.28),
93
were related to life satisfaction. Satisfaction with supervision (r = 0.20), satisfaction
with promotional opportunities (r = 0.16), and satisfaction with co-workers
(r = -0.04) were not significantly related to job satisfaction.
Studies generally provide support for the spillover model, and most
researchers tend to rely on this model (Rain, Lane & Steiner, 1991). Although it has
been suggested that this model may not be appropriate for everyone, Judge and
Watanabe (1994) concluded that job satisfaction and life satisfaction were positively
related for approximately 80% of the participants in their study.
Although these studies have supported the spillover model, the methodology
has been criticised. First, common method variance has been identified as a problem
as both job satisfaction and life satisfaction are measured by self-report (Rain et al.,
1991). This issue is extremely difficult to avoid however as there is no acceptable
way to measure attitudes other than self-report. Objective measures of life
satisfaction correlate poorly with self-reported life satisfaction (Cummins, 2000a),
and behavioural measures of job satisfaction correlate only weakly with self-reported
measures of job satisfaction (Iaffaldano & Muchinsky, 1985).
The second methodological limitation concerns the cross-sectional study
designs, which cannot determine the direction of causality between two variables.
Cramer (1995) used cross-lagged correlations to examine a time-related relationship
between job satisfaction and life satisfaction over 13 months. Job satisfaction and
life satisfaction were positively related at the initial testing and also 13 months later,
suggesting that the two variables may be causally related.
94
In summary, the direction of the relationship between job satisfaction and life
satisfaction continues to be debated (Iverson & Maguire, 2000). It is generally
assumed that job satisfaction contributes to life satisfaction, but it is possible that life
satisfaction influences job satisfaction, or that the relationship is reciprocal. It is
clear however, that life satisfaction and job satisfaction are positively related. This
suggests that people with high job satisfaction will also have high life satisfaction,
and that people with low job satisfaction will also have low life satisfaction.
However, the relationship may not be quite so straightforward as life satisfaction is
held under homeostatic control.
1.11.2.3
Consistency of Life Satisfaction
Recent publications have proposed a model for the homeostatic maintenance
of life satisfaction (Cummins, 2000b). The basis of this model is the finding that life
satisfaction, when measured either by a single question about “satisfaction with life
as a whole” or by satisfaction averaged across a number of domains, is remarkably
predictable. The demonstration of this phenomenon has rested on a statistic called a
percentage of scale maximum (%SM) which converts Likert scale data into a range
from 0 to 100. Applying this statistic to the combined mean values from large
population surveys has revealed that they average 75 + 2.5%SM. In other words,
using two standard deviations to define the normative range, it can be predicted that
the mean level of life satisfaction of Western population samples will lie within the
range 70-80%SM (Cummins, 1995).
95
The consistency of these data provides a basis for the proposal that life
satisfaction is held under homeostatic control. The model that describes how such
homeostasis can be achieved proposes two levels of influence. The first involves an
affective “set-point range” which is determined by personality. The second level
involves a buffering system comprising the three processes of perceived control,
optimism, and self-esteem (Cummins, 2000b). Thus, it is proposed, through the
interaction of these mechanisms, the average life satisfaction for normative
population samples is held within the range 70-80%SM.
This model of homeostasis can be used to make predictions about the life
satisfaction of employees with low job autonomy and employees with high job
autonomy. Provided that their homeostatic systems are operating normally, their life
satisfaction is predicted to lie within the normal range. However, the homeostatic
system can be defeated by a substantial source of negative input, and the low job
autonomy group may have an increased probability of encountering such
circumstances. This may be, for example, through exposure to circumstances of
reduced personal control. Thus it is predicted that a sample of employees with low
job autonomy will contain more people experiencing homeostatic defeat than a
sample of employees with high job autonomy. The employees experiencing such
defeat are expected to report an average level of life satisfaction that approximates
the lower boundary of the normative range (70%SM) or even falls below this level.
96
1.11.2.4
Summary
As postulated by the spillover model, life satisfaction and job satisfaction are
positively related. Although they are expected to co-vary, life satisfaction is held
under homeostatic control and may not be free to vary. The average level of life
satisfaction reported by employees is expected to lie within 70-80%SM. They may
report a lower level of job satisfaction however, if they are experiencing homeostatic
defeat.
97
1.12 Model of Job Satisfaction
This review has identified five main predictors of job satisfaction. As
demonstrated in Figure 2, these include job autonomy, primary control, secondary
control, personality, and life satisfaction. The major proposal of the model is that
primary and secondary control mediate the relationship between job autonomy and
job satisfaction. This is represented by the arrows from job autonomy, through
primary and secondary control, to job satisfaction. Primary and secondary control
may not account for all of the variance in job autonomy, and thus job autonomy is
also directly related to job satisfaction.
It is expected that job autonomy influences the use and adaptiveness of the
control strategies. In terms of the use of the control strategies, employees with high
job autonomy are expected to rely on more primary control and less secondary
control than employees with low job autonomy. Employees with high job autonomy
are expected to be more successful when implementing primary control, and thus
have less need for secondary control, which serves to compensate for primary control
failure. In Figure 2, this relationship is represented by the arrow from job autonomy
to the control strategies.
In regard to the adaptiveness of the control strategies, it is proposed that
employees who match their level of job autonomy with their control strategies will
be most satisfied with their jobs. It is expected that primary control will be more
adaptive for employees with high job autonomy and that secondary control will be
more adaptive for employees who cannot control their environment. It is thus
98
expected that job autonomy moderates the relationship between the control strategies
and job satisfaction. In Figure 2, this moderation effect is represented by the
interaction terms (i.e., job autonomy x primary control, job autonomy x secondary
control). These interaction terms are expected to predict job satisfaction.
In addition to the control strategies, personality and life satisfaction are
expected to directly influence job satisfaction. People higher on extroversion and
lower on neuroticism are expected to report a higher level of job satisfaction and life
satisfaction. Life satisfaction and job satisfaction are also proposed to influence one
another.
In summary, the model proposes that job satisfaction can be predicted from
job autonomy, primary and secondary control, personality and life satisfaction. This
model will be tested in study one, with employees that are low in job autonomy and
employees that are high in job autonomy.
99
Figure 2- Model of Job Satisfaction
Job Autonomy x
Primary Control
Primary Control
Job Satisfaction
Job Autonomy
Secondary Control
Job Autonomy x
Secondary Control
Personality
Life Satisfaction
100
2 Chapter 2 - Study One
101
2.1
Abstract
This study tests the model of job satisfaction developed in chapter 1. The major
proposal of this model is that job autonomy influences the use of the control
strategies and the relationship between the control strategies and job satisfaction
(refer to Figure 2). Employees with high job autonomy are expected to rely on more
primary control strategies and less secondary control strategies than employees with
low job autonomy. Furthermore, primary control is expected to be the most adaptive
strategy for employees with high job autonomy, whilst secondary control is expected
to be the most adaptive strategy for workers with low job autonomy. These
propositions were tested by comparing a sample of high job autonomy employees
(university academic staff) with a sample of low job autonomy employees
(supermarket register operators). As hypothesised, the academics reported higher job
autonomy and lower secondary control than the supermarket workers, however the
two groups did not report different levels of primary control. Additionally, primary
control appeared to be the most adaptive strategy for both occupational groups, and
secondary control was not related to job satisfaction. These findings are discussed in
relation to the life span theory of control and the discrimination model.
102
2.2
Proposal for Study One
The model of job satisfaction developed in chapter 1 proposes that job
autonomy relates to job satisfaction through influencing the use of the control
strategies, and the relationship between the control strategies and job satisfaction.
The model also proposes that personality and life satisfaction predict job satisfaction
(refer to Figure 2). In order to test these propositions, study one will compare
workers with low job autonomy with workers with high job autonomy. The first step
is to identify what type of employees fit into these two groups.
2.2.1
Identifying Employees with Low/High Job Autonomy
According to Ganster’s (1989, cited in Dwyer & Ganster, 1991) scale,
employees with high job autonomy can exert choice in several domains of their
work, such as in the scheduling of their rest breaks, procedures and policies, and in
the variety of tasks they perform. An occupational group that appears to exemplify
high job autonomy, is university academic staff. Their level of job autonomy has
rarely been assessed (Leung, Siu, & Spector, 2000), however academics have
traditionally had flexibility in their work, and freedom to pursue their own research
interests (Winefield, 2000). They can often choose among a variety of tasks,
including research, teaching, and administration (Fisher, 1994).
It is particularly important to study academics’ level of job autonomy as
researchers have recently suggested that “although in theory, the freedom indicative
of high control still exists, in practice, there has been a steady erosion of job control”
103
(Fisher, 1994, p. 61). This has been attributed to the increasing demands placed on
academics, where their workloads have increased and there is increasing pressure to
attract external funding (Winefield, 2000). However, the current study proposes that
even if their level of job autonomy is decreasing, they should still be in the upper
range.
Employees in the lower range of job autonomy are those that have little
opportunity to exert choice in their work. They tend to have “routinised” jobs and
have few tasks from which to choose. Supermarket register operators were selected
as representing such low autonomy workers. These workers are expected to have
little control over many aspects of their job, such as their rest breaks, the tasks they
work on, and their working pace.
104
2.3
Aims and Hypotheses
This study aims to compare the levels of job autonomy, control strategies,
and job satisfaction of supermarket register operators with academics. A number of
hypotheses have been developed as follows:
1) Job autonomy will be positively related to job satisfaction, and academics will
report higher job autonomy than supermarket workers.
This hypothesis tests the basic assumptions of the study. The study aims to
extend Karasek’s (1979) job demand-control model, proposing that employees with
high job autonomy have high job satisfaction because they rely on different control
strategies. As such, it needs to be demonstrated that job autonomy is related to job
satisfaction, and that the two groups selected for this study differ in their levels of job
autonomy as expected.
2) The academic group will report less secondary control and more primary control
than the supermarket workers.
This hypothesis examines how job autonomy influences the use of the control
strategies. As the academics are expected to have more control over their working
environment than the supermarket workers, they are more likely to successfully
change it using primary control. As secondary control is used to compensate for, and
avoid future primary control failure, it is expected that the supermarket workers will
report more secondary control than the academics.
105
3) Job autonomy is positively related to primary control and negatively related to
secondary control.
This hypothesis also tests whether job autonomy influences the use of the
control strategies, however, unlike hypothesis two, it is based on the measured level
of job autonomy rather than the assumed level.
4) Primary control will be more positively related to job satisfaction than secondary
control for the academics, and secondary control will be more positively related to
job satisfaction than primary control for the supermarket workers.
This hypothesis tests whether job autonomy influences the relationship
between the control strategies and job satisfaction. According to the discrimination
model (Thompson et al., 1998), primary control is most adaptive in controllable
situations and secondary control is most adaptive in uncontrollable situations.
Although empirical studies have generally failed to support this model, the studies
have been criticised for their measurement of primary and secondary control
strategies.
5) The relationship between primary and secondary control and job satisfaction will
be moderated by job autonomy.
It is expected that the relationship between primary and secondary control
and job satisfaction will change depending on the level of measured job autonomy.
106
This hypothesis is similar to hypothesis four, however, rather than being based on the
assumed level of job autonomy, it is based on the measured level of job autonomy.
6) The relationship between job autonomy and job satisfaction is mediated by
primary and secondary control strategies.
This hypothesis tests the proposed explanation for the relationship between
job autonomy and job satisfaction. It offers an alternative to Karasek and Theorell’s
(1990) explanation of the job demand-control model. They propose that job decision
latitude (i.e., similar to job autonomy) is positively related to job satisfaction because
it gives workers the freedom to choose how they complete their work and thereby
reduces the arousal produced from job demands. An alternative explanation, to be
tested here, is that workers with high job autonomy mostly rely on the preferred
control strategies, namely primary control.
7) Academics will report higher job satisfaction and higher life satisfaction than the
supermarket workers.
As the academics are expected to report higher job autonomy and more
primary control than the supermarket workers, they are expected to report a higher
level of job satisfaction. This level of job satisfaction is expected to be positively
related to life satisfaction.
8) Primary control, secondary control, job autonomy, personality, and life
satisfaction will predict job satisfaction.
107
These variables are assumed to be the major predictors of job satisfaction, as
depicted in Figure 2.
108
2.4
2.4.1
Method
Participants
The sample consisted of 104 university academic staff, and 96 supermarket
register operators. The academic group was obtained from seven Schools within
Deakin University. The response rate was 32%. The supermarket workers group
was obtained from two supermarket chains, with 16 stores being involved. As these
employees only collected a questionnaire if they were interested in participating in
the study, a response rate could not be calculated.
2.4.2
Materials
Both the academics and the supermarket workers received a plain language
statement (refer to Appendix A) and an anonymous questionnaire. The questionnaire
consisted of several scales, which measured job autonomy, job related primary
control and secondary control, job satisfaction, life satisfaction and personality.
2.4.2.1
Job Autonomy
Although this study is examining the job demand-control model, Karasek and
Theorell’s (1990) scale of job decision latitude was not used. This scale is criticised
for confounding job control with skill level, skill variety, and job complexity
(Ganster, 1989). In response to this criticism, Ganster (1989, cited in Dwyer &
Ganster, 1991) developed an work control scale which examined the amount of
choice an employee has in several areas of their work, such as their work tasks, work
109
pacing, work scheduling, physical environment, decision making, interaction, and
mobility.
Although the scale has good reliability (Fox, Dwyer & Ganster, 1993;
Ganster, Dwyer & Fox, 2001; Schaubroeck & Merritt, 1997), a factor analysis
demonstrated that two factors emerged (Smith et al., 1997). One factor included
items on job autonomy (16 items), while the other factor included items on
predictability (5 items). The predictability items include “how much can you
generally predict the amount of work you will have to do on a given day” and “how
much are you able to predict what the results of decisions you make on the job will
be.” As these predictability items load on a different factor from the job autonomy
items, they should be excluded from the scale. Hence, for the purpose of the present
study, only the former items were used.
A further potential problem with this scale is that some of the items directly
refer to control. In an attempt to disguise the purpose of the scale, these items were
changed from “control” to “choice.” For example, the item “how much control do
you have over the quality of your work” was changed to “In my job, I can choose the
quality of my work.”
Furthermore, to reduce the number of items in the scale, two repetitive items
were deleted. The items “how much control do you have over when you come to
work and leave” and “how much control do you have over when you take vacations
or days off” were deemed to be too similar to the following item; “how much control
do you have over the scheduling and duration of your rest breaks.” All items refer to
the timing and scheduling of rest breaks, and as such, only the latter item was
110
retained. This revised job autonomy scale consisted of 14 items (refer to Appendix
B). They were rated on a 10-point scale ranging from 1 (do not agree at all) to 10
(agree completely).
2.4.2.2
Primary Control and Secondary Control
The control strategies were assessed by the Primary and Secondary Control
Scale, developed by Heeps et al., (2000). This includes five items assessing primary
control strategies and 14 items assessing secondary control strategies. All of these
items were revised to be relevant to the workplace (refer to Appendix C). They were
measured on a 10-point scale, ranging from 1 (do not agree at all) to 10 (agree
completely). Although this scale was only developed recently, early factor analyses
suggest that two factors emerge (Maher & Cummins, 2001; Misajon & Cummins, in
press).
This scale was deemed to be superior to Heckhausen et al’s., (1997)
Optimisation in Primary and Secondary Control Scale (OPS). The reason this scale
was not selected merits discussion, as the proposed model of job satisfaction is partly
based on the life span theory of control. In this theory, Heckhausen and Schulz
(1995) propose that humans face two challenges in life; the need to be selective and
the need to compensate for failure. On this basis, the Optimisation in Primary and
Secondary Control Scale (Heckhausen et al., 1997) measures two types of primary
and secondary control; selective and compensatory.
Selective primary control is the investment of resources to reach goals, whilst
compensatory primary control is used when internal resources are insufficient
111
(i.e., others help, technical aids). Selective secondary control refers to selfmanagement directed at enhancing commitment to goals, and compensatory
secondary control serves to buffer the negative effects of failure.
An alternative simpler explanation for the use of primary and secondary
control is offered. Rather than assisting with the need to be selective and the need to
compensate for failure, it is proposed that people use control strategies whenever
they risk losing control. Examples of such situations are when people are unable to
solve a problem, or when something bad happens to them. Primary control provides
a sense of control derived from changing one’s realities, whereas secondary control
provides a sense of control derived from accepting or adjusting to one’s realities
(Halliday & Graham, 2000; Thompson et al., 1994).
As this new explanation is not consistent with the Optimisation in Primary
and Secondary Control Scale (Heckhausen et al., 1997), this scale was not
appropriate for this study. The scale includes some situations, which do not appear
to prompt the use of control strategies, such as “when I have decided on something.”
The scale also includes statements that are assessing general beliefs rather than
strategies. For example, “I invest my time in developing broad skills that can be
used in many areas”, “I stay active and involved in several different areas of life”,
and “many life goals become important to me because it is the right time for them.”
As a new explanation for the use of control strategies has been developed, the
Optimisation in Primary and Secondary Control Scale (Heckhausen et al., 1997) is
no longer appropriate. As such, the Primary and Secondary Control Scale (Heeps et
112
al., 2000) will be used in this study. This scale examines how people react to
situations where they risk losing control.
2.4.2.3
Job Satisfaction
Two scales of job satisfaction were administered; a facet scale and a global
scale. The facet scale is a revision of the Job Descriptive Index (Smith, Kendall &
Hulin, 1969). This scale, reported to be the most frequently used measure of job
satisfaction (Ironson, Smith, Brannick, Gibson & Paul, 1989), assesses five facets of
job satisfaction. The scale contains 72 items assessing nature of work, supervision,
pay, co-workers, and opportunity for promotion. This scale is reliable and
convergent and discriminant validity has been demonstrated (Gillet & Schwab, 1975;
Johnson, Smith & Tucker, 1982). This scale has been criticised however, as the
items have not been revised since the scale was developed.
In response to this criticism, Roznowski (1989) developed a revised scale by
calculating the discriminating power of the existing items, as well as some new
items. This revised scale had higher reliability with the alpha coefficient ranging
from 0.81 to 0.91. Although this revised scale may be more relevant to today’s
workforce, it still contains 72 items. To reduce the number of items for the current
study, a further revision was made. Only three items were selected to measure each
facet (refer to Appendix D). These items were selected as they had the highest
discrimination power.
This facet measure of job satisfaction is useful to diagnose the strengths and
weaknesses of organisations, however it cannot be summed to produce an overall
113
measure of job satisfaction (Ironson et al., 1989). Many researchers continue to use
facet scales to obtain an overall measure of job satisfaction (e.g., O’Driscoll &
Beehr, 2000; Schappe, 1998), however facet scales have been criticised as they may
exclude areas that are important to the respondent, or include areas that are
unimportant to the respondent. Therefore, in addition to the facet measure, a global
item of job satisfaction was also used.
The global measure of job satisfaction is a one-item measure. The item is
“taking into consideration all the things about your job, how satisfied are you with
it?” This item was rated on a 10-point scale ranging from 1 (completely dissatisfied)
to 10 (completely satisfied). This global scale requires the respondent to combine
their reactions to various aspects of the job into a single response. When answering
this question, the respondent may incorporate aspects of their job not included in the
facet scale. Although internal consistency cannot be established, a meta-analysis of
single-item measures of job satisfaction has demonstrated that single-item measures
correlate with other measures, such as the Job Diagnostic Survey (Hackman &
Oldham, 1976), the Job in General Scale (Ironson et al., 1989), and the Minnesota
Satisfaction Questionnaire (Weiss et al., 1967). On average, the correlation between
other scales and single item scales was r = 0.63 (Wanous, Reichers & Hudy, 1997).
2.4.2.4
Life Satisfaction
The subjective dimension of the Comprehensive Quality of Life Scale (Com-
QOL) developed by Cummins (1997) assesses satisfaction with seven domains of
life, including material well-being, health, productivity, intimacy, safety, community,
114
and emotional well-being (refer to Appendix E). An 11-point scale was utilised,
ranging from 0 (completely dissatisfied) to 10 (completely satisfied). The scale is
psychometrically sound, with internal reliability ranging from 0.5 to 0.8 (Cummins,
1997) and validity has been established using data from a review of the QOL
domains (Cummins, 1996).
2.4.2.5
Personality
The neuroticism and extroversion subscales of the NEO Five Factor
Inventory (short form; Costa & McCrae, 1992) was used to measure personality.
This scale contains 12 items to measure extroversion and 12 items to measure
neuroticism (refer to Appendix F). Six facet scales are measured in each factor.
Neuroticism is the sum of scales measuring anxiety, angry hostility, depression, selfconsciousness, impulsiveness, and vulnerability. Extroversion is the sum of warmth,
gregariousness, assertiveness, excitement-seeking, and positive emotions.
Convergent and discriminant validity of both of these factors has been established
(Costa & McCrae, 1992, Leong & Dollinger, 1991; Tinsley, 1994).
2.4.3
Procedure
Ethics approval was obtained from Deakin University, and consent was
obtained from the Heads of School for the academics, and the Human Resource
Managers for the supermarket workers. The recruitment procedure differed
depending on the group. The academics were sent the questionnaire via the internal
mail system. They returned the questionnaire by post. The supermarket workers
115
questionnaires were left in the staff room. As they were expected to complete the
questionnaire outside of work time, a $5 lottery ticket was given to each participant
to thank them for their time. The lottery tickets were given to the managers of the
store. The supermarket workers that returned the questionnaire collected their lottery
ticket from the service desk. At the conclusion of the study, the participating Heads
of School and the Human Resource Managers received a summary of the aggregated
results for all academics and supermarket workers.
116
2.5
2.5.1
Results
Data Screening and Checking of Assumptions
Procedures for data screening, and checking the procedural analytic
assumptions for all dependent variables followed those appropriate for group data.
The data set was initially examined for missing values, acquiescence, outliers,
normality and linearity. In regard to missing values, less than 4% of the values for
academics, and less than 5% of the values for supermarket workers were missing for
any one item. As there was no pattern to these missing values, they were replaced
with the group mean. Although this reduces the variance of the variables and
bivariate correlations (Tabachnick & Fidell, 1996), the replacement is a conservative
estimate.
Once the missing values were replaced, the data set was examined for
participants consistently reporting extreme scores (i.e., 1 or 10), in an attempt to
reduce the influence of acquiescence. One participant was omitted from the entire
sample for consistently reporting extreme scores on every scale. Other participants
reporting extreme scores on just one scale were excluded from that particular
analysis. Specifically, seven participants’ (all supermarket workers) responses were
deleted from the life satisfaction analyses, and nine participants’ responses were
omitted from the primary control analyses (three academics, six supermarket
workers).
117
Univariate outliers were identified on the facet job satisfaction scale, the life
satisfaction scale, and the control scales. Specifically, five cases of job satisfaction,
12 cases of life satisfaction, three cases of primary control, and nine cases of job
autonomy, lay outside three standard deviations from the mean. As these cases are
from the intended population, yet have more extreme values than the normal
distribution, they were recoded to three standard deviations from the mean.
On completion of the screening process, normality was assessed using the
skew/standard error <3, Kolmogorov-Smirnof values, frequency histograms, and
normal probability plots. In the academic group, overall life satisfaction (-3.91) and
the one-item measure of job satisfaction (-5.60) were mildly negatively skewed. In
the supermarket workers group, primary control was negatively skewed (-3.27). As
transformations are not recommended for data that are mildly and naturally skewed
(Tabachnick & Fidell, 1996), these data were not transformed. Finally,
homoscedasticity and linearity were assessed through bivariate scatterplots and these
appeared to demonstrate reasonable linear relationships between the variables.
118
2.5.2
Descriptive Statistics and Inter-Correlations
Table 3 contains the means and standard deviations for the major variables in
the study for each occupational group. In this table, and the tables thereafter, all
mean scores are converted to a percentage of scale maximum (%SM) which ranges
from 0-100. The formula is:
%SM = (mean score for the original domain – 1) x 100/ (number of scale points – 1).
Table 3 demonstrates that the academics report slightly higher job
satisfaction, primary control and job autonomy, and lower secondary control than the
supermarket workers. Table 4 displays the correlations among all of the major
variables for the academics and the supermarket workers. This demonstrates that job
autonomy and primary control are positively related to job satisfaction for both
occupational groups.
Table 3- Means and Standard Deviations of Major Variables for Academics
and Supermarket Workers
Variable
Job Satisfaction - 1 item
Job autonomy
Primary Control
Secondary Control
Life Satisfaction
Neuroticism
Extroversion
M
66.05
51.94
71.56
36.63
78.22
36.60
61.71
Academics
SD
21.09
14.63
11.95
15.64
10.96
15.78
12.41
Supermarket Workers
M
SD
59.71
25.69
34.50
20.24
67.06
18.62
46.74
19.77
73.30
15.97
39.25
17.88
65.28
13.99
119
Table 4- Inter-Correlations for the Academics and the Supermarket Workers
JS
JS
JA
PC
SC
LS
Neu
Ext
0.41**
0.44**
0.04
0.20*
-0.27
0.09
JA
PC
SC
LS
Neu
Ext
0.25*
0.38**
0.43**
0.14
0.03
0.07
0.07
0.06
-0.02
-0.07
-0.23*
-0.16
-0.03
0.04
-0.50**
0.17
0.17
0.42**
0.02
0.29**
-0.30**
0.43**
0.08
0.15
-0.19*
0.09
-0.02
0.11
-0.14
0.25**
-0.01
0.05
-0.12
-0.59**
0.25**
-0.17*
* p<0.05 , ** p>0.01; Correlations for supermarket workers are bolded.
JS = Job satisfaction; JA = Job autonomy; PC = Primary control; SC = Secondary
control; LS = Life satisfaction; Neu = Neuroticism; Ext = Extroversion
120
2.5.3
Factor Analyses
Prior to testing the hypotheses, factor analyses were conducted on the revised
scales of job satisfaction, primary and secondary control, and job autonomy.
2.5.4
Factor Analysis of the Job Descriptive Index
To ensure the 15 job satisfaction items represented each of the five facets, a
principle components analysis with direct oblimin rotation was conducted. The
assumptions of sample size, normality, outliers, linearity, and the factorability of the
correlation matrix were initially examined.
Factor analysis requires a minimum of five subjects per variable (5 x 15 = 75)
(Coakes & Steed, 1999), hence the sample size of 199 was adequate. Some of the
job satisfaction items were not normally distributed however the solution is still
worthwhile if normality is not met (Tabachnick & Fidell, 1996). Five outlying cases
were recoded to three standard deviations from the mean. Reasonably linear
relationships existed among the variables. In regard to the factorability of the
correlation matrix, all of the correlations exceeded 0.30. The measures of sampling
adequacy (MSA) were > 0.50. Bartlett’s test of Sphericity was large and significant
(1574.89), and Kaiser-Meyer-Olkin (KMO) Measure of Sampling Adequacy
exceeded 0.60.
A principal components analysis, with direct oblimin rotation, yielded four
eigenvalues over 1. With this four factor solution 22% of the nonredundant residuals
had absolute values > 0.05, suggesting the presence of another factor. When a
121
principal components analysis with direct oblimin rotation was conducted with five
factors, only 7% of the nonredundant residuals had absolute values > 0.05. As such,
a five-factor model was deemed to be most appropriate. The loadings of the items on
each of the five factors are presented in Table 5. These loadings demonstrate that
Factor 1 refers to promotion, Factor 2 to nature of work, Factor 3 to pay, Factor 4 to
supervisors, and Factor 5 to co-workers.
Table 5- Factor Analysis of Job Satisfaction Scale
Item
There is a good chance for promotion in
my job.
There is a fairly good chance for
promotion in my job.
There are good opportunities for
advancement in my job.
My work is dull.
My work is boring.
My work is interesting.
My pay is bad.
I am well-paid.
My pay is unfair.
My supervisors know how to supervise.
My supervisors are bad.
My supervisors are annoying.
My co-workers are stupid.
My co-workers are responsible.
My co-workers are a waste of time.
Eigenvalues
% of variance
Cumulative %
Cronbach's Alpha (total scale)
F1
0.94
F2
F3
F4
F5
0.94
0.90
0.97
0.90
0.81
0.91
0.82
0.81
0.81
0.79
0.78
4.32
28.77
28.77
0.82
Items with loadings less than 0.30 are not shown.
2.78
18.56
47.32
1.87
12.46
59.79
1.65
10.99
70.77
0.79
0.73
0.79
0.94
6.27
77.04
122
2.5.5
Factor Analysis of the Primary and Secondary Control Scale
To ensure that the primary and secondary control items loaded on two
separate factors, a factor analysis was conducted on the primary and secondary
control scale. The sample size was adequate (N = 190), and the secondary control
items were normally distributed for both groups. The primary control items were
mildly negatively skewed for both the academics (pc4 = -4.41, pc5 = -3.17) and the
supermarket workers (pc1 = -3.5, pc2 = -3.77, pc4 = -4.26, pc5 = -3.17; refer to
Table 7 for items). However, as factor analysis is robust to violations of normality,
the resulting solution was still deemed to be worthwhile (Tabachnick & Fidell,
1996). Linearity among the variables as assessed through scatterplots was
reasonable. The correlation matrix was factorable with all correlations exceeding
0.30. The measures of sampling adequacy exceeded 0.50 for all variables. Bartlett’s
test of sphericity was large and significant (1429.34), and Kaiser-Meyer-Olkin
(KMO) Measure of Sampling Adequacy exceeded 0.60.
A principal components analysis with direct oblimin rotation yielded 5
factors. The total variance explained by these five factors is demonstrated in Table
6.
123
Table 6- Total Variance Explained by a Five-Factor Solution
Initial Eigenvalues
Factor
1
2
3
4
5
Total
5.125
2.846
1.615
1.476
1.121
% of variance
26.975
14.980
8.498
7.766
5.901
Cumulative %
26.971
41.956
50.454
64.120
69.269
This five-factor solution demonstrated that four of the five primary control items
loaded on one factor, and that the rest of the secondary control items loaded on the
other four factors. However, as there was no clear pattern in the other four factors, a
four-factor and three-factor solution were also requested. In both of these analyses
however, many of the items loaded on more than one factor.
To investigate the hypothesised two-factor solution, a principal components
analysis with direct oblimin rotation was requested. More than two factors are
present however, as 67% of the nonredundant residuals had absolute values > 0.05.
As demonstrated in Table 7, all of the primary control items loaded on Factor 2.
Seven of the 14 secondary control items loaded on Factor 1, and the remaining seven
secondary control items loaded only on Factor 2 or on both factors. As such, in
subsequent analyses, the scale will include all five primary control items and only the
seven non-complex secondary control items. With only these items, a factor analysis
reveals that the primary control factor accounts for 19.84% of the variance, and the
secondary control factor accounts for 28.30% of the variance.
124
Table 7- Factor Analysis of Primary and Secondary Control Scale
No.
pc1
pc2
pc3
pc4
pc5
sc1*
sc2
sc3
sc4
sc5
sc6
sc7
sc8
sc9
sc10
sc11
sc12
sc13
sc14
Item
When I have a goal at work that is difficult to reach, I
think about different ways to achieve it.
When I want something at work to change, I think I
can make it happen.
When a work task really matters to me, I think about
it a lot.
When I really want to reach a goal at work, I believe I
can achieve it.
When faced with a difficult work situation, I believe I
can overcome it.
I can see that something good will come of it.
I remember you can't always get what you want.
I know things will work out okay in the end.
I remember I am better off than many other people.
I remember I have already accomplished a lot in life.
I remember the success of my family or friends.
I think nice thoughts to take my mind of it.
I remind myself the situation will change if I am just
patient.
I tell myself it doesn’t matter.
I think about my success in other areas.
I don’t feel disappointed because I knew it might
happen.
I can see it was not my fault.
I ignore it by thinking about other things.
I realise I didn’t need to control it anyway.
Eigenvalues
% of variance
Cumulative variance
Cronbach's Alpha (for revised scale)
F1
F2
0.70
0.39
0.51
0.71
0.51
0.32
0.32
0.44
0.36
0.53
0.55
0.74
0.65
0.65
0.45
0.67
0.51
0.39
0.34
0.38
0.72
0.74
0.61
0.40
0.69
0.61
5.13
14.98
14.98
0.82
2.85
26.98
41.96
0.70
Items with loadings less than 0.30 are not shown.
pc= primary control; sc=secondary control; Bolded items are included in the scale.
*All secondary control items preceded by “When something bad happens that I
cannot change”
125
2.5.6
Factor Analysis of the Job Autonomy Scale
To ensure the items on the job autonomy scale were measuring a single
construct, a factor analysis was conducted. The assumption of normality was
violated with items 1, 2, 4, and 10 being mildly negatively skewed for academics.
Items 7, 8, 9, and 11 were mildly positively skewed for the supermarket workers. As
before, these variables were not transformed. Nine univariate outliers were recoded
to three standard deviations from the mean. All correlations exceeded 0.30 and all of
the variable MSA exceeded 0.50. Barlett's test of sphericity was significant
(1077.97), and Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy exceeded
0.60.
A principal components analysis with direct oblimin rotation demonstrated
that the job autonomy items loaded on three factors. There was no meaningful
pattern within these factors however, and all but three items loaded on more than one
factor. In an attempt to find a pattern among the items, a two-factor principle
components analysis with direct oblimin rotation was conducted. This analysis,
presented in Table 8, demonstrated that six items loaded only on the first factor, one
item loaded only on the second factor, and the remaining seven loaded on both
factors. Factor 1 contained items that were directly related to the nature of the work
(i.e., tasks, order of work, working pace), whereas the items that loaded on Factor 2
related to organisational structure (i.e., pay, evaluation). Although this factor
analysis demonstrates that two factors emerge, all items will be retained in this scale
126
as the overall measure of job autonomy should be based on the nature of the work
and the organisational structure.
Table 8- Factor Analysis of Job Autonomy Scale
No
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Item
In my job I can choose among a variety of tasks or
projects to do.
In my job I can choose the order in which I do my work.
In my job I can choose how quickly I do my work.
In my job I can choose how I schedule my rest breaks.
In my job I can choose the physical conditions of my
workstation.
In my job I can choose when I interact with others.
In my job I can choose the amount I earn.
In my job I can choose the number of times I am
interrupted at work.
In my job I can choose how my work is evaluated.
In my job I can choose the quality of my work.
In my job I can choose the policies and procedures in my
work unit.
In my job I can choose among a variety of methods to
complete my work.
In my job I can choose how much work I get done.
In general, how much are you able to influence work and
work-related matters.
Eigenvalues
% of variance
Cumulative variance
Cronbach's Alpha (revised scale)
Items with loadings less than 0.30 are not shown.
F1
0.76
F2
0.33
0.84
0.71
0.71
0.56
0.46
0.74
0.42
0.66
0.58
0.76
0.56
0.54
0.59
0.82
0.36
0.67
0.64
0.41
0.50
0.58
41.37
41.37
0.86
1.27
9.04
50.41
127
2.6
Hypothesis Testing
In order to test the proposed model of job satisfaction, multivariate analyses
of variance were conducted to investigate how the academics and the supermarket
workers differed in their levels of job autonomy, control strategies, and job
satisfaction. Multiple regression analyses were also conducted to predict job
satisfaction from job autonomy, control strategies, personality, and life satisfaction.
In order to test these hypotheses, 22 p-values must be computed, and as such,
familywise error rate must be considered. Familywise error rate is the probability of
making at least one Type I error in a set of analyses (Keppel, 1991). Increasing the
number of statistical tests can potentially increase the familywise error. The formula
for familywise error is FW= (alpha level) x (number of comparisons). In this study,
the familywise error rate is (0.05) x (22) = 1.1. One solution to reduce this
familywise error rate is adjust the alpha level using the Bonferroni test (Keppel,
1991). The desired alpha level (0.05) is divided by the number of tests (22), yielding
a recommended alpha level of 0.002. Although reducing the alpha level decreases
the probability of Type I errors, it also increases the probability of Type II errors
(Keppel, 1991). The solution therefore is to strike a balance between the two errors.
Thus, the alpha level will be reduced to 0.01.
2.6.1
Hypothesis One- Assumption Testing
In order to test the first part of hypothesis one, proposing that job autonomy
and job satisfaction are positively related, the correlation coefficients for each
128
occupational group were examined. Consistently, job autonomy was positively
related to job satisfaction for both the academics (r = 0.41) and the supermarket
workers (r = 0.25).
In order to test the second part of hypothesis one, proposing that the
academics would report higher job autonomy than the supermarket workers, an
analysis of variance was employed. The assumption of univariate homogeneity of
variance, assessed using Levene’s test was not met, F (1, 197) = 12.77, p = 0.00,
however as this assumption is of little concern when the sample sizes are similar
(Tabachnick & Fidell, 1997), the analysis proceeded with caution using an alpha
level of 0.01. The univariate test of significance demonstrated that, consistent with
hypothesis one, the academics (M = 51.94, SD = 14.63) reported significantly higher
job autonomy than the supermarket workers, (M = 34.50, SD = 20.24),
F (1, 197) = 49.51, p = 0.00.
2.6.2
Hypothesis Two- Occupational Differences in the Use of the Control
Strategies
In order to examine hypothesis two proposing that the academics will report
less secondary control and more primary control than the supermarket workers, a
multivariate analysis of variance was performed. The assumptions of normality,
linearity, multicollinearity, and homogeneity of variance-covariance were examined
for the variables.
All of the variables were normally distributed, and reasonably linear
relationships were evident. Two univariate outliers were recoded to three standard
129
deviations from the mean and there were no multivariate outliers. There was no
evidence of multicollinearity, as the determinant of the within-cell correlation
was > 0.0001 (i.e., 0.798).
The assumption of univariate homogeneity of variance, as assessed by
Levene’s test, was met for secondary control, F (1, 188) = 2.84, p > 0.05. Equality of
error variance was not found however for primary control, F (1, 188) = 17.07,
p < 0.05. Levene’s test is sensitive to non-normality however, and this can lead to
overly conservative rejection (Tabachnick & Fidell, 1997). As such, the analyses
will proceed with caution using an alpha level of 0.01. The assumption of
multivariate homogeneity of variance-covariance, assessed through Box’s M test was
also violated. Box’s M test is a notoriously sensitive test of homogeneity of
variance-covariance, and it is recommended that if the test is violated, the
multivariate tests be examined by Pillai’s criterion rather than Wilk’s lamba.
The multivariate test of significance, using Pillai’s criterion, demonstrated
that occupational differences existed, F (2, 187) = 10.03, p = 0.00. As demonstrated
in Table 9, the supermarket workers reported significantly higher secondary control
than the academics, F (1, 188) = 15.50, p = 0.00. The two groups did not report
significantly different levels of primary control, F (1, 188) = 3.99, p = 0.04. It must
be noted however that the difference in primary control was significant at 0.05, but
not at the more stringent alpha level of 0.01. Hence, only partial support was
provided for the second hypothesis.
130
Table 9- Means and Standard Deviations of Control Measures for Academics
and Supermarket Workers
Variable
Primary Control
Secondary Control
M
71.56
36.63
Academic
SD
11.95
15.64
Supermarket
M
SD
67.06
18.62
46.74
19.77
Bolded constructs demonstrate significant occupational differences.
2.6.2.1
Summary
Multivariate analyses of variance demonstrated that the academics report
higher job autonomy, and lower secondary control than the supermarket workers.
The two groups did not report significantly different levels of primary control.
2.6.3
Hypothesis Three- Examining how Job Autonomy Relates to the
Control Strategies
To examine hypothesis three, proposing that job autonomy will be positively
related to primary control and negatively related to secondary control, the correlation
coefficients were examined. It was necessary to examine whether job autonomy
influences the control strategies using the measured level of job autonomy because
there was some variability in the level of job autonomy reported within occupational
groups. A median split was conducted on job autonomy and the employees were
split into two groups. The majority of academics were in the high job autonomy
group (66%), however 34% were in the low job autonomy group. Similarly, 70% of
131
the supermarket workers were in the low job autonomy group, however 30% were in
the high job autonomy group.
Job autonomy was positively related to primary control (r = 0.46), but not
related to secondary control (r = - 0.18). These results provide partial support for
hypothesis three, suggesting that job autonomy influences primary, but not secondary
control.
2.6.4
Hypothesis Four- Examining how Job Autonomy Influences the
Adaptiveness of the Control Strategies
To examine hypothesis four, proposing that i) primary control will be more
positively related to job satisfaction than secondary control for the academics, and ii)
secondary control will be more positively related to job satisfaction than primary
control for the supermarket workers, a standard multiple regression analysis was
conducted on each occupational group. The assumptions of normality, linearity and
homoscedasticity of residuals were assessed through examination of the residual
scatterplots. These assumptions were met, and there was no evidence of
multicollinearity.
As demonstrated in Table 10, R was significantly different from zero for both
the academics, R = 0.44, F (2, 102) = 12.53, p = 0.00, and the supermarket workers,
R = 0.31, F (2, 82) = 4.37, p = 0.01. Primary control predicted job satisfaction for
both groups, accounting for 20% of the variance in job satisfaction for the academics
and almost 10% for the supermarket workers. Secondary control did not predict job
satisfaction for either group. Hence, consistent with hypothesis four, primary control
132
was more positively related to job satisfaction than secondary control for the
academics. Inconsistently however, secondary control was not related to job
satisfaction for the supermarket workers.
Table 10- Multiple Regression of Primary and Secondary Control on Job
Satisfaction for Academics and Supermarket Workers
Group
Acad
Variable
JS
PC
SC
0.44
0.04
PC
B

sr2 (unique)
0.44
0.05
19.62**
-0.02
0.78
0.006
R =0.44**
R2=0.20
Adj R2=0.18
0.43
-0.05
0.31
-0.04
9.61**
R =0.31*
R2=0.10
Adj R2=0.07
Super
PC
SC
0.31
-0.02
0.08
p<0.01; Acad – Academics; Super – Supermarket workers; JS – Job satisfaction;
PC – Primary control; SC – Secondary control
**
For academics, R2 composed of shared variance (1.9%) and unique variance (98.1%)
For supermarket workers, R2 composed of shared variance (3.9%) and unique
variance (96.1%).
2.6.5
Hypothesis Five- Does Job Autonomy Moderate the Relationship
Between the Control Strategies and Job Satisfaction?
In order to examine hypothesis five, proposing that the relationship between
the control strategies and job satisfaction is moderated by job autonomy, two
hierarchical multiple regression analyses were conducted. Job autonomy is proposed
to be a moderator, which means that it affects the direction and/or the strength of the
relationship between the control strategies and job satisfaction. Specifically, the
133
relationship between primary control and job satisfaction is expected to be positive
when job autonomy is high and negative when job autonomy is low. Furthermore,
the relationship between secondary control and job satisfaction is expected to be
positive when job autonomy is low, and negative when job autonomy is high. These
expected relationships are demonstrated below in Figure 3.
Figure 3- Expected Moderated Effect of Job Autonomy on a) Primary Control
and Job Satisfaction and b) Secondary Control and Job Satisfaction
High Job Autonomy
a)
JS
Low Job Autonomy
Primary Control
b)
Low Job Autonomy
JS
High Job Autonomy
Secondary Control
A moderation effect can be tested in a number of ways depending on whether
the variables are continuous or discrete (Baron & Kenny, 1986). In this hypothesis,
134
the moderator variable and the independent variable are both continuous. When both
variables are continuous, and when the effect of the independent variable on the
dependent variable varies linearly with respect to the moderator, a hierarchical
multiple regression analysis is conducted to test the presumed relationship (Baron &
Kenny, 1986). As demonstrated in Figure 4, the dependent variable is regressed on
the independent variable, the moderator variable, and the product of the independent
variable and the moderator (Baron & Kenny, 1986). Moderator effects are
demonstrated if the interaction term is significant when the independent variable and
the moderator variables are controlled (Baron & Kenny, 1986).
135
Figure 4- Job autonomy Moderates the Relationship between a) Primary
control and b) Secondary Control, and Job Satisfaction.
Order of Variable Entry
a)
Step 1
Primary control
Step 2
Job autonomy
Step 3
Primary control x Job
autonomy
1.1.1.1
J
ob
b)
Step 1
Secondary control
Step 2
Job autonomy
Step 3
Job
Satisfaction
Satisf
Jobactio
Satisfaction
n
Secondary control x
Job autonomy
In order to test the moderating effect of job autonomy on primary control and
secondary control, two hierarchical multiple regression analyses were conducted on
the combined sample. By using the combined sample, there was more range in the
levels of job autonomy. In these analyses the control strategies were entered first,
then job autonomy, and then the interaction term. For primary control, R was
significantly different from zero after the first step (i.e., primary control), R = 0.37,
F (1, 188) = 30.34, p = 0.00, and the second step (i.e., job autonomy), R = 0.40,
Finc (1, 187) = 4.89, p = 0.03. However, the addition of the interaction term was not
significant, R =0.40, Finc (1,186)= 0.33, p = 0.57.
136
For secondary control, R was not significantly different from zero after the
first step, R = 0.01, F (1, 197) = 0.04, p = 0.85. After job autonomy was entered, the
value of R increased, R = 0.33, F (1, 196) = 24.01, p = 0.00. There was no further
increase however when the interaction term was entered in step three, R = 0.34,
F (1, 195) = 1.58, p = 0.21. These analyses, displayed in Table 11, demonstrate that
inconsistent with hypothesis five, job autonomy did not moderate the relationship
between the control strategies and job satisfaction.
137
Table 11- Moderating Role of Primary and Secondary Control on the
Relationship Between Job Autonomy and Job Satisfaction
Step
IV
DV
B

sr2 (unique)
1
2
Primary control
Primary control
Job Autonomy
Primary control
Job autonomy
Primary control x job
autonomy
JS
0.57
0.46
0.20
0.55
0.41
-0.003
0.37
0.30
0.17
0.36
0.34
-0.21
13.91**
6.96**
2.19*
3.46**
R =0.40
R2=0.16
AdjR2=0.15
0.02
0.02
0.40
0.26
0.63
-0.005
0.01
0.02
0.33
0.21
0.52
-0.28
R =0.34
R2=0.12
3
1
2
3
Secondary control
Secondary control
Job Autonomy
Secondary control
Job autonomy
Secondary control x job
autonomy
JS
10.89**
4.54*
AdjR2=0.10
p<0.01, *p<0.05; JS – Job satisfaction
**
2.6.6
Hypothesis Six- Do the Control Strategies Mediate the Relationship
Between Job Autonomy and Job Satisfaction?
Hypothesis six proposes that the relationship between job autonomy and job
satisfaction is mediated by the control strategies. In this hypothesis, the control
strategies are acting as mediators because they are explaining why job autonomy is
related to job satisfaction. That is, employees with high job autonomy are expected
to rely on more primary control and less secondary control than employees with low
job autonomy. As primary control strategies are more positively related to job
138
satisfaction than secondary control strategies, employees with higher job autonomy
report higher job satisfaction.
It must be noted that although secondary control strategies are less positively
related to job satisfaction than primary control, it is proposed that for workers with
low job autonomy, secondary control strategies are superior to primary control
strategies. If these workers use primary control, they are expected to experience
primary control failure.
According to Baron and Kenny (1986), in order to establish mediation, three
standard regression analyses must demonstrate that: a) job autonomy predicts
primary and secondary control; b) primary and secondary control and job autonomy
together predict job satisfaction; and c) job autonomy predicts job satisfaction. For a
mediation effect to be significant, all three regression equations must be significant,
and the effect of the independent variable on the dependent variable must be less in
b) than in c) (Baron & Kenny, 1986). This mediation analysis is demonstrated in
Figure 5.
Figure 5- Mediating Role of Control Strategies on the Relationship Between Job
Autonomy and Job Satisfaction
a
Primary/Secondary Control
b
Job autonomy
c
Job Satisfaction
139
This method will not be used however as there is an easier way to test the
mediating role of the control strategies. Rather than conducting three regression
analyses, only one hierarchical regression analysis is needed (M. Stokes, personal
communication, August 16, 2002). In this analysis, primary and secondary control
strategies are entered first, followed by job autonomy. It is expected that once
primary and secondary control strategies have been entered, there would be no
relationship between job autonomy and job satisfaction. As such, primary and
secondary control would explain the relationship between job autonomy and job
satisfaction.
The assumptions of normality, linearity and homoscedasticity of residuals
were met, and there was no evidence of multicollinearity. As demonstrated in Table
12, R was significantly different from zero after primary and secondary control were
entered, R = 0.37, F (2, 187) = 15.23, p = 0.00. Primary control accounted for 13%
of the variance in job satisfaction, and secondary control was not significant. R did
not significantly increase after job autonomy was added to the equation, R = 0.40,
Finc (3, 186) = 11.92, p = 0.03. Even if the less stringent alpha level of 0.05 was
used, job autonomy only accounts for 2% of the unique variance in job satisfaction.
As such, it appears that when primary and secondary control are entered first, there is
no relationship between job autonomy and job satisfaction. This suggests that partial
support is provided for hypothesis six as primary control, but not secondary control,
mediates the relationship between job autonomy and job satisfaction.
140
Table 12 -Hierarchical Multiple Regression Testing the Mediating Role of the
Control Strategies
Step
1
2
IV
Primary Control
Secondary Control
Primary Control
Secondary Control
Job Autonomy
DV
JS
JS
B
0.57
-0.04

0.37
-0.03
sr2 (unique)
13.91**
R =0.37**
R2=0.14
AdjR2=0.13
0.46
-0.02
0.20
0.30
-0.02
0.17
6.96**
R =0.40**
R2=0.16
AdjR2=0.15
2.01*
**p>0.01 *p>0.05; JS - Job satisfaction
It must be noted that although the results demonstrate that primary control is
a partial mediator of the relationship between job autonomy and job satisfaction, the
use of multiple regression to estimate a mediational model is based on the
assumption that there is no measurement error in the mediator. This assumption is
particularly concerning as the mediator is likely to be measured with error. The
presence of such error tends to produce “an underestimation of the effects of the
mediator, and an overestimation of the effects of the independent variable on the
dependent variable” (Baron & Kenny, 1986, p. 1177).
One statistical method that models the measurement error is structural
equation modeling. Structural equation modeling is based on the analysis of sample
variances and covariances rather than individual cases. This approach is particularly
useful for latent variables, which are hypothetical constructs that cannot be directly
measured, such as job satisfaction.
141
Although structural equation modelling has some advantages over multiple
regression, it will not be used in this thesis for a number of reasons. First, unlike
hierarchical multiple regression, structural equation modelling is a confirmatory
technique. The current study, although grounded in theory, is exploratory,
attempting to combine the propositions of the job demand-control model (Karasek &
Theorell, 1990) with propositions of the life span theory of control (Heckhausen &
Schulz, 1995). As the theory is exploratory, there are a variety of different models
that can be examined. If numerous modifications of a model were tested, the
analysis would be exploratory, and there would be an increased risk of Type I errors
(Ullman, 1996). As this thesis is attempting to develop and explore the proposed
model of job satisfaction and search for unexpected relationships, structural equation
modelling may be problematic. Once the model is more established however,
structural equation modelling may be required.
A further problem with using structural equation modelling is that it requires
large sample sizes. The issue of an adequate sample size continues to be debated,
however Boomsma (1983) suggested that as a general rule, samples of 200 are
required to give parameter estimates with any degree of confidence. As the
relationship between the variables is expected to be different for academics and
supermarket workers, two models would need to be conducted, thus there would
need to be 200 in each occupational group.
A sample size of 200 is problematic due to time constraints, but also because
of the particular workers selected for this study. The data collection process
undertaken in study one demonstrated that workplaces, particularly those employing
142
low autonomy workers, such as call centres, factories, or supermarkets, were
reluctant to become involved in any research. The employers refused to participate
in the surveys for a variety of reasons. Some mentioned that the majority of their
employees were from Non-English speaking backgrounds and as such would be
unable to understand the survey. Others admitted that work motivation was very
low, and as such, the response rate would be poor. Still others were concerned that
the employees would expect changes to be made to the workplace on the basis of
their responses. These employers’ reactions indicate that is difficult to obtain a
sample size of 400.
2.6.6.1
Summary
In summary, it appears that primary control mediates the relationship between
job autonomy and job satisfaction. This finding was based on multiple regression
analyses however, which assumes that there is no measurement error in the mediator.
Although this measurement error can be accounted for in structural equation
modeling, it is concluded that such a method is not appropriate whilst the proposed
model of job satisfaction is in an exploratory stage.
2.6.7
Hypothesis Seven- Occupational Differences in Job and Life
Satisfaction
Hypothesis seven proposes that the academics will report higher job
satisfaction and higher life satisfaction than the supermarket workers. A univariate
analysis of variance was conducted on the global one-item measure of job
143
satisfaction. The assumptions of normality, linearity, and homogeneity of variance
were examined. Job satisfaction was normally distributed for the supermarket
workers, however it was negatively skewed for the academics (-5.60). The
assumption of homogeneity of variance was violated, F (1, 197) = 6.06, p < 0.05, and
as such, the analysis will proceed with caution using an alpha level of 0.01.
Inconsistent with hypothesis seven, there were no occupational differences in the
one-item measure of job satisfaction, F (1, 197) = 3.66, p = 0.04. The levels of job
satisfaction reported by the two groups are provided in Table 13.
To examine whether the two groups differed on the facets of job satisfaction,
a multivariate analysis of variance was conducted on the five facets of job
satisfaction, namely nature of work, co-workers, pay, supervisors, and opportunities
for promotion. Normality was assessed using skew/standard error < 3,
Kolmogorov-Smirnof values, and normal probability plots. Although the nature of
work facet (-3.49) and the co-workers facet (-4.87) were negatively skewed for the
academic group, the remainder of the variables were normally distributed for both
groups. Five univariate outliers were recoded to three standard deviations from the
mean, and no multivariate outliers were identified. Examination of bivariate
scatterplots, and correlations revealed reasonably linear relationships. There was no
evidence of multicollinearity as the determinant of the within-cell correlation
was >0.0001. Univariate homogeneity of variance, assessed by Levene’s test,
demonstrated that equality of error variance was evident for the supervision facet,
F (1, 197) = 0.40, p > 0.05. Equality of error variance was not found however for
pay, F (1, 197) = 5.53, p < 0.05, nature of work, F (1, 197) = 27.14, p < 0.05,
144
co-workers, F (1, 197) = 5.82, p < 0.05, and promotion, F (1, 197) = 4.20, p < 0.05.
As such, the univariate tests will be examined with caution. The assumption of
multivariate homogeneity of variance-covariance, as assessed through Box’s M test,
was also violated, F (15, 146782) = 4.90, p < 0.001.
The multivariate tests were examined using Pillai’s criterion. The job facets
were affected by occupation, F (5, 193) = 35.10, p = 0.00. As demonstrated in Table
13, academics reported significantly higher satisfaction with nature of work,
F (1, 197)= 95.59, p = 0.00, and co-workers, F (1, 197) = 32.51, p = 0.00, than
supermarket workers. However, the supermarket workers reported higher
opportunity for promotion than the academics, F (1, 197) = 9.21, p = 0.00.
Table 13- Means and Standard Deviations of Job Satisfaction Scale for
Academics and Supermarket Workers
Variable
Nature
Co-Workers
Pay
Supervisors
Promotion
One-item measure
M
85.97
82.30
51.20
60.29
38.31
66.05
Academic
SD
13.74
15.75
27.55
24.47
27.40
21.09
M
56.98
67.85
54.17
67.03
51.12
59.71
Supermarket
SD
26.95
20.01
23.56
24.49
32.18
25.69
Bolded variables indicate significant occupational differences
It was expected that levels of job satisfaction would be related to levels of life
satisfaction, and that the academics reporting higher job satisfaction than the
supermarket workers would also report higher life satisfaction. To examine this
hypothesis, a univariate analysis of variance was conducted to examine overall life
145
satisfaction, and a multivariate analysis of variance was conducted to examine on
which domains the groups differed.
To compare their overall life satisfaction, a univariate analysis of variance
was conducted. Life satisfaction was normally distributed for the supermarket
workers however it was mildly negatively skewed for the academics (-3.94). The
assumption of homogeneity of variance, as assessed through Levene’s test of
equality of error variance was violated, F (1, 190) = 15.24, p < 0.05, and as such, the
analysis proceeded with caution using an alpha level of 0.01. Consistent with
hypothesis seven, the academics reported higher life satisfaction than the
supermarket workers, F (1, 190) = 6.38, p = 0.01.
A multivariate analysis of variance was conducted on the seven domains of
life satisfaction to examine where these differences lay. The assumptions of
normality, linearity, multicollinearity and homogeneity of variance-covariance were
examined for the seven domains. The emotional well-being domain was mildly
negatively skewed for the academics, and the intimacy domain was mildly negatively
skewed for the supermarket workers. 12 univariate outliers were recoded to three
standard deviations from the mean. Four multivariate outliers were examined and
recoded to the next less extreme score. The assumption of linearity, examined
through bivariate scatterplots, was met. Equality of error variance was demonstrated
only for satisfaction with health, F (1, 190) = 2.28, p > 0.05, and as such, the analysis
will proceed with caution. As the assumption of multivariate homogeneity of
variance, examined through Box’s M test was also violated, F (28, 110823) = 2.54,
p < 0.001, Pillai’s criterion was used to examine the multivariate test.
146
Pillai’s criterion was significant, F (7, 184) = 1039.56, p < 0.01. Academics
reported significantly higher productivity satisfaction, F (1, 190) = 7.63, p = 0.006,
and safety satisfaction, F (1, 190) = 12.62, p = 0.00, than supermarket workers. The
means and standard deviations for the satisfaction domains are provided in Table 14.
Table 14- Means and Standard Deviations of Life Satisfaction for Academics
and Supermarket Workers
Occupation
Material Satisfaction
Health Satisfaction
Productivity Satisfaction
Intimacy Satisfaction
Community satisfaction
Safety Satisfaction
Emotional Satisfaction
OVERALL SATISFACTION
M
80.56
75.03
76.00
81.08
75.71
83.52
78.86
78.22
Academic
SD
13.18
20.07
13.91
18.44
16.22
15.13
15.27
10.96
Supermarket
M
SD
78.56
19.43
69.16
23.85
69.05
21.67
76.63
24.21
70.07
20.13
74.46
20.85
76.63
22.10
73.30
15.97
Bolded variables indicates occupational differences
2.6.8
Hypothesis Eight- Predictors of Job Satisfaction
In order to evaluate hypothesis eight, which proposes that primary control,
secondary control, job autonomy, personality and life satisfaction predict job
satisfaction, a multiple regression analysis was conducted on both occupational
groups. The correlations among the variables are displayed in Table 4.
For both groups, the assumptions of normality, linearity and homoscedasticity
of residuals were met, and there was no evidence of multicollinearity. R was
significantly different from zero after all the variables had been added for both the
academics, R = 0.54, F (6,98) = 6.69, p = 0.00, and the supermarket workers,
147
R =0.46, F (6, 73) = 3.25, p = 0.00. For the academics, the unique predictors of job
satisfaction were job autonomy and primary control. As demonstrated in Table 15,
primary control and job autonomy accounted for 4% and 8% of the variance in job
satisfaction respectively. It must be noted however that job autonomy was not
significant at the more stringent alpha level of 0.01. For the supermarket workers,
there was only one unique predictor of job satisfaction, namely primary control.
Primary control accounted for 8% of the variance in job satisfaction.
These results suggest that hypothesis eight is partially supported as primary
control and job autonomy predicted job satisfaction. However, secondary control,
personality and life satisfaction were not unique predictors of job satisfaction.
Furthermore, even when all the variables were included in the equation, R2 was small
(R2 = 0.29, R2 = 0.21).
148
Table 15- Multiple Regression of Job Autonomy, Control Strategies,
Personality, and Life Satisfaction for Academics and Supermarket Workers
Group
IV
DV
B

Acad
Job autonomy
Primary Control
Secondary Control
Neuroticism
Extroversion
Life Satisfaction
JS
0.34
0.58
0.03
-2.30
-1.02
0.10
0.24
0.33
0.03
-0.16
-0.05
0.05
R =0.54**
R2=0.29
0.08
0.46
0.15
-3.63
-1.14
-0.04
0.06
0.36
0.12
-0.25
-0.06
-0.03
R =0.46**
R2=0.21
Super
Job autonomy
Primary Control
Secondary Control
Neuroticism
Extroversion
Life Satisfaction
JS
sr2
(unique)
4.41*
8.12**
Adj2=0.25
8.35**
Adj2=0.15
** p<0.01, * p<0.05; Acad – Academics; Super- Supermarket workers; JS – Job
satisfaction
2.6.8.1
Summary
The academics reported significantly higher life satisfaction than the
supermarket workers, but similar levels of job satisfaction. The major predictors of
job satisfaction were job autonomy and primary control strategies.
2.6.9
Conclusion
The major propositions of this study were that job autonomy influences the
use of the control strategies and the relationship between the control strategies and
149
job satisfaction. As hypothesised, the academics reported higher job autonomy,
higher life satisfaction and lower secondary control than the supermarket workers.
Inconsistent with the hypotheses, the two groups reported similar levels of primary
control and job satisfaction. However, job autonomy was positively correlated with
primary control and not correlated with secondary control.
In regard to the proposal that job autonomy influences the relationship
between the control strategies and job satisfaction, the findings were less supportive.
Primary control was the most adaptive strategy for both groups, and secondary
control was not related to job satisfaction for either group. The implications of these
findings will now be discussed.
150
2.7
Discussion
This study tested a new explanation for the relationship between job
autonomy and job satisfaction, namely that job autonomy influences the use and
adaptiveness of primary and secondary control strategies. In regard to the use, the
findings demonstrated that the supermarket workers reported more secondary control
than the academics, but that only primary control was related to job autonomy. In
regard to the adaptiveness, primary control was the most adaptive strategy for
academics and supermarket workers. These findings are discussed in terms of the
life span theory of control (Heckhausen & Schulz, 1995) and the discrimination
model (Thompson et al., 1998). Before these propositions are explained in detail, the
basic assumptions of the study will be examined.
2.7.1
Assumption Testing
The basic assumptions of the study were that job autonomy was positively
related to job satisfaction, and that the study used two occupational groups that
differed in their level of job autonomy. Consistently, job autonomy was positively
related to job satisfaction for the academics and the supermarket workers
(r = 0.41, r = 0.25, respectively). These correlations are slightly lower than those
reported in other studies using Ganster’s (1989, cited in Dwyer & Ganster, 1991)
scale. For example, Munro, Rodwell and Harding (1998) demonstrated that the
correlation between job autonomy and job satisfaction was r = 0.69, whilst Fox et al.,
151
(1993) demonstrated that r = 0.46. It must be noted however that these studies relied
on the original version of the scale, which included items on predictability.
Consistent with the second part of hypothesis one, the academics’ levels of
job autonomy (M = 52%SM) were significantly higher than the supermarket workers
(M = 32%SM). For the purpose of this study, this difference in job autonomy should
be sufficient to examine the differences in primary and secondary control. It is
expected that the relationship between job autonomy and the control strategies is
linear, and that with increasing job autonomy, the use of primary control will
increase, and the use of secondary control will decrease. As such, even if the
difference between the group is not extremely large, the differences in the use of the
control strategies should still exist, however they may be less extreme.
In order to understand the meaning of these levels of job autonomy, it is
useful to compare them with other studies. Although normative data on Ganster’s
(1989, cited in Dwyer & Ganster, 1991) scale are not available, a few studies have
relied on this scale. They have shown that nurses scored 46%SM (Ganster et al.,
2001), and 57%SM (Munro et al., 1998). Furthermore, manufacturing employees
scored 57%SM (Dwyer & Ganster, 1991). It is difficult to make comparisons with
past studies however, as these studies have generally altered the scale in some way
(e.g., Ganster et al., 2001; Munro et al., 1998). Indeed the current study made an
important change to the scale, as the items on predictability were excluded.
152
2.7.2
Does Job Autonomy Influence the Use of the Control Strategies?
Partial support was provided for the second hypothesis as there was a
significant occupational difference in secondary control (M = 36%SM academics;
M = 46%SM supermarket workers), but not primary control (M = 71%SM
academics; M = 67%SM supermarket workers). Partial support was also provided
for hypothesis three as primary control was positively related to job autonomy
(r = 0.46), however secondary control was not correlated (r = -0.18). These findings
are somewhat inconsistent, with the former suggesting that job autonomy influences
secondary control but not primary control, and the latter suggesting that job
autonomy influences primary but not secondary control. More emphasis is placed on
hypothesis three as it is based on the measured level of job autonomy rather than the
assumed level. Thus, these findings demonstrate that as job autonomy increases,
primary control increases.
These findings appear to be inconsistent with Abouserie’s (1996) study on
academics’ coping strategies. In this study, academics were given a list of strategies
and required to indicate which ones they use to handle stress. The following coping
strategies emerged as the most common; acceptance of the problem (58%), talking
with others (57.7%), and trying to come to terms with each problem (55.8%).
Although the most common strategy, “acceptance of the problem” appears to
be a secondary control strategy, it is different to secondary control. Secondary
control is often referred to as acceptance however it is not acceptance that the
problem exists; it is acceptance that the problem cannot be overcome. Acceptance of
153
the problem may be interpreted as recognising that the problem exists, which is not
secondary control. Thus, although Abouserie’s (1996) results suggest that academics
mostly use secondary control, this may not be the case.
One study partially supports the findings from the current study. Narayanan,
Marian and Spector (1999) studied the coping strategies reported by academics, sales
employees, and clerical workers. They used an open-ended questionnaire where
participants were asked how they handled a stressful event at work. The academics
tended to handle their problems at work by taking direct action (24% of sample), and
talking to the chair of department (26%). The clerical workers and the sales
employees, on the other hand, reported that they talked with their co-workers (22%,
29%, respectively), or friends (24%, 29%, respectively).
Although Narayanan et al., (1999) did not measure job autonomy, their
findings demonstrate that the employees expected to have higher job autonomy
(i.e., academics) tended to rely on primary control-type strategies. The employees
expected to have lower job autonomy (i.e., sales employees, clerical workers) tended
to rely on secondary control-type strategies. These findings were partially consistent
with the current study.
The finding that job autonomy is positively related to primary control
provides some support for the proposed model of job satisfaction presented in Figure
2. This model, based on the job demand-control model (Karasek & Theorell, 1990)
and the life span theory of control (Heckhausen & Schulz, 1995), proposes that
employees with high job autonomy are more likely to successfully change the
154
environment using primary control. Thus, as job autonomy increases, primary
control increases.
However, the findings in the present study must be examined cautiously as a
limitation has been identified. The primary and secondary control scale required
respondents to indicate their agreement with each type of strategy, from 1 (do not
agree at all) to 10 (agree completely). It is now recognised that the only information
this scale provides is whether the respondents have ever used the strategies, and not
how often they are using the strategies. The current findings only demonstrate that
as job autonomy increases, employees’ agreement with the primary control strategies
increases, not the frequency.
2.7.3
Does Job Autonomy Influence the Relationship Between the Control
Strategies and Job Satisfaction?
In addition to testing whether job autonomy influences the use of the control
strategies, the current study also tested whether job autonomy influenced the
relationship between the control strategies and job satisfaction. Consistently,
primary control (r = 0.44) was more positively related to job satisfaction than
secondary control (r = 0.04) for the academics. However, secondary control
(r = 0.14) was not more positively related to job satisfaction than primary control
(r = 0.38) for the supermarket workers. Further analyses demonstrated that job
autonomy did not moderate the relationship between the control strategies and job
satisfaction. As such, it appears that primary control is more adaptive than secondary
control for all employees, whether they have low or high job autonomy.
155
These findings do not support the discrimination model (Thompson et al.,
1998) which proposes that primary control is the most adaptive strategy in
controllable situations, and that secondary control is the most adaptive strategy in
uncontrollable situations. Rather, these findings support the primacy/back-up model
(Thompson et al., 1998), which proposes that primary control strategies are more
adaptive than secondary control strategies for people in low-control or high-control
situations.
Although these results appear to support the primacy/back-up model, closer
examination of the Primary and Secondary Control Scale (Heeps et al., 2000)
reveals several limitations. The most notable is that some of the primary control
items examined whether the employees believed that they could change their
situation, rather than examining how they could change their situation. For example,
the items “I think I can make it happen”, “I believe I can achieve it” and “I believe I
can overcome it” measure whether a person believes that they can change a situation.
These general and non-specific thoughts were assessed rather than specific perceived
behaviours (i.e., work harder) because it was assumed that there could be an
unlimited number of specific behaviours.
However, it is now questioned whether believing that one can change a
situation is a measure of primary control. A person may report that they can change
a situation for a variety of reasons, not just if they use primary control strategies
when they face difficulties. For example, a person may report that they can change
their environment because they have high optimism. Alternatively, they may be
using the secondary control strategy “illusory optimism” where they tell themselves
156
that “everything will work out okay in the end.” These examples serve to illustrate
that people who believe that they can change their environment may not necessarily
use primary control.
To overcome these limitations, the measure of primary control may need to
be more specific. Rather than assessing whether people generally believe they can
change their environment, the primary control scale needs to assess how people
change their environment using primary control strategies. Thus the scale needs to
examine perceived strategies (e.g., exerting more effort, working harder) rather than
beliefs. This would make the scale consistent with the secondary control scale,
which assesses specific strategies.
In summary, although the findings suggest that the relationship between the
control strategies and job satisfaction is not influenced by job autonomy, several
problems have been identified in the primary and secondary control scale. The
primary and secondary control scale needs to be revised so that the primary control
items refer to perceived strategies rather than beliefs, and the rating scale needs to
assess frequency.
2.7.4
Do the Control Strategies Mediate the Relationship Between Job
Autonomy and Job Satisfaction?
Hypothesis six tested an alternative explanation to Karasek and Theorell’s
(1990) proposal for the relationship between job autonomy and job satisfaction. This
explanation, developed in chapter 1, proposes that employees with high job
autonomy rely on more primary control strategies which are positively related to job
157
satisfaction, whereas employees with low job autonomy rely on more secondary
control strategies which are less positively related to job satisfaction. It must be
noted however that although secondary control strategies are less positively related to
job satisfaction, it is proposed that for workers with low job autonomy, secondary
control strategies are superior to primary control strategies. If these workers use
primary control, they are expected to experience primary control failure.
The results demonstrated that primary control, but not secondary control,
mediated the relationship between job autonomy and job satisfaction. This provides
empirical evidence supporting one mechanism by which job autonomy may
influence job satisfaction. The importance of these findings must not be
overemphasised however, as problems have been identified with the primary and
secondary control scale. As such, the mediating role of primary and secondary
control needs to be re-examined using a revised scale.
2.7.4.1
Summary
The major aim of this study was to test an explanation for the relationship
between job autonomy and job satisfaction. The explanation proposes that job
autonomy influences the use and adaptiveness of the control strategies. The results
from the current study have offered some support for job autonomy influencing the
use of primary control strategies, but less support for job autonomy influencing the
adaptiveness of the control strategies. However, as there are some methodological
problems with the scale, the proposition requires further examination.
158
2.7.5
Examining Occupational Differences in Job Satisfaction
The differences in job autonomy and primary and secondary control were
expected to influence job satisfaction, where the academics were expected to report
higher job satisfaction than the supermarket workers. This proposal was not
supported however, as the academics reported similar levels of job satisfaction
(M = 66%SM) as the supermarket workers (M = 59%SM). In order to understand
these levels of job satisfaction, past studies will be examined.
2.7.5.1
Past Studies on Job Satisfaction
As few studies have examined academics’ or supermarket workers’ levels of
job satisfaction, and as there does not appear to be any consensus as to what is the
normative level of job satisfaction, a review was conducted. A range of studies
(N=36), which examined the levels of job satisfaction reported by different
occupational groups, were selected from psychology databases. These studies,
displayed in Appendix G, examine several occupational groups including nurses,
teachers, managers, manufacturing employees, and social workers. Although these
studies relied on several different scales, including global and facet scales, they were
reasonably consistent. The average level of job satisfaction was 66.75%SM, and the
scores ranged from 44.75%SM (Laschinger, Finegan & Shamain, 2001) to 87%SM
(Fisher, 2000). This average is similar to the academics and supermarket workers
levels of job satisfaction.
159
A few studies have specifically examined academics’ and supermarket
workers’ levels of job satisfaction. For academics, researchers have reported the
following levels of job satisfaction; 57%SM (Leung et al., 2000), 65%SM (Hill,
1986), 66%SM (Lahey & Vihtelic, 2000), 74% (Carson, Lanier & Carson, 2001),
82% (Olsen, 1993) and 83%SM (Niemann & Dovidio, 1998).
Although these scores vary widely, it must be recognised that these studies
have relied on different scales of job satisfaction. Some relied on facets scales of job
satisfaction (Hill, 1986; Lahey & Vihtelic, 2000) whilst others relied on global scales
on job satisfaction (Carson, Lanier & Carson, 2001; Leung et al., 2000; Niemann &
Dovidio, 1998; Olsen, 1993). However, the facet versus global distinction does not
necessarily explain the differing levels of job satisfaction, as facet and global scales
of job satisfaction have been shown to be moderately correlated (Wanous et al.,
1997). Rather within the facet and global scales, there is extensive variability that
may account for the inconsistent levels of job satisfaction.
There are differences among the facet scales of job satisfaction. For example,
Hill’s (1986) facet scale of job satisfaction measures satisfaction with several
dimensions including economic, teaching administrative, collegial, recognitionsupport, and convenience. In contrast, Lahey and Vihtelic (2000) focussed on the
work itself, pay, recognition, co-workers, and supervision. The difference between
Hill’s (1986) facets and Lahey and Vihtelic (2000) facets may be important. Hill’s
(1986) facets were designed to be specific to academia, however it appears that they
are focussing on the areas that academics traditionally cite as a source of stress, such
as recognition, finances (Leung et al., 2000), and administration (Abouserie, 1996).
160
As such, the academics in Hill’s (1986) study may have a reported a lower level of
job satisfaction than those in Lahey and Vihtelic (2000) study because the scale was
focussed on the more negative aspects of the job.
There are also differences among the global scales of job satisfaction. For
example, Niemann and Dovidio (1998) relied on a 3-item measure of job
satisfaction, which included the following items; “I am satisfied with my job”, “I find
fulfillment in my work” and “I feel free to do the work that is important to me.” The
level of job satisfaction reported by the academics in this study may have been
higher because of the inclusion of the third item, which may be confounded with job
autonomy.
A more valid global measure of job satisfaction was used in Olsen’s (1993)
study. Measuring job satisfaction through one-item (i.e., “All things considered, how
satisfied are you with your position”), they found academics reported a high level of
job satisfaction (M = 82%SM). It must be noted however that this level of job
satisfaction was reported by academics in their first year of appointment.
Interestingly, they re-tested these academics at the end of their third year, and found
that their level of job satisfaction had declined to 71.66%SM. This lower level is
more consistent with other studies.
In summary, it is extremely difficult to produce an average level of job
satisfaction for academics. Only a few studies have examined academics job
satisfaction and these have tended to rely on different scales. The level of job
satisfaction found in the current study fits within the range found by past studies. It
must be noted however that this range is reasonably large.
161
In regard to supermarket workers, the only studies that can be compared with
the current findings are those conducted on retail workers. These studies have
generally reported a higher level of job satisfaction than that found for the
supermarket workers. For example, Doran, Stone, Brief and George’s (1991) study
demonstrated that retail workers given the Minnesota Satisfaction Questionnaire
(Weiss et al., 1967) reported a level of job satisfaction that was 73%SM.
Furthermore, Leung’s (1997) study on retail workers reported similar findings using
Hackman and Oldham’s (1975) scale (70% SM).
Although these studies report higher levels of job satisfaction, it must be
noted that the workers in these studies were obtained from a department store (Doran
et al., 1991) and a casual apparel store (Leung, 1997), and as such, may have more
job autonomy that the supermarket workers. The supermarket register operators are
required to work on the cash register for the majority of their shift whereas retail
assistants can often choose among different task to complete. Thus, it is difficult to
compare these studies with the current findings.
2.7.5.2
Explaining the Levels of Job Satisfaction Reported by the Academics
and the Supermarket Workers
The finding that such two distinct occupational groups report similar levels of
job satisfaction is surprising. However, there may be differences between the groups
that can account for this. First of all, the nature of the work is very different for these
two groups. The supermarket workers engage in repetitive work, and as such, they
may face few novel difficulties. The academics, on the other hand, are expected to
162
be involved in several complicated activities and face many varied difficulties. As
such, although the current study was proposing that supermarket workers would have
lower job satisfaction because they have less autonomy, they may also have fewer
difficulties to overcome.
Another difference between the two occupational groups that may explain
their similar levels of job satisfaction concerns their different investments and
expectations. Whereas the supermarket workers have invested little time into
training, the academics have invested at least seven years studying at university. The
number of years invested in training or education may be particularly important, as it
has been suggested that education is positively correlated with expectations (Clark &
Oswald, 1996).
For example, Clark’s (1996) study of British employees demonstrated that the
percentage of employees who reported that they were very satisfied with their job
was greatest for the group with the lowest education (M = 78%SM). The next
highest reported a level which was 74%SM, and the highest educated group reported
a level of job satisfaction that was 73%SM. Although the differences between these
groups are small, it is surprising that the group with the lowest level of education
would report a level of job satisfaction that equalled those with a higher education,
let alone surpassed it. As such, the academics may have higher job expectations than
the supermarket workers.
In summary, inconsistent with the hypotheses, the academics and teachers
reported similar levels of job satisfaction. This finding may be partly attributed to
163
the supermarket workers experiencing fewer difficulties than the academics, or the
academics having higher job expectations than the supermarket workers.
2.7.6
Examining Occupational Differences in Life Satisfaction
Consistent with hypothesis seven, the academics reported higher overall life
satisfaction (M=78.22) than the supermarket workers (M=73.30). The academics’
levels of life satisfaction were expected to be higher because job satisfaction was
expected to be positively related to life satisfaction. Although the academics did not
report higher job satisfaction than the supermarket workers, they did report higher
life satisfaction.
In regard to the normative levels of life satisfaction, Cummins’ (1995, 2000b)
homeostatic theory of life satisfaction proposes that the mean life satisfaction across
population samples lies within the 70-80%SM range. This is because people have a
“set-point range” for their life satisfaction. This set-point range is determined by
personality variables, namely neuroticism and extroversion. Together, these two
variables provide an affective balance, where the mid-point for the set-point range is,
on average, 75%SM. This affective balance influences the second-order buffers (i.e.,
optimism, self-esteem and control) so that, on average, their set-point is also
75%SM. These second-order buffers can however be influenced by the external
world. Hence, the mid-point for the set-point range can range between 70-80%SM.
Consistent with this prediction, both the academics’ (M = 78.22%SM) and the
supermarket workers’ levels of life satisfaction (M = 73.30%SM) lay within this
range.
164
The academics’ level of life satisfaction was at the higher end of the
normative range. According to the homeostatic theory of life satisfaction, the ceiling
for population sample means is approximately 80%SM (Cummins, 2000b). This
value represents the theoretical maximum for sample means grouped as data where
the distribution of set-ranges is normal, and each person has achieved the upper value
of their set-range. As such, the academics’ level of life satisfaction, in relative terms,
is extremely high.
The supermarket groups’ level of life satisfaction was at the lower end of the
normative range. Cummins (2000b) proposes that when life satisfaction falls
towards the 70%SM mark, homeostatic devices operate to prevent it from falling
further. When the sample mean approaches 70%SM however, the homeostatic
machinery is defeated for a significant proportion of the sample. As this happens,
the distribution collapses and the standard deviations increase. Consistent with this
prediction, the standard deviation of life satisfaction for the supermarket workers
group (SD = 15.97) was greater than for the academic group (SD = 10.96). Hence, a
greater proportion of the supermarket workers may be experiencing homeostatic
defeat. In summary, the academics reported higher life satisfaction than the
supermarket workers, however both means lay within the normative range.
2.7.7
Predicting Job Satisfaction from Job Autonomy, Control Strategies,
Personality, and Life Satisfaction
Partial support was provided for hypothesis eight, as job autonomy and
primary control predicted job satisfaction for the academics, and primary control
165
predicted job satisfaction for the supermarket workers. Inconsistent with the
proposed model of job satisfaction however, secondary control did not predict job
satisfaction for either group, and job autonomy did not predict job satisfaction for the
supermarket workers. The finding that secondary control did not predict job
satisfaction clearly needs to be re-examined as there are several methodological
problems with the secondary control scale. The finding that job autonomy did not
predict job satisfaction for the supermarket workers requires further examination.
The finding that job autonomy did not predict job satisfaction for the
supermarket workers may reflect problems with the job autonomy scale. The job
autonomy scale was a multidimensional scale. The scale was thought to be superior
to other scales as it prompted employees to consider several aspects of their work
environment (Ganster & Fusilier, 1989). However, the scale may also be
problematic, as although it ensures that respondents think of the same facets, some
facets may not be appropriate for some employees.
An alternative is to use a global scale of job autonomy. For example, the Job
Descriptive Survey (Hackman & Oldham, 1975) measures job autonomy through
assessing whether the employee has the opportunity for independence and freedom in
their job. Using this scale, the respondents can just consider the areas that are
important to them. They can include facets that are not specified in the facet scale,
and exclude facets that are not relevant to their workplace. As such, the supermarket
workers, although reporting low job autonomy on the multidimensional scale, may
have higher levels of global job autonomy. As such, future studies will need to
assess job autonomy using a global measure.
166
2.7.8
Conclusion
This study has contributed to the development of the proposed model of job
satisfaction (refer to Figure 2). This model, adapted from Karasek’s (1979) job
demand-control model, proposes that job autonomy relates to job satisfaction through
influencing the way employees manage their work difficulties. The findings
demonstrated that workers with higher job autonomy do manage their work
difficulties differently from workers with lower job autonomy. Specifically, as job
autonomy increases, primary control increases.
In addition to examining how job autonomy influences the use of control
strategies, this study also proposed that job autonomy influences the adaptiveness of
the control strategies. Primary control strategies were, as predicted, the most
adaptive strategies for the academics, however secondary control strategies were not
the most adaptive strategies for the supermarket workers. These findings supported
the primacy/back-up model, suggesting that all employees, whether they have low or
high job autonomy, should rely on primary control strategies when they face a
difficulty at work. However, as problems have now been identified with the primary
and secondary control scale and job autonomy scale, further research needs to
re-examine these hypotheses.
167
3 Chapter 3 - Study Two
168
3.1
Abstract
This study aims to re-test the proposal that job autonomy influences the amount of
control strategies that employees use, the relationship between the control strategies
and job satisfaction. This study attempted to overcome the major limitations
identified in study one, concerning the primary and secondary control scale and the
job autonomy scale. Furthermore, this study examined the influence of two new
variables, namely need for job autonomy and social support at work. Two
occupational groups that were expected to differ in their levels of job autonomy
(i.e., secondary school teachers and academics) were compared. It was expected that
the academics would report higher job autonomy, higher primary control, and lower
secondary control than the teachers. Furthermore, it was expected that primary
control would be more adaptive for the academics, whereas secondary control would
be more adaptive for the teachers. These hypotheses were not supported however, as
both groups reported equally high levels of primary and secondary control, and
primary and secondary control were not related to job satisfaction. These
inconsistent results prompted a review of the underlying assumptions of the study.
Some methodological limitations were identified in the hypotheses examining job
autonomy and the control strategies. Despite this, support for the remaining
hypotheses highlighted the importance of social support at work in predicting job
satisfaction.
169
3.2
Proposal for Study Two
This study re-examines the proposal that job autonomy influences the use and
the adaptiveness of primary and secondary control strategies. It attempts to
overcome the limitations identified in study one. This study uses: a) a revised
version of the Primary and Secondary Control Scale; b) a new measure of job
autonomy; and c) different occupational groups for comparison. Furthermore, this
study incorporates recent research suggesting that the need for job autonomy
mediates the relationship between job autonomy and job satisfaction, and examines
how social support at work influences job satisfaction. These changes will now be
discussed.
3.2.1
a) The Primary and Secondary Control Scale
The Primary and Secondary Control Scale, developed by Heeps et al., (2000)
was implemented in study one because it was one of the best scales that concurred
with Rothbaum et al’s., (1982) and Heckhausen and Schulz’s (1995) definition of
control. However, the scale was exploratory, and study one highlighted some
problems with the scale. As such, a review was conducted on the scale in
collaboration with RoseAnne Misajon. This review, which was based on factor
analyses of the scale, highlighted several problems with the scale. These problems
involved: i) the stem of the item; ii) the content of the item; and iii) the rating scale.
From this review, a third and fourth edition of the Primary and Secondary Control
Scale was developed.
170
Factor analyses conducted on the first and second edition of the Primary
Control and Secondary Control Scale were reviewed (e.g., Cahill, 1998; Cousins,
2001; Maher & Cummins, 2001, Misajon, 2002; Misajon & Cummins, in press;
Spokes, 1998). These analyses were based on a variety of samples, including elderly
people, people with arthritis, people with multiple sclerosis, and academics. Each
researcher tended to make minor changes to the scale, where they may have excluded
some items, or changed the wording of others, to make the scale more suitable to
their sample. These researchers then conducted exploratory factor analyses on the
scale, and found that the items initially loaded on 3, 4 or 5 factors. As they were
often unable to explain these factors, they then requested two factors. The resulting
analyses are displayed in Table 16. In this table, items that were excluded from that
particular version of the scale are represented by NA. Items that did not load on any
factors, or alternatively loaded on both factors are represented by a dotted line. Items
that loaded on the primary control factor are bolded, whilst items that loaded on the
secondary control factor are not bolded. This table demonstrates that the primary
control items generally factored well, however the secondary control items often
loaded on both factors. The primary control scale will be discussed first.
171
Table 16- Factor Analysis of Primary and Secondary Control Scale
Study
Primary Control Items
New ways to achieve goal
Persistence
Remove obstacles
Invest time
Learn skills
Ask for help or advice
Effort to make it happen
Secondary Control Items
Positive Re-interpretation
Wisdom
Illusory optimism
Downward social
comparison
Past success
Vicarious
Positive approach
Goal disengagement
Present success
Predictive negative
Attribution
Behavioural avoidance
Active avoidance
Sour grapes
Support
Give up
1
2
3
4
5
6
7
0.56
0.75
0.71
0.60
0.59
0.37
0.31
0.80
0.73
0.70
0.61
0.74
0.56
0.53
0.60
0.64
0.66
0.72
0.76
0.36
----
0.53
0.62
0.56
0.66
0.65
0.61
0.54
0.56
0.84
0.74
0.81
0.49
0.68
0.82
0.70
0.71
0.51
0.51
NA
NA
0.39
0.79
0.81
NA
0.72
0.67
NA
0.65
0.46
0.47
0.54
0.65
0.32
0.58
---0.50
0.60
0.49
0.52
0.68
0.57
0.51
0.67
0.43
0.44
0.41
0.37
0.77
-------------
0.42
0.47
---0.46
0.71
0.76
0.63
---0.57
---0.43
0.30
---0.39
---NA
0.52
0.60
NA
NA
0.73
0.66
0.76
---NA
NA
0.32
0.59
0.71
0.41
---NA
0.62
0.37
0.41
---NA
NA
----0.49
0.31
0.98
0.65
NA
0.72
---0.43
------NA
0.57
-0.58
0.81
0.82
0.54
0.82
0.58
0.74
0.74
0.49
0.67
NA
0.77
NA
---------0.72
0.74
0.61
0.40
0.69
0.74
0.61
-------
---------0.76
---0.54
---0.76
0.76
0.54
-------
Studies 1 = Maher (2001); 2 = Misajon (2000); 3 = Spokes (1998); 4 = Cahill (1998);
5 = Misajon (2001); 6 = Study one; 7 = Cousins (2001)
Bolded factor loadings refer to the primary control factor.
172
3.2.1.1
Primary Control Scale
3.2.1.2
i) Stem of Primary Control Items
The factor analyses in Table 16 demonstrate that the primary control items
generally factor well. However, one reason why the primary control items may have
loaded on a different factor to the secondary control items is that the primary and
secondary control items were presented separately in the scale. The primary control
items began with “when I find a goal that is difficult to reach”, “when I really want
something” and “when something gets in the way of a goal”, whereas the secondary
control items all began with “when something bad happens that I cannot change.”
The reason the two strategies have different stems is that it was originally
assumed that primary control strategies were only used when a person faced a
difficulty that they could change, and that secondary control strategies were only
used when the difficulty could not be changed. This assumption may be incorrect
however, as it is possible for people to use secondary control when they face a
situation that they can change. For example, an employee may be upset that a coworker is always late. They may know that if they use primary control and talk to
their supervisor about the problem, the co-worker will be reprimanded, and as such
begin to arrive on time. However, they may choose not to use primary control as
they may then lose their friendship with the co-worker. Rather, they may implement
secondary control, and tell themselves that the problem “doesn’t matter.”
173
Similarly, it is possible that people use primary control when they face
situations that they cannot change. For example, an employee may dislike their work
times, yet be aware that the work times cannot be changed. Even so, they may
attempt to change their working times through using primary control, and discussing
solutions with their supervisor. The supervisor would presumably reject their
proposal, and the primary control strategy would have failed. Despite knowing the
possibility of primary control failure however, the employee may have decided to
take a risk.
As it is possible for primary and secondary control to be used in controllable
and uncontrollable situations, the scale was changed so that the stems of the items are
the same. The revised scale includes the primary and secondary control items
together, with the following introductory sentence; “Here are ways people deal with
difficult situations in their lives. How often have you had these thoughts when
facing a difficulty over the past week?” Examples of these thoughts are “it will work
out okay in the end” and “I knew it would happen.” The other control items which
involved actions rather than thoughts had an alternative introductory paragraph;
“How often have you done these things when facing a difficulty over the past week?”
(i.e., “I kept trying”, “I told someone about it”, “I worked to overcome it”).
3.2.1.3
ii) Primary control Item Content
As demonstrated in Table 16, all studies found that the items assessing new
ways to achieve goals, persistence, remove obstacles, learn skills and invest time,
loaded on the primary control factor. The items measuring effort to make it happen,
174
and ask others for help or advice occasionally loaded on the secondary control factor.
These two items were deleted as they were criticised for being similar to secondary
control strategies. Specifically, the item referring to effort to make it happen,
generally worded as “I think I can make it happen” does not actually refer to the
person putting in effort, and rather is similar to the secondary control strategy of
illusory optimism (i.e., “I know it will work out okay in the end”). The other item
referring to asking for help or advice was also deleted as it difficult to separate it
from secondary control. Indeed asking the boss or someone who has some power
over the problem for help or advice may be a means of changing the environment.
However, discussions with people who have less power over the situation, such as
friends, may only serve to make the person accept the problem.
Although the remainder of the primary control items loaded on the primary
control factor, there were still conceptual problems with the items. For example, the
item referring to learning skills was deleted from the scale, as it is only relevant if the
person is attempting to achieve something, and cannot be applied to the new stem,
namely difficult situations. Furthermore, the item assessing investing time was
omitted, as it was not necessarily indicative of primary control. A person may spend
lots of time on a problem, yet not attempt to change the environment.
The remainder of the items were criticised as they examined whether the
employees believed that they could change their situation, rather than examining how
they change their situation. For example, the items “I think I can make it happen”, “I
believe I can achieve it” and “I believe I can overcome it” measure whether a person
believes that they can change a situation. As discussed in chapter 2, it is questioned
175
whether believing that one can change a situation is a measure of primary control. A
person may report that they can change a situation for a variety of reasons, not just if
they use primary control strategies when they face difficulties. For example, a
person may report that they can change their environment because they have high
optimism. Alternatively, they may be using the secondary control strategy illusory
optimism, where they tell themselves that “everything will work out okay in the
end.” As such, the revised primary control scale, displayed in Table 17, is changed
to examine perceived strategies (e.g., exerting more effort, working harder) rather
than beliefs.
Table 17- Original and Revised Primary Control Items
2nd Edition (Heeps et al.,
2000)
New ways to achieve goal I think about different
ways to achieve it
Effort to make it happen
I think I can make it
happen
Invest time
I think about it a lot
Persistence
I believe I can achieve it
Remove obstacles
I believe I can overcome
it
Primary Control Strategy
3.2.1.4
4th Edition (Maher et al.,
2001)
I looked for different
ways to overcome it
I worked to overcome it
NA
I kept trying
I worked out how to
remove obstacles
iii) Rating Scale
The primary control items were originally rated on a 10-point scale ranging
from 1 (do not agree at all) to 10 (agree completely). This rating scale indicates
whether an individual agrees that they have used a strategy, not how much they have
used a strategy. Two people may report that they agree completely that they have
used a strategy, however one may use it 10 times a day, whilst the other may use it
176
once a week. As the scale did not differentiate between these people, the primary
control rating scale was changed to assess frequency.
In order to reduce inaccuracies, the scale was changed from measuring the
control strategies that people generally use when they face a difficulty to examining
the strategies people have used over the past week. As such, the rating scale ranged
from 0 (never) to 10 (every time).
3.2.1.5
Secondary Control
3.2.1.6
i) Stem of Secondary Control Items
As previously discussed, it was thought that the primary control items may
have loaded on a different factor to the secondary control items because the stems of
the items were different. In order to overcome this, the secondary control items were
placed with the primary control items. The stem of the item was changed from
“when something bad happens that I cannot change” to “how often have you done
these things when facing a difficulty over the past week.”
3.2.1.7
ii) Item Content
As demonstrated in Table 16, a few secondary control items loaded on both
the primary control factor and the secondary control factor. There did not appear to
be a consistent pattern in these studies however, with some studies finding that an
item loaded on a secondary control factor, whilst others found that it loaded on a
primary control factor. It was originally expected that the secondary control items
177
would form one factor, however it is now proposed that each item measures a
different strategy and that these strategies are independent. One person may use one
secondary control strategy in all situations, and so not use any of the others. This
proposal has implications for the scoring of the secondary control scale, and also for
factor analyses of the scale.
In regard to scoring, the proposal that respondents’ scores on secondary
control items may not be consistent suggests that the secondary control items cannot
be aggregated. However, secondary control can still be measured by using the
highest scoring item. This scoring procedure will be explained in detail later.
In regard to factor analyses, the proposal that respondents’ scores on
secondary control items are not consistent may explain why the secondary control
items loaded on more than one factor. Respondents may report different scores on
all the secondary control items, and thus they would not be expected to cluster
together. As such, rather than eliminating any items which loaded on a primary
control factor, the items were examined in terms of their theoretical usefulness.
Many of the items were similar to others, such as past success and present
success, and positive approach and behavioral avoidance. Present success (“I think
about my success in other areas”) encompasses past success (“I remember I have
accomplished a lot in life”). Furthermore, positive approach (“I do something nice to
take my mind off things”) and behavioral avoidance (“I do some physical exercise or
try to relax”) could be combined to measure active avoidance (“I do something to
take my mind off things”).
178
Two items that had been deleted from the first edition of the scale, namely
denial and support, were reinstated. Denial, measured by the item “I ignored it” was
deleted from previous versions of the scale as it was thought to be similar to the item
for goal disengagement (i.e., “It doesn’t matter”). However, telling oneself that a
problem is not important is clearly different from denying that the problem exists.
Intuitively, goal disengagement may be more adaptive than denial.
Support, measured by the item “told someone about it” was also added to the
scale. It was originally deleted from the first edition of the scale as it was a
behavioural strategy. It was assumed that all secondary control strategies had to be
cognitive strategies. This is not the case however, and Heckhausen and Schulz
(1995) recommend that the distinction between primary and secondary control
should not be based on behavioural versus cognitive, rather whether it involves
changing the environment versus changing the self. Support allows the person to
change themselves and become more likely to accept a situation.
After this theoretical analysis, 12 secondary control strategies remained (refer
to Table 18). These strategies were grouped according to their purpose. All of the
strategies are designed to make the person feel better about their situation, however
they may do this by reducing negative feelings (i.e., self-protective) or by increasing
positive feelings (i.e., self-affirmative). As demonstrated in Table 19, people may
reduce negative feelings by telling themselves that a difficult situation is not their
fault, that they knew it would happen, or that it doesn’t matter. People may increase
positive feelings however by thinking that they are better off than many other people,
and thinking about areas of their life in which they have been successful.
179
Table 18- Original and Revised Secondary Control Items
Secondary control Second Edition
strategy
Positive reI can see that something good will
interpretation
come if it
Wisdom
Illusory-optimism
Downward social
Omparison
Past success
Vicarious
Positive approach
Goal
disengagement
Predictivenegative
Attribution
Active avoidance
Sour grapes
Present success
Denial
Support
I remember you can’t always get
what you want
I know thing will work out OK in
the end
I remember I am better off than
many other people
I remember I have already
accomplished a lot in life
I remember the success of my
family and friends
I think nice thoughts to take my
mind off it
I tell myself it doesn’t matter
I don’t feel disappointed because I
knew it might happen
I can see it is not my fault
I ignore it by thinking about other
things
I realise I didn’t need to control it
anyway
I think about my success in other
areas
NA
NA
4th Edition
I looked for something
else that was positive in
the situation
I can’t always get what I
want
It will work out okay in
the end
I am better off than many
other people
NA
I thought of the success
of my family or friends
NA
It doesn’t matter
I knew it would happen
It was not my fault
I did something different,
like going for a walk
NA
I thought about my
success in other areas.
I ignored it
Told someone about it
Table 19- Functions of the Secondary Control Strategies
Use
Selfprotective
Selfaffirmation
Definition of use
Reduces the
negative impact of
the situation
Increases positive
feelings about self
Secondary Control Strategy
Illusory optimism, goal disengagement,
predictive negative, attribution, denial,
wisdom
Downward social comparison, vicarious,
present success, support, positive reinterpretation, active avoidance
180
3.2.1.8
iii) Rating Scale
As with the primary control items, the secondary control items were changed
to measure frequency. Each strategy was rated on an 11-point scale ranging from
0 (never) to 10 (every time).
3.2.1.9
Summary
Following a review of the factor analyses conducted on the primary and
secondary control scale, and an investigation of the item stem, the item content, and
the rating scale, a revised scale was developed. This scale, presented in Appendix H
will be implemented in the second study. One further point that requires discussion
however, is the scoring of the control scale.
3.2.1.10
Scoring the primary and secondary control scale
Previous versions of the Primary and Secondary Control Scale (Heeps et al.,
2000) averaged across the strategies to obtain an overall score for primary control
score and an overall score for secondary control. The problem with this method
however, is that a person may report that they use one secondary control strategy
every time (10) and report never (0) for the remaining strategies. Calculating the
average level of secondary control in this situation would result in a low score. As
they used a secondary control strategy every time they faced a difficulty in the
previous week, a low score is not representative of their secondary control use.
181
One solution to this problem is to take the highest score for primary control
and the highest score for secondary control. Using this method, a person who reports
10 for one secondary control strategy and 0 for the rest would receive a score of 10
for secondary control. Another person may report different scores for all secondary
control strategies, including a 5, 7, 8, 10, 2, 3, 4. This person would also receive a
score of 10, as it is the highest score. The fact that the second person has higher
scores on other secondary control items does not mean that the person uses more
secondary control strategies, only that they use a greater variety of secondary control
strategies.
3.2.2
b) Job Autonomy Scale
The next limitation that was identified in study one concerns the
measurement of job autonomy. The autonomy scale implemented in study one was
a multidimensional scale. The scale examined specific facets of the work, such as
variety of work, pace of work, scheduling of rest breaks, and interaction with others.
This multidimensional scale was advantageous as it prompted the employees to
consider all aspects of their work. This is important as employees may fail to
consider some facets of their work. They may have accepted for example that they
cannot change their pay, policies and amount of interruptions, and thus no longer
expect to be able to make choices in these areas. The multidimensional scale ensures
that all workers think about the same job facets.
However, it is now recognised that the multidimensional scale is also
problematic. Although the scale prompts employees to consider all aspects of their
182
work, some of the facets may not be appropriate for them, or important to them.
With a global scale, the respondent can include facets that are not specified in the
facet scale, and exclude facets that are not relevant to their workplace. As such, their
response is only based on facets that they think are important. They may exclude
some facets because they have lowered their expectations, however if they have
accepted them, then they are not expected to influence their levels of job satisfaction.
3.2.3
c) Occupational Groups
This study will compare two occupational groups that have different levels of
job autonomy, low and high. As in study one, university academic staff have been
selected for the high job autonomy group. Academics traditionally have flexibility in
their work and freedom to pursue their own research interests (Winefield, 2000).
They can often choose among a variety of tasks, including research, teaching, and
administration (Fisher, 1994). Whether this theoretical expectation existed in
practice was tested in study one. The results demonstrated that the academics
reported a level of job autonomy which was 53%SM. It could not be ascertained
whether this score was high however, as there was few comparative studies for
Ganster’s (1989, cited in Dwyer & Ganster, 1991) multidimensional scale of job
autonomy. Study two will overcome this problem by relying on a scale, which has
been used more extensively.
Secondary school teachers have been selected for the low job autonomy
group. Teachers have been selected rather than supermarket workers because this
study is attempting to minimise the differences between the groups. In study one, it
183
was demonstrated that although the supermarket operators reported higher job
autonomy, and higher secondary control than the academics, the two groups reported
similar levels of job satisfaction. However, there were differences between the two
occupational groups that may have accounted for the similar levels of job
satisfaction. The supermarket workers would have experienced fewer difficulties at
work than the academics, and may have had lower job expectations than the
academics.
Study two attempts to examine two groups which have similar experiences at
work, but which have differing levels of job autonomy, namely secondary school
teachers and academics. Although both occupational groups deliver education to
students and have similar roles, it is expected that teachers will report lower job
autonomy.
Although few studies have examined Australian teachers’ levels of job
autonomy, a recent report proposes that although the Government attempted to
empower schools and teachers through providing schools with more responsibility,
teachers are experiencing reduced autonomy (Senate Employment, Education and
Training References Committee, 1998). Teachers are reporting that they want to
have more involvement in decision-making. One study which interviewed 956
Australian teachers about the changes they felt were necessary to reduce stress
(Teacher Stress in Victoria, 1990) found that the most common change (80%) was to
increase staff collaboration and communications. They also mentioned increasing
consultations before major decisions are made.
184
The type of decisions that the teachers want to be consulted on concern
curriculum selection, development and implementation (Senate Employment,
Education and Training References Committee, 1998). It is particularly important
that the teachers are involved in curriculum selection so that they can have control
over the means of producing the results by which they will then be judged (Cole,
1989).
In summary, this study will test the major hypotheses by comparing
academics and teachers. Although little research has examined these two groups, it
is expected that the academics will report higher job autonomy than the teachers.
3.2.4
d) Need for Job Autonomy
In study one, it was assumed that high job autonomy was beneficial for all
employees. This assumption was based on Karasek and Theorell’s (1990, p. 12)
proposal that “if jobs were redesigned with high job decision latitude…demands
would be seen as challenges and would be associated with increased learning and
motivation, with more effective performance and less risk of illness.” However, it
must be noted that other researchers have suggested that people may differ in the
extent to which they like to exercise control over their environment (Burger &
Cooper, 1979; Parkes, 1989). This difference in need for autonomy may influence
the relationship between job autonomy and job satisfaction, where job autonomy
may have greater influence on job satisfaction when need for job autonomy is high.
Only a few studies have examined the moderating role of need for job
autonomy on the relationship between job autonomy and job related outcomes (e.g.,
185
de Jonge, Landeweerd & Breukelen, 1994, cited in de Rijk, Le Blanc, Schaufeli, &
de Jonge, 1998; Nicolle, 1994). These studies have tended to produce inconsistent
findings. For example, de Rijk et al., (1998) cite de Jonge et al’s., (1994) study as
providing evidence that the need for autonomy moderated the relationship between
job autonomy and emotional exhaustion and health complaints. When de Rijk et al.,
(1998) replicated the study however, they failed to find support for the moderating
role.
One other study, conducted by Nicolle (1994) provides some support for the
moderating role of need for autonomy. This study demonstrated that for nurses with
a low need for autonomy, job autonomy was positively related to absenteeism,
however for nurses with a high need for autonomy, job autonomy was not related to
absenteeism. These results must be interpreted with caution however, as only 3 of
the 36 analyses were significant.
One further study has been reported to provide evidence for the moderating
role of need for autonomy. De Rijk et al., (1998) cited Gaziel’s (1989) study on
school administrators as being supportive of the hypothesis. According to De Rijk et
al., (1998) this study demonstrated that for administrators who had a low need for
autonomy, job autonomy was not related to job satisfaction, whereas for
administrators who had a high need for autonomy, there was a positive relationship
between job autonomy and job satisfaction. Examination of the study demonstrates
that this is not the case however. Gaziel’s (1989) study did not examine the
relationship between autonomy and job satisfaction for workers with differing levels
186
of job autonomy. Rather, the study examined the major predictors of a perceived
deficiency in autonomy.
In summary, it has been suggested that employees may differ in their need for
autonomy and that the relationship between job autonomy and job satisfaction may
differ depending on this need. As only a few studies have examined this proposed
moderating effect, and as the studies tend to be inconsistent, clearly further research
is needed.
3.2.5
e) Addition of Social Support
As mentioned in chapter 1, the job demand-control model was extended to
include social support (Johnson & Hall, 1988; 1994; Johnson, Hall & Theorell, 1989;
Karasek & Theorell, 1990). Social support at work refers to “overall levels of
helpful social interaction available in the job from both co-workers and supervisors”
(Karasek & Theorell, 1990, p.69). Two major types of social support have been
identified, namely emotional support and instrumental support. According to
Karasek and Theorell (1990, p. 70), emotional support refers to the “degree of social
and emotional integration and trust between co-workers, supervisors and others”,
whereas instrumental support refers to “extra resources or assistance with work tasks
given by co-workers or supervisors.” The job demand-control-support model
proposes that social support at work predicts job satisfaction.
Study one did not examine social support at work as it focussed on
understanding how job autonomy influences the control strategies, and on personality
and life satisfaction. However, after examining research on the relationship between
187
social support and job satisfaction further, social support appears to be an extremely
important predictor, and as such, study two will examine social support at work in
more detail.
Social support at work has been shown to directly and indirectly increase job
satisfaction. In regard to the direct effects, several researchers have demonstrated
that social support at work is positively related to job satisfaction (r = 0.52; Dollard
et al., r = 0.66, Munro et al., 1998), and negatively related to job dissatisfaction
(r = -0.29, r = -0.28; LaRocco, House & French, 1980). These studies suggest that
workers who report higher social support tend to be more satisfied with their jobs.
One possible explanation for the positive relationship between social support
and job satisfaction is that social support reduces the negative effects of work
demands. This explanation, known as the buffering hypothesis, proposes that social
support at work buffers the potentially negative effects of high demands on job
satisfaction. Only a few studies have examined the buffering hypothesis for job
satisfaction.
A review of these studies, conducted by Van Der Doef and Maes (1999)
demonstrated that only two (i.e., Karasek, Triantis & Chaudry, 1982; Landsbergis,
Schnall, Dietz, Friedman & Pickering, 1992) of the six studies (Chay, 1993; de Jonge
& Landeweerd, 1993, cited in Van der Doef & Maes, 1999; Melamed, Kushnir &
Meir, 1991; Parkes & von Rabenau, 1993) that examined the buffering hypothesis
were supportive. Their review found no major differences among the studies to
account for the inconsistent findings except that both supportive studies used male
samples and the others used mixed or female samples.
188
One difference among the studies that may explain the findings is the
operationalisation of social support at work. For example, Karasek et al., (1982)
measured tolerance of supervisor, attentiveness of supervisor, instrumental support of
supervisor, demands of supervisor, number of co-workers, instrumental co-worker
support, and emotional co-worker support. Alternatively, Chay (1993) relied on the
Interpersonal Support Evaluation List (Cohen, Kamarack, Mermelstein & Hoberman,
1985) which measures appraisal support, belonging support, tangible support and
esteem support. A similar and briefer scale was employed by Landsbergis et al.,
(1992), who relied on Karasek and Theorell’s (1990) scale. This scale measures
emotional and instrumental support from co-workers and supervisors. There is
certainly no agreed upon way of measuring social support at work (Unden, 1996),
and as such, it is unclear if the operationalisation of social support influenced the
results. What is clear is that the buffering role of social support requires more
investigation.
In summary, although it is intuitively expected that social support at work
would reduce the negative effects of job demands or job stressors, the results are far
from consistent. As there are such few studies however, more research is required.
189
3.3
Model of Job Satisfaction
A revised model of job satisfaction, displayed in Figure 6, will be tested.
This model is similar to that presented in Figure 2, as the major proposal of the
model is that primary and secondary control mediate the relationship between job
autonomy and job satisfaction. However, this model includes new sections on social
support and need for job autonomy. In Figure 6, these changes are represented by
bolded variables and arrows. These new proposals will be discussed.
It is now proposed that the relationship between job autonomy and job
satisfaction is moderated by need for job autonomy. It should not be assumed that all
employees desire high autonomy. Indeed, some workers may have low job
autonomy yet still report high job satisfaction because they do not desire freedom
and independence in their job. Need for job autonomy and job autonomy predict the
interaction term (i.e., need for job autonomy x job autonomy), which in turn predicts
job satisfaction.
It is also proposed that social support at work influences job satisfaction
directly and indirectly. It is expected to be positively correlated with job satisfaction,
and to also moderate the effect of work difficulties on job satisfaction. In Figure 6,
this is represented by the interaction term. Difficulties at work and social support at
work together predict the interaction term (i.e., difficulties x social support), which in
turn predicts job satisfaction.
190
Figure 6 -Revised Model of Job Satisfaction for Study 2
Difficulties
at work
Social
Support at
work
Difficulties x
Social
Support
Job Autonomy x
Primary Control
Primary Control
Job Satisfaction
Job Autonomy
Secondary Control
Job Autonomy x
Secondary Control
Personality
Job Autonomy x
Need for job
autonomy
Need for job
autonomy
Life Satisfaction
191
3.4
Aims and Hypotheses
This study will compare levels of job autonomy, control strategies, and job
satisfaction reported by university academic staff and secondary school teachers. It
will also test the extent to which job autonomy mediates the relationship between
primary and secondary control strategies using the revised measures of job autonomy
and control strategies. The hypotheses are as follows:
1) Job autonomy will be positively related to job satisfaction, and the academics will
report higher job autonomy than the teachers.
This hypothesis tests the basic assumptions of the study. It needs to be
demonstrated that job autonomy is related to job satisfaction, and that comparisons
made between the two occupational groups are valid.
2) The academics will report more primary control, and less secondary control than
the teachers.
As the academics have higher job autonomy, they are expected to be more
likely to successfully implement primary control strategies than the teachers. As
secondary control is used to compensate for, and avoid future primary control failure,
it is expected that the teachers will report more secondary control than the academics
3) Job autonomy will be positively related to primary control, and negatively related
to secondary control.
192
As in hypothesis two, this hypothesis is examining whether job autonomy
influences the use of primary and secondary control. However, unlike hypothesis
two, it is based on the measured level of autonomy rather than the expected
occupational level
4) Primary control will be more positively related to job satisfaction than secondary
control for the academics, and secondary control will be more positively related to
job satisfaction than primary control for the teachers.
This study proposes that job satisfaction results from a match between job
autonomy and control strategies. Based on the discrimination model, it is proposed
that primary control is most adaptive for employees who can control their work
environment (i.e., high job autonomy), and that secondary control is most adaptive
for employees who have little control over their environment (i.e., low job
autonomy). Although primary control is generally more adaptive than secondary
control, the teachers have a high probability of experiencing primary control failure
when implementing primary control strategies, and thus it is expected that, for them,
secondary control strategies will be more adaptive.
5) The relationship between the control strategies and job satisfaction is moderated
by perceived job autonomy.
This hypothesis, like hypothesis four, is testing whether job autonomy
influences the relationship between the control strategies and job satisfaction. Unlike
193
hypothesis four however, it is based on the measured level of job autonomy rather
than the assumed level of autonomy based on the occupation.
6) The relationship between job autonomy and job satisfaction is mediated by
primary and secondary control strategies.
This hypothesis is testing an explanation for the relationship between job
autonomy and job satisfaction. This explanation proposes that people who have high
job autonomy have high job satisfaction because of their use of the control strategies.
These workers use more primary control and less secondary control, and are thus
able to overcome their difficulties.
7) The academics will report higher job satisfaction and higher life satisfaction than
the teachers.
The academics are expected to report higher job satisfaction than the teachers
as they have higher job autonomy, and use more primary control and less secondary
control. This level of job satisfaction is expected to influence their level of life
satisfaction.
8) The influence of work difficulties on job satisfaction is moderated by levels of
social support at work.
This hypothesis is based on the job demand-control-support model (Karasek,
1979) which proposes that social support can reduce the effect of demands at work.
194
9) The relationship between perceived job autonomy and job satisfaction is
moderated by need for autonomy.
People may differ in their need for autonomy, and this will influence the
relationship between job autonomy and job satisfaction.
10) Job autonomy, control strategies, life satisfaction, personality, difficulties at
work, and social support at work, predict job satisfaction.
These are all of the variables included in Figure 6. These are the major
predictors of job satisfaction.
195
3.5
3.5.1
Method
Participants
The sample consisted of 108 university academic staff, and 97 secondary
school teachers. The academics were obtained from one university, whereas the
secondary school teachers were obtained from 20 Government schools. For the
academics, the response rate was 21%. The response rate of the teachers could not
be calculated as the questionnaires were collected from the staff room only if the
teachers were interested in completing the survey. The demographic characteristics
of the sample are displayed in Table 20. The bolded values demonstrate where the
largest proportion of the sample lies, which tends to be fairly consistent across the
groups.
196
Table 20- Demographics of the Academics and Teachers
Variable
Gender
Age
Years in Occupation
Hours worked per week
3.5.2
Male
Female
18-25
26-35
36-45
46-55
56+
0-5
6-10
11-15
16-20
20+
21-30
31-40
41-50
51-60
61+
% Academic
49
51
0
8.3
32.4
44.4
14.8
13
27.8
22.2
7.4
29.6
6.5
7.4
47.2
32.4
6.5
% Teachers
47
53
2.1
15.5
37.1
39.2
6.2
11.3
11.3
13.4
18.6
45.4
4.1
12.4
45.4
27.8
10.3
Materials
Both the academics and the teachers received a plain language statement
(refer to Appendix I) and an anonymous questionnaire. The questionnaire consisted
of several scales, which measured job autonomy, need for job autonomy, primary
and secondary control, work difficulties, job satisfaction, life satisfaction, personality
and social support at work.
3.5.2.1
Job Autonomy
As discussed in the rationale for study two, a global measure of job autonomy
was administered. This scale developed by Hackman and Oldham (1975) is part of a
larger scale, the Job Diagnostic Survey. This scale is the most commonly used
instrument for measuring job autonomy (Spector, 1986). It consists of three items
197
that assess overall perceived job autonomy, such as “in my job, I can decide on my
own how to go about doing my work” (refer to Appendix J).
Although the psychometric properties of the scale have been questioned in the
past (Fried, 1991; Fried and Ferris, 1991), a major review which examined 15 years
of empirical research on the psychometric properties of the scale provided some
support. Taber and Taylor (1990) demonstrated that the average test-retest
correlations for the scale were moderate (r = 0.63), internal consistency was
moderate (0.69), and there was good discriminant validity.
Although these psychometric statistics are not exceptional, the use of the
scale has been supported in a recent review conducted by Boonzaier, Flicker and
Rust (2001). Furthermore, it must be noted that as mentioned by Breaugh (1989), a
better alternative is not available. Breaugh (1989, Breaugh, 1998) actually
developed a new measure of job autonomy, however this scale was deemed not to be
appropriate for this study as like Ganster’s (1989, cited in Dwyer & Ganster, 1991)
scale, it is multidimensional. As such, this study used the autonomy items of the Job
Diagnostic Survey (Hackman & Oldham, 1975). In this study, Cronbach’s Alpha
was 0.83.
3.5.2.2
Need for Job Autonomy
As there are only a few studies that have examined need for job autonomy,
the measures of need for job autonomy were reviewed. First, Fung-Kam (1998)
tested preference for job autonomy using Edwards (1959) Personal Preference
Schedule. This scale consists of 28 sets of paired statements representing different
198
personality traits and a score is given to the respondent who chooses the statement
representing the personality trait of need for autonomy. The major problem with this
scale is that it refers to general autonomy, and not specifically to autonomy at work.
Another need for job autonomy scale is Algera’s (1981, cited in Landeweerd
& Boumans, 1994) scale. This scale asks the respondents to rate the attractiveness of
various work situations. Although this scale may have been adequate, to date it has
only been published in Dutch, and as such was not viable.
An exploratory scale was developed by de Rijk et al., (1998). This scale
consists of four items which examine how important it is for the person to set the
pace of their tasks, have control over what they do at work and the way that they do
it, doing their own planning at work, and giving orders instead of receiving them.
This scale was selected for the current study even though psychometric statistics
have not been produced, as the items have face validity. These items were rated on a
10-point scale, ranging from 1 (not at all important) to 10 (could not be more
important; refer to Appendix K).
As this scale is exploratory, a factor analysis was conducted on the scale to
ensure that the items were measuring need for job autonomy. The assumptions were
met, where Bartlett’s test of sphericity was large and significant, and Kaiser-MeyerOlkin (KMO) measure of sampling adequacy exceeded 0.6. A principal components
analysis with direct oblimin rotation yielded one factor. Examination of the
eigenvalues however demonstrated that the second factor had an eigenvalue of 0.999,
and as such a two-factor solution was tested. This analysis, displayed in Table 21,
demonstrates that item four (i.e., “How important is it for you to give orders instead
199
of receiving them”) loaded on the second factor. Item four is different to the other
three items as it may also measure need for authority. As a result, item four was
deleted from the scale. When all four items were included in the scale, Cronbach’s
alpha was low (0.56), however when item four was deleted, Cronbach’s alpha was
adequate (0.77).
Table 21- Factor Analysis of the Need for Job Autonomy Scale
No.
1
2
3
4
Item
How important is it for you to set the pace of your
tasks at work.
How important is it for you to have control over
what you do at work and the way that you do it.
How important is it for you to do your own
planning at work.
How important is it for you to give orders to work
instead of receiving them.
Eigenvalues
% of variance
Cumulative variance
Cronbach's Alpha (for revised scale)
F1
0.83
F2
0.87
0.80
0.99
2.098
52.45
52.45
0.77
0.999
24.98
77.43
Loadings less than 0.40 are excluded; Bolded items are included in the scale
3.5.2.3
Primary control and Secondary Control
As discussed in the rationale, the Primary and Secondary Control Scale
developed by Heeps et al., (2000) was revised for this study (Maher et al., 2001).
The scale now includes four primary control items and 12 secondary control items
(refer to Appendix H). These items are rated on a 11-point scale ranging from 0
(never) to 10 (every time). Although the control strategies were aggregated in study
one, it now appears that this scoring method is flawed. The items cannot be
aggregated as people may use one strategy all the time, and never use the others.
200
Using the average, they would receive a score that is not representative of the
frequency of secondary control strategies (i.e., every time). As such, an alternative
solution used here is to record the highest frequency for primary control strategies
and the highest frequency for secondary control strategies.
3.5.2.4
Work Difficulties
Work difficulties were measured in the Primary and Secondary Control Scale
(Maher et al., 2001; refer to Appendix H). Prior to assessing how the employees deal
with their work difficulties, the scale assesses the frequency of work difficulties.
Specifically, the item is “how often do you have difficulty doing something at work.”
The rating scale ranges from 1 (never) to 10 (all the time).
3.5.2.5
Job Satisfaction
Two scales of job satisfaction were administered; a facet scale and a global
scale. The facet scale was changed from the Job Descriptive Index (JDI) in study
one to the Minnesota Satisfaction Questionnaire (MSQ; Weiss et al., 1967; refer to
Appendix L). The MSQ was used because unlike the JDI, which only examines five
facets of the job, the MSQ examines 20 facets. It was thought that a greater
understanding of the groups could be obtained by using the MSQ. Furthermore, the
items in the MSQ can be aggregated to measure intrinsic and extrinsic job
satisfaction. Intrinsic job satisfaction refers to how people feel about the nature of
the tasks, whereas extrinsic job satisfaction refers to how people feel about aspects of
the work situation that are external to the work itself (Spector, 1997). This scale has
201
adequate reliability where Cronbach’s alpha ranges from 0.82 to 0.88 and
discriminant validity has been demonstrated (Hirschfeld, 2000).
The facet measure is only useful to gain insight into the teachers’ and
academics’ level of job satisfaction. It cannot be used as the dependent variable
however as a facet scale cannot be aggregated (Ironson et al., 1989). It may exclude
areas that are important to the respondent, or include areas that are unimportant to the
respondent. As such, a one-item measure of job satisfaction was used as the
dependent variable. Although internal consistency cannot be established with a
single-item measure, the single item measure of job satisfaction has been shown to
correlate with other measures of job satisfaction, where r = 0.63 (Wanous et al.,
1997).
3.5.2.6
Life Satisfaction
As in study one, the subjective dimension of the Comprehensive Quality of
Life Scale (Com-QOL) developed by Cummins (1997) was used to assess
satisfaction with seven domains of life, including material well-being, health,
productivity, intimacy, safety, community and emotional well-being (refer to
Appendix E). An 11-point scale was utilised, ranging from 0 (completely
dissatisfied) to 10 (completely satisfied)
3.5.2.7
Personality
The extroversion and neuroticism subscales of the NEO Five Factor
Inventory, developed by Costa and McCrae (1992) were used to measure personality.
202
This scale, discussed in study one, contains 12 items to measure extroversion and 12
items to measure neuroticism (refer to Appendix F). Convergent and discriminant
validity of both of these personality factors has been established (Costa & McCrae,
1992, Leong & Dollinger, 1991; Tinsley, 1994).
3.5.2.8
Social Support at Work
Social support at work was measured by Karasek and Theorell’s (1990) scale
which has two components; supervisor support and co-worker support (refer to
Appendix M). Each component is measured by 4 items, and rated on a scale from
1 (not true at all) to 10 (could not be more true). Two items measure emotional
support, and two measure instrumental support. Emotional support measures the
degree of social cohesion in the work group, whilst instrumental support measures
the amount of assistance given with work tasks. Although the scale measures
emotional and instrumental support, the four items are summed to provide an overall
support score.
The items in the scale were changed slightly to ensure that they referred to
the employee. Some of them were quite ambiguous, such as “my supervisor shows
concern” and “my supervisor pays attention.” As these items could be interpreted in
regard to work tasks or other employees, they were changed to “my supervisor shows
concern for me” and “my supervisor pays attention to me.”
Past studies using the original scale have demonstrated that the scale has
adequate reliability with Cronbach’s alpha ranging from 0.69 to 0.89 (Karasek et al.,
1998), and 0.81 to 0.87 (Pelfrene, Vlerick, Mak, De Smets, Kornitzer & De Backer,
203
2001). Furthermore, factor analyses have demonstrated that the supervisor support
items load on a different factor to the co-worker support items (Pelfrene et al., 2001).
3.5.3
Procedure
Ethics approval was obtained from Deakin University, and the Department of
Education, Employment and Training (DEET). Consent was obtained from the
Heads of School to recruit the academics, and from the Principals for the teachers.
The recruiting procedure differed depending on the group. Five hundred academics
within one University were sent a questionnaire package. If they chose to participate
in the study, they completed the questionnaire and returned it using a reply paid
envelope. For the teachers, each Principal that agreed to assist with the study was
sent 10-15 questionnaires. They then discussed the questionnaires in their staff
meetings, and left them in the staff room for the teachers to collect. On occasion, the
Principals chose to distribute the questionnaires to a selection of staff members.
These questionnaires were sent back to Deakin University using a reply-paid
envelope. At the conclusion of the study, the participating Heads of School and the
Principals received a summary of the results.
204
3.6
3.6.1
Results
Data Screening and Checking of Assumptions
The data set for each occupational group was initially examined for missing
values, acquiescence, outliers, normality and linearity. Less than 5% of the values
for academics and teachers were missing for any one item. As there was no pattern
to these missing values, they were, as in study one, replaced with the group mean.
Univariate outliers were identified in the primary and secondary control scale (5), the
job autonomy scale (1), the facet job satisfaction scale (2), and the life satisfaction
scale (18). These values were recoded to lie within three standard deviations of the
mean.
Normality was assessed using the skew/standard error<3,
Kolmogorov-Smirnof values, frequency histograms, and normal probability plots.
For the academics, job autonomy (-5.76), and co-worker support (-3.24) were mildly
negatively skewed. For the teachers, job satisfaction (-3.79), supervisor support
(-3.44), and co-worker support (-5.16) were mildly negatively skewed. As in study
one, these variables were not transformed as transformations are not recommended
for data that are mildly and naturally skewed (Tabachnick & Fidell, 1996). Rather,
these were examined using the more conservative alpha level of 0.01. Linearity was
assessed through bivariate scatterplots, and these appeared to demonstrate reasonable
linear relationships.
205
3.6.2
Descriptive Statistics and Inter-Correlations
Table 22 contains the means and standard deviations for the major variables
in the study for each occupational group. Whilst the teachers reported lower job
autonomy than the academics, they reported similar levels of job satisfaction, and
primary and secondary control. Table 23 displays the correlations among all of the
major variables for the academics and the teachers. This table demonstrates that
although job autonomy is correlated with job satisfaction, primary and secondary
control strategies are not.
Table 22- Means and Standard Deviations of Major Variables for Academics
and Teachers
Variable
Job Satisfaction-One item
Intrinsic Job Satisfaction
Extrinsic Job Satisfaction
Job Autonomy
Primary Control
Secondary Control
Life Satisfaction
Neuroticism
Extroversion
Supervisor Support
Co-worker Support
Difficulties at work
Academics
M
SD
64.09
21.72
72.76
13.24
44.46
20.45
74.93
16.81
81.58
14.42
83.33
13.37
74.20
11.34
35.07
16.60
61.62
13.06
44.57
29.44
71.11
18.90
49.59
24.08
Teachers
M
SD
68.79
20.23
77.07
14.80
56.15
20.89
66.32
17.79
80.64
15.82
82.93
11.90
75.61
14.48
33.33
16.04
63.92
15.69
64.31
26.56
77.49
15.57
46.05
23.01
All scores have been converted to a percentage of scale maximum (%SM) which
ranges from 0-100. The formula is (mean score for the original domain-1) x 100/
(number of scale points –1).
206
Table 23- Inter-Correlations for the Academics and Teachers
JS
JS
JA
PC
SC
LS
N
E
SS
CS
Di
JA
0.51**
0.37**
-0.07
-0.06
0.38**
-0.38**
0.22*
0.46**
0.42**
-0.31**
0.12
0.00
0.20*
-0.38**
0.16
0.26**
0.15
-0.20*
PC
0.14
0.06
0.40**
0.09
-0.07
0.20*
-0.15
0.07
-0.07
SC
0.14
0.06
0.56**
0.03
0.03
0.13
-0.09
0.05
0.03
LS
0.46**
0.38**
0.15
0.27**
-0.48**
0.26**
0.19*
0.23**
-0.33**
N
-0.29**
-0.14
-0.19
-0.08
-0.39**
-0.44**
-0.06
-0.31**
0.25**
E
0.23*
0.17
0.01
0.19*
0.52**
-0.32**
0.08
0.38**
-0.09
SS
0.64**
0.39**
0.01
-0.05
0.36**
-0.15
0.16
0.39**
-0.13
CS
0.39**
0.34**
-0.14
-0.06
0.26**
0.03
0.18
0.55**
-0.26**
* p<0.05 , ** p>0.01; Correlations for teachers are bolded.
JS = Job satisfaction; JA = Job autonomy; PC = Primary control; SC = Secondary
control; LS = Life satisfaction; N = Neuroticism; E = Extroversion; SS = Supervisor
support; CS = Co-worker support; Di = Difficulties at work
3.6.3
Preliminary Examination of the Primary Control and Secondary
Control Scale
As the primary and secondary control scale is exploratory, it will be
examined here before the hypotheses are tested. The descriptive statistics displayed
in Table 22 and 23 indicate that primary and secondary control did not behave as
expected. Both the academics and the teachers reported high levels of primary
control (M = 82%SM, M = 81%SM), and secondary control (M = 83%SM,
M = 83%SM). Furthermore, the control strategies did not correlate with job
satisfaction. One interesting finding however is that primary control was positively
correlated with secondary control for both groups (r = 0.40, r = 0.56).
Overall however, these statistics are inconsistent with study one, where the
supermarket workers (M = 46%SM) reported significantly higher levels of secondary
207
control than the academics (M = 36%SM). Furthermore, primary control was
moderately correlated with job satisfaction (r = 0.31, r = 0.44).
A major difference between these two studies is the edition of the Primary
and Secondary Control Scale. The scale was changed for study two, where a new
scoring procedure was implemented. In the first study, the items were simply
aggregated for each control strategy, however in the current study, the highest
frequency for primary and secondary control was recorded. This method does not
appear to have been successful however in differentiating the respondents. The
frequency distribution, displayed in Table 24, demonstrates that 76% of the subjects
reported a level of primary control between 77%SM and 100%SM, and that 84% of
the subjects reported a level of secondary control between 77%SM and 100%SM.
This range is concerning, suggesting that there may have been a ceiling effect.
Table 24- Frequency of Primary and Secondary Control
Value
33.33
44.44
55.56
66.67
77.78
88.89
100.00
Primary Control Frequency
%
2.0
2.4
4.4
15.1
25.4
32.2
18.5
Secondary Control Frequency %
0.5
1.5
3.4
9.3
33.7
32.7
19.0
In order to examine how much the scoring procedure in this study influenced
the resulting levels of primary and secondary control, the Primary and Secondary
Control Scale was also examined as in study one, where the average was calculated.
Before the items were aggregated, a factor analysis was conducted on the scale.
208
The assumptions were met, where Bartlett’s test of sphericity was large and
significant, and Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy
exceeded 0.60. A principal components analysis with direct oblimin rotation yielded
four factors. These four factors accounted for 57% of the variance. Only 9 of the 16
items loaded on only one factor. Two of the primary control items loaded on the first
factor, however a secondary control item also loaded on this factor. The remaining
secondary control items (i.e., 6) were equally distributed among the second, third and
fourth factor. There was no pattern to these items, and the four-factor solution could
not be interpreted. The analysis was repeated requesting three factors to determine if
a three-factor solution could be useful. This analysis was no clearer however, with
the primary control and secondary control items loading on all three factors. The
secondary control items that loaded on different factors did not appear to be
measuring different functions of secondary control.
In response to these analyses, a two-factor solution was requested. This
analysis provided a much clearer solution, with the four primary control items
loading on the first factor, and six of the 12 secondary control items loading on the
second factor. It must be noted however, that this two-factor solution only accounted
for 40% of the variance. Another factor analysis was conducted with only the bolded
items in Table 25. This analysis demonstrated that the two factors accounted for
49% of the variance.
209
Table 25- Factor Analysis of the Revised Primary and Secondary Control Scale
No.
pc1
pc2
pc3
pc4
sc1
sc2
sc3
sc4
sc5
sc6
sc7
sc8
sc9
sc10
sc11
sc12
Item
I looked for different ways to overcome it.
I kept trying.
I worked to overcome it.
I worked out how to remove obstacles.
It will work out okay in the end.
I knew it would happen.
I can’t always get what I want.
It doesn’t matter.
I am better off than many other people.
It was not my fault.
I told someone about it.
I thought of the success of my family or friends.
I thought about my success in other areas.
I did something different, like going for a walk.
I ignored it.
I looked for something else that was positive in the
situation.
Eigenvalues
% of variance
Cumulative variance
Cronbach's Alpha (for revised scale)
F1
0.67
0.77
0.82
0.59
0.41
0.40
F2
0.53
0.71
0.68
0.64
0.40
0.59
0.61
0.58
0.41
0.54
0.71
4.41 2.11
26.98 13.19
27.59 40.78
0.82
Loadings less than 0.40 are excluded; Bolded items are included in the scale
When the control items were aggregated rather than separated into the highest
score, both groups still reported similar levels of the control strategies. For primary
control, the academics reported a mean of 79.09, whilst the teachers reported 80.18.
For secondary control, the academics reported 46.21, and the teachers reported
46.56. Furthermore, the control strategies were not strongly related to job
satisfaction. Primary control was not related at all to job satisfaction, whilst
secondary control was slightly negatively related to job satisfaction (r = -0.24).
210
These descriptive statistics suggest that even if the scales were aggregated, the results
would still not be significant.
As the scoring procedure does not appear to have drastically altered the
results, the hypotheses will be tested using the intended scoring procedure
(i.e., highest number). This scoring method, although problematic because of its
small range, is theoretically superior to the aggregated measure. However,
preliminary analyses using this scoring method clearly demonstrate that primary and
secondary control strategies are not behaving as expected, and as such, many of the
hypotheses will not be supported. In order to reduce the repetitiveness of these
findings, the hypotheses examining primary and secondary control will be examined
collectively. Specifically, this refers to hypotheses 2, 3, 4, 5, and 6.
211
3.7
Hypothesis Testing
In order to test the proposed model of job satisfaction, multivariate analyses
of variance were conducted to compare the levels of control strategies, job
autonomy, job satisfaction and life satisfaction reported by the academics and the
teachers. Multiple regression analyses were also conducted to examine the major
predictors of job satisfaction, and the moderating role of need for autonomy and
social support at work. As in study one, the alpha level was reduced to 0.01 in order
to reduce the risk of Type I errors.
3.7.1
Hypothesis One: Levels of Job Autonomy and Job Satisfaction
In order to test the first part of hypothesis one, proposing that job autonomy is
positively related to job satisfaction, the correlation coefficients for each
occupational group were examined. Consistently, job autonomy was positively
related to job satisfaction for both the academics (r = 0.37) and the teachers
(r = 0.51).
In order to test the second part of hypothesis one, proposing that the
academics would report higher job autonomy than the teachers, an analysis of
variance was employed. The assumption of univariate homogeneity of variance,
using Levene’s test, was met, F (1, 203)= 2.58, p > 0.05. The univariate test of
significance demonstrated that, as hypothesised, the academics (M = 75.13,
SD = 16.12) reported significantly higher job autonomy than the teachers,
(M = 66.34, SD = 17.74), F (1, 203) = 13.82, p = 0.00.
212
3.7.2
Hypotheses Two and Three: Examining how Job Autonomy Influences
the Amount of Primary and Secondary Control Strategies
Hypotheses two and three examine how job autonomy influences the amount
of control strategies that employees use. In order to test hypothesis two, that the
academics would report more primary control and less secondary control than the
teachers, a multivariate analysis of variance was employed. The variables were
normally distributed, and reasonably linear relationships were evident. There was no
evidence of multicollinearity as the determinant of the within-cell correlation
> 0.0001 (i.e., 0.774). Univariate homogeneity of variance, as assessed through
Levene’s test was met for primary control, F (1, 203) = 0.50, p > 0.05, and secondary
control, F (1, 203) = 1.68, p > 0.05. The assumption of multivariate homogeneity of
variance was also met using Box’s M test.
The multivariate test using Pillai’s criterion was not significant,
F (2, 202) = 0.09, p = 0.91. Inconsistent with hypothesis two, the academics
(M = 81.58, SD = 14.42) reported similar level of primary control as the teachers
(M = 80.64, SD = 15.81), F (1, 203) = 0.20, p = 0.66. Furthermore, the academics
(M = 83.33, SD = 13.37) reported similar levels of secondary control as the teachers
(M = 82.93, SD = 11.90), F (1, 203) = 0.05, p = 0.82.
In order to test hypothesis three that job autonomy is positively related to
primary control and negatively related to secondary control, job autonomy was
correlated with primary and secondary control. This analysis demonstrated that,
213
inconsistent with hypothesis three, primary control (r = 0.09) and secondary control
(r = 0.03) were not significantly related to job autonomy.
3.7.3
Hypotheses Four and Five: Examining how Job Autonomy Influences
the Relationship Between the Control Strategies and Job Satisfaction
Hypothesis four and five test the proposal that job autonomy influences the
relationship between job autonomy and job satisfaction. In order to test hypothesis
four, proposing that primary control will be more positively related to job satisfaction
than secondary control for the academics, and secondary control will be more
positively related to job satisfaction than primary control for the teachers, two
standard multiple regression analyses were conducted.
The assumptions of normality, linearity, and homoscedasticity of residuals, as
assessed through examination of the residual scatterplots, were met for both groups.
As demonstrated in Table 26, R was not significantly different from zero for the
academics, R = 0.08, F (2, 105) = 0.32, p = 0.73, or for the teachers, R = 0.16,
F (2, 94) = 1.16, p = 0.32. Inconsistent with hypothesis four, primary and secondary
control were not related to job satisfaction for either group.
214
Table 26- Multiple Regression of Primary and Secondary Control on Job
Satisfaction for Academics and Teachers
Group
Acad
IV
JS
PC
SC
-0.07
-0.06
PC
B

sr2 (unique)
0.40
-0.08
-0.07
-0.05
-0.04
0.22
0.14
R =0.08
R2=0.006
AdjR2=-0.013
0.12
0.15
0.09
0.09
0.55
0.50
R =0.16
R2=0.02
AdjR2=-0.003
Teach
PC
SC
0.14
0.14
0.56
Acad – Academics; Teach – Teachers; PC - Primary control; SC – Secondary
control; JS – Job satisfaction
For the academics, R is composed of 0.36% unique variance and 99.64% shared
variance. For the teachers, R is composed of 1.05% unique variance and 98.95%
shared variance.
Hypothesis five, similar to hypothesis four, examines how job autonomy
influences the adaptiveness of the control strategies, however it is based on the
measured level of job autonomy rather than the presumed level. Hypothesis five
proposes that the relationship between the control strategies and job satisfaction is
moderated by job autonomy. As discussed in study one, job autonomy is a
moderator rather than a mediator as it specifies when certain effects will hold. That
is, when job autonomy is high, primary control will be more strongly correlated with
job satisfaction, and when job autonomy is low, secondary control will be more
strongly correlated with job satisfaction.
215
In order to test the moderating effect of job autonomy on primary control and
secondary control, two hierarchical multiple regression analyses were conducted. In
these analyses the control strategy was entered first, then job autonomy, and then the
interaction term. The assumptions of normality, linearity, and homoscedasticity of
residuals, as assessed through examination of the residual scatterplots were met.
As demonstrated in Table 27, for primary control, R was not significantly different
from zero after the first step (i.e., primary control), R = 0.03, F (1, 203)= 0.15,
p = 0.70. The addition of job autonomy did result in an increase in R, where
R = 0.39, Finc (1, 202) = 36.29, p = 0.00. However, the addition of the interaction
term was not significant, R = 0.39, Finc (1, 201) = 0.12, p = 0.73.
For secondary control, R was not significantly different from zero after the
first step, R = 0.02, F (1, 203)= 0.09, p = 0.77. After job autonomy was entered, the
value of R increased, R = 0.39, F (1, 202) = 36.35, p = 0.00, however the addition of
the interaction term in step three was not significant, R = 0.39, F (1, 201) = 0.21,
p = 0.65. Thus, hypothesis five was not supported.
216
Table 27- Hierarchical Multiple Regression testing the Moderating role of the
Control Strategies on the Relationship Between Job Autonomy and Job
Satisfaction
Step
1
2
3
Step
1
2
3
IV
Primary control
Primary control
Job Autonomy
Primary control
Job autonomy
Primary control x job
autonomy
IV
Secondary control
Secondary control
Job Autonomy
Secondary control
Job autonomy
Secondary control x job
autonomy
DV
JS
DV
JS
B
0.04
-0.01
0.48
0.12
0.62
-0.001

0.03
-0.009
0.39
0.08
0.51
-0.16
sr2 (unique)
R =0.39
R2=0.15
AdjR2=0.14
sr2 (unique)
0.03
0.01
0.47
0.23
0.72
-0.002

0.02
0.009
0.39
0.14
0.59
-0.24
R =0.39
R2=0.15
AdjR2=0.14
15.21**
15.21**
*p<0.05, ** p<0.01; JS – Job satisfaction
3.7.4
Hypothesis Six: Examining the Proposed Explanation for the
Relationship Between Job Autonomy and Job Satisfaction
In order to test hypothesis six, which proposes that the relationship between
job autonomy and job satisfaction is mediated by the control strategies, a hierarchical
multiple regression analysis was conducted.
217
The assumptions of normality, linearity and homoscedasticity of residuals
were met, and there was no evidence of multicollinearity. As demonstrated in Table
28, R was not significantly different from zero after primary and secondary control
were entered, R = 0.03, F (2, 202) = 0.08, p = 0.92. R did significantly increase after
job autonomy was added to the equation, R = 0.39, Finc (3, 201)= 36.16, p = 0.00.
Only job autonomy predicted job satisfaction accounting for 15% of the variance in
job satisfaction. As such, when primary and secondary control were controlled for,
job autonomy still predicted job satisfaction. Inconsistent with hypothesis six,
primary and secondary control did not mediate the relationship between job
autonomy and job satisfaction.
Table 28- Hierarchical Multiple Regression Testing the Mediating Role of the
Control Strategies
Step
1
2
IV
Primary control
Secondary control
Primary control
Secondary control
Job autonomy
DV
JS
JS
**p>0.01; JS – Job satisfaction
B
0.03
0.02

0.02
0.01
R =0.03
R2=0.001 AdjR2=-0.009
-0.03
0.03
0.48
-0.02
0.02
0.39
15.21**
R =0.39**
R2=0.15
AdjR2=0.14
sr2(unique)
218
3.7.5
Hypothesis Seven: Occupational Differences in Job Satisfaction and
Life Satisfaction
In order to test hypothesis seven, that academics would report higher job
satisfaction than the teachers, an analysis of variance was conducted on the one-item
measure of job satisfaction. Years working in occupation and age were entered as
covariates in this analysis as the academics tended to be older, and had worked less
years than the teachers (i.e., refer to Table 20). Life satisfaction was normally
distributed, and reasonably linear relationships were evident. Univariate
homogeneity of variance, as assessed through Levene’s test was met,
F (1, 203)= 0.33, p > 0.05. The univariate test demonstrated that the academics
(M = 64.09, SD = 21.72) reported similar levels of job satisfaction as the teachers
(M = 68.79, SD = 20.23), F (1, 201) = 3.39, p = 0.07.
Although the one-item measure of job satisfaction is the dependent variable
in this study, the facet measure of job satisfaction was also explored to gain a greater
understanding of the two occupational groups. A multivariate analysis of variance
was conducted on the intrinsic and extrinsic facets of job satisfaction. The variables
were normally distributed, and reasonably linear relationships were evident. There
was no evidence of multicollinearity as the determinant of the within-cell correlation
> 0.0001 (i.e., 0.526). Univariate homogeneity of variance, as assessed through
Levene’s test was met for intrinsic job satisfaction, F (1, 203) = 1.63, p > 0.05, and
extrinsic job satisfaction, F (1, 203)= 0.03, p > 0.05. The assumption of multivariate
homogeneity of variance was also met using Box’s M test.
219
The multivariate test, using Pillai’s criterion was significant,
F (1, 199) = 5.96, p = 0.00. Examination of the univariate tests demonstrated that the
teachers (M = 56.15, SD = 20.89) reported higher extrinsic job satisfaction than the
academics (M = 44.46, SD = 20.45), F (1, 203)= 16.35, p = 0.00. The teachers
(M = 77.07, SD = 14.80) also reported higher intrinsic job satisfaction than the
academics (M = 72.76, SD = 13.24), F (1, 203)= 4.84, p = 0.03, however this finding
was not significant as the more stringent alpha level of 0.01. The means and
standard deviation for the intrinsic and extrinsic items for academics and teachers are
provided in Table 29.
220
Table 29- Means and Standard Deviations of the Intrinsic and Extrinsic Job
Satisfaction Items for Academics and Teachers
Item
Being able to keep busy all the time
The chance to work alone on the job
The chance to do different things from time
to time
The change to be somebody in the
community
The way my boss handles his/her work
The competence of my supervisor in
making decisions
Being able to do things that don’t go against
my conscience
The way my job provides for steady
employment
The chance to do things for other people
The chance to tell people what to do
The chance to do something that makes use
of my abilities
The way company policies are put into
practice
My pay and the amount of work that I do
The chance for advancement on the job
The freedom to use my own judgement
The chance to try my own methods of doing
the job
The praise I get for doing a good job
The feeling of accomplishment I get from
the job
Academics
M
SD
71.70
25.35
80.45
16.49
77.77
21.69
Teachers
M
SD
81.21
19.60
73.88
22.51
78.24
21.75
65.74
24.87
69.64
21.12
47.32
51.95
30.62
30.60
67.35
70.10
25.85
24.23
68.72
25.01
78.92
21.90
77.47
25.21
87.51
21.35
77.77
59.16
77.05
18.10
22.52
20.05
87.05
67.12
80.06
14.85
20.34
20.15
30.34
24.28
48.68
24.91
49.59
43.41
74.18
73.25
26.46
28.64
20.42
20.50
45.02
53.15
72.62
76.51
28.79
29.88
22.22
21.51
44.14
69.86
30.96
22.73
52.57
72.05
28.31
21.76
Bolded items measure extrinsic job satisfaction. Non-bolded items measure
intrinsic job satisfaction.
In order to test the second part of hypothesis seven, that the academics would
report higher life satisfaction than the teachers, a univariate analysis of variance was
employed. Life satisfaction was normally distributed using the skewness/standard
221
error < 3 criterion. The assumption of univariate homogeneity of variance, as
assessed by Levene’s test, was not met, F (1, 203) = 0.96, p < 0.05, and as such, this
analysis proceeded with caution using an alpha level of 0.05.
The univariate test of significance demonstrated that, inconsistent with
hypothesis seven, there were no occupational differences in levels of life satisfaction,
F (1, 203) = 0.51, p = 0.11. As demonstrated in Table 30, the teachers average level
of life satisfaction was 74.21 (SD = 11.34) and the academics average level was
75.61 (SD = 14.48). The means and standard deviations for the seven domains of
life satisfaction are also presented to demonstrate that the two groups appear to be
more satisfied with safety, intimacy and material well-being, and less satisfied with
health and community.
Table 30- Means and Standard Deviations of the Domains of Life Satisfaction
for Academics and Teachers
Domain
Material Satisfaction
Health Satisfaction
Productivity Satisfaction
Intimacy Satisfaction
Safety Satisfaction
Community Satisfaction
Emotional Satisfaction
Overall life satisfaction
Academics
M
SD
76.87
17.91
68.73
21.37
75.03
14.58
76.77
19.98
81.80
16.87
70.24
18.72
72.91
18.79
74.35
11.16
Teachers
M
SD
76.08
16.92
67.35
23.22
73.03
18.84
78.97
20.33
84.56
16.67
72.98
20.57
76.59
19.62
75.64
14.45
222
3.7.6
Hypothesis Eight: Examining how Social Support at Work Moderates
the Relationship between Difficulties at Work and Job Satisfaction
Hypothesis eight proposes that social support at work moderates the effect of
work difficulties on job satisfaction. Social support at work is proposed to be a
moderator, that is, a variable that affects the direction and/or strength of the
relationship between an independent variable (i.e., work difficulties) and a dependent
variable (i.e., job satisfaction). It is a moderator rather than a mediator because it
affects the relationship between work difficulties and job satisfaction, however it
does not explain why work difficulties and job satisfaction are related.
In order to test this moderation effect, a hierarchical multiple regression is
required. In the first step the independent variable is entered (i.e., work difficulties).
In the second step the moderator variable (i.e., social support) is entered. Finally, in
the third step the interaction term is entered (i.e., independent variable multiplied by
the moderator variable). Moderator effects are evident if the interaction term
predicts the dependent variable after the independent variable and the moderator
variables have been entered in steps one and two.
Two hierarchical multiple regression analyses were conducted for supervisor
support, and co-worker support. For both analyses, the assumptions of normality,
linearity, and homoscedasticity of residuals were met, and there was no evidence of
multicollinearity.
223
The moderating role of supervisor support on the relationship between work
difficulties and job satisfaction was tested first. R was significantly different from
zero at the end of the first step (i.e., work difficulties), R = 0.26, F (1, 203) = 14.94,
p = 0.00. The addition of supervisor support resulted in a significant increment in
R2, where R = 0.57, Finc (1, 203) = 78.86, p = 0.00. Supervisor support accounted
for 26% of the variance in job satisfaction. As demonstrated in Table 31, the
addition of the interaction term (difficulties x supervisor support) did result in a
significant increment in R2, where R =0.59, Finc (1, 201) = 4.29, p = 0.04. It must be
noted that this finding was not significant using the more stringent alpha level of
0.01.
Table 31- Hierarchical Multiple Regression Analysis Examining if Supervisor
Support Moderates the Relationship between Work Difficulties and Job
Satisfaction
Step
IV
DV
B

sr2 (unique)
1
Freq. of Diff
JS
-0.23
-0.26
6.86**
R =0.26**
R2=0.07
Adj R2=0.06
-0.18**
0.37**
-0.20
0.52
3.84
26.11
R =0.57**
R2=0.33
Adj R2=0.32
-0.36
0.16
0.04
-0.40
0.22
0.36
4.04**
R =0.59*
R2=0.34
Adj R2=0.33
2
3
Freq. of Diff
Supervisor
Freq. of Diff
Supervisor
Diff x Sup
JS
JS
p<0.05 , ** p>0.01; JS – Job satisfaction
1.39*
224
Although not significant at 0.01, the interaction between difficulties and
supervisor support will be examined further for two reasons. First, only a few
studies have examined the moderating role of social support at work, and as such, the
pattern of the interaction requires investigation. Second, it is difficult to achieve
statistical significance in moderation analyses as the power is low (Bobko, 2001).
By having the independent variable, the moderator and the interaction term
(independent variable x moderator), there is an increased chance of multicollinearity
(Bobko, 2001). As the correlation between predictors increases, the standard
deviation of the regression weights increases, and it becomes less likely that the null
hypothesis will be rejected. In order to reach significance therefore, the analysis
needs to have large effects of large sample sizes (Bobko, 2001). Thus, as the
analysis was significant at 0.05, it will be examined further.
Work difficulties were regressed on job satisfaction separately for those with
low supervisor support, and those with high supervisor support. As proposed by
Cohen and Cohen (1983), the low and high distinction was defined as scores that fell
one standard deviation below or above the mean for supervisor support. As
demonstrated in Figure 7, the regression lines were consistent with the hypothesis,
where the slope of the regression line of work difficulties on job satisfaction was
steeper for high supervisor support than for low supervisor support.
225
Figure 7 - Relationship Between Work Difficulties and Job Satisfaction for
Employees with Low/High Supervisor Support
100
90
80
y = -0.4973x + 87.963
Job Satisfaction
70
Low Supervisor Support
High Supervisor Support
60
50
y = -3.9712x + 72.123
40
30
20
10
0
0
2
4
6
Work Difficulties
8
10
226
In order to test the moderating role of co-worker support, another hierarchical
multiple regression was conducted. R was significantly different from zero at the
end of the first step (i.e., work difficulties), R = 0.26, F (1, 202)= 14.91, p = 0.00.
The addition of co-worker support resulted in a significant increment in R2, where
R = 0.46, Finc (1, 202)= 36.78, p = 0.00. As demonstrated in Table 32, co-worker
support accounted for 7% of the variance in job satisfaction. The interaction term
(difficulties x co-worker support) did not result in a significant increment in R2,
R = 0.46, Finc (1, 202) = 0.06, p = 0.81. As such, co-worker support does not appear
to moderate the effect of work difficulties on job satisfaction.
Table 32- Hierarchical Regression Analyses examining whether Co-worker
Support Moderates the Relationship between Work Difficulties and Job
Satisfaction.
Step
IV
DV
B

sr2 (unique)
1
Freq. of Diff
JS
-0.23
-0.26
6.86**
R =0.26**
R2=0.07
Adj R2=0.06
-0.18
0.46
-0.20
0.38
4.00**
14.36**
R =0.46**
R2=0.21
Adj R2=0.20
-0.13
0.50
-0.008
-0.14
0.42
-0.07
0.12
2.25*
0.023
R =0.46
R2=0.21
Adj R2=0.20
2
3
Freq. of Diff
Co-worker
Freq. of Diff
Co-worker
Diff x Co-worker
JS
JS
p<0.05 , ** p>0.01; JS – Job satisfaction
227
3.7.7
Hypothesis Nine: The Moderating Role of Need for Autonomy on the
Relationship Between Job Autonomy and Job Satisfaction
In order to test hypothesis nine that need for autonomy moderates the
relationship between job autonomy and job satisfaction, a hierarchical multiple
regression analysis was conducted. In this case, need for autonomy is a moderator
variable, as it specifies when the relationship between job autonomy and job
satisfaction will be strong or weak.
The assumptions of normality, linearity, and homoscedasticity of residuals
were met, and there was no evidence of multicollinearity. In the first step, job
autonomy was entered, followed by need for autonomy in the second step. Finally,
the interaction term (i.e., autonomy x need for autonomy) was entered. R was
significantly different from zero at the end of the first step, R = 0.39,
F (1, 203)= 36.61, p = 0.00. The addition of need for job autonomy did not result in
a significant increment in R2, where R = 0.39, Finc (1, 203) = 0.02, p = 0.89. The
addition of the interaction term did not result in a significant increment in R2, where
R = 0.39, Finc (1, 201) = 0.23, p = 0.63. As demonstrated in Table 33, need for job
autonomy does not moderate the relationship between job autonomy and job
satisfaction.
228
Table 33- Hierarchical Regression Analyses examining whether Need for Job
Autonomy Moderates the Relationship between Job Autonomy and Job
Satisfaction.
Step
IV
DV
B

sr2(unique)
1
Job autonomy
JS
0.47
0.39
15.29**
R =0.39**
R2=0.15
Adj R2=0.15
0.48
-0.19
0.39
-0.01
14.51**
R =0.39
R2=0.15
Adj R2=0.15
0.78
2.35
-0.03
0.64
0.12
-0.31
R =0.39
R2=0.15
2
3
Job Autonomy
Need for job autonomy
Job Autonomy
Need for autonomy
Job Autonomy x Need
for autonomy
JS
JS
Adj R2=0.14
** p<0.01; JS – Job satisfaction
3.7.8
Hypothesis Ten: Major Predictors of Job Satisfaction
In order to test hypothesis ten, which examines several major predictors of
job satisfaction, two standard multiple regression analyses were conducted for the
academics and the teachers. The following predictors were included: primary and
secondary control; job autonomy; personality (neuroticism and extroversion); life
satisfaction; social support at work (supervisors and co-workers); and difficulties at
work. For both analyses, the assumptions of normality, linearity, and
homoscedasticity of residuals were met, and there was no evidence of
multicollinearity.
229
R was significantly different from zero for both the academics, R = 0.65,
F (9, 98) = 7.92, p = 0.00, and the teachers, R = 0.75, F (9, 87) = 12.49, p = 0.00.
The major predictors of job satisfaction were the same for both occupational groups,
namely job autonomy and supervisor support at work. As demonstrated in Table 34,
job autonomy accounted for approximately 2% and 5% of the variance in job
satisfaction for the academics and the teachers respectively. The finding for the
academics must be examined cautiously however, as it was not significant at the
more stringent alpha level of 0.01. Supervisor support at work accounted for 5% and
13% of the variance in job satisfaction for the academics and teachers respectively.
230
Table 34- Standard multiple Regression Predicting Job Satisfaction for
Employees with Low Autonomy and Employees with High Autonomy
Group
Acad
Variable
B

Primary Control
Secondary Control
Job Autonomy
Life Satisfaction
Neuroticism
Extroversion
Co-worker Support
Supervisor Support
Difficulties at Work
-0.13
-0.01
0.23
0.29
-0.19
0.02
0.20
0.21
-0.10
-0.09
-0.01
0.17
0.15
-0.15
0.01
0.18
0.28
-0.16
R =0.65**
R2=0.42
0.03
0.19
0.28
0.15
-0.14
-0.01
0.04
0.36
-0.11
0.02
0.11
0.25
0.11
-0.11
-0.008
0.03
0.47
-0.12
R =0.75**
R2=0.56
sr2 (unique)
2.28*
5.61**
AdjR2=0.37
Teach
Primary Control
Secondary Control
Job Autonomy
Life Satisfaction
Neuroticism
Extroversion
Co-Worker Support
Supervisor Support
Difficulties at Work
4.70**
13.54**
AdjR2=0.52
p<0.05 , ** p>0.01; Acad – Academics; Teach- Teachers
For academics, R is composed of 14.46% unique variance and 85.54% shared
variance. For teachers, R is composed of 21.97% unique variance and 78.03%
shared variance
As supervisor support appeared to account for the largest proportion of the
variance in job satisfaction for both occupational groups, a further regression
analysis was conducted to examine the value of R with only supervisor support. R
231
was significantly different from zero for the academics, R = 0.46, F (1, 106) = 27.81,
p = 0.00, and the teachers, R = 0.64, F (1, 95) = 67.29, p = 0.00. Supervisor support
at work accounted for 21% and 41% of the variance in job satisfaction for the
academics and the teachers respectively.
3.7.9
Conclusion
This study tested whether job autonomy influenced the use and adaptiveness
of primary and secondary control strategies. Inconsistent with the hypotheses, the
teachers and academics reported similar levels of primary and secondary control.
Furthermore, primary and secondary control strategies were not correlated with job
satisfaction. Although the control strategies did not mediate the relationship between
job autonomy and job satisfaction, the findings highlighted the importance of
supervisor support in predicting job satisfaction. These findings will now be
discussed.
232
3.8
Discussion
This study was designed to extend the job demand-control model (Karasek &
Theorell, 1990), testing an explanation for the positive relationship between job
autonomy and job satisfaction. This explanation proposed that job autonomy
influences the use and adaptiveness of the control strategies. Employees who
reported higher job autonomy were expected to successfully implement primary
control. The hypotheses were generally not supported however, and as such, a
thorough review of the assumptions and the hypotheses is required.
3.8.1
Assumption- The Academics Represent a High Job Autonomy Group
and the Teachers Represent a Low Job Autonomy Group
The major assumption underlying this study is that the academics were
expected to report higher job autonomy than the teachers as they have more freedom
and choice in many aspects of their work. Consistent with this expectation, the
academics (M = 75%SM) reported significantly higher job autonomy than the
teachers (M = 66%SM). Although this difference was significant, it must be
demonstrated that the difference is meaningful. One way to determine if the
difference is meaningful is to calculate the standard error of measurement (SEM;
Wyrwich, Nienaber, Tierney & Wolinsky, 1999). In the past, researchers have used
the SEM to determine clinically meaningful standards. It is estimated by multiplying
the standard deviation of the scale by the square root of one minus the reliability
coefficient, or
SEM


1  rx x
233
Although there is no consensus about how many SEMs an individuals score
must change for it to be considered significant, Wyrwich et al., (1999) suggest that a
2.77 SEM criterion is the safest. In their study however, they demonstrated that a
one SEM criterion reflected a minimal clinically important difference. In the current
study, the SEM is estimated to be 7.18. Thus, on average there appears to be one
SEM difference between the levels of job autonomy reported by academics and
teachers. This may not necessarily be meaningful as Wyrwich et al (1999) stress that
their results should not be generalised to other populations or tests.
Another way to examine if the difference is meaningful is to compare the
current levels of job autonomy with that reported by other occupational groups
(Hackman & Oldham, 1980). As demonstrated in Table 35, the means range from
58%SM to 73%SM. Compared to these occupational groups, the academics are in
the higher range and the teachers are in the lower range. However, it must be noted
that these data are relatively old. More recent studies have administered Hackman
and Oldham’s (1975) scale to different occupational groups, however they do not
separate the occupational groups (e.g., Renn & Vandenberg, 1995; Tiegs et al.,
1992).
234
Table 35- Normative Data for Hackman and Oldham’s (1980) Autonomy Scale
Occupation
Professional
Management
Clerical
Sales
Service
Processing
Machine Trades
Bench Works
Structural Works
Normative Data (%SM)
73.33
73.33
58.33
63.33
66.66
58.33
65.00
60.00
66.66
Although past studies do not shed much light on whether the differences in
levels of job autonomy reported by the academics and teachers are meaningful, it is
clear that the academics are reporting significantly higher job autonomy than the
teachers. These two groups may not represent the extremes of job autonomy,
however the difference should be great enough to demonstrate the expected
differences in the control strategies. It is assumed that the use of the control
strategies varies linearly with job autonomy over this range of values. As such, even
if the academics and teachers do not represent extremes of job autonomy, the
expected findings should be evident, albeit weaker.
In summary, it appears that the academics report higher job autonomy than
the teachers. It is difficult to ascertain whether this difference is meaningful,
however it is concluded that the difference should be great enough to demonstrate the
expected effects.
235
3.8.2
Hypothesis Testing
The major hypotheses tested in the study were that job autonomy influences
the use of primary and secondary control strategies, and also the relationship between
the control strategies and job satisfaction (i.e., adaptiveness). These hypotheses were
not supported, and as such possible explanations for the findings will be considered,
and the methodology will be re-examined.
3.8.3
Job Autonomy Influences the Amount of the Control Strategies
It was hypothesised that job autonomy would influence the amount of
primary and secondary control strategies. Academics, who have higher job
autonomy than the teachers, were expected to report using more primary control and
less secondary control. Furthermore, it was expected that job autonomy would be
positively correlated with primary control and negatively correlated with secondary
control. Inconsistent with these hypotheses however, the two occupational groups
reported similar levels of control strategies, and job autonomy was not related to the
control strategies.
These findings do not support the model of job satisfaction presented in
Figure 6. Specifically, the findings do not support the arrow from job autonomy to
primary and secondary control. This proposal was based on an extension of the life
span theory of control (Heckhausen & Schulz, 1995), which essentially proposes that
if a person can control a situation, they will attempt to change it. If they cannot
control the situation however, it is more likely that their attempts to change it would
236
fail, and thus they will seek to accept the situation. It is most surprising therefore
that the academics reported high secondary control, and that the teachers reported
high primary control. Explanations for these unexpected findings will be discussed.
3.8.3.1
Why did the Academics Report High Secondary Control?
The academics reported equally high levels of primary and secondary control
as the teachers. This means that when they face a difficulty at work, they use both
primary and secondary control strategies. It is interesting that they rely on secondary
control however, because theoretically, they should have less need for it than the
teachers. As they have reasonably high control over their environment, they are
expected to successfully implement primary control most of the time and rarely rely
on secondary control strategies. As this is clearly not the case, the use of secondary
control may need to be re-examined.
In addition to using secondary control to compensate for primary control
failure, secondary control may be used as a means of temporarily avoiding primary
control. If employees were faced with a large number of difficulties at work, they
may initially use secondary control. For example, workers may tell themselves that
it will work out okay in the end, or that it doesn’t matter. This may be necessary for
workers, such as academics, who face many difficulties, and must delay dealing with
some of them. Once they can deal with them however, it is expected that they do so
using primary control. Thus, secondary control may be used prior to primary
control.
237
The explanation that secondary control can be used prior to primary control
may explain why the academics reported high levels of secondary control. However,
this explanation does not account for the lack of occupational differences in
secondary control. Even if both groups use secondary control prior to dealing with
their difficulties, the teachers would be expected to rely on more secondary control
than the academics. After initially delaying dealing with a problem using secondary
control, it is expected that the academics would then use primary control, but that the
teachers would continue using secondary control.
3.8.3.2
Why did the teachers report high primary control?
Although the teachers were expected to rely mostly on secondary control,
they reported equally high levels of primary and secondary control. One explanation
for this finding is that the teachers may have avoided repeated primary control
failure. If they implemented primary control and failed, they were expected to rely
mostly on secondary control. Through relying on secondary control, and accepting
their situation rather than trying to change it however, they can then maintain their
perceptions of primary control. Thus, the teachers’ levels of primary control may be
explained, in part, by their reliance on secondary control strategies.
3.8.3.3
Summary
Both the academics and the teachers reported similar levels of primary and
secondary control. The academics reported higher secondary control than expected,
and the teachers reported higher primary control than expected. The academics’
238
level of secondary control may be explained by the proposal that secondary control
may also be used prior to primary control. The teachers’ levels of primary control
may be explained by the proposal that they can avoid primary control failure through
relying on secondary control strategies.
3.8.4
Job Autonomy Influences the Relationship Between the Control
Strategies and Job Satisfaction
It was hypothesised that job autonomy influences the adaptiveness of the
control strategies, such that primary control would be more positively related to job
satisfaction than secondary control for the academics, and that secondary control
would be more positively related to job satisfaction than primary control for the
teachers. Furthermore, it was hypothesised that the relationship between the control
strategies and job satisfaction would be moderated by job autonomy. Inconsistent
with both of these hypotheses however, primary and secondary control strategies
were not related to job satisfaction. This suggests that workers can use primary or
secondary control strategies to deal with their difficulties.
These findings are inconsistent with the primacy/back-up model and the
discrimination model (Thompson et al., 1998). The discrimination model proposes
that primary control is the most adaptive strategy in controllable situations, and that
secondary control is the most adaptive strategy in uncontrollable situations. The
primacy/back-up model proposes that primary control is the most adaptive strategy in
low-control and high-control situations. The current results demonstrated that
primary and secondary control strategies were not related to job satisfaction.
239
However, a major limitation has now been identified in the study that may render
these findings invalid.
3.8.5
Limitations in the Hypotheses Examining Job Autonomy and Control
Strategies
There was a methodological problem in this study that may have limited the
findings examining how job autonomy influences the use and the adaptiveness of the
control strategies. This problem concerns the specificity of the hypotheses. The
hypotheses tested whether job autonomy influenced the relationship between the
control strategies and job satisfaction at a general level, however the
primacy/back-up model and the discrimination model actually only refer to one
situation. These models propose that the controllability of a situation influences the
control strategies used to handle that situation. Thus, the hypotheses need to be
measured at a situational level, rather than at an occupational level.
This lack of consistency between the definition of the discrimination model
and the testing of the discrimination model is not limited to this study. All of the
studies that Thompson et al., (1998) claimed to test the discrimination model actually
fail to test it as specified in the definition.
To test the discrimination model, researchers need to correlate a measure of
perceived control over one situation with the control strategies used in that situation.
Past researchers have relied on specific measures of the controllability of the
situation, and the control strategies, however they aggregated them rather than
examining them separately.
240
For example, Thomson et al., (1996, 1994) developed a list of specific facets
relevant to living with HIV, such as progression of HIV infection, family
relationships, and quality of medical care. For each of these facets, respondents rated
how much control they had over them, and the extent to which they used primary and
secondary control to handle them. If they correlated each facet with the control
strategies used to handle that facet, they would be testing the discrimination model.
However, they aggregated the items to obtain an overall measure of perceived
control, an overall measure of primary control and an overall measure of secondary
control.
Another study measured specific controllability and control strategies, yet
failed to use this information to test the discrimination model. Thompson et al.,
(1998) examined control over physical appearance, measuring how much control
people had over the attractiveness of their hair, body strength and agility, weight and
body shape, skin and overall physical appearance. They also measured the primary
and secondary control strategies in relation to age-related changes over physical
appearance. Although they measured these specific variables however, they added
the perceived control scale to the primary control items to measure primary control.
Thus, they failed to examine whether the controllability of a situation influenced the
control strategies used in that situation.
It thus appears as though the current study and previous studies have failed to
adequately test the discrimination model. In order to do so, future studies need to
examine the controllability of the situation and the control strategies at a situational
level. It should then be tested whether the amount of autonomy an employee has
241
over a situation predicts the use and the adaptiveness of the control strategies in that
situation. As such, it is not necessary to examine the control strategies that
employees with high/low job autonomy are using, rather to examine which control
strategies all workers use in low-control and high-control situations.
3.8.5.1
Summary
Job autonomy did not influence the use or adaptiveness of the control
strategies. These findings were inconsistent with the proposed model of job
satisfaction, which attempted to explain the relationship between job autonomy and
job satisfaction. However, a major limitation was identified in the study, where it
appears as though the study has failed to test the discrimination model.
3.8.6
Other Predictors of Job Satisfaction
The remainder of the hypotheses will now be examined. These hypotheses
examine occupational differences in job and life satisfaction, the buffering role of
social support at work, and the moderating role of need for autonomy. Additionally,
the major predictors of job satisfaction that are included in the proposed model of job
satisfaction are examined.
3.8.7
Occupational Differences in Job Satisfaction and Life Satisfaction
Inconsistently, the academics did not report higher job satisfaction or life
satisfaction than the teachers. The finding that there was no difference in life
satisfaction is not surprising given that there were no occupational differences in job
242
satisfaction. Furthermore, the levels of life satisfaction reported by both groups were
within the normative range according to the homeostatic theory of life satisfaction
(Cummins, 1995, 2000b). It is surprising however that the academics (M = 64.09)
and the teachers (M = 68.79) reported similar levels of job satisfaction. The levels
of job satisfaction reported by both groups will firstly be compared with past studies.
3.8.7.1
Comparisons with Past Studies
Past research has reported varying levels of job satisfaction for academics.
Researchers have reported the following levels of job satisfaction; 57%SM (Leung et
al., 2000), 65%SM (Hill, 1986), 66%SM (Lahey & Vihtelic, 2000), 74%SM (Carson
et al., 2001), 82%SM (Olsen, 1993) and 83%SM (Niemann & Dovidio, 1998).
Furthermore, study one reported a level of 66%SM. The scales used in some of these
studies were criticised in study one, however a normative level of job satisfaction
was not established. All that can be concluded is that the academics in this study, as
with those in study one, report a level of job satisfaction that is within the range
found by other researchers.
In regards to teachers, researchers have reported several different levels of
job satisfaction. For example, Klecker and Loadman (1999) found a similar level of
job satisfaction (M = 68%SM) as the present study, however others have reported a
higher level of job satisfaction of 80%SM (Ma & Macmillan, 1999; Schonfeld,
2000). All of these studies are flawed however as they relied on poor measures of
job satisfaction.
243
For example, Ma and MacMillan (1999) included the following items to
measure job satisfaction; “I find my professional role satisfying”, “I look forward to
each day”, “I am committed to making our school one of the best in the province”
and “If I could start over, I would become a teacher again.” The item “I look
forward to each day” may actually be dependent on personality and quality of life, as
well as job satisfaction. Furthermore, the item “If I could start over, I would become
a teacher again”, is likely to be dependent on the teachers’ age and how much they
have invested into becoming a teacher. A teacher who has spent 20 years teaching
may agree with this item because they are satisfied with their job, or because they
want to justify why they are still in the profession. Until psychometric data are
obtained for this scale, the results are questionable.
Other researchers have relied on facet scales of job satisfaction. For example,
Klecker and Loadman (1999) measured satisfaction with salary, professional
advancement, professional challenge, autonomy, work conditions, interactions with
colleagues, and interactions with students. Facet scales are criticised however for
excluding facets that are important to the individual, or including facets that are not
important. Furthermore, it is expected that the teachers’ level of job satisfaction
would be lower in this scale, as it is dependent on their level of job autonomy. This
is problematic because although job autonomy is expected to be related to job
satisfaction, job autonomy may not be a domain of job satisfaction.
In general, it appears that the academics’ level of job satisfaction is
reasonably consistent with past studies. The teachers’ level of job satisfaction tends
244
to be slightly lower than previous studies, however these studies have relied on
inadequate measures of job satisfaction.
As few studies have examined academics’ and teachers’ levels of job
satisfaction, the findings can also be compared to the review conducted in study one,
which demonstrated an average level of job satisfaction of 66%SM. This average
level is consistent with both the academics and the teachers. Possible explanation for
the similar levels of job satisfaction will be presented.
3.8.7.2
Explaining the Similar Levels of Job Satisfaction
There are a number of explanations for the academics’ and teachers’ similar
levels of job satisfaction. First, the recruitment process was different for the teachers
and the academics. The academics were sent their questionnaire through internal
mail, whereas the teachers were, in some cases, given their questionnaire by the
Principal of the school. This may be problematic for the teachers, as the Principal, in
order to obtain positive results, may have given the questionnaires to happier
workers. Alternatively, even if social desirability was not important, the Principals
may have given the questionnaires to teachers that were more likely to agree, or were
more organised. Thus, it must be questioned whether the teachers included in the
sample are representative of the average teacher.
It is possible however that the teachers’ average level of job autonomy is
representative of the average teacher. Although teachers report a lower level of job
autonomy than the academics, there are certainly many other determinants of job
satisfaction. One major factor that was highlighted in this research was social
245
support at work. Examination of the descriptive statistics demonstrates that the
teachers report higher satisfaction with their supervisor support, and that supervisor
support was strongly correlated with job satisfaction.
3.8.8
The Influence of Social Support at Work on the Relationship Between
Work Difficulties and Job Satisfaction
Supervisor support, but not co-worker support moderated the relationship
between work difficulties and job satisfaction. The relationship between work
difficulties and job satisfaction was weaker when supervisor support was high. It
must be noted however that this hypothesis was only significant at an alpha level of
0.05, and not 0.01, suggesting that the effect may not be large. However, even if the
effect is not large, the findings suggest that supervisors should ensure that they
provide emotional and instrumental support to their employees. To do this, the
supervisor needs to show concern for their employees, and provide tangible
assistance (Karasek & Theorell, 1990).
The finding that supervisor support plays a greater role than co-worker
support has been reported by previous researchers (e.g., Beehr, 1985; Fenlason &
Beehr, 1994; Russell, Altmaier & Van Velzen, 1987). Co-workers have less
influence at work, and as such may have less influence over difficulties at work
(Fenlason & Beehr, 1994).
Consistent with the current findings, a few studies have demonstrated that
social support at work has positive moderating effects on job satisfaction (i.e.,
Karasek et al., 1982; Landsbergis et al., 1992). However, there are studies that have
246
failed to demonstrate the moderating role of social support (Chay, 1993; de Jonge &
Landeweerd, 1993, cited in Van Der Doef & Maes, 1999; Melamed at al., 1991;
Parkes & Von Rabenau, 1993).
One difference between these supportive and non-supportive studies is in the
measure of social support. Two of the supportive studies (i.e., current study and
Landsbergis et al., 1992) relied on Karasek and Theorell’s (1990) scale, whereas the
non-supportive studies relied on several different scales. However, there are too few
studies to draw conclusions about the influence of the scales. It is clear that further
research is needed to examine the moderating role of social support.
3.8.9
The Influence that Need for Job Autonomy has on the Relationship
Between Job Autonomy and Job Satisfaction
Need for job autonomy did not moderate the relationship between job
autonomy and job satisfaction. Although past empirical studies were equivocal, it
seemed intuitive that differences in need for autonomy would influence the
relationship between job autonomy and job satisfaction. The non-supportive finding
is consistent with a few past studies. For example, de Rijk et al., (1998) failed to
demonstrate that need for autonomy moderated the relationship between job
autonomy and emotional exhaustion and health complaints. Furthermore, Nicolle
(1994) found need for autonomy moderated the relationship between job autonomy
and absenteeism in only 3 of 36 analyses.
As the current study and previous studies generally fail to demonstrate that
need for job autonomy moderates the relationship between job autonomy and job
247
satisfaction/job stress, this hypothesis will no longer be investigated. Too few
studies have been conducted to simply conclude the effect does not exist, however it
seems that more testing is required to develop a valid measure of need for job
autonomy. Studies are relying on exploratory measures, and as such may not be
adequately measuring the need for job autonomy construct. As such, it is
recommended that researchers firstly attempt to develop a need for job autonomy
scale that is psychometrically sound.
3.8.10
Major predictors of Job Satisfaction
Inconsistent with the proposed model of job satisfaction, only job autonomy
and supervisor support uniquely predicted job satisfaction for both groups. The
relationship between these variables and job satisfaction was consistent with past
research. For job autonomy, r = 0.43 (Tiegs et al., 1992), and for social support at
work, r = 0.52 (Dollard et al., 2000), and r = 0.66 (Munro et al., 1998).
The control strategies, personality, life satisfaction, co-worker support, and
difficulties at work were moderately correlated with job satisfaction, however they
did not uniquely predict job satisfaction. As such, several changes need to be made
to the variables included in the model. The control strategies will be retained in the
model, however as discussed earlier, several changes will be made to the Primary
and Secondary Control Scale. Personality was a poor predictor of job satisfaction,
and as such, it will be excluded from study three. Life satisfaction was also a poor
predictor of job satisfaction, however it will be retained in the model as it is acting as
both an independent variable and a dependent variable. Supervisor support
248
explained much of the variance in job satisfaction, and as such both types of social
support (i.e., co-worker and supervisor) will be re-examined. Difficulties at work
will also be retained in the model, as it is necessary to demonstrate what the primary
and secondary control strategies are used for.
In summary, as a result of the model of job satisfaction only being partially
supported, several changes will be made in study three. These changes will be
explained further in chapter 4.
3.8.11
Conclusion
Although the findings demonstrated that job autonomy did not predict the use
or adaptiveness of the control strategies, one major limitation was identified in this
study. The hypotheses were criticised for being too general, and it was suggested
that study three should examine job autonomy and the control strategies at the
situational level rather than at the occupational level. Furthermore, this study
highlighted the importance of supervisor support at work, which will be examined
further in study three.
249
4 Chapter 4 - Study Three
250
4.1
Abstract
The major proposal of this study is that the controllability of a work difficulty
influences the use and adaptiveness of the control strategies used to handle that
difficulty. It was expected, based on the discrimination model, that in controllable
situations, employees would use more primary control than secondary control, and
that primary control would be the most adaptive. In uncontrollable situations
however, it was expected that employees would use more secondary than primary
control, and that secondary control would be the most adaptive strategy. These
proposals were not supported as employees reported using similar strategies for
controllable and uncontrollable difficulties. Furthermore, primary control strategies
were more adaptive than secondary control strategies for both types of difficulties.
These findings challenge the belief that control strategies are influenced by
situational variables and also question the assumption that primary control failure
negatively affects job satisfaction. The implications of these findings for the
proposed model of job satisfaction are discussed.
251
4.2
Proposal for Study Three
This study continues to test the proposal that job autonomy influences the use
and the adaptiveness of primary and secondary control strategies, however unlike
previous studies, it will be examined at a situational level, rather than an
occupational level. As such, changes are made to the specificity of the hypotheses
and the primary and secondary control scale. Further changes are made to the model
of job satisfaction, where it is proposed that the control strategies and social support
at work moderate the relationship between controllable and uncontrollable work
difficulties and job satisfaction.
4.2.1
Specificity of Hypotheses Testing the Proposal that Job Autonomy
Influences the Control Strategies
In this study, the proposal that job autonomy influences the use and
adaptiveness of the control strategies is examined at a more specific level. As
discussed in chapter 3, studies one and two were criticised as they were not
consistent with the definition of the discrimination model. The discrimination model
proposes that when the situation is controllable, primary control is more adaptive,
and when the situation is uncontrollable, secondary control is more adaptive.
However, empirical tests of the model, including studies one and two, have examined
perceived control and control strategies at a general level (i.e., Thompson et al.,
1996; Thompson et al., 1994; Thompson et al., 1998).
252
The difference between the definition and empirical tests of the
discrimination model may be important. If the discrimination model is tested at a
more specific level, the relationship between the two variables may be stronger. All
employees, whether they be low autonomy or high autonomy, are expected to have
high control over some aspects of their job and less control over other aspects.
Furthermore, all employees, whether they be low or high job autonomy, are expected
to use primary control in some situations and secondary control in others. By
correlating their general level of job autonomy with their general level of control
strategies, the results become less extreme, the low and high autonomy groups
become more similar, and the correlations become weaker.
In order to accurately test the discrimination model, study three will examine
whether the controllability of a situation influences the use and adaptiveness of the
control strategies used in that situation. Past research examining these proposals will
be examined.
4.2.2
Examining how the Controllability of a Difficulty Influences the Use of
the Control Strategies
As proposed by the life span theory of control, it is expected that all
individuals will implement primary and secondary control strategies (Heckhausen &
Schulz, 1995). However, the ratio of these strategies is expected to be influenced by
the controllability of the situation. If the situation is appraised as being controllable,
it is expected that people will try to change it using primary control. The situation is
253
amenable to change, and as such, it is expected that attempts to change the
environment using primary control would be successful.
If however the situation is appraised as being uncontrollable, it is expected
that people will attempt to change themselves using secondary control. If the person
tried to change the environment using primary control, they would be likely to
experience primary control failure. In order to avoid primary control failure
therefore, it is expected that people would rely on more secondary control. Hence, it
is proposed that the controllability of the situation is inversely related to the
probability of primary control failure, which in turn, influences the use of secondary
control strategies.
4.2.3
Empirical Studies Examining if the Controllability of a Situation
Influences the Use of Control Strategies
Although chapter 1 identified some studies that examined the amount of
general primary and secondary control reported by employees to handle general work
difficulties, no studies have been located which report the amount of primary and
secondary control people use in controllable and uncontrollable situations. One
study has examined the control strategies reported by people only in low-control
situations (i.e., HIV-positive men in prison). According to the discrimination model,
it would be expected that these men would rely on more secondary control than
primary control. This was not the case however, as Thompson et al., (1996)
demonstrated that the men reported slightly more primary control (M = 48.5%SM)
than secondary control (M = 45%SM).
254
Although there are no studies examining the use of the control strategies in
controllable and uncontrollable situations, there are some studies that have been
conducted on coping strategies. These studies generally examine the amount of
problem-focussed and emotion-focussed coping strategies reported by people in
controllable and uncontrollable situations (e.g., Bowman & Stern, 1995; Folkman et
al., 1986; Forsythe & Compas, 1987; Terry & Hynes,1998; Valentiner, Holahan &
Moos, 1994; Vitaliano, DeWolfe, Maurio, Russo & Katon, 1990).
One study has specifically examined coping strategies at work. Bowman and
Stern (1995) asked nurses to describe two stressful events, one that they “could
control, could change or could do something about” and one that was “difficult to
control, that you had to accept or had to get used to.” Participants then rated the
controllability of the situation, and completed Lazarus and Folkman’s (1984) coping
scale. Unfortunately however, Bowman and Stern (1995) did not examine the mean
coping strategies separately for each stressful situation. Instead they aggregated
them, providing mean scores for avoidance coping, problem-reappraisal coping, and
problem solving coping. It must be noted however that even if the means were
provided, the validity of the research design is questioned. There are problem with
using the terms “change” and “do something about” for controllable situations and
“accept” and “get used to” for uncontrollable situations. These terms may bias the
employees to respond in ways that are consistent with the discrimination model. By
their nature, situations that have been changed are those where primary control
strategies have been used, and situations that have been accepted are those where
secondary control strategies have been used.
255
Other studies have examined coping strategies in controllable and
uncontrollable non-work situations. These studies have provided somewhat mixed
support. For example, Forsythe and Compas (1987) demonstrated that people used
more problem-focussed coping for events appraised as controllable (M = 13.82) than
uncontrollable (M = 10.81), however there were no differences in emotion-focussed
coping. Furthermore, Folkman et al., (1986) found that married couples tend to use
more problem-focussed coping in situations perceived as changeable, and more
emotion-focussed coping in situations perceived as having to be accepted.
However, Valentiner et al., (1994) demonstrated that college students did not report
more problem-focussed type coping (M = 55.39) than emotion-focussed type coping
(M = 57.78) in a controllable event. Furthermore, they did not report more emotionfocussed type coping (M = 53.50) than problem-focussed type coping in an
uncontrollable event (M = 54.34).
Instead of reporting the average level of coping strategies in controllable and
uncontrollable situations, other studies have examined the correlations between
perceived control and coping strategies. In this case, it would be expected that
perceived control would be positively correlated with problem-focussed coping and
negatively correlated with emotion-focussed coping. Overall however, these studies
have tended to be inconsistent.
For example, Osowiecki and Compas (1999) demonstrated that problemfocussed and emotion-focussed coping were not significantly related to perceived
control. A similar result was found by Conway and Terry (1992) where problemfocussed coping, self-denigration and escapism did not correlate with the
256
controllability of an event. However, Park, Folkman and Bostrom (2001)
demonstrated that controllability appraisal was positively correlated with problemfocussed strategies (i.e., planful problem solving, r = 0.23), and negatively correlated
with emotion-focussed strategies (distancing, r = -0.29).
When examining these studies on coping strategies, the flaws in the
conceptualisation of problem-focussed and emotion-focussed coping must be
considered. As discussed in chapter 1, the theory underlying problem-focussed and
emotion-focussed coping and the questionnaire designed to assess these strategies
(i.e., Ways of Coping Questionnaire; WCQ) has methodological limitations
(Edwards & O’Neill, 1998). The most concerning problem is that there is overlap
among the coping dimensions, where some problem-focussed coping strategies
resemble emotion-focussed coping strategies (Edwards & O’Neill, 1998).
The conceptualisation of primary and secondary control is superior to
problem-focussed and emotion-focussed coping because it maintains the distinction
between changing the environment (i.e., primary control), and changing the self (i.e.,
secondary control). Thus, the proposal that the controllability of the situation
influences the amount of control strategies will be tested in this study.
4.2.3.1
Summary
Based on the proposals of the life span theory of control, it is expected that
when employees have a controllable difficulty, they use more primary control than
secondary control. When they have an uncontrollable difficulty, it is expected that
they will attempt to avoid primary control failure, and thus report more secondary
257
control than primary control. Although no studies have examined the control
strategies reported by people in controllable and uncontrollable situations, studies
using coping strategies have offered, at best, mixed support.
4.2.4
Examining how Controllability Influences the Adaptiveness of the
Control Strategies
Based on the discrimination model (Thompson et al., 1998), it is expected
that primary control is the most adaptive strategy in controllable situations, and that
secondary control is the most adaptive strategy in uncontrollable situations. This is
consistent with the life span theory of control which proposes that although primary
control is the more adaptive strategy, primary control failure can have negative
consequences (Heckhausen et al., 1997). It is postulated that the controllability of
the situation is inversely related to the probability of primary control failure.
As discussed in chapter 1, only a few studies have examined the most
adaptive control strategies in low-control situations (Thompson et al., 1996; 1994;
1993; 1998). These studies suggest that primary control strategies are more adaptive
in controllable and uncontrollable situations, however these studies were criticised
for their measurement of perceived control, and primary and secondary control
strategies.
It must be noted however that a similar hypothesis was being developed in
the coping literature. Several researchers have tested this proposition, referred to as
the “goodness of fit” hypothesis (Carver, Scheier & Weintraub, 1989; Conway &
Terry, 1992; Folkman et al., 1986; Roberts, 1995; Vitaliano et al., 1990). They
258
recognise that “it is not the coping response per se that is the key to reduce emotional
distress, but rather how well the coping strategy fits the perceived situation”
(Osowiecki & Compas, 1998, p. 485).
Although there is significant overlap among the coping studies and control
strategy studies, researchers are yet to integrate the results. As there are few such
studies, this integration is essential to gain a greater understanding about whether the
controllability of the situation influences the use and adaptiveness of the
control/coping strategies in that situation. It must be noted however that the majority
of studies have relied on Lazarus and Folkman’s (1984) problem-focussed coping
and emotion-focussed coping, which was criticised in chapter 1.
4.2.4.1
Integrating Empirical Studies on the Discrimination Model and the
Goodness of Fit Model
Empirical studies examining whether the controllability of the situation
influences the coping/control scales used in that situation provide mixed support.
Generally, these studies demonstrate the primary control-type strategies are more
adaptive than secondary control-type strategies in controllable situations. However,
some of the studies fail to demonstrate that secondary control-type strategies are
more adaptive than primary control-type strategies in uncontrollable situations (e.g.,
Bowman & Stern, 1995; Conway & Terry, 1992; Osowieki & Compas, 1998, 1999;
Park, Folkman & Bostrom, 2001; Vitaliano et al., 1990).
For example, Thompson et al’s., (1996) study on HIV-positive men
demonstrated that primary control was negatively related to distress and secondary
259
control was positively related to distress. However, other studies have demonstrated
that secondary control is more adaptive than primary control. For example, Terry
and Hynes (1998) demonstrated that for women coping with in vitro fertilization,
problem management coping (i.e., trying to solve the problem) was related to more
distress. Secondary control-type strategies (i.e., problem appraisal, and emotional
approach) were related to less distress.
Integration of the studies testing the goodness of fit model and the
discrimination model highlights the inconsistencies in the area. Generally, it appears
as though primary control is adaptive in controllable situations, but that secondary
control may not be the most adaptive strategy in uncontrollable situations. These
results may be limited, as problems have been identified in the design of the studies.
4.2.4.2
Research Design
Researchers have typically relied on two major types of designs to measure
the goodness of fit hypothesis and the discrimination model. The first design
assesses how people handle one stressful situation (e.g., Carver et al., 1989; Conway
& Terry, 1992; Folkman et al., 1986; Roberts, 1995; Vitaliano et al., 1990).
Typically, the person reports on the most stressful encounter they had during the
previous week, indicating how much they could control the situation, and what they
did to handle the situation. The researcher then correlates the controllability of the
situation with the coping strategies. This measure is problematic however as the
respondent chooses whether they report a controllable or uncontrollable situation and
the researcher cannot influence this variable.
260
The second type of design, much like studies one and two, examines the
control strategies used by people in low-control situations. Researchers have studied
various groups such as cancer patients (Osowiecki & Compas, 1998, 1999),
HIV- positive men (Thompson et al., 1996; Thompson et al., 1994), people
experiencing age-related physical changes (Thompson et al., 1998), and children
experiencing homesickness (Thurber & Weisz, 1997). These studies generally assess
how much perceived control the person has over the situation (e.g., cancer) and then
assesses which control/coping strategies they used to handle the situation
(e.g., Osowiecki & Compas, 1998). This design is criticised however as it is
inconsistent with the proposed models. Both the discrimination model and the
goodness of fit model refer to one situation. In order to test whether the
controllability of a situation influences the control strategies used in that situation,
the scale needs to be more specific.
4.2.4.3
Summary
Research examining the discrimination model and the goodness of fit
hypothesis is equivocal. It appears however that the majority of studies find that
primary control is adaptive in controllable situations, but less demonstrate that
secondary control is adaptive in uncontrollable situations. The validity of these
findings are questioned however, as the research designs are criticised. In order to
accurately test the discrimination model, a more specific scale is required which
assesses how people handle controllable and uncontrollable difficulties.
261
4.2.5
Developing a Situation Specific Primary and Secondary Control Scale
In order to test the proposal that the controllability of a situation influences
the use and adaptiveness of the control strategies in that situation, a situation specific
primary and secondary control scale is developed. This scale overcomes many of the
limitations identified in previous scales, as it: a) assesses how people react in
controllable and uncontrollable situations; b) can be used by workers in any
occupation; and c) contains few items.
The Situation Specific Primary and Secondary Control Scale (Maher &
Cummins (2002) is an extension of the 4th edition of the Primary and Secondary
Control Scale (Maher et al., 2001). The scale includes four primary control
strategies and 12 secondary control strategies. Respondents are asked to indicate
how often, during the past week, they have used various strategies when facing a
difficulty at work. The major change made to this scale is that rather than thinking
about any difficulty at work, the respondents now think about one difficulty that they
can control and one difficulty that they cannot control.
The major issue in designing a scale to measure low-control and high-control
situations was deciding on the wording of the situation. Only one study was located
that tested respondents in controllable situation and uncontrollable situations at work
(Bowman & Stern, 1995). For the controllable situation, the employee was told to
consider a situation that they “could control, could change, or do something about.”
For the uncontrollable situation, the employee was told to consider a situation that
was “difficult to control, that you had to accept or get used to.”
262
Other researchers examining one situation have asked “how much control do
you have over” (Thompson et al., 1994; Thurber & Weisz, 1997), “how much
influence do you have over” (Conway & Terry, 1992) and “how much can you
change” (Carver et al., 1989, Folkman et al., 1986).
As mentioned previously, there are problem with using the terms “change”,
“do something about”, and “influence” for controllable situations and “accept” and
“get used to” for uncontrollable situations. These terms may bias the employees to
respond in ways that are consistent with the discrimination model.
A viable alternative that does not imply that the situation has been changed is
“control.” The employees could be asked to think of a difficulty that they can
control and a difficulty that they cannot control. Control is superior to the other
constructs, as it does not bias the respondents to nominate primary control strategies
in a high-control situation.
Changes were also made to the primary control items. The primary control
items were designed to be general, assessing whether the person looks for different
ways to overcome difficulties, persists, puts in effort, and works out how to remove
obstacles. Closer examination of these items however revealed several problems.
For example, the item, “looked for different ways to overcome it” may not actually
represent primary control. A person who looks for different ways to overcome their
difficulties does not necessarily attempt to change the environment to suit their
needs. They may think about the different ways, decide that they are all too risky,
and then resort to secondary control strategies. To demonstrate primary control, a
person needs to do more than just think of different ways to overcome the difficulty,
263
rather they need to act on their environment. As such, this item was changed to a
strategy identified by Folkman and Lazarus (1980) termed “choose and act on a
potential solution.”
The item, “work hard to overcome it” is also criticised as it implies that the
strategy must be successful for it to represent primary control. According to this
item, the person must not only work harder, but must also overcome the difficulty.
Primary control does not necessarily involve overcoming the problem, only that the
person attempts to change the environment. As such, this item was changed to
“work harder.”
Another problematic item is “work out how to remove obstacles.” This item,
adapted from Heckhausen et al’s., (1997) scale, refers to goals rather than
difficulties. This item is appropriate for goals as, if a goal is not obtained, there must
be obstacles in the way of it. However, there may not necessarily be obstacles in the
way of a difficulty. This item does not appear to fit with the control scale, which
focuses on difficulties at work, and as such was deleted from the scale.
Based on these analyses, the following three items were included in the
primary control scale; “choose and act on a potential solution”; “keep trying”; and
“work harder.” One extra item was developed to account for the fact that other
people, such as management staff, often control many problems in the workplace. A
major way that a person may change difficulties in the workplace is through
discussions or confrontations. An item developed in Latack’s (1986) coping scale to
measure this is “discussing the problem with the people involved.” As discussing the
264
problem may not necessarily mean that the environment is changed, the item was
revised to “discuss solutions with the people involved.”
The addition of this item led to a review of the “support” item in the
secondary control scale to ensure that the two were different. Indeed, it is difficult to
separate support as a primary control strategy and support as a secondary control
strategy. The main distinction between the two, however, is that support as a
primary control strategy involves the person changing the environment, whereas
support as a secondary control strategy involves the person changing themselves to
accept the problem.
The secondary control strategy of support was measured by the item “told
someone about it” in previous editions of the scale. In order to ensure that this item
is distinct from the primary control strategy, it was changed to clearly demonstrate
that it involves changing the self (i.e., “I told someone about the difficulty to make
me feel better”).
One final change to the primary and secondary control scale concerns the
scoring. In study two, the highest score for primary and secondary control was
recorded. The items were not aggregated because the resulting score was deemed to
be unrepresentative of the control strategies. For example, using an aggregated
score, a person who reported that they use one secondary control strategy every time
(10), and reported never (0) for the remaining strategies would receive a low score.
In order to demonstrate that this person is using one secondary control strategy all
the time, the highest score for secondary control was recorded (10), and the person
received a high score. However, as demonstrated in study two, using the highest
265
score does not appear to be the answer. This method did not differentiate the
respondents, with 76% of the subjects reporting a level of primary control between
77%SM and 100%SM, and 84% of the subjects reporting a level of secondary
control between 77%SM and 100%SM. This range is concerning, suggesting that
there may have been a ceiling effect. As a result, the aggregated scoring procedure
implemented in study one will be used in the current study. However, it must be
noted that the aggregated scoring method, although used by the majority of
researchers in the field, is biased towards people who use a greater variety of
strategies.
4.2.5.1
Summary
In order to test the proposal that the controllability of a situation influences
the use and adaptiveness of the control strategies in that situation, a situation specific
primary and secondary control scale was developed. The major change made to the
scale is that rather than thinking about any difficulty at work, the employees are now
required to think about one difficulty that they can control and one difficulty that
they cannot control. Changes were also made to the wording of the primary control
items and the scoring procedure.
4.2.6
Examining the Moderating Role of Primary and Secondary Control
Strategies
In addition to examining the amount and adaptiveness of the control
strategies in controllable and uncontrollable situations, this study also tests whether
266
the control strategies moderate the relationship between work difficulties and job
satisfaction. Although no other studies have examined the moderating role of the
control strategies, several researchers in the coping literature have suggested that it is
not the stressor that predicts job satisfaction, but rather how the person deals with the
stressor (Aldwin & Revenson, 1987, Ashford, 1988; Parkes, 1990, 1994; Perrewe &
Zellars, 1999; Osipow, Doty & Spokane, 1985).
For example, Ashford (1988) demonstrated that coping moderated the effect
of organisation transitions on job stress, where employees who shared emotions
experienced less stress after organisational change. Parkes (1990) also demonstrated
that coping moderated the effect of work demands on general health, however they
found that employees who reported more direct coping (i.e., problem-focussed
coping) had better health. As only a few studies have examined this proposal in the
workplace, and as they have relied on varying measures of coping, this proposal will
be examined further.
It is proposed that primary control strategies are only useful in reducing stress
when the situation is controllable. In these situations, primary control can be
implemented successfully, and the negative effects of the difficulty can be reduced.
When the situation is uncontrollable however, secondary control strategies may be
useful in helping the person adjust to the situation. If they use primary control
strategies, they are likely to experience primary control failure, which may increase
their stress. However, if they use secondary control, they can reduce their stress by
accepting the situation. These exploratory proposals will be tested.
267
4.2.7
Examining the Moderating Role of Social Support at Work
In addition to primary and secondary control, social support at work may
moderate the relationship between controllable and uncontrollable work difficulties
and job satisfaction. In study two, the moderating roles of co-worker and supervisor
support were examined. This study will also examine the major types of social
support, namely instrumental and emotional support. These two types of social
support are expected to play different roles (Ducharme & Martin, 2000; Wong,
Cheuk & Rosen, 2000). Instrumental support is expected to buffer work difficulties
because it helps workers to cope effectively with problems, whereas emotional
support is not expected to buffer work difficulties as it does not directly alter the
stressor (Wong et al., 2000). Some support has been provided for these proposals,
where instrumental supervisor support, but not emotional supervisor support, has
been shown to buffer the effects of job stress on job satisfaction (Wong et al., 2000).
A more specific explanation is developed for this study, where it is proposed
that both instrumental and emotional support buffer the effects of work difficulties.
Specifically, it is proposed that instrumental support buffers the effects of
controllable difficulties and emotional support buffers the effects of uncontrollable
difficulties. Instrumental support is useful if the difficulty is controllable as other
people may help the person to overcome the problem. However, when the difficulty
is uncontrollable, instrumental social support may not be useful as there is nothing
that can be done to overcome the difficulty. Rather, in these situations, emotional
social support may help the person to accept these difficulties. As this study is
268
examining both controllable and uncontrollable difficulties, it must be considered
whether emotional and instrumental social support moderate both types of
difficulties.
4.2.7.1
Summary
This study proposes that the control strategies and social support at work
moderate the relationship between work difficulties and job satisfaction.
Specifically, it is expected that primary control and instrumental support moderate
the relationship between controllable work difficulties and job satisfaction.
Furthermore, it is expected that secondary control and emotional support moderate
the relationship between uncontrollable work difficulties and job satisfaction.
269
4.3
Revised Model of Job Satisfaction
This study aims to test the model of job satisfaction presented in Figure 8,
which is fundamentally different to the previous two models. The model proposes
that employees experience controllable and uncontrollable difficulties, which are
negatively related to job satisfaction.
In response to these difficulties, employees can implement primary and
secondary control strategies. It is expected that the ratio of control strategies will
vary depending on whether the difficulty is controllable or uncontrollable. For
controllable difficulties, it is expected that workers will rely on primary more than
secondary control. For uncontrollable difficulties, it is expected that workers will
rely on secondary more than primary control.
The adaptiveness of the control strategies is also expected to vary depending
on whether the difficulty is controllable or uncontrollable. For controllable
difficulties, it is expected that primary control will be more adaptive and therefore
more positively related to job satisfaction than secondary control. As the situation is
controllable, it is likely that a person can change it.
For uncontrollable difficulties, secondary control will be more positively
related to job satisfaction than primary control. As the situation is uncontrollable, it
is unlikely that a person can change the situation using primary control, and thus
secondary control would be more adaptive than primary control failure.
Both types of difficulties (controllable and uncontrollable) are expected to be
directly and indirectly related to job satisfaction. Employees who report more
270
difficulties at work are expected to report lower job satisfaction. However two
variables that may moderate the relationship between work difficulties and job
satisfaction are primary and secondary control and social support at work. These
moderation effects are represented by the interaction terms in Figure 8.
In regard to primary and secondary control, it is proposed that primary
control strategies are useful in reducing the effects of work difficulties when the
situation is controllable. When the situation is uncontrollable, secondary control
strategies may reduce the effects of work difficulties on job satisfaction.
In regard to social support at work, it is expected that instrumental support
will buffer the effects of controllable difficulties on job satisfaction. Emotional
support is expected to buffer the effects of uncontrollable difficulties on job
satisfaction.
In addition to work difficulties, job autonomy and life satisfaction are
expected to directly predict job satisfaction. Both of these relationships have been
demonstrated in studies one and two.
In summary, controllable and uncontrollable work difficulties, the primary
and secondary control strategies used to handle such difficulties, and social support
at work, are expected to determine job satisfaction. Job satisfaction is, in turn,
expected to influence, and be influenced by, life satisfaction.
271
Figure 8- Revised Model of Job Satisfaction
Secondary Control
Instrumental
Support
Primary Control
Controllable
Diff x Primary
Control
Life Satisfaction
Uncontrollable
Diff x
Secondary
Control
Controllable Difficulties
Job Satisfaction
Uncontrollable Difficulties
Control Diff x
Instrumental
Support
Job
Autonomy
Uncontrol
Diff x
Emotional
Support
Secondary Control
Primary Control
Emotional Support
272
4.4
Hypotheses
1) When workers face controllable difficulties, they are expected to use primary
control more than secondary control. Conversely, when workers face uncontrollable
difficulties, they are expected to use secondary control more than primary control.
Workers should match their control strategies to the controllability of the
situation. When the situation is controllable, the workers will be likely to change it
using primary control. When the situation is uncontrollable, the workers will be
likely to accept the situation using secondary control.
2) When workers face controllable difficulties, primary control is expected to be
more positively related to job satisfaction than secondary control. Conversely, when
workers face uncontrollable difficulties, secondary control is expected to be more
positively related to job satisfaction than primary control.
This hypothesis is based on the discrimination model, which proposes that
primary control is more adaptive in controllable situations, and that secondary
control is more adaptive in uncontrollable situations.
3) Primary control will moderate the effect of controllable difficulties on job
satisfaction and secondary control will moderate the effect of uncontrollable
difficulties on job satisfaction.
This hypothesis proposes that primary control strategies are useful in
reducing the influence of work difficulties on job satisfaction when the situation is
273
controllable. When the situation is uncontrollable however, secondary control
strategies are expected to reduce the effects of work difficulties on job satisfaction.
4) Instrumental social support will moderate the effects of controllable work
difficulties on job satisfaction.
When the difficulty is controllable, it is proposed that other people may help the
person to overcome the problem.
5) Emotional social support will moderate the effects of uncontrollable work
difficulties on job satisfaction.
Emotional social support is expected to help reduce the influence of uncontrollable
difficulties on job satisfaction. In these cases, instrumental social support may not be
useful as the situation cannot be overcome, however emotional social support may
help reduce the severity of these difficulties.
6) Work difficulties, the control strategies used to handle work difficulties, social
support at work, job autonomy and life satisfaction, will predict job satisfaction.
These variables are expected to be major predictors of job satisfaction, as
demonstrated in Figure 8.
274
4.5
4.5.1
Method
Participants
The sample consisted of 214 general employees, obtained from a database
and through convenience sampling. The age of the participants ranged from 21-73
years, with the average being 44.78 years (SD=15.18). The demographic
characteristics of the sample are displayed in Table 36.
Table 36- Demographic Characteristics of the Sample
Variable
Gender
% sample
Male
Female
47.7
47.7
Professional
Business
Trade
Clerical
Retail
Labourer
Other
46.7
13.1
10.7
13.1
6.1
1.9
3.3
1-20
21-30
31-40
41-50
51-60
61+
13.6
14.0
27.6
27.1
12.1
4.7
Occupation
Hours worked per week
Bolded values indicate the largest proportion.
275
4.5.2
Materials
All of the respondents received a plain language statement (refer to Appendix
N) and a questionnaire. The questionnaire consisted of scales measuring controllable
and uncontrollable difficulties, primary and secondary control for such difficulties,
job autonomy, job satisfaction, life satisfaction and social support at work.
4.5.2.1
Controllable and Uncontrollable Difficulties
Controllable and uncontrollable difficulties were measured in the Situation
Specific Primary and Secondary Control Scale (Maher & Cummins, 2002; refer to
Appendix O). Although the most direct way to measure this variable would be to
ask, “how often do you experience difficulties that you can control/cannot control”,
this item was deemed to be too cognitively taxing and prone to errors. Rather the
employees were given a list of potential difficulties that they may face at work such
as supervisors, co-workers, kind of work, pay, work-place rules, promotion, time
management and others. They indicated which difficulties they experienced that
they could control.
In order to determine how frequently they experienced controllable
difficulties, they were asked to consider the difficulty that they experienced most
often and could control, and indicate how often they experienced it. This process
was repeated for uncontrollable difficulties.
Although this question only refers to one difficulty, it was deemed to be the
best method. One alternative is to ask them on average how often they experience
276
the difficulties they ticked. However, it is extremely difficult for employees to
mentally calculate how often they experience each difficulty they ticked and then
calculate the average. Another alternative would be to get them to indicate how
often they experienced each difficulty. This was problematic however, as there are
an unlimited number of work difficulties, particularly as this study is relying on a
general sample of employees.
4.5.2.2
Primary Control and Secondary Control
The Situation Specific Primary and Secondary Control Scale (Maher &
Cummins, 2002; refer to Appendix O) was used in this study. The main difference
between this scale and earlier scales is that the respondents now indicate which
control strategies they use for controllable difficulties and uncontrollable difficulties.
There are four primary control, and 12 secondary control strategies from which to
choose. There is also the option to list other strategies that they use. Each strategy is
rated on a five-point scale ranging from 0 (never) to 4 (always).
4.5.2.3
Job Autonomy
Job autonomy was measured by the job autonomy items in the Job Diagnostic
Survey (Hackman & Oldham, 1975). This scale consists of three items that assess
overall perceived job autonomy, such as “In my job, I can decide on my own how to
go about doing my work” (refer to Appendix B). The items were rated on a 10-point
scale ranging from 1 (do not agree at all) to 10 (agree completely). As discussed in
study two, the psychometric statistics are adequate.
277
4.5.2.4
Job Satisfaction
Job satisfaction was measured by a one-item measure. The item “taking into
consideration all the things about your job, how satisfied are you with it” was rated
on an 11-point scale. Although internal consistency cannot be established with a
single-item measure, this measure of job satisfaction has been shown to correlate
with other measures of job satisfaction, where r = 0.63 (Wanous et al., 1997).
In studies one and two, a facet measure of job satisfaction was also used. As
this facet measure was only used to gain insight into the particular occupational
groups, it is not necessary in this study which is relying on a general sample of
employees.
4.5.2.5
Life Satisfaction
Life satisfaction was measured by the Personal Well-being Index developed
recently by Cummins et al., (2001). This scale attempts to overcome some
methodological problems that were identified in the Comprehensive Quality of Life
Scale, which was used in studies one and two. Most of these problems involve the
objective scale, or the importance scale, rather than the satisfaction scale. However,
some problems were identified with the seven domains of life satisfaction, such as
the domain of emotional well-being or happiness refers to an affective state rather
than a domain of life (Cummins, 2002). Furthermore, the wording of some of the
items were not optimal (Cummins, 2002). As such, rather than happiness, future
278
security is included as a domain of life satisfaction. Thus, the Personal Well-being
Index consists of seven domains of satisfaction which are rated on an 11-point scale
(refer to Appendix P).
4.5.2.6
Social Support at Work
In study two, social support at work was measured by Karasek and Theorell’s
(1990) scale. This scale consisted of both instrumental and emotional support,
however the items were designed to be summed to provide an overall support score.
These items were examined more closely in this study to ensure that they were
appropriate for a range of occupations.
Some problems were identified with the instrumental support scale. For
example, the item, “my supervisor creates a good teamwork environment for me”
may not measure instrumental support. It is not necessary that an employees works
in a teamwork environment for them to receive instrumental support. To provide
instrumental support, the employer only needs to offer some material assistance. A
further problem with the scale concerns the item “my co-workers are competent.”
This item, although intended to measure instrumental support, only assesses the coworkers competence. Co-workers may indeed be competent, however this does not
mean that they offer assistance when required. These problems, although they were
not recognised in study two, will be rectified in this study.
A review of the other major measures of social support at work was
undertaken. This review demonstrated that although there are many scales that claim
to measure social support at work, few adequately measure emotional and
279
instrumental support. Some researchers rely on a one item measure, such as “how
true is it that your supervisors are warm/friendly when you have problems” and “how
true is it that your supervisor helps you complete a given task” (Himle & Jayaratne,
1991; Wong et al., 2000). These scales are criticised however for failing to capture
the different ways that emotional or instrumental support can be offered.
Some studies have relied on scales which only focus on emotional support
and fail to measure instrumental support (Dollard et al., 2000, Rodriguez, Bravo,
Peiro & Schaufeli, 2001). Others do not claim to measure either component (Caplan
et al., 1975; Van der Doef & Maes, 1999), whilst others still claim to measure five
types of social support (Unden, 1996).
One exploratory scale recently developed by Ducharme and Martin (2000)
does not appear to suffer from any of these problems. Their scale, developed only to
assess co-worker support, includes five items assessing instrumental support and five
items assessing emotional support. Consistently, a factor analysis on the scale
demonstrated that two factors emerged. To make this scale appropriate for the
current study, it was posited that the co-worker items could also be applied to
supervisors. If this were done, the scale would consist of 20 items. As this may be
unnecessarily long, some items were deleted. Specifically, only the three highest
loading items were selected each for instrumental and emotional support. For
emotional support, these were “your co-workers really care about you”, “you feel
close to your co-workers” and “your co-workers take a personal interest in you.”
These three items were also changed to be applicable to supervisors.
280
For instrumental support, the items were “your co-workers would fill in while
you’re absent”, “your co-workers are helpful in getting the job done” and “your
co-workers give useful advice on job problems.” These three items also needed to be
applicable to supervisors. As one of them was not (i.e., “your supervisors would fill
in while you were absent”), the next highest loading item in the factor analysis was
selected. This was “your co-workers assist with unusual work problems.” The
resulting scale is a six-item scale for co-worker support and a six-item scale for
supervisor support, that both assess instrumental and emotional support (refer to
Appendix Q).
4.5.3
Procedure
Ethics approval was obtained from Deakin University. The majority of the
sample was obtained from a database developed by Australian Unity and Deakin
University. This database contains information for 900 people that have been
randomly selected from the population, and have agreed to participate in a survey.
The employment status of these people was unknown, and as such two
questionnaires were sent to them, one if they were employed, and one if they were
unemployed. A total of 250 (27%) questionnaires were returned however only 130
(14.44%) of these were completed by people that were employed. The remainder of
the sample was obtained through convenience, and snowballing.
281
4.6
4.6.1
Results
Data Screening and Checking of Assumptions
The data set was examined for missing values, outliers, normality and
linearity. There were very few missing values for measures of life satisfaction, job
satisfaction, job autonomy, job demands, and co-worker support (i.e., < 5%). There
was a higher rate of missing values for supervisor support (19%) and primary and
secondary control (6%, 11%, respectively). The treatment of these values depended
on their context. If the participant had completed the majority of the scale, the
missing values were replaced with the group mean. If however, the person had failed
to complete any of the scale, they were excluded from analyses using that scale.
Overall, this treatment resulted in less than 5% of the missing values being replaced
with the group mean.
Univariate outliers were identified in the measures of life satisfaction
(2 cases), job satisfaction (3 cases), job autonomy (1 case), co-worker support
(4 cases), and primary and secondary control (3 cases). These values were re-coded
to lie within three standard deviations of the mean.
Many of the scales were negatively skewed, where the skew/standard
error > 3. These include life satisfaction (-4.44), job satisfaction (-5.09), job
autonomy (-7.07), and co-worker and supervisor support (-4.56, -5.15, respectively).
As transformations are not recommended for variables that are mildly and naturally
282
skewed (Tabachnick & Fidell, 1997), these variables were not transformed.
Reasonably linear relationships were evident among the variables.
4.6.2
Descriptive Statistics and Inter-Correlations
Table 37 contains the means and standard deviations of the major variables.
This table demonstrates that employees are using primary more than secondary
control in controllable and uncontrollable situations. Other interesting findings are
that employees report that their co-workers offer more instrumental and emotional
support than their employers. Additionally, the level of life satisfaction is in the
expected range.
Table 38 displays the correlations among the major variables. Several
variables correlate well with job satisfaction including job autonomy, life
satisfaction, social support at work, and difficulties at work. Strong correlations are
also observed among primary control in controllable situations and primary control
in uncontrollable situations. Similarly, secondary control in controllable situations is
strongly correlated to secondary control in uncontrollable situations.
283
Table 37- Means and Standard Deviations of the Major Variables
Variable
Job Satisfaction
Job Autonomy
Life Satisfaction-domain
Co-worker emotional support
Co-worker instrumental support
Supervisor emotional support
Supervisor instrumental support
Frequency of controllable difficulties
Frequency of uncontrollable difficulties
Primary control for controllable difficulty
Secondary control for controllable difficulty
Primary control for uncontrollable difficulty
Secondary control for uncontrollable difficulty
M
71.67
78.18
73.68
74.63
78.03
65.86
70.55
58.70
57.07
71.98
53.74
65.56
53.10
SD
18.41
20.34
13.76
20.81
21.30
26.66
26.32
20.33
22.25
13.05
12.12
16.73
12.68
All scores have been converted to a percentage of scale maximum (%SM) which
ranges from 0-100. The formula is (mean score for the original domain-1) x 100/
(number of scale points –1)
Table 38- Inter-Correlations among Major Variables
JS
JA
Cdif
Udif
PcC
ScC
PcU
ScU
LS
Sup
Cow
JS
JA
Cdif
Udif
PcC
ScC
PcU
ScU
LS
Sup
0.57
-0.37
-0.35
0.27
-0.12
0.23
-0.04
0.46
0.37
0.48
-0.22
-0.32
0.34
-0.25
0.18
-0.16
0.34
0.40
0.41
0.42
-0.13
-0.03
-0.12
-0.05
-0.14
-0.21
-0.14
-0.14
0.10
-0.11
0.10
-0.13
-0.24
-0.14
-0.03
0.69
0.07
0.13
0.10
0.07
0.06
0.64
-0.13
0.04
-0.06
0.20
0.19
0.08
0.09
0.08
0.04
0.26
0.31
0.57
Bolded items p<0.01
JS - job satisfaction; JA - job autonomy; Cdif - controllable difficulties;
Udif -uncontrollable difficulties; PcC - primary control for controllable difficulties;
ScC - secondary control for controllable difficulties; PcU - primary control for
uncontrollable difficulties; ScU - secondary control for uncontrollable difficulties;
LS - life satisfaction; Sup - supervisor support; Cow - co-worker support.
284
4.6.3
Factor Analyses
Factor analyses were conducted on the two exploratory scales in the study
measuring primary and secondary control strategies and social support at work.
4.6.4
Primary and Secondary Control Scale
Two factor analyses were required to examine the primary and secondary
control items for a controllable difficulty and an uncontrollable difficulty. For both
of these analyses, the assumptions were met where Bartlett’s test of sphericity was
significant, and Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy
exceeded 0.60.
For controllable difficulties, a principle component analysis with direct
oblimin rotation yielded five factors. These five factors accounted for 53.8% of the
variance, however only 6 of the 16 items loaded on only one factor, and as such, this
five-factor solution could not be interpreted. The analysis was repeated using a fourfactor solution. Ten items loaded on only one factor, and the primary control items
loaded on a separate factor to the secondary control items. However, there was no
pattern to the secondary control items that loaded on the remaining three factors. In
response to this, a three-factor analysis was conducted. This analysis was much
clearer, demonstrating that all of the primary control items loaded on Factor 1, and
the secondary control items were mainly divided among Factor 2 and Factor 3.
As demonstrated in Table 39, Factor 1 consists of all four primary control
items. Two of the secondary control items negatively loaded on this factor as well,
285
and as such, they will be excluded from the analyses. Factors 2 and 3 contain the
remaining secondary control items. The items that loaded on Factor 2 were support,
vicarious, past success, behavioural avoidance and positive re-interpretation, whilst
the items that loaded on Factor 3 were predictive-negative, wisdom, and attribution.
There is a theoretical distinction among these items that was introduced in chapter 3.
As demonstrated in Table 19, two functions of secondary control are posited, namely
self-protective and self-affirmative.
Self-protective secondary control strategies reduce the negative impact of the
situation, whilst self-affirmative secondary control strategies increase positive
feelings about the self. All of the items that loaded on Factor 2 involve
self-affirmation, whilst the items that loaded on Factor 3 involve self-protection. It
must be noted however that some strategies that were expected to load on the two
factors did not. The self-affirmative strategy of downward social comparison did not
load on Factor 2. Furthermore, the self-protective strategies of goal disengagement,
illusory optimism, and denial did not load on Factor 3. Despite this however, overall
the factor analysis supports the two types of secondary control.
286
Table 39- Factor Analysis of Primary and Secondary Control Item in
Controllable Situations
Item
Pc1
Pc4
Pc7
Pc10
Sc2
Sc3
F1
0.56
0.70
0.42
0.60
-0.61
F2
F3
Discuss solutions with the people involved.
Choose a solution and act on it.
Work harder.
Keep trying.
Think that the difficulty doesn’t matter.
Think that this difficulty will work out okay
in the end.
Sc5 Think that I knew this difficulty would
0.68
happen.
Sc6 Think that I can’t always get what I want.
0.73
Sc8
Think that I am better off than many other
people.
Sc9 Think that this difficulty is not my fault.
0.42
Sc11 Tell someone about this difficulty to make me
0.54
feel better.
Sc12 Think of the success of my family/friends.
0.61
Sc13 Think about my success in other areas.
0.74
Sc14 Do something different, like going for a walk.
0.66
Sc15 Ignore this difficulty.
-0.63
Sc16 Look for something else that is positive in the
0.65
situation.
Eigenvalues 2.40
2.25
1.46
% Variance 15.01 14.04 9.12
Cumulative variance 15.01 29.05 36.17
Items with loadings less than 0.30 are not shown.
Self-protective secondary control items are bolded.
Self-affirmative secondary control items are italicised.
Factor 1- Primary Control, Factor 2- Self-affirmative, Factor 3- Self-protective
In addition to the controllable difficulties, a principal component analysis
with direct oblimin rotation was conducted on the strategies used for uncontrollable
difficulties. This analysis yielded six factors, however as only one item loaded on
one factor, a five-factor solution, and a four-factor solution were requested. Both
analyses could not be interpreted, as there was no pattern to the secondary control
287
items. As such, a three-factor solution was conducted. This solution was remarkably
similar to the analysis of controllable situations. As demonstrated in Table 40, all
four primary control items loaded on Factor 1, and the secondary control items
loaded on their respective types of secondary control (i.e., self-protective and
self-affirmative). As with the analysis for controllable situations however, there are
a few exceptions, where goal disengagement and denial loaded on Factor 1. Unlike
the controllable analysis, Factor 2 included the self-affirmative strategy of downward
social comparison, however it also included illusory optimism, which is a selfprotective strategy. Finally, Factor 3 was the same in both analyses where it
excluded goal disengagement, illusory optimism, and denial.
288
Table 40- Factor Analysis of Primary and Secondary Control Items in
Uncontrollable Situation
Item
Pc1
Pc4
Pc7
Pc10
Sc2
Sc3
F1
0.66
0.70
0.59
0.61
-0.68
F2
F3
Discuss solutions with the people involved.
Choose a solution and act on it.
Work harder.
Keep trying.
Think that the difficulty doesn’t matter.
Think that this difficulty will work out okay
0.41
in the end.
Sc5 Think that I knew this difficulty would
0.63
happen.
Sc6 Think that I can’t always get what I want.
0.73
Sc8
Think that I am better off than many other
0.58
people.
Sc9 Think that this difficulty is not my fault.
0.54
Sc11 Tell someone about this difficulty to make me
0.39
feel better.
Sc12 Think of the success of my family/friends.
0.57
Sc13 Think about my success in other areas.
0.70
Sc14 Do something different, like going for a walk.
0.63
Sc15 Ignore this difficulty.
-0.65
Sc16 Look for something else that is positive in the
0.66
situation.
Eigenvalues 2.90
2.29
1.60
% Variance 18.16 14.33 10.02
Cumulative variance 18.16 32.49 42.51
Items with loadings less than 0.30 are not shown.
Self-protective secondary control items are bolded.
Self-affirmative secondary control items are italicised.
Factor 1- Primary Control, Factor 2- Self-affirmative, Factor 3- Self-protective
Commonalties among the two factor analyses were examined. This study
aims to compare the primary and secondary control strategies for controllable with
uncontrollable difficulties, and as such, the comparisons should be based on the same
items. For primary control, all four items loaded on Factor 1 in both analyses, and as
289
such they will all be included. For self-affirmative secondary control, the common
items were sc11, sc12, sc13, sc14, and sc16. As such, self-protective secondary
control will be measured by these items (i.e., support, vicarious, present success,
active avoidance, and positive re-interpretation). For self-protective secondary
control, the common items were sc5, sc6, and sc9, which are predictive-negative,
wisdom, and attribution. Table 41 demonstrates which self-protective and selfaffirmative strategies were included in the analyses.
Table 41-Secondary Control Items included in Analyses
Type of Strategy
Self-protective
Self-protective
Self-protective
Self-protective
Self-protective
Self-protective
Strategy Item
Attribution
Predictive-Negative
Wisdom
Goal Disengagement
Illusory Optimism
Denial
Current Study
Self-protective
Self-protective
Self-protective
----
Self-affirmative
Self-affirmative
Self-affirmative
Self-affirmative
Self-affirmative
Self-affirmative
Support
Vicarious
Present Success
Active Avoidance
Positive Re-interpretation
Downward Social Comparison
Self-affirmative
Self-affirmative
Self-affirmative
Self-affirmative
Self-affirmative
--
4.6.5
Social Support at Work
A principal components factor analysis with direct oblimin rotation was
conducted on the social support scale to ensure that the items were measuring four
types of support, namely co-worker instrumental support, co-worker emotional
support, supervisor instrumental support and supervisor emotional support. The
assumptions were met where Bartlett’s test of sphericity was significant, and
290
Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy exceeded 0.60.
As demonstrated in Table 42, all of the supervisor items loaded on Factor 1,
and all of the co-worker items loaded on Factor 2. The instrumental support and the
emotional support items for supervisors and co-workers did not load on different
factors. As these factor loadings are very high, the scale should only be divided into
co-worker support and emotional support. However, as the hypotheses refer to the
specific types of support, the instrumental and emotional support items will remain
separate.
Table 42- Factor Analysis of the Social Support at Work Scale
Item
My co-workers really care about me.
I feel close to my co-workers.
My co-workers take a personal interest in me.
My co-workers assist with unusual work problems.
My co-workers are helpful in getting the job done.
My co-workers give useful advice on job problems.
My supervisor really cares about me.
I feel close to my supervisor.
My supervisor takes a personal interest in me.
My supervisor assists with unusual work problems.
My supervisor is helpful in getting the job done.
My supervisor gives useful advice on job problems.
Eigenvalues
% Variance
Cumulative % variance
Items with loadings less than 0.30 are not shown.
F1
0.86
0.91
0.88
0.91
0.93
0.93
7.34
61.97
61.97
F2
0.82
0.88
0.93
0.88
0.91
0.77
2.05
17.05
79.02
291
4.7
Hypothesis Testing
In order to test the hypotheses, repeated measures analyses of variance were
conducted to compare the use of control strategies in different situations, and
multiple regression analyses were used to examine the adaptiveness of the control
strategies. Hierarchical regression analyses were conducted to examine the
moderating role of the control strategies and social support on the relationship
between work difficulties and job satisfaction. Finally, a standard multiple
regression analysis was conducted to examine the major predictors of job
satisfaction. As in study two, the alpha level was reduced to 0.01 in order to reduce
the risk of Type I errors.
4.7.1
Hypothesis One- Use of Control Strategies for Controllable and
Uncontrollable Difficulties
Hypothesis one proposes that when workers face controllable difficulties,
they use more primary control than secondary control, and when workers face
uncontrollable difficulties, they use more secondary control than primary control.
Before examining this hypothesis, the types of difficulties that employees are
reporting as being controllable or uncontrollable are examined. As demonstrated in
Table 43, the most common controllable difficulties were with time management,
motivation and co-workers. The most uncontrollable difficulties were with pay,
amount of work, work-place rules, and promotion.
292
Table 43 - Controllable and Uncontrollable Difficulties Reported by Employees
Difficulty
Difficulties with supervisors
Difficulties with co-workers
Difficulties with kind of work
Difficulties with pay
Difficulties with work-place rules
Difficulties with promotion
Difficulties with time management
Difficulties with motivation
Difficulties with work times
Difficulties with amount of work
Other
% yes and
controllable
38.8
60.7
52.8
23.4
30.4
12.1
62.1
62.6
49.5
42.1
4.7
% yes and
uncontrollable
35.5
17.8
29.9
46.3
44.4
40.2
19.6
11.2
23.4
44.9
4.2
To test whether workers use more primary control when they face
controllable difficulties and more secondary control when they face uncontrollable
difficulties, two repeated-measures analyses of variance were conducted. The
variables were normally distributed, and the homogeneity of variance was met.
Keppel (1991) proposes that the adequacy of the sphericity tests has been questioned
and as such, researchers should assume that sphericity does not hold, and make the
adjustments.
For the controllable difficulties, Mauchly’s test of sphericity was not violated.
However, Greenhouse-Geisser epsilon was greater than 0.75, and as such, the
degrees of freedom were calculated using the Huynh-Feldt epsilon. The difference
was significant, F (1.984, 400.712) = 99.64, p = 0.00. As demonstrated in Table 44,
employees reported significantly more primary control than affirmative secondary
control, F (1, 202) = 187.94, p = 0.00, and self-protective secondary control,
293
F (1, 202) = 32.27, p = 0.00. Employees did not report more protective secondary
control than affirmative secondary, F (1, 202) = 1.20, p = 0.27. Thus, consistent with
the first part of hypothesis one, employees reported more primary control than
secondary control for a controllable difficulty.
For the uncontrollable difficulties, Greenhouse-Geisser epsilon was greater
than 0.75, and as such, Huynh-Feldt epsilon was used. This demonstrated that the
difference was significant, F (2, 386) = 38.57, p = 0.00. As demonstrated in Table
44, employees reported more primary control than self-affirmative secondary control,
F (1, 193) = 80.27, p = 0.00, and self-protective secondary control,
F (1, 193) = 29.94, p = 0.00. Employees also reported more self-protective
secondary control than self-affirmative secondary control, F (1, 193) = 8.78,
p = 0.003. Overall however, inconsistent with the second part of hypothesis one,
employees reported more primary than secondary control for uncontrollable
difficulties.
Table 44- Employees Use of Primary and Secondary Control in Controllable
and Uncontrollable Situations
Situation
Controllable
Strategy
Primary Control
Secondary Control- Self affirmation
Secondary Control- Self protective
M
71.98
52.88
54.60
SD
13.05
16.70
16.21
Uncontrollable
Primary Control
Secondary Control-Self-affirmation
Secondary Control- Self-protective
64.56
50.74
55.45
16.73
16.76
16.90
294
To determine the most commonly used control strategies to handle
controllable and uncontrollable difficulties, the means for each individual strategy
were examined. As demonstrated in Table 45, the most commonly used primary
control strategy for controllable and uncontrollable difficulties is “keep trying” and
the least common strategy is “discuss solutions with the people involved.” The most
common secondary control strategy for controllable and uncontrollable difficulties is
“think that I am better off than many other people.” The least common secondary
control strategy for controllable difficulties is “ignore this difficulty” and for
uncontrollable difficulties is “think that the difficulty doesn’t matter.”
295
Table 45- Means and Standard Deviations of Individual Control Strategies
Strategy
Discuss solutions with the people involved.
(pc)
Choose a solution and act on it. (pc)
Work harder. (pc)
Keep trying. (pc)
Think that the difficulty doesn’t matter. (sc)
Think that this difficulty will work out okay in
the end. (sc)
Think that I knew this difficulty would happen.
(sc)
Think that I can’t always get what I want. (sc)
Think that I am better off than many other
people. (sc)
Think that this difficulty is not my fault. (sc)
Tell someone about this difficulty to make me
feel better. (sc)
Think of the success of my family/friends. (sc)
Think about my success in other areas. (sc)
Do something different, like going for a walk.
(sc)
Ignore this difficulty. (sc)
Look for something else that is positive in the
Situation. (sc)
Controllable
M
SD
66.00 23.64
Uncontrollable
M
SD
55.03 27.00
73.28
67.49
77.58
33.99
53.33
19.35
20.56
18.52
26.72
27.78
61.73
63.92
77.58
30.15
49.36
25.21
24.40
18.52
26.19
25.63
55.67
25.19
55.67
25.19
51.42
67.49
24.83
22.98
51.42
67.49
24.83
22.98
56.16
56.40
23.44
27.51
56.16
56.44
23.44
26.39
46.43
54.56
45.81
27.30
23.48
26.69
41.23
53.22
44.20
28.00
24.53
28.14
24.87
61.21
22.59
23.08
30.54
58.63
26.78
22.61
Bolded strategies have highest frequencies
4.7.2
Hypothesis Two- Adaptiveness of the Control Strategies for
Controllable and Uncontrollable Difficulties
Hypothesis two proposes that primary control is more positively related to job
satisfaction than secondary control for controllable difficulties and that secondary
control is more positively related to job satisfaction than primary control for
uncontrollable difficulties. In order to test this hypothesis, two standard multiple
regression analyses were conducted.
296
The assumptions of normality, linearity, and homoscedasticity of residuals
were met for both analyses, and there was no evidence of multicollinearity. For the
controllable situation, R was significantly different from zero, R = 0.26,
F (3, 199) = 4.63, p = 0.004. As demonstrated in Table 46, only primary control
predicted job satisfaction, accounting for 5% of the variance. Thus, consistent with
the first part of hypothesis two, primary control was more positively correlated with
job satisfaction than secondary control for controllable difficulties.
For the uncontrollable difficulties, R was also significantly different from
zero, R = 0.29, F (3, 190) = 5.87, p = 0.001. As with the controllable difficulty, only
primary control predicted job satisfaction, accounting for almost 6% of the variance.
Thus, inconsistent with the second part of hypothesis two, primary control was more
positively correlated with job satisfaction than secondary control for uncontrollable
difficulties.
297
Table 46- Standard Multiple Regression Analysis Predicting Job Satisfaction
From Primary and Secondary Control
Difficulty
Control
Var
JS
PC
PC
SC- A
SC- P
0.24**
0.03
-0.01
0.12
-0.01
SC-A
B

sr2(unique)
0.24
0.006
-0.08
5.38**
0.09
0.33
0.007
-0.08
R =0.26**
R2=0.07
AdjR2=0.05
0.28
0.11
-0.11
0.25
0.01
0.08
5.95**
R =0.29**
R2=0.09
AdjR2=0.07
Uncontrol
PC
SC-A
SC-P
0.26**
0.13
-0.08
0.18
0.05
0.14
**p<0.01
Control - controllable difficulty; Uncontrol - uncontrollable difficulty; JS- job
satisfaction; PC - primary control; SC-A - self-affirmative secondary control;
SC-P- self-protective secondary control.
In order to determine the most adaptive control strategies for controllable and
uncontrollable difficulties, the correlations between each individual strategy and job
satisfaction are presented in Table 47. This table demonstrates that the most adaptive
primary control strategy for both types of difficulties is “keep trying” and the least
adaptive is “work harder.” The most adaptive secondary control strategy for a
controllable difficulty is “think that I am better off than many other people” and in an
uncontrollable situation is “look for something else that is positive in the situation.”
The least adaptive secondary control strategy in both situations is “ignore this
difficulty.”
298
Table 47- Correlations between Individual Control Strategies and Job
Satisfaction for Controllable and Uncontrollable Difficulties
Strategy
Controllable Uncontrollable
Discuss solutions with the people involved. (pc)
Choose a solution and act on it. (pc)
Work harder. (pc)
Keep trying. (pc)
Think that the difficulty doesn’t matter. (sc)
Think that this difficulty will work out okay in the
end. (sc)
Think that I knew this difficulty would happen.
(sc)
Think that I can’t always get what I want. (sc)
Think that I am better off than many other
people. (sc)
Think that this difficulty is not my fault. (sc)
Tell someone about this difficulty to make me feel
better. (sc)
Think of the success of my family/friends. (sc)
Think about my success in other areas. (sc)
Do something different, like going for a walk. (sc)
Ignore this difficulty. (sc)
Look for something else that is positive in the
situation. (sc)
0.11
0.23**
0.09
0.24**
-0.22**
0.13
0.19**
0.15
0.14
0.29**
-0.03
0.12
-0.09
-0.08
-0.02
0.22**
0.006
0.17
-0.10
-0.08
-0.09
-0.03
-0.04
0.12
-0.02
-0.25**
0.15
0.004
0.19**
-0.009
-0.19
0.31**
**p<0.01
4.7.3
Hypothesis Three- The Moderating Role of Primary and Secondary
Control
In order to test hypothesis three proposing that primary control moderates the
effect of controllable difficulties on job satisfaction and secondary control moderates
the effect of uncontrollable difficulties on job satisfaction, two hierarchical multiple
regression analyses were conducted.
299
The assumptions of normality, linearity, and homoscedasticity of residuals
were met for both analyses, and there was no evidence of multicollinearity. As
demonstrated in Figure 9, difficulties (i.e., controllable or uncontrollable) were
entered in the first step. In the second step the moderator variable was entered (i.e.,
primary or secondary control), and in the third step the interaction term was entered.
300
Figure 9 – Primary and Secondary Control Moderate the Relationship between
Work Difficulties and Job Satisfaction
Order of Variable Entry
a)
Step 1
Controllable difficulties
Step 2
Primary control
Step 3
Controllable difficulties x
primary control
Job Satisfaction
b)
Step 1
Uncontrollable difficulties
Step 2
Secondary control
Step 3
Uncontrollable Difficulties
x secondary control
Job Satisfaction
In the primary control analysis, R was significantly different from zero after
controllable difficulties were entered, R = 0.32, F (1, 199) = 22.91, p = 0.00. As
demonstrated in Table 48, R increased after primary control was added, R = 0.39,
Finc (1, 198) = 12.05, p = 0.001. The addition of the interaction term however did
not increase R, where R = 0.41, Finc (1, 197) = 2.99, p = 0.09. Thus, inconsistent
with hypothesis three, primary control did not moderate the effect of controllable
work difficulties on job satisfaction.
In the secondary control analysis, R was significantly different from zero
after uncontrollable difficulties were entered, R = 0.35, F (1, 189) = 26.68, p = 0.00.
301
As demonstrated in Table 48, R did not increase when secondary control was added,
R = 0.36, Finc (1, 188) = 1.73, p = 0.19, or when the interaction term was added,
R = 0.37, F (1, 187) = 1.35, p = 0.25. As such, inconsistent with hypothesis three,
secondary control did not moderate the effect of uncontrollable work difficulties on
job satisfaction.
302
Table 48- Hierarchical Multiple Regression testing the Moderating Role of
Control Strategies on the Relationship between Work Difficulties and Job
Satisfaction
Step
1
2
3
1
2
3
IV
Controllable difficulties
Controllable difficulties
Primary control
Controllable difficulties
Primary Control
Controllable difficulties x
Primary control
Uncontrollable difficulties
Uncontrollable difficulties
Secondary Control
Uncontrollable difficulties
Secondary control
Uncontrollable difficulties
x Secondary control
**p<0.01; JS – Job satisfaction
DV
JS
JS
JS
JS
JS
JS
B
-0.29

-0.32
sr2(unique)
10.30**
R =0.32**
R2=0.10
AdjR2=0.10
-0.28
0.32
-0.31
0.23
9.24**
5.15**
R =0.39**
R2=0.16
AdjR2=0.15
-0.81
-0.08
0.007
-0.89
-0.06
0.65
2.82**
R =0.41
R2=0.17
AdjR2=0.16
-0.29
-0.35
12.39**
R =0.35**
R2=0.12
AdjR2=0.12
-0.30
-0.36
12.89**
R =0.36
R2=0.13
AdjR2=0.12
-0.56
-0.13
0.004
-0.68
-0.09
0.39
R =0.37
R2=0.14
AdjR2=0.12
303
4.7.3.1
Summary
Employees reported using more primary than secondary control strategies for
both controllable and uncontrollable difficulties. Primary control was more
positively related to job satisfaction than secondary control for both types of
difficulties. Although primary control was positively related to job satisfaction, it
did not moderate the effect of controllable difficulties on job satisfaction. Similarly,
secondary control did not moderate the effect of uncontrollable work difficulties on
job satisfaction.
4.7.4
Hypothesis Four - Moderating Role of Instrumental Support
To test hypothesis four, proposing that instrumental support moderates the
relationship between controllable difficulties and job satisfaction, hierarchical
regression analyses were conducted for co-workers and supervisors. The
assumptions of normality, linearity, and homoscedasticity of residuals were met for
both analyses, and there was no evidence of multicollinearity. In the first step
difficulties were entered (i.e., controllable or uncontrollable). In the second step the
moderator variable was entered (i.e., co-worker instrumental or supervisor
instrumental), and in the third step the interaction term was entered.
For co-workers instrumental support, R was significantly different from zero
after step one, R = 0.32, F (1, 198) = 22.74, p = 0.00, where controllable difficulties
accounted for 10% of the variance in job satisfaction. R increased after step two,
304
R = 0.49, Finc (1, 197) = 30.72, p = 0.00, where difficulties and co-worker
instrumental support accounted for 8% and 14% of the variance, respectively. R did
significantly increase when the interaction term was added in step 3, R = 0.50,
Finc (1, 196) = 4.30, p = 0.03, where the interaction term accounted for 1.64% of the
variance. This analysis, displayed in Table 49, is consistent with hypothesis four.
Although not significant at 0.01, this analysis, as in study two, will be
examined further as only a few studies have examined the moderating role of social
support. Furthermore, as discussed in study two, it is difficult to achieve statistical
significance in moderation analyses as the power is low (Bobko, 2001).
Controllable work difficulties were regressed on job satisfaction separately
for those with low co-worker support, and those with high co-worker support. As
proposed by Cohen and Cohen (1983), the low and high distinction was defined as
scores that fell one standard deviation below or above the mean for supervisor
support. As demonstrated in Figure 10, the regression lines were consistent with the
hypothesis, where the slope of the regression line of controllable work difficulties on
job satisfaction was steeper for high co-worker instrumental support than for low coworker instrumental support.
305
Figure 10 - Regression of Controllable Work Difficulties on Job Satisfaction for
Employees with Low Instrumental Co-Worker Support and Employees with
High Instrumental Co-Worker Support
100
90
y = 0.0231x + 76.84
80
70
Job Satisfaction
y = -0.0058x + 64.518
60
50
Low co-worker
instrumental
support
40
High co-worker
instrumental
support
30
20
10
0
0
10
20
30
40
50
60
70
Controllable Difficulites
80
90 100
306
For supervisors instrumental support, R was significantly different from zero
after step one, R =0.36, F (1, 176) = 25.92, p = 0.00. As demonstrated in Table 49, R
increased after supervisors instrumental support was added, R = 0.42,
Finc (1, 175) = 10.73, p = 0.001, however it did not increase further when the
interaction term was added, R = 0.42, F (1, 174) = 0.06, p = 0.80. Thus, inconsistent
with hypothesis four, supervisor instrumental support did not moderate the effect of
controllable work difficulties on job satisfaction.
307
Table 49- Hierarchical Regression Analyses Testing the Moderating Role of
Instrumental Support
Step
1
2
3
1
2
3
IV
Controllable difficulties
Controllable difficulties
Co-workers instrumental
Controllable difficulties
Co-workers instrumental
Controllable difficulties x
Co-workers instrumental
Controllable difficulties
Controllable difficulties
Supervisors instrumental
Controllable difficulties
Supervisors instrumental
Controllable difficulties x
Supervisors instrumental
DV
JS
JS
JS
JS
JS
JS
B
-0.28

-0.32
sr2(unique)
10.30**
R =0.32**
R2=0.10
AdjR2=0.10
-0.26
0.31
-0.29
0.37
8.49**
14.56**
R =0.49**
R2=0.24
AdjR2=0.23
-0.64
0.01
0.005
-0.72
0.02
0.55
4.20
.
1.64*
R =0.50*
R2=0.25
AdjR2=0.24
-0.33
0.17
12.81**
R =0.36**
R2=0.13
AdjR2=0.12
-0.28
0.16
-0.31
0.23
8.94**
5.02**
R =0.42**
R2=0.18
AdjR2=0.17
-0.32
0.13
0.0005
-0.35
0.18
0.06
R =0.42
R2=0.18
AdjR2=0.17
*p<0.05, **p<0.01; JS- Job satisfaction
4.7.5
Hypothesis Five- Moderating Role of Emotional Support
In order to test hypothesis five, proposing that emotional support moderates
the effect of uncontrollable work difficulties on job satisfaction, two hierarchical
308
multiple regression analyses were conducted. The assumptions of normality,
linearity, and homoscedasticity of residuals were met for both analyses, and there
was no evidence of multicollinearity.
For co-workers, R was significantly different from zero after uncontrollable
difficulties had been entered, R = 0.33, F (1, 192) = 24.48, p = 0.00. R significantly
increased after co-worker emotional support was entered, R = 0.55,
Finc (1, 191) = 51.32, p = 0.00. However, as demonstrated in Table 50, the addition
of the interaction term in step three was not significant, R = 0.56,
Finc (1, 190) = 1.85, p = 0.18. Thus, inconsistent with hypothesis five, co-worker
emotional support did not moderate the effect of uncontrollable work difficulties on
job satisfaction.
For supervisors, R was significantly different from zero after step one,
R = 0.36, F (1, 173) = 24.95, p = 0.00. As demonstrated in Table 50, R did
significantly increase after supervisors emotional support was entered, R = 0.45,
Finc (1, 172) = 16.90, p = 0.00, but did not increase further when the interaction term
was added, R =0.46, F (1, 171) = 0.76, p = 0.39. Inconsistent with hypothesis five,
co-worker and supervisor emotional support did not moderate the effect of
uncontrollable work difficulties on job satisfaction.
309
Table 50- Hierarchical Regression Analyses Testing the Moderating Role of
Emotional Support
Step
1
2
3
1
2
3
IV
Uncontrollable difficulties
Uncontrollable difficulties
Co-workers emotional
Uncontrollable difficulties
Co-workers emotional
Uncontrollable difficulties
x Co-workers emotional
Uncontrollable difficulties
Uncontrollable difficulties
Supervisors emotional
Uncontrollable difficulties
Supervisors emotional
Uncontrollable difficulties
x Supervisors emotional
DV
JS
JS
JS
JS
JS
JS
*p<0.05, **p<0.01; JS – Job satisfaction
B
-0.28

-0.34
sr2(unique)
11.29**
R =0.34**
R2=0.11
AdjR2=0.11
-0.23
0.38
-0.28
0.43
7.67**
18.75**
R =0.55**
R2=0.30
AdjR2=0.29
-0.45
0.20
0.003
-0.55
0.23
0.33
2.58**
R =0.56
R2=0.31
AdjR2=0.30
-0.30
-0.36
11.22**
R =0.36**
R2=0.13
AdjR2=0.12
-0.24
0.20
-0.29
0.29
7.95**
7.84**
R =0.45**
R2=0.20
AdjR2=0.20
-0.37
0.09
0.002
-0.44
0.13
0.20
2.62*
R =0.46
R2=0.21
AdjR2=0.19
310
4.7.5.1
Summary
In summary, only co-worker instrumental support moderated the effect of
difficulties on job satisfaction. Supervisor instrumental support, supervisor
emotional support and co-worker emotional support did not act as moderators.
4.7.6
Hypothesis Six- Major Predictors of Job Satisfaction
In order to test hypothesis six, proposing that job autonomy, difficulties at
work, control strategies, social support at work, and life satisfaction predict job
satisfaction, a standard regression analysis was conducted. The assumptions of
normality, linearity, and homoscedasticity of residuals were met for both analyses,
and there was no evidence of multicollinearity.
When all of the variables were entered, R was significantly different from
zero, R = 0.74, F (12, 142) = 13.92, p = 0.00. As demonstrated in Table 51,
controllable difficulties, job autonomy, life satisfaction, and co-workers emotional
support uniquely predicted job satisfaction. Inconsistent with hypothesis six
however, control strategies, uncontrollable difficulties, and supervisor social support
did not uniquely predict job satisfaction.
311
Table 51- Standard Multiple Regression Predicting Job Satisfaction
Predictor
Controllable difficulties
Uncontrollable difficulties
Job autonomy
Secondary control- Controllable situation
Secondary control- Uncontrollable
Situation
Primary control- Controllable situation
Primary control-Uncontrollable situation
Life Satisfaction
Co-workers emotional support
Co-workers instrumental support
Supervisors emotional support
Supervisors instrumental support
B
-0.18
-0.11
0.24
1.30
-0.49

-0.18
-0.13
0.27
0.03
-0.01
3.15
0.72
0.31
0.17
0.14
-0.04
-0.02
0.09
0.03
0.19
0.19
0.16
-0.06
-0.02
R =0.74**
R2=0.54
sr2(unique)
2.62**
4.08**
4.41**
1.49*
AdjR2=0.50
*p<0.05, **p<0.01; R is composed of 15.52% unique variance and 84.48% shared
variance.
4.7.6.1
Summary
When all of the variables in the hypothesised model of job satisfaction were
entered into a regression equation, they accounted for 54% of the variance.
However, only controllable difficulties, job autonomy, life satisfaction, and
co-worker emotional support uniquely predicted job satisfaction.
4.7.7
Conclusion
Employees reported using more primary than secondary control in both
controllable and uncontrollable situations. Primary control was more adaptive than
secondary control in both situations and was positively correlated with job
satisfaction. However, primary and secondary control did not moderate the effect of
312
work difficulties on job satisfaction. Co-worker instrumental support did moderate
the effect of controllable work difficulties on job satisfaction, however supervisor
instrumental support did not. Furthermore, emotional support did not moderate the
effect of uncontrollable difficulties on job satisfaction. The major predictors of job
satisfaction were controllable difficulties, job autonomy, life satisfaction and
co-worker instrumental support. These findings will now be discussed.
313
4.8
Discussion
The study proposed that the controllability of a work difficulty influences the
use and adaptiveness of the control strategies used to handle that difficulty. The
findings demonstrated, however, that for both controllable and uncontrollable
difficulties, primary control strategies were used more than secondary control
strategies, and primary control strategies were more adaptive than secondary control
strategies. These findings, which are inconsistent with the discrimination model,
suggest that trait control strategies may exist. The proposal that employees use
similar control strategies in all situations questions the assumption that employees
using primary control in uncontrollable situations will experience primary control
failure.
The results from this study also question the importance of the control
strategies, as they, along with social support, did not moderate the effect of work
difficulties on job satisfaction. These findings must be regarded with caution
however as limitations have now been identified in the operationalisation of work
difficulties. These hypotheses will now be examined.
4.9
Hypotheses Testing
The hypotheses can be grouped into three major proposals. The first proposal
is that the controllability of a difficulty influences the use and adaptiveness of the
control strategies used to handle that difficulty. The second proposal is that the
control strategies and social support at work moderate the effects of work difficulties
314
on job satisfaction. The third proposal is that general job autonomy, difficulties at
work, control strategies, social support at work and life satisfaction predict job
satisfaction. Before these proposals are discussed, the conceptualisation of the
control strategies requires explanation.
4.9.1
Primary Control, Self-Protective Secondary Control, and SelfAffirmative Secondary Control
Factor analyses of the control strategies demonstrated that employees were
using three types of control strategies, namely primary control, self-protective
secondary control and self-affirmative secondary control. Although two types of
secondary control were identified in chapter 2, it was not known whether the
differences between them would be great enough to form separate factors. As this
was the case however, the two types of secondary control require further exploration.
All of the secondary control strategies involve people changing themselves to
fit in with their situation, however there are two ways that this can be done. Selfprotective secondary control strategies reduce negative feelings about the situation.
The strategies that loaded on the self-protective factor were attribution (“Think that
this difficulty is not my fault”), predictive negative (“Think that I knew this difficulty
would happen”), and wisdom (“Think that I can’t always get what I want”). These
strategies make the situation less concerning, and people conclude that a situation is
not as bad as it seems.
The second type of secondary control, self-affirmative, promotes positive
feelings. The strategies that were identified as being self-affirmative were support
315
(“Tell someone about this difficulty to make me feel better”), vicarious (“Think of
the success of my family or friends”), present success (i.e., “Think about my success
in other areas”), active avoidance (“Do something different, like going for a walk”)
and positive re-interpretation (“Look for something else that is positive in the
situation”). These strategies make people feel good about themselves and their lives.
It must be noted that this conceptualisation of self-protective and selfaffirmative secondary control was not completely supported. Four items did not load
on the expected factors. Specifically, downward social comparison (“Think that I am
better off than many other people”) did not load on the self-affirmative factor.
Furthermore, goal disengagement (“Think that the difficulty doesn’t matter”),
illusory optimism (“Think that this difficulty will work out okay in the end”) and
denial (“Ignore this difficulty”) did not load on the self-protective factor. There is no
ready explanation as to why these items did not load on the expected factors. Clearly
however, the majority of items were consistent with the conceptualisation of selfprotective and self-affirmative secondary control.
Although this is a novel approach to secondary control strategies, it must be
noted that the conceptualisation of these three strategies is still consistent with
Heckhausen and Schulz’s (1995) proposals. Specifically, primary control strategies
involve attempts to change the environment to fit in with the self, and both types of
secondary control strategies involve attempts to change the self to fit in with the
environment. The new idea however is that some secondary control strategies
reduce negative feelings, whilst others promote positive feelings.
316
Although factor analyses have not been conducted on other primary and
secondary control scales as they generally contain only one item (i.e., Thompson et
al., 1996, 1994), they have been conducted on coping scales. The factors emerging
from these analyses can be compared to the three factors found in this study. As the
most common coping scale is the Ways of Coping Questionnaire (Folkman &
Lazarus, 1985; Folkman, Chesney, Cooke, Boccellari & Collette, 1994), factor
analyses of this scale will be examined.
Unlike the Situation Specific Primary and Secondary Control Scale (Maher et
al., 2002), which uses one item for each strategy, the Ways of Coping Questionnaire
(Folkman & Lazarus, 1985) uses multiple items for each strategy. As such, when
factor analyses are conducted on the scale, the items cluster according to the type of
strategy. For example, Folkman et al., (1986) demonstrated that a factor analysis,
averaged across several samples yielded eight factors, including confrontative
coping, distancing, self-controlling, seeking social support, accepting responsibility,
escape-avoidance, planful problem solving and positive reappraisal. It must be noted
however, that as discussed in chapter 1, factor analyses conducted on this scale are
far from consistent (Edwards & O’Neill, 1998).
Factor analyses of the Ways of Coping Questionnaire are not comparable to
those conducted on the Situation Specific Primary and Secondary Control Scale.
Factor analyses of the Ways of Coping Questionnaire identify which items measure a
particular strategy, whereas factor analyses of the Situation Specific Primary and
Secondary Control Scale (Maher et al., 2002) identify which strategies cluster
together. As such, the factor analyses in the current study are theoretically different
317
from previous analyses. Rather than just examining whether items measure a
strategy, they demonstrate how the strategies are related to each other. This means
that the underlying purpose of the strategies can be examined.
The development of three types of control strategies is exploratory, and as
such, further research is required. However, this conceptualisation may be useful in
determining the best type of secondary control. It could be hypothesised that selfaffirmative secondary control would be more positively correlated with job and life
satisfaction than self-protective secondary control, as rather than just decreasing
negative feelings, they increase positive feelings. This proposal is not supported in
the current study as both self-protective and self-affirmative secondary control
strategies were not related to job satisfaction. Despite this however, further research
may benefit from recognising there may be two types of secondary control. The
three major proposals of this study will now be examined.
4.9.2
Proposal One: The Controllability of the Difficulty Influences the
Amount and Adaptiveness of the Control Strategies Used to Manage that
Difficulty
4.9.2.1
The Amount of Control Strategies Used for Controllable and
Uncontrollable Difficulties
It was hypothesised that employees would use more primary than secondary
control for controllable difficulties, and more secondary than primary control for
uncontrollable difficulties. Support was found for the former part of the hypothesis,
318
however no support was found for the latter as employees reported using more
primary than secondary control for uncontrollable difficulties.
These finding are partially consistent with the life span theory of control
(Heckhausen & Schulz, 1995). This theory proposes that people prefer primary
control and that it has primacy over secondary control. Consistently, employees
reported more primary than secondary control for controllable difficulties.
However, the theory also proposes that when people are faced with
uncontrollable situations, the probability of primary control failure increases, and
control strategies focus on changing oneself rather than changing ones environment
(Heckhausen & Schulz, 1995). This does not appear to be the case for the employees
in this study however, as they report more primary than secondary control in
uncontrollable situations.
Only one other study has reported the amount of control strategies used in an
uncontrollable situation. Thompson et al., (1996) demonstrated that HIV positive
men in prison (i.e., low-control situation) reported slightly more primary control (M
= 48.5%SM) than secondary control (M = 45%SM). This study, which is also
inconsistent with the life span theory of control, was criticised in chapter 1 for
measuring primary control using perceived control and secondary control using
acceptance. However, it now appears that even when a new measure of primary and
secondary control is used, employees report using more primary than secondary
control for uncontrollable difficulties. Three explanations have been developed for
this finding.
319
4.9.2.2
Why is Primary Control Used more than Secondary Control for
Uncontrollable Difficulties?
There are three possible explanations for the employees reporting more
primary than secondary control for uncontrollable difficulties. First, it may be that
when completing the questionnaire, the respondents were unable to conceptualise
uncontrollable difficulties. Second, primary control may be used first for all
difficulties and secondary control may only be used if primary control fails. Third,
the controllability of the situation may not influence the control strategies people use,
and rather people may have trait control strategies. These explanations will be
discussed.
4.9.2.3 a) Conceptualisation of Controllable and Uncontrollable Difficulties
One reason why the employees may have reported higher primary control than
secondary control in uncontrollable situations is that the employees were unable to
conceptualise the distinction between controllable and uncontrollable difficulties.
The terms “controllable” and “uncontrollable” were used because, although being
abstract in nature, they did not bias the respondents as much as other constructs such
as change, influence, do something about, or accept.
Despite being abstract, it appears that the participants generally did understand
these terms and the distinction between them. The results demonstrated that the
majority of participants indicated that difficulties with time management, motivation
and co-workers were controllable and difficulties with pay, amount of work,
320
work-place rules and promotion were uncontrollable. As such, it appears that the
participants understood what constituted a controllable and an uncontrollable
difficulty, and hence this proposal does not explain why employees reported higher
primary than secondary control in uncontrollable situations.
4.9.2.4
b) Primary control is Implemented First for Controllable and
Uncontrollable Difficulties
Another explanation for the finding that primary is used more than secondary
control for uncontrollable difficulties is that primary control is always implemented
first. It was assumed that employees would rely on secondary control for
uncontrollable difficulties in an attempt to avoid primary control failure. However, it
must be noted that Heckhausen and Schulz (1995) proposed that primary control
strategies are used first and it is possible that this applies in controllable and
uncontrollable situations. Perhaps people attempt to change all situations using
primary control, and if they fail, they then rely on secondary control strategies. If
this is the case, it would be expected that people would use comparable amounts of
primary control in controllable and uncontrollable situations, but that they would use
more secondary control in uncontrollable situations.
As demonstrated by the mean levels of primary and secondary control
however, this does not appear to be case. The primary control levels were similar for
controllable situations (M = 71.98) and uncontrollable situations (M = 64.56),
however there was no difference in their levels of secondary control
321
(controllable, M = 53.74, uncontrollable, M = 53.10). Hence, the proposal that
employees report more primary control than secondary control in uncontrollable
situations because they use primary control first in such situations and only use
secondary control when primary control fails, does not appear to be supported.
4.9.2.5
c) Trait Control Strategies
Another explanation for the finding that the employees reported more
primary control than secondary control for uncontrollable difficulties is that trait
control strategies may exist. People may have a set of strategies that they
consistently use to handle their difficulties, and they may not consider the usefulness
of the strategy within that situation. The correlations between the control strategies
used in controllable situations with the control strategies used in uncontrollable
situations supports this proposal. Primary control for a controllable difficulty was
strongly correlated with primary control for an uncontrollable difficulty (r = 0.69).
Furthermore, secondary control for a controllable difficulty was strongly correlated
with secondary control for an uncontrollable difficulty (r = 0.64). The correlations
between primary and secondary control strategies were much weaker. Primary and
secondary control strategies for controllable difficulties were not correlated. Primary
and secondary control for uncontrollable difficulties were only weakly correlated
(r = 0.20).
The idea that peoples’ responses to difficulties are stable has been discussed
in the coping literature. It is proposed that people have coping “styles”,
“dispositions”, or “traits” that they bring to the situation (Carver et al., 1989).
322
Accordingly, “people do not approach each coping context anew, but rather bring to
bear a preferred set of coping strategies that remains relatively fixed across time and
circumstances” (Carver et al., 1989, p. 270).
Few researchers have examined trait coping, perhaps because Folkman and
Lazarus (1986) disputed the idea, proposing that coping is contextual, and that it is
influenced by the person’s appraisal of the situation. However, other studies besides
the current research dispute this proposition. A study conducted by Schwartz, Neale,
Marco, Shiffman and Stone (1999) assessed trait coping using the Daily Coping
Questionnaire (Stone & Neale, 1984) and the Ways of Coping Questionnaire
(Folkman & Lazarus, 1984). The question at the beginning of each scale was
changed to “how do you typically deal with stressful situations.” They also
measured coping using a momentary measure where participants were given a
programmable palm-top computer. Participants would type in their stressful events
and indicate how they coped with them immediately after the event.
They examined how much of the variance in the momentary scales was due
to differences among individuals. For example, for escape coping, they examined
how much of the variance was due to the tendency for some individuals to report
escape coping more than others. The results demonstrated that 42% of the variability
in the momentary assessments was due to individual differences in escape coping.
The other coping strategies accounted for less of the variance, ranging between
20-30% for the Ways of Coping Scale, and for 15-19% of the Daily Coping Scale.
These findings suggest that a person’s coping response could be partially predicted
from a general coping scale, and thus supports trait coping.
323
The proposal that coping is a trait or disposition can be used to explain the
current findings. Employees may have reported using primary control for an
uncontrollable difficulty because primary control strategies are within their
disposition. Thus, rather than evaluating the situation, they evaluate the coping
strategies they have in their repertoire.
4.9.2.6
Summary
Consistent with the life span theory of control, employees reported using
more primary than secondary control for controllable difficulties. Inconsistently
however, they also reported using more primary than secondary control for
uncontrollable difficulties. Three explanations were developed to account for these
findings. The first, proposing that employees did not understand uncontrollable
difficulties, was dismissed, as employees seemed to classify their difficulties as
expected. The second explanation proposed that people use more primary control for
uncontrollable difficulties because they implement primary control first for all
difficulties, and only use secondary control if primary control fails. This was not
supported by the data, as the levels of secondary control were the same. The third
explanation proposed that the controllability of the situation did not influence the
control strategies the employees used. Rather, it is proposed that employees have
‘trait’ control strategies. Employees may fail to evaluate the situation and rather
simply use the strategies in their repertoire. The relationship between these control
strategies and job satisfaction will now be examined.
324
4.9.2.7
Adaptiveness of Primary and Secondary Control for Controllable and
Uncontrollable Difficulties
It was hypothesised that primary control would be more adaptive than
secondary control for controllable difficulties and that secondary control would be
more adaptive than primary control for uncontrollable difficulties. Partial support
was provided for this hypothesis, as primary control was more positively related to
job satisfaction than secondary control for controllable difficulties. Inconsistently
however, primary control was also more positively related to job satisfaction than
secondary control for uncontrollable difficulties.
These findings are inconsistent with the discrimination model (Thompson et
al., 1998), which proposes that primary control is the most adaptive strategy only for
controllable situations. Rather, the findings support the primacy/back-up model
(Thompson et al., 1998), which proposes that primary control is more adaptive than
secondary control in both controllable and uncontrollable situations.
As with the current study, past empirical studies have supported the
primacy/back up model (Thompson et al., 1996; 1994; 1993; 1998). As limitations
were identified in these studies however, it was thought that when these limitations
were addressed, the discrimination model would be supported. These limitations,
discussed in chapter 1, concern the measurement of perceived control and primary
and secondary control strategies. A more notable flaw however is that these studies
failed to adequately test the discrimination model and the primacy/back-up model.
Rather than correlating the controllability of a situation with the control strategies
325
used to handle that situation, these studies examined general levels of perceived
control and control strategies. Some of the studies did examine the constructs at a
more specific level (e.g., Thompson et al., 1996, 1994), however they then
aggregated the items to obtain an overall measure of perceived control and an overall
measure of primary and secondary control. It appears however that even when all of
the limitations were addressed, the findings still supported the primacy/back up
model.
The current findings, although referring to control strategies, can also be
compared to the empirical studies on the goodness of fit hypothesis for coping
strategies. These studies generally demonstrate that, consistent with the current
findings, problem-focussed strategies are more adaptive than emotion-focussed
strategies in controllable situations. They also demonstrate that emotion-focussed
strategies are not more adaptive than problem-focussed strategies in uncontrollable
situations (e.g., Bowman & Stern, 1995; Conway & Terry, 1992; Osowieki &
Compas, 1998, 1999; Park, Folkman & Bostrom, 2001; Vitaliano et al., 1990).
These studies were criticised for their research designs in the introduction. It appears
however, that even when these problems are addressed, similar results are obtained.
In summary, it appears that consistent with past studies, primary control is
more adaptive than secondary control for both controllable and uncontrollable
difficulties. As many flaws were identified with the past studies, it was expected that
when these flaws were addressed, the results would be more consistent with the
discrimination model. This is not the case however, and as such further exploration
326
is needed to explain why primary control is more adaptive than secondary control in
uncontrollable situations.
4.9.2.8
Why is Primary Control more Adaptive than Secondary Control in
Uncontrollable Situations?
The finding that primary control is adaptive in uncontrollable situations is
contrary to intuition. As such, it is important that this finding can be explained
theoretically. It was expected that if employees tried to change an uncontrollable
situation using primary control, they would fail and this failure would negatively
influence perceived competence, self-efficacy, self-esteem (Heckhausen et al., 1997),
and job satisfaction.
The current findings, which demonstrate that primary control is positively
related to job satisfaction for uncontrollable difficulties, challenge the assumptions
regarding primary control failure. Primary control failure has not been measured in
the past, or in the current study, as it is extremely difficult to assess. It requires the
person to indicate how often they used each of the control strategies and indicate the
successfulness of each strategy. This is cognitively taxing for the respondents, and if
completed for primary and secondary control, would add another 17 items to each
control scale (controllable and uncontrollable). More importantly however, it may
not even be possible for people to recall this information. Whilst they may
remember whether they solved a problem, it is unlikely that they can recall which
strategy was more successful than others. Furthermore, it may actually be a
327
combination of strategies that contributes to the problem being overcome. For these
reasons, the successfulness of the strategies was not assessed.
As primary control failure was not measured in the current study however, it
is possible that it did not behave as expected. Firstly, it may be that the employees
who are implementing primary control are not experiencing primary control failure.
Secondly, employees may be experiencing primary control failure, yet experiencing
few negative consequences. These explanations will be discussed.
4.9.2.9
Primary Control does not lead to Primary Control Failure
In regard to the first explanation, employees who reported high primary
control for “uncontrollable” difficulties may have reported high job satisfaction
because they successfully implemented the strategies. Perhaps people only use
primary control when they know that they will be successful. Indeed, it seems
maladaptive for people to use primary control if they know that it will lead to
primary control failure.
If it proposed that employees only use primary control when they know they
will be successful, it must still be questioned how they could successfully change an
uncontrollable situation using primary control. One possibility is that the difficulties
reported by employees as being uncontrollable are only low-control difficulties.
Most of the difficulties reported, such as pay, promotion and workplace-rules may
not be completely uncontrollable. Other people determine them, and it is possible for
the people to be influenced, and thus for primary control to be successful. Perhaps
328
different results would be obtained if people were given difficulties that are clearly
uncontrollable such as the death of a loved one or a natural disaster.
4.9.2.10
Primary Control Failure does not Negatively Influence Job
Satisfaction
In regard to the second explanation, it may be that the employees are
experiencing primary control failure, but that the primary control failure is not
having negative effects. The life span theory of control proposes that primary
control failure will threaten perceived competence, self-efficacy, and self-esteem
(Heckhausen et al., 1997). It must be noted however that these effects have not been
tested. Perhaps it is better to have tried to implement primary control and failed than
to have not tried at all. Employees can tell themselves that there was nothing more
they could do, and thus they may feel better about their own control efforts.
Both of these explanations are speculative, and indeed require empirical
validation. To do this, future studies need to invest time in developing and
measuring the successfulness of primary and secondary control strategies.
4.9.2.11
Summary
Although primary control was more adaptive than secondary control in
controllable situations, it was also more adaptive in uncontrollable situations. These
findings are inconsistent with the discrimination model and the goodness of fit
hypothesis. It is difficult to explain as it was expected that employees who used
primary control for uncontrollable difficulties would experience primary control
329
failure. However, it may be that the employees only use primary control when they
know that they will be successful. Alternatively, the employees may experience
primary control failure, yet the consequences of primary control failure may be less
damaging than not attempting at all. Further empirical research is required to
examine these proposals.
4.9.3
Proposal Two: Moderators of Controllable and Uncontrollable
Difficulties on Job Satisfaction
Moderators of work difficulties were examined, as these variables may be
more amenable to change than work difficulties. Employers may be reluctant to
reduce work difficulties, where both the job and the organisation would need to
undergo a thorough assessment. Furthermore, it may be impossible to reduce some
work difficulties if they are inherent in the nature of the work.
It was hypothesised that the control strategies and social support at work
would moderate the relationship between work difficulties and job satisfaction.
Inconsistently however, primary control did not moderate the effect of controllable
difficulties, and secondary control did not moderate the effect of uncontrollable
difficulties. For the social support variables, only co-worker instrumental support
moderated the effect of controllable work difficulties on job satisfaction. It must be
noted that this finding was significant at 0.05, however it was not significant at the
more stringent alpha level of 0.01. As few studies have examined the moderating
role of social support in the workplace however, this finding was examined further.
330
The finding that co-worker support played a greater role than supervisor
support is inconsistent with other studies (e.g., Beehr, 1985; Fenlason & Beehr,
1994; Russell, Altmaier & Van Velzen, 1987). It was expected that as co-workers
have less influence over difficulties at work, their support would not be as beneficial
as supervisor support (Fenlason & Beehr, 1994). It must be noted however that the
measure of supervisor support used in this scale was exploratory. Although the scale
has face validity, there are no independent psychometric data for the scale. The
findings will be compared to past studies.
4.9.3.1
Past Studies Examining the Moderators of Stress
In regard to the control strategies, no other studies have examined whether
the control strategies moderate the effect of work difficulties. However, a few
studies have demonstrated that coping strategies moderate the effect of stressors on
stress (Aldwin & Revenson, 1987, Ashford, 1988; Parkes, 1990, 1994; Perrewe &
Zellars, 1999; Osipow et al., 1985). These studies are inconsistent with the current
findings, demonstrating that some coping strategies do moderate work stress. There
is no conclusive evidence however, as to which control strategies moderate work
stress.
In regard to social support, a few studies, including study two, have
demonstrated that social support at work has a moderating effect on job satisfaction
(i.e., Karasek et al., 1982; Landsbergis et al., 1992). However, other studies have
failed to find the moderating role of social support (Chay, 1993; de Jonge &
Landeweerd, 1993; Melamed at al., 1991; Parkes & Von Rabenau, 1993). As
331
discussed in chapter 3, one difference between the supportive and non-supportive
studies is the measure of social support.
Two of the supportive studies (i.e., current study and Landsbergis et al.,
1992) relied on Karasek and Theorell’s (1990) scale. Although some of the items in
Karasek and Theorell’s (1990) scale were criticised in this chapter, there is certainly
no agreed way of measuring social support at work (Unden, 1996). The current study
does not shed light on the problem however, as the social support scale did not factor
as expected. The scale was only measuring two variables, supervisor support and
co-worker support. Further research is needed on the operationalisation of social
support to ensure that all four types of social support are assessed.
In general, the findings on the moderating role of the control strategies and
social support are somewhat inconsistent with other similar studies. One major
difference between the current study and the other studies however is the
independent variable. Other studies have relied on job stress or work demands,
whereas this study used work difficulties. This may have been problematic since
work difficulties, controllable and uncontrollable, did not strongly predict job
satisfaction.
4.9.3.2
Limitations in the Moderation Hypotheses
The finding that work difficulties did not strongly predict job satisfaction is a
concern for the robustness of this analysis. A moderation analysis tests whether the
relationship between two variables (i.e., work difficulties and job satisfaction) varies
depending on a moderator variable (i.e., control strategies or social support). A
332
median split conducted on the moderator produces a low group and a high group.
The relationship between work difficulties and job satisfaction for each group is then
examined. If the relationship between the two variables is not strong for the average
group however, it is unlikely that it will be strong when the moderator is low or high.
It was expected that work difficulties would strongly predict job satisfaction,
and as such, two explanations have been developed to account for the weak
relationship. These concern the nature of work difficulties and the operationalisation
of work difficulties.
4.9.3.3
Nature of Work Difficulties
Researchers that have examined the moderating role of social support have
examined job demands rather than work difficulties. Job demands are the
psychological stressors in the work environment (i.e., high pressure of time, high
working pace, difficult and mentally exacting work; Karasek & Theorell, 1990).
Work difficulties are much broader than job demands, and refer to any problems that
employees face at work.
4.9.3.4
Operationalisation of Work Difficulties
Work difficulties were measured by asking the employees to indicate how
often they experienced their most commonly occurring difficulty. This is a difficult
question to answer, as the employee needs to consider all of the difficulties that they
face, think about how often they face each one and identify the one that they face the
most.
333
This item was useful in that it led people into thinking about how they handle
that difficulty, however it may not have accurately assessed work difficulties. One
person may experience one difficulty all the time yet rarely experience any other
difficulties. Another person may experience ten difficulties all the time. Using the
current scale however, these respondents would receive the same score. Thus, the
difficulty at work scale requires revision. Perhaps the primary and secondary control
scale could still include the item assessing the most common difficulty as this helps
respondents to focus on a specific situation, however another measure of work
difficulties is required.
Developing a valid measure of work difficulties for a general sample of
employees is problematic. The obvious solution is to ask respondents on average
how often they face controllable and uncontrollable difficulties at work. These items
may be prone to errors however as they are cognitively taxing, requiring the
employee to mentally average their work difficulties.
Another solution is to ask respondents to indicate how often they experience
each difficulty that they select from a list. Hence, as with the current scale, the
respondents would be given a list of general work difficulties. They would tick
which ones they experience and could control and then indicate how often they
experience each difficulty. They would then do the same for uncontrollable
difficulties. The problem however is that with the addition of the frequency item, the
length of the scale doubles. Furthermore, there are an unlimited number of work
difficulties and as such, some would be omitted.
334
Another solution is to develop a list of occupational specific difficulties and
ask employees how often they experience them. This solution, although it would
enable the testing of the moderation hypotheses, is discouraged however, as different
occupational groups cannot be compared.
One final solution is offered. An open-ended format could be used, where
respondents are asked to list their top five difficulties at work, and for each one,
indicate how often they face it. This solution may be more time-consuming for the
researcher to code, however it is not too cognitively taxing and it can be applied to a
general sample of employees.
4.9.3.5
Summary
There was little support for the moderating role of the control strategies and
social support on the relationship between work difficulties and job satisfaction.
These findings are limited however by the operationalisation of work difficulties.
The scale only examined the most frequently occurring difficulty and as such, did not
provide an accurate assessment of work difficulties. Future researchers may need to
use an open-ended format, where respondents are asked to list their difficulties at
work and indicate how often they face each one.
4.9.4
Proposal Three: Predictors of Job Satisfaction
It was hypothesised that job autonomy, difficulties at work, control strategies,
social support at work, and life satisfaction would predict job satisfaction. Only
controllable difficulties, job autonomy, life satisfaction, and co-worker emotional
335
support uniquely predicted job satisfaction (15%). Thus, primary and secondary
control strategies, uncontrollable difficulties, co-worker instrumental support, and
supervisor emotional support and instrumental support did not uniquely predict job
satisfaction. Possible explanations for these findings are discussed.
The finding that the control strategies did not uniquely predict job satisfaction
is particularly difficult to explain. It is intuitive that the control strategies that
employees use to handle work difficulties influence their level of job satisfaction.
One possibility is that it may not be primary and secondary control alone that predict
job satisfaction, rather the effectiveness of the control strategies. Future studies may
need to assess the control strategies and the effectiveness of them.
The finding that uncontrollable difficulties did not uniquely predict job
satisfaction may reflect the operationalisation of work difficulties. Employees were
asked how often they face their most commonly occurring controllable and
uncontrollable difficulty. As discussed previously, this measure may be flawed and
as such, more research is required to understand the importance of work difficulties
in predicting job satisfaction.
The finding that supervisor support did not uniquely predict job satisfaction
may also be explained by its measurement. As discussed previously, the social
support scale did not factor as expected, and it appeared as though the scale was only
measuring two variables, supervisor support and co-worker support. The scale was a
co-worker scale that was extended to supervisors. Perhaps separate scales are
required for the different roles. As such, further research is needed on the
336
operationalisation of social support to ensure that all four types of social support are
assessed.
4.9.4.1
Summary
Partial support was provided for the proposed predictors of job satisfaction,
as controllable difficulties, job autonomy, life satisfaction and co-worker emotional
support uniquely predicted job satisfaction. The finding that primary and secondary
control strategies, uncontrollable difficulties, co-worker instrumental support and
supervisor emotional and instrumental support did not uniquely predict job
satisfaction may be due to operationalisation issues.
4.9.5
Conclusion
The study tested three major proposals, which centered on job satisfaction,
control strategies and the controllability of the situation. The first proposal that the
controllability of the difficulty influences the use and adaptiveness of the control
strategies used for that difficulty, was not supported. Employees reported using
more primary control than secondary control for controllable and uncontrollable
difficulties. Three explanations were developed to account for these findings,
however the most plausible was that people have trait control strategies.
In addition to being used more than secondary control, primary control was
also more adaptive than secondary control for controllable and uncontrollable
difficulties. These findings, which are inconsistent with the discrimination model,
challenge the assumptions about primary control failure. It is possible that primary
337
control was adaptive for uncontrollable difficulties because it was being
implemented successfully. If it was being implemented successfully, perhaps the
work difficulties at work were low-control rather than being uncontrollable.
Alternatively, the employees may have been experiencing primary control failure,
however that failure may not have negatively affected job satisfaction.
The second major proposal, that the control strategies and social support at
work moderated the effects of work difficulties on job satisfaction, was generally not
supported. The findings tended to be inconsistent with previous studies examining
job stress, and the replacement of job stress with work difficulties was questioned.
Specific problems with the operationalisation of work difficulties were identified that
may have limited the findings.
The third proposal, that general job autonomy, difficulties at work, control
strategies, social support at work and life satisfaction predict job satisfaction was
partially supported. On the basis of these findings, it was clear that measures of
primary and secondary control, work difficulties, and social support require further
exploration.
In summary, these findings suggest that a satisfied worker has high job
autonomy, high social support, high life satisfaction, few work difficulties, and uses
primary control to deal with controllable and uncontrollable difficulties. The
implications of these findings will be discussed in chapter 5.
338
5 Chapter 5 - Final Discussion
339
5.1
Abstract
This thesis tested a model of job satisfaction that includes environmental and
dispositional predictors. The major proposal of the model is that job autonomy
influences the use and adaptiveness of primary and secondary control strategies. The
model also examines other predictors of job satisfaction, including life satisfaction,
work difficulties, and social support at work. Additionally, it proposes that the
control strategies and social support at work moderate the relationship between work
difficulties and job satisfaction. Empirical support offered for these proposals in
chapters two, three and four are reviewed and a revised model of job satisfaction is
presented. This model continues to include job autonomy, primary and secondary
control, life satisfaction and work difficulties, however it also includes the
successfulness of primary and secondary control and re-introduces personality
variables.
340
5.2
The Development of A New Model of Job Satisfaction
This thesis developed a model of job satisfaction that includes environmental
(i.e., job autonomy, social support at work, and work difficulties) and dispositional
predictors (i.e., primary and secondary control, personality and life satisfaction).
This model extended the job demand-control model (Karasek & Theorell, 1990),
offering an alternative explanation for the positive relationship between job
autonomy and job satisfaction. The job demand-control model was selected for
further investigation because, unlike other dominant theories, it is highly applicable
to the workplace and attractive to employers.
The job demand-control model proposes that job demands and job decision
latitude interact to predict job satisfaction, and that the most satisfied workers are
those who have high job decision latitude and high job demands. The implication of
this proposal is that employers can increase job satisfaction without reducing work
demands.
According to Karasek and Theorell (1990), employees with high job decision
latitude can translate the physiological arousal produced from job demands into
action through effective problem solving. They propose that workers with high job
autonomy are “given the freedom to decide what is the most effective course of
action in response to a stressor” (Karasek & Theorell, 1990, p. 36). However, this
explanation has been criticised for being tautological. It proposes that job control, or
the ability to choose at work, increases job satisfaction because it allows people to
choose how they deal with their demands at work.
341
An alternative explanation for the positive link between job autonomy and
job satisfaction is that employees with high job autonomy have higher job
satisfaction because they respond differently to work difficulties. Employees can
respond to work difficulties in two ways; they can either change the situation using
primary control or they can change themselves using secondary control.
It is expected that these primary and secondary control strategies mediate the
relationship between job autonomy and job satisfaction. Job autonomy is expected to
influence the amount of control strategies that employees use and the adaptiveness of
those strategies.
In regard to the amount of control strategies, employees with high job
autonomy are expected to rely on more primary control and less secondary control
than employees with low job autonomy. As primary control strategies are preferred
over secondary control strategies, employees with higher job autonomy have higher
job satisfaction than employees with lower job autonomy.
In regard to the adaptiveness of the strategies, it is expected that primary
control strategies are more adaptive than secondary control only when the situation is
controllable. When the situation is uncontrollable, secondary control is expected to
be the most adaptive strategy.
The major proposal of the model of job satisfaction is thus that: 1) primary
and secondary control mediate the relationship between job autonomy and job
satisfaction. However, several other propositions are also examined, including that;
2) social support at work and life satisfaction are positively related to job satisfaction
and; 3) the control strategies and social support at work moderate the relationship
342
between work difficulties and job satisfaction. Empirical tests of these proposals will
be examined.
1) Primary and Secondary Control Strategies Mediate the Relationship Between
Job Autonomy and Job Satisfaction
It is expected that primary and secondary control strategies explain the
relationship between job autonomy and job satisfaction. Job autonomy is expected to
influence the amount of control strategies than employees report, and the
adaptiveness of the control strategies.
5.2.1.1
Job Autonomy Influences the Use of Primary and Secondary Control
Strategies
It is expected that all employees, with either low or high job autonomy,
implement primary and secondary control strategies. According to the life span
theory of control (Heckhausen & Schulz, 1995), primary control has primacy over
secondary control as it is preferred and is implemented first. If primary control is
implemented successfully, the problem is resolved. If primary control fails however,
the person is expected to implement secondary control strategies to compensate for,
and avoid, future primary control failure.
When these propositions are applied to the workplace, it is expected that job
autonomy influences the likelihood of primary control failure. It is proposed that job
autonomy is inversely related to the probability of primary control failure, which in
343
turn, influences the use of secondary control strategies. Thus, employees with high
job autonomy are expected to experience less primary control failure and as such, use
less secondary control than employees with low job autonomy.
This proposal was tested by comparing the control strategies reported by low
job autonomy workers with those reported by high job autonomy workers. Study
one compared supermarket workers and academics, whilst study two compared
teachers and academics. Both of these studies provided minimal support.
In study one, the supermarket workers reported similar levels of primary
control and more secondary control, than the academics. Although these results
suggest that job autonomy influences the use of secondary control, but not primary
control, it must be noted that these results are based on the levels of job autonomy
inferred from type of occupation. Thus, it is assumed that supermarket workers are
low in job autonomy and academics are high in job autonomy. When the same
analysis was conducted with the reported levels of job autonomy, the results
changed, in that only primary control was related to job autonomy. Thus the findings
from study one suggest that job autonomy influences the use of primary control, but
not secondary control. These findings were limited however, as the primary and
secondary control scale used in this study was flawed. The rating scale did not
assess how much control strategies the person was using, rather how much they
agreed with the strategies presented to them in the scale.
The primary and secondary control scale was revised for study two and
administered to teachers and academics. This study was not supportive of the
proposals however, as the groups reported similar levels of primary and secondary
344
control. When studies one and two are considered together, it appears as though
there is little support for the proposal that job autonomy influences the use of the
control strategies.
These findings were attributed to, in part, the specificity of the hypotheses.
Studies one and two examined the proposal that job autonomy influences the control
strategies at a general level, measuring how much control employees have over their
work environment and how they generally handle work difficulties. It was expected
that this relationship may increase in strength however if the hypotheses were more
specific. In this case, the controllability of one situation would be correlated with the
control strategies used to handle that situation.
As such, study three examined the amount of control strategies that
employees used for controllable and uncontrollable difficulties. It was hypothesised
that employees would use more primary than secondary control for controllable
difficulties and more secondary than primary control for uncontrollable difficulties.
Inconsistently however, employees reported more primary than secondary control
strategies for controllable and uncontrollable difficulties.
One explanation for this finding is that employees have trait control
strategies. The use of primary and secondary control for controllable difficulties was
highly correlated with the use of primary and secondary control for uncontrollable
difficulties. Thus, people may have a set of strategies that they consistently use to
handle their difficulties. Rather than evaluating the controllability of each work
difficulty, employees may simply use the strategies that they know.
345
If the amount of primary and secondary control used by employees remains
stable across situations, then a dispositional factor, such as personality may predict
the control strategies. A few researchers have previously examined how personality
variables relate to coping strategies (Brebner, 2001; Carver et al., 1989; Gunthert,
Armeli & Cohen, 1999; Saklofske & Kelly, 1995; Scheier, Weintraub & Carver,
1986). It is proposed that people with high extroversion use active coping strategies
where they talk out their problems and people with high neuroticism use passive
coping strategies where they tend to blame themselves, and also other people (Costa,
Somerfield & McCrae, 1996). These proposals are extended to the control strategies,
where it is expected that extroversion is positively related to primary control and
neuroticism is positively related to secondary control.
The correlations between extroversion and neuroticism and the control
strategies were examined in studies one and two. In regard to extroversion, study
one demonstrated that primary control was positively correlated with extroversion
for the academics, r = 0.25, and the supermarket workers, r = 0.42. Furthermore, in
study two, primary control was positively related to extroversion for the academics,
r = 0.20. Thus, these findings suggest that people high on extroversion use more
primary control.
In regard to neuroticism, study one demonstrated that there was no
relationship between secondary control and neuroticism. Study two provided some
support, as teachers’ levels of neuroticism were positively related to secondary
control (r = 0.19). Although these findings suggest that neuroticism is at best, only
weakly correlated with secondary control, studies using coping strategies have
346
demonstrated that neuroticism is strongly correlated with emotion-focussed coping
(i.e., Brebner, 2001; Saklofske & Kelly, 1995). These higher correlations may
reflect the difference between emotion-focussed coping strategies and secondary
control strategies. In general emotion-focussed strategies tend to be more negative
than secondary control strategies and thus may be more positively correlated with
neuroticism. As the secondary control strategies developed for this study included
more positive strategies, further research may be required to examine which
personality variables predict secondary control. It might be useful to examine how
the remaining personality variables (i.e., conscientiousness, agreeableness and
openness) relate to secondary control.
In regard to the proposed model of job satisfaction, the finding that
employees reported similar levels of control strategies in controllable and
uncontrollable situations suggests that changes need to be made to the model. As
such, rather than job autonomy, it is proposed that personality predicts the use of the
control strategies.
5.2.1.2
Summary
There was marginal support for the proposal that job autonomy predicts the
use of the control strategies. When this proposal was changed to be more specific,
the controllability of the difficulty did not influence the use of primary and secondary
control strategies. The finding that employees reported similar levels of primary and
secondary control for controllable and uncontrollable difficulties suggests that trait
control strategies may exist. Employees may have a set of control strategies that they
347
regularly use, irrespective of the controllability of the problem. As such, the model of
job satisfaction is changed so that personality predicts the control strategies rather
than job autonomy or the controllability of the situation.
5.2.1.3
Job Autonomy Influences the Adaptiveness of Primary and Secondary
Control
The relationship between the control strategies and job satisfaction is
expected to change depending on the level of job autonomy. This hypothesis is
based on the discrimination model, which proposes that primary control is the more
adaptive strategy in controllable situations and that secondary control is the more
adaptive strategy in uncontrollable situations. This model underlies the philosophy
of the serenity prayer; “Grant me the strength to change what I can, the patience to
accept what I cannot, and the wisdom to know the difference” (Thompson et al.,
1998, p. 587). An alternative model has also been developed, namely the
primacy/back-up model. This model proposes that primary control is more adaptive
than secondary control in controllable and relatively uncontrollable situations.
Previous empirical studies have supported the primacy/back-up model
(Thompson et al., 1996; 1994; 1993; 1998), however these studies were criticised for
their measurement of the controllability of the situation and the control strategies. In
studies one and two, the correlations between the control strategies and job
satisfaction for the low job autonomy group were compared to the correlations for
the high job autonomy group. Study one supported the primacy/back up model,
demonstrating that primary control was the most adaptive strategy for both the
348
academics and the supermarket workers. As the scale of primary control was
subsequently criticised, the proposal was re-tested with a revised scale in study two.
Study two did not support the primacy/back-up model or the discrimination model,
demonstrating that primary and secondary control strategies were not related to job
satisfaction.
As mentioned previously, in both of these studies, the hypotheses were not
consistent with the definition of the discrimination model or the primacy/back-up
model. The hypotheses were too general and as such were made more specific in
study three. In this study, the relationships between the control strategies for
controllable and uncontrollable difficulties and job satisfaction were examined.
The findings from study three refuted the discrimination model and supported
the primacy/back-up model. Primary control was more adaptive than secondary
control for controllable and uncontrollable difficulties. These findings suggest that
employees should use primary control whenever they face a difficulty at work, even
if it is uncontrollable.
The proposal that primary control is adaptive in uncontrollable situations is
difficult to explain as is it is assumed that they are likely to experience primary
control failure. It must be noted however that primary control failure was not
measured, and as such, the assumption that primary control in uncontrollable
situations results in primary control failure may be inaccurate. It is possible that
employees using primary control for uncontrollable difficulties report higher job
satisfaction because they implemented it successfully. As such, the successfulness of
the control strategies must be measured in future studies.
349
This proposal is incorporated in the revised model of job satisfaction. It is
now proposed that primary and secondary control strategies are not directly related to
job satisfaction, rather that they indirectly influence job satisfaction through the
successfulness of the control strategies. For example, suppose two employees report
having primary control strategies in their repertoire, however only one of them
implements primary control successfully. It would be expected that the employee
who is successfully implementing primary control would report higher job
satisfaction than the employee experiencing primary control failure. As such, the
successfulness of primary control may be a better predictor of job satisfaction than
primary control directly. It is expected that if employees successfully implement the
strategies, they will report higher job satisfaction.
5.2.1.4
Summary
The controllability of the difficulty did not influence the relationship between
the control strategies and job satisfaction. Even when the situation was
uncontrollable, primary control was the most adaptive strategy. These findings,
along with previous research, refute the discrimination model and support the
primacy/back-up model. The primacy/back-up model is difficult to explain as people
who use primary control in uncontrollable situations are expected to experience
primary control failure. Primary control failure was not measured however, and as
such, the successfulness of the control strategies must also be measured in future
studies. The model of job satisfaction is revised where it is proposed that the control
350
strategies are indirectly related to job satisfaction through the successfulness of the
control strategies.
5.2.2
Conclusion: Do the Control Strategies Mediate the Relationship
Between Job Autonomy and Job Satisfaction?
The above findings demonstrate that the control strategies do not explain the
relationship between job autonomy and job satisfaction. As such, the question of
why job autonomy is related to job satisfaction remains unanswered. One possibility
is self-determination. According to DeCharms (1968) and Deci and Ryan (1986),
humans have an innate need for competence and self-determination. Individuals
attempt to seek out situations that challenge them. They find these activities
rewarding and experience positive emotions such as enjoyment and excitement (Fay
& Frese, 2001; Ryan & Deci, 2001).
Another possibility is job status. Employees with high job autonomy have
jobs that generally involve more responsibility and job status than employees with
low job autonomy. Although few studies have examined the relationship between
job status and job satisfaction, one study has demonstrated that female employees
with higher job status tend to report higher job satisfaction than females employees
with lower job status (Secret & Green, 1998).
Another possibility is self-esteem. Employees with high job autonomy may
feel that their employer trusts them and thus may value themselves more than
employees with low job autonomy. Self-esteem has been shown to be positively
related to job satisfaction, where the average correlation is r = 0.26 (Judge & Bono,
351
2001). Thus, job status or self-esteem may mediate the relationship between job
autonomy and job satisfaction.
It is important that researchers continue to examine why job autonomy is
related to job satisfaction as the explanation offered by Karasek and Theorell (1990)
in the job demand-control model (Karasek & Theorell, 1990) is tautological and
vague. It is necessary that researchers understand the mechanism underlying the
proposal that job autonomy can reduce the influence of job demands.
5.2.3
2) Social Support at Work and Life Satisfaction Directly Predict Job
Satisfaction
The next major proposal of the model of job satisfaction is that social support
at work and life satisfaction predict job satisfaction.
5.2.3.1
Social Support at Work
Social support is expected to be directly related to job satisfaction. In study
two, supervisor support (r = 0.64, r = 0.46), and co-worker support (r = 0.39,
r = 0.42) were positively correlated with job satisfaction for the teachers and
academics, respectively. Furthermore, study three demonstrated that supervisor
support (r = 0.37) and co-worker support (r = 0.48) were moderately correlated with
job satisfaction. In regard to the proposed model of job satisfaction, social support at
work appears to be an important predictor.
352
5.2.3.2
Life Satisfaction
Life satisfaction is expected to be positively related to job satisfaction. In
study one, life satisfaction was not related to job satisfaction for the supermarket
workers, however it was weakly related for the academics, r = 0.20. The results were
stronger in study two, where life satisfaction was moderately correlated with job
satisfaction for the academics, r = 0.38 and the teachers, r = 0.46. Study three also
demonstrated, using a general sample of employees that r = 0.46. On the basis of
these findings, it is concluded that life satisfaction is a direct predictor of job
satisfaction.
The positive correlations between life satisfaction and job satisfaction support
the spillover model, which proposes that satisfaction in one domain of an
individual’s life extends into other areas. Life satisfaction may spillover into job
satisfaction or job satisfaction may spillover into life satisfaction. Thus, employers
need to ensure that their employees are satisfied with all major areas of their lives,
not just the workplace. Employees also need to be satisfied with their standard of
living, their health, their personal relationships, their safety, and feeling part of their
community.
The levels of life satisfaction reported by the employees are particularly
interesting. According to Cummins (2000b), life satisfaction is held under
homeostatic control. Using two standard deviations to define the normative range, it
is predicted that the mean subjective life satisfaction of Western population samples
353
lay within the range 70-80%SM (Cummins, 1995). Consistently, all mean levels lay
within the 70-80%SM range, M = 78.31, M = 73.09, M = 74.20, M = 75.61,
M = 73.68. The finding that life satisfaction can be predicted within such a small
range is remarkable. Even the employees with low job autonomy (i.e., supermarket
workers and teachers) reported levels of life satisfaction that were within the
normative range.
The mechanisms that underlie this prediction involve personality, perceived
control, optimism, and self-esteem (Cummins, 2000b). More empirical studies are
needed to examine how these predictors are related to life satisfaction. These results
are not only important in developing a theory of life satisfaction, but these predictors
are important for employers attempting to increase job satisfaction.
5.2.3.3
Summary
Social support at work and life satisfaction both directly predicted job
satisfaction and are included in the revised model of job satisfaction. It is proposed
that social support influences job satisfaction and that life satisfaction and job
satisfaction influence each other.
354
5.2.4
3) The Control Strategies and Social Support at Work Moderate the
Relationship Between Work Difficulties and Job Satisfaction
5.2.4.1
Moderating Role of Control Strategies
Previous researchers have suggested that it is not the stressor that predicts job
satisfaction, but rather how the person deals with the stressor (Aldwin & Revenson,
1987, Ashford, 1988; Parkes, 1990, 1994; Perrewe & Zellars, 1999; Osipow, Doty &
Spokane, 1985). Thus, it is expected that if employees match their control strategies
to the situation, the negative influence of work difficulties on job satisfaction is
lessened.
Study three did not support this proposal however, demonstrating that
primary and secondary control did not act as moderators. These findings suggest that
even if employees match their control strategies to the situation, the negative
influence of work difficulties on job satisfaction is not lessened. As such, this part of
the model of job satisfaction requires revision.
An alternative proposal is offered. Rather than the control strategies
moderating the effect of work difficulties on job satisfaction, the successfulness of
the strategies may be important. Thus, it is expected that if employees successfully
implement the matching control strategies, the influence of work difficulties on job
satisfaction decreases. The model of job satisfaction is thus altered, where the
successfulness of primary control moderates the effect of controllable difficulties,
355
and the successfulness of secondary control moderates the effect of uncontrollable
difficulties.
5.2.4.2
Moderating Role of Social Support
In regard to social support at work, it is expected that co-worker and
supervisor support moderate the effect of work difficulties on job satisfaction. Study
two demonstrated that supervisor support, but not co-worker support moderated work
difficulties. As supervisor support increased, the relationship between work
difficulties and job satisfaction decreased.
Study three examined the types of social support, proposing that instrumental
support buffers the effects of controllable difficulties and emotional support buffers
the effects of uncontrollable difficulties. Marginal support was found for this
proposal, as co-worker instrumental support moderated the relationship between
controllable work difficulties and job satisfaction.
These findings in study two and three are somewhat inconsistent. This
inconsistency may be attributed to, in part, the measurement of work difficulties. In
study two, general work difficulties were measured by the item “how often do you
face difficulties at work?” This item is prone to errors as it is cognitively taxing,
requiring the employee to mentally average their work difficulties.
In study three, controllable and uncontrollable work difficulties were
measured by asking employees how often they face their most commonly occurring
controllable and uncontrollable difficulty. This scale was also criticised as it only
focused on one difficulty. One person may experience one difficulty all the time yet
356
rarely experience any other difficulties. Another person may experience ten
difficulties all the time. Using the current scale however, these respondents would
receive the same score.
It is thus concluded that further research is needed to examine the variables
that moderate the relationship between work difficulties and job satisfaction. This
research needs to measure work difficulties using an open-ended format, where
respondents list their top five difficulties at work and indicate how often they face
each one. This scale is not expected to be excessively taxing and can be
administered to a general sample of employees. Using this scale, it is expected that
instrumental support will moderate the effect of controllable difficulties and
emotional support will moderate the effect of uncontrollable difficulties.
5.2.4.3
Summary
There was no support for the moderating role of the control strategies on the
relationship between work difficulties and job satisfaction, however there was some
support for social support at work. As primary and secondary control did not act as
moderators, the model of job satisfaction was revised to examine the successfulness
of the control strategies. Although there was only minimal support for the
moderating role of social support, the indirect relationship is retained in the model of
job satisfaction as the operationalisation of work difficulties was criticised.
357
5.3
Revised Model of Job Satisfaction
This discussion has combined the results from three studies to develop a
revised model of job satisfaction (refer to Figure 11). The revisions are based on the
current results, findings from other research, or are purely speculative. The bolded
arrows and variables represent changes made to the model.
The first proposal that personality influences the use of the control strategies
is based on past studies of coping and personality. Although further research needs
to be conducted to determine which personality variables predict secondary control,
it is proposed that extroversion is positively related to primary control. Past research
has also demonstrated that extroversion is also positively related to job satisfaction
and life satisfaction, and that neuroticism is negatively related to job satisfaction and
life satisfaction. Thus, personality influences the control strategies, job satisfaction,
and life satisfaction.
Primary and secondary control strategies are no longer directly related to job
satisfaction; rather it is speculated that they are indirectly related to job satisfaction
through the successfulness of the control strategies. It is expected that if employees
successfully implement the strategies, they will report higher job satisfaction.
In addition to the successfulness of the control strategies, job autonomy and
social support at work are expected to be positively related to job satisfaction. Job
satisfaction is also expected to be reciprocally related to life satisfaction. These
relationships have all been demonstrated in the current findings.
358
Based on the findings of study three, controllable and uncontrollable
difficulties are expected to be negatively related to job satisfaction. The relationship
between work difficulties and job satisfaction is hypothesised to be moderated by the
successfulness of the control strategies and social support at work.
Specifically, it is expected that the successfulness of primary control
moderates the effect of controllable difficulties, and the successfulness of secondary
control moderates the effect of uncontrollable difficulties. The effect of controllable
work difficulties on job satisfaction is expected to be less when primary control is
successful, and the effect of uncontrollable difficulties on job satisfaction is expected
to be less when secondary control is successful.
Social support at work is also expected to moderate the effect of work
difficulties on job satisfaction. As demonstrated in study three, instrumental support
is expected to moderate the effect of controllable difficulties on job satisfaction. It is
also hypothesised that emotional support will moderate the effect of uncontrollable
difficulties on job satisfaction. Although this proposal was not supported in study
three, it is expected that when a new measure of work difficulties is used, it will be
supported.
359
Figure 11- Revised Model of Job Satisfaction
Secondary Control
Instrumental
Support
Primary Control
Success of SC
Success of PC
Controllable
Diff x
Success PC
Personality
Job Satisfaction
Life Satisfaction
Uncontrol
Diff x
Success of
SC
Success of PC
Controllable
Diff x
Instrumental
Support
Controllable
Difficulties
Uncontrollable
Difficulties
Job
Autonomy
Uncontrol
Diff x
Emotional
Support
Success of SC
Secondary Control
Primary Control
Emotional Support
360
5.4
Conclusion
This thesis extended the job demand-control model (Karasek & Theorell,
1990), offering an alternative explanation for the relationship between job autonomy
and job satisfaction. A model of job satisfaction was developed which included job
autonomy, primary and secondary control, life satisfaction, work difficulties and
social support at work. The major proposal of this model was that job autonomy
influences the use and adaptiveness of primary and secondary control strategies.
Empirical testing of the model demonstrated that primary and secondary
control did not mediate the relationship between job autonomy and job satisfaction.
Employees reported using more primary control than secondary control for
controllable and uncontrollable difficulties. Furthermore, primary control was more
adaptive than secondary control for both types of difficulties.
Using these findings, a revised model of job satisfaction was developed. This
model proposes that rather than job autonomy, personality influences the use of the
control strategies. Furthermore, it is proposed that the control strategies do not
directly relate to job satisfaction, rather they are indirectly related through the
successfulness of the control strategies. In addition to these variables, job autonomy,
social support at work, life satisfaction and work difficulties continue to be included
as predictors of job satisfaction.
361
5.5
Final Word
This study developed a model of job satisfaction that offered an alternative
explanation to Karasek and Theorell (1990) for the relationship between job
autonomy and job satisfaction. Based on the life span theory of control (Heckhausen
& Schulz, 1995) and the discrimination model (Thompson et al., 1998), it was
proposed that employees with high job autonomy reported high job satisfaction
because they relied on more primary control and less secondary control strategies
than employees with low job autonomy. These proposals were not supported, as
primary control was the most commonly used and most adaptive strategy for
controllable and uncontrollable difficulties. These findings suggest that the serenity
prayer might best be changed to
“Grant me the strength to change the things I can…. and the things I cannot.”
362
5.6
References
Abouserie, R. (1996). Stress, coping strategies and job satisfaction in
university academic staff. Educational Psychology, 16, 49-57.
Agho, A.O., Price, J.L., & Mueller, C.W. (1992). Discriminant validity of
measures of job satisfaction, positive affectivity and negative affectivity. Journal of
Occupational and Organizational Psychology, 65, 185-196.
Alderfer, C.P. (1969). An empirical test of a new theory of human needs.
Organizational Behavior and Human Performance, 4, 142-175.
Aldwin, C.M., & Revenson, T.A. (1987). Does coping help? A
reexamination of the relation between coping and mental health. Journal of
Personality and Social Psychology, 53, 337-348.
Algera, J.A. (1980). Kenmerken van werk. Unpublished Ph.D. Thesis,
University of Leiden.
Ambrose, M.L., & Kulik, C.T. (1999). Old friends, new faces: Motivation
research in the 1990s. Journal of Management, 25, 231-237.
Armstrong, T.B. (1971). Job content and context factors related to
satisfaction for different occupational levels. Journal of Applied Psychology, 55, 5765.
Arnold, H.J., & House, R.J. (1980). Methodological and substantive
extensions to the Job Characteristics Model of motivation. Organizational Behavior
and Human Decision Processes, 25, 161-183.
363
Arvey, R.D., Bouchard, T.J., Segal, N.L., & Abraham, L.M. (1989). Job
satisfaction: Environmental and genetic components. Journal of Applied
Psychology, 74, 187-192.
Arvey, R.D., McCall, B.P., Bouchard, T.J., Taubman, P., & Cavanaugh, M.A.
(1994). Genetic influences on job satisfaction and work values. Personality and
Individual Differences, 17, 21-33.
Ashford, S.J. (1988). Individual strategies for coping with stress during
organisational transitions. Journal of Applied Behavioral Science, 24, 19-36.
Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable
distinction in social psychological research: Conceptual, strategic and statistical
considerations. Journal of Personality and Social Psychology, 51, 1173-1182.
Beehr, T.A. (1985). The role of social support in coping with organizational
stress. In T.A. Beehr., & R.S. Bhagat (Eds.), Human Stress and Cognition in
Organizations: An Integrated Perspective. (pp. 375-398). New York: Wiley
Beutell, N.L., & Wittig-Berman, U. (1999). Predictors of work-family
conflict and satisfaction with family, job, career, and life. Psychological Reports, 85,
893-903.
Bobko, P. (2001). Correlation and Regression: Applications for IndustrialOrganizational Psychology and Management. London: Sage.
Boey, K.W. (1998). Coping and family relationships in stress resistance: A
study of job satisfaction of nurses in Singapore. International Journal of Nursing
Studies, 35, 353-361.
364
Bogg, J., & Cooper, C. (1995). Job satisfaction, mental health, and
occupational stress among senior civil servants. Human Relations, 48, 327-341.
Boomsma, A. (1983). On the robustness of LISREL (maximum likelihood
estimation) against small sample sizes of nonnormality. Unpublished Ph.D. thesis,
University of Groningen, The Netherlands.
Boonzaier, B., Ficker, B., & Rust, B. (2001). A review of research on the Job
Characteristics Model and the attendant job diagnostic survey. South African
Journal of Business Management, 32, 11-35.
Bowman, G.D., & Stern, M. (1995). Adjustment to occupational stress: The
relationship of perceived control to effectiveness of coping strategies. Journal of
Counseling Psychology, 42, 294-303.
Brayfield, A.H., & Rothe, H.F. (1951). An index of job satisfaction. Journal
of Applied Psychology, 35, 307-311.
Breaugh, J.A. (1989). The Work Autonomy Scales: Additional validity
evidence. Human Relations, 42, 1033-1056.
Breaugh, J.A. (1998). The development of a new measure of global work
autonomy. Educational and Psychological Measurement, 58, 119-129.
Brebner, J. (2001). Personality and stress coping. Personality and Individual
Differences, 31, 317-327.
Brenner, V.C., Carmack, C.W., & Weinstein, M.G. (1971). An empirical test
of the motivation-hygiene theory. Journal of Accounting Research, 9, 359-366.
Burger, J.M., & Cooper, H.M. (1979). The desirability of control.
Motivation and Emotion, 3, 381-393.
365
Burke, R.J., & Greenglass, E.R. (2000). Hospital restructuring and nursing
staff well-being: The role of coping. International Journal of Stress Management, 7,
49-59.
Cahill, J. (1998). Stories of success and survival: The role of primary and
secondary control mechanisms in the subjective well-being of people on methadone
maintenance treatment programs. Unpublished Honours thesis, Deakin University,
Melbourne, Australia.
Cammann, C., Fichman, M., Jenkins, D., & Klesh, J. (1979). The Michigan
Organizational Assessment Questionnaire. Unpublished manuscript, University of
Michigan: Ann Arbor.
Campbell, J.P., & Pritchard, R.D. (1976). Motivation theory in Industrial and
Organizational Psychology. In M. Dunnette (Ed.), Handbook of Industrial and
Organizational Psychology. (pp 63-130). Chicago: Rand McNally.
Caplan, R.D., Cobb, S., French, J.R.P. Jr., Harrison, R.U., & Pinneau, S.R. Jr.
(1975). Job Demands and Worker Health. U.S. Department of Health, Education,
and Welfare Publication No. 75-160. Washington D.C.: U.S. Government Printing
Office, The Institute for Social Research.
Carson, K.D., Lanier, P.A., & Carson, P.P. (2001). A glimpse inside the
ivory tower: A cross-sectional comparison of work orientations in academia.
International Journal of Public Administration, 24, 479-492.
Carver, C.S., Scheier, M.F., & Weintraub, J.K. (1989). Assessing coping
strategies: A theoretically based approach. Journal of Personality and Social
Psychology, 56, 267-283.
366
Champoux, J.E. (1980). A three sample test of some extensions to the Job
Characteristics Model of work motivation. Academy of Management Journal, 23,
466-478.
Chay, Y.W. (1993). Social support, individual differences and well-being: a
study of small business entrepreneurs and employees. Journal of Occupational and
Organizational Psychology, 66, 285-302.
Chipperfield, J.G., Perry, R.P., & Menec, V.H. (1999). Primary and
secondary control-enhancing strategies. Journal of Aging and Health, 11, 517-539.
Clark, A.E. (1996). Job satisfaction in Britain. British Journal of Industrial
Relations, 34, 189-217.
Clark, A.E., & Oswald, A.J. (1996). Satisfaction and comparison income.
Journal of Public Economics, 69, 57-81.
Coakes, S.J., & Steed, L.G. (1999). SPSS: Analysis without Anguish.
Brisbane: John Wiley.
Cohen, J., & Cohen, P. (1983). Applied Multiple Regression/Correlation
Analysis for the Behavioral Sciences. Hillsdale, N.J: Lawrence Erlbaum.
Cohen, S., & Kamarack, T., Mermelstein, R., & Hoberman, H.M. (1985).
Measuring the functional components of social support. In I.G. Sarason., & B.R.
Sarason (Eds.), Social Support: Theory, Research and Applications. The Hague,
Netherlands: Nijhoff.
Cole, M. (1989). The politics of stress in teaching. In M.Cole., & S. Walker
(Eds.), Teaching and Stress. (pp.161-170). Milton Keynes: Open University Press.
367
Constantinople, A. (1967). Perceived instrumentality of the college as a
measure of attitudes towards college. Journal of Personality and Social Psychology,
5, 196-201.
Conway, V.J., & Terry, D.J. (1992). Appraised controllability as a moderator
of the effectiveness of different coping strategies: A test of the goodness of fit
hypothesis. Australian Journal of Psychology, 44, 1-7.
Cooper, C.L., Sloan, S., & Williams, S. (1988). Occupational Stress
Indicator. Windsor: NFER-Nelson.
Costa, P.T. Jr., & McCrae, R.R. (1985). The NEO Personality Inventory
Manual. Odessa, FL: Psychological Assessment Resources.
Costa, P.T. Jr., & McCrae, R.R. (1989). Personality as a life long
determinant of wellbeing. In L.Z. Malatesta., & C.E. Izzard (Eds.), Emotion in Adult
Development. (p.141-157). Beverly Hills, CA: Sage.
Costa, P.T. Jr., & McCrae, R.R. (1992). NEO PI-R professional manual.
Odessa, FL: Psychological Assessment Resources.
Costa, P.T. Jr., Somerfield, M.R., & McCrae, R.R. (1996). Personality and
coping: A reconceptualisation. In M. Zeidner., & N.S. Endler (Eds.), Handbook of
Coping: Theory, Research, Applications. New York: Wiley
Cousins, R. (2001). Predicting subjective quality of life: The contributions of
personality and perceived control. Unpublished Doctorate thesis, Deakin University,
Melbourne, Australia.
Cramer, D. (1995). Life and job satisfaction: A two-wave panel study. The
Journal of Psychology, 129, 261-268.
368
Cummins, R.A. (1995). On the trail of the gold standard for subjective wellbeing. Social Indicators Research, 35, 179-200.
Cummins, R.A. (1996). The domain of life satisfaction: An attempt to order
chaos. Social Indicators Research, 38, 303-328.
Cummins, R.A. (1997). Comprehensive Quality of Life Scale-Adult Manual
(ComQol-A5). Melbourne, Australia: Deakin University.
Cummins, R.A. (2000a). Objective and subjective quality of life: An
interactive model. Social Indicators Research, 52, 55-72.
Cummins, R.A. (2000b). Normative life satisfaction: Measurement issues
and a homeostatic model. In B. Zumbo (Ed.), Social Indicators and Quality of Life
Research Methods: Methodological Development and Issues. Amsterdam: Klumer.
Cummins, R.A. (2002). Caveats to the Comprehensive Quality of Life Scale.
http://acqol.deakin.edu.au.
Cummins, R.A., Eckersley, R., Pallant, J., Van Vugt, J., Shelley, J., Pusey,
M., & Misajon, R. (2001). The Australian Unity Index of Well-Being: Survey 1:
Report 1. (on-line). Available: http://acqol.deakin.edu.au.
Cummins, R.A., & Nistico, H. (in press). Maintaining life satisfaction: The
role of positive cognitive bias. Journal of Happiness Studies
DeCharms, R. (1968). Personal Causation. New York: Academic Press.
Deci, E.L., & Ryan, R.M. (1985). Intrinsic Motivation and SelfDetermination in Human Behaviour. New York: Plenum Press.
de Jonge, J., Breukelen, G.J.P., Landeweerd, J.A., & Nijhuis, F.J.N. (1999a).
Comparing group and individual level assessments of job characteristics in testing
369
the Job Demand-Control Model: A multilevel approach. Human Relations, 52, 9597.
de Jonge, J., Mulder, M.J.G.P., & Nijhuis, F.J.N. (1999b). The incorporation
of different demand concepts in the job demand-control model: Effects of health care
professionals. Social Science and Medicine, 48, 1149-1160.
DeNeve, K.M. (1999). Happy as an extroverted clam? The role of
personality for subjective well-being. Psychological Science, 8, 141-144.
De Rijk, A.E., Le Blanc, P., Schaufeli, W.B., & de Jonge, J. (1998). Active
coping and need for control as moderators of the job demand-control model: Effects
on burnout. Journal of Occupational and Organizational Psychology, 71, 1-18.
Diener, E., Suh, E.M., Lucas, R.E., & Smith, H.L. (1999). Subjective wellbeing: Three decades of progress. Psychological Bulletin, 125, 276-302.
Dollard, M.F., Winefield, H.R., Winefield, A.H., & Jonge, J.D. (2000).
Psychosocial service workers: A test of the demand control-support model. Journal
of Occupational and Organizational Psychology, 73. 501-510.
Doran, L.I., Stone, V.K., Brief, A.P., & George, J.M. (1991). Behavioral
intentions as predictors of job attitudes: The role of economic choice. Journal of
Applied Psychology, 76, 40-45.
Ducharme, L.J., & Martin, J. (2000). Unrewarding work, co-worker support,
and job satisfaction. Work and Occupations, 27, 223-244.
Dwyer, D.H., & Ganster, D.C. (1991). The effects of job demands and
control in employee attendance and satisfaction. Journal of Organizational Behavior,
12, 595-608.
370
Edwards, A.L. (1959). Edwards Personal Preference Schedule. New York:
Psychological Corporation.
Edwards, J.R., & O’Neill, R. (1999). The construct validity of scores on the
Ways of Coping Questionnaire: Confirmatory analysis of alternative factor
structures. Educational and Psychological Measurement, 58, 955-983.
Evans, B.K., & Fischer, D.G. (1992). A hierarchical model of participatory
decision-making, job autonomy and perceived control. Human Relations, 45, 11691190.
Evans, M.G. (1991). The problem of analyzing multiplicative composites:
interactions revisited. American Psychologist, 46, 6-15.
Ewen, R.B. (1964). Some determinants of job satisfaction: A study of the
generalisability of Herzberg’s theory. Journal of Applied Psychology, 48, 161-163.
Fay, D., & Frese, M. (2000). Self-starting behavior at work: Toward a theory
of personal initiative. In J. Heckhausen (Ed.), Motivational Psychology of Human
Development. Amsterdam: Elservier Science.
Felce, D., & Perry, J. (1995). Quality of life: Its definition and measurement.
Research in Developmental Disabilities, 16, 51-74.
Fenlason, K.J., & Beehr, T.A. (1994). Social support and occupational stress:
Effects of talking to others. Journal of Organizational Behavior, 15, 157-175.
Ferris, K.R. (1977). A test of the expectancy theory of motivation in an
accounting environment. The Accounting Review, 3, 605-615.
371
Finlay, W., Martin, J.K., Roman, P.M., & Blum, T.C. (1995). Organizational
structure and job satisfaction: Do bureaucratic organizations produce more satisfied
employees? Administration and Society, 27, 427-451.
Fisher, C.D. (2000). Mood and emotions while working: missing pieces of
job satisfaction? Journal of Organizational Behavior, 21, 185-202.
Fisher, S. (1994). Stress in Academic Life: The Mental Assembly Line.
Buckingham: Open University Press.
Fletcher, B.C., & Jones, F. (1993). A refutation of Karasek’s DemandDiscretion Model of Occupational Stress with a range of dependent measures.
Journal of Organizational Behavior, 14, 319-330.
Folkman, S., Chesney, M.A., Cooke, M., & Bocccellari, A. (1994).
Caregiver burden in the HIV-positive and HIV-negative partners of men with AIDS.
Journal of Consulting and Clinical Psychology, 62, 746-756.
Folkman, S., & Lazarus, R.S. (1980). An analysis of coping in a middle-aged
community sample. Journal of Health and Social Behavior, 21, 219-239.
Folkman, S., & Lazarus, R.S. (1985). If it changes, it must be a process:
Study of emotion and coping during three stages of a college examination. Journal
of Personality and Social Psychology, 48, 150-170.
Folkman, S., Lazarus, R.S., Dunkel-Schetter, C., DeLongis, A., & Gruen, R.J.
(1986). Dynamics of a stressful encounter: Cognitive appraisal, coping and
encounter outcomes. Journal of Personality and Social Psychology, 50, 992-1003.
372
Forsythe, C.J., & Compas, B.E. (1987). Interaction of cognitive appraisals of
stressful events and coping: Testing the goodness of fit hypothesis. Cognitive
Therapy and Research, 11, 473-485.
Fox, M.L., Dwyer, D.J., & Ganster, D.C. (1993). Effects of stressful job
demands and control on physiological and attitudinal outcomes in a hospital setting.
Academy of Management Journal, 36, 289-319.
Fox, S., & Feldman, G. (1988). Attention state and critical psychological
states as mediators between job dimensions and job outcomes. Human Relations, 41,
229-245.
Fried, Y. (1991). Meta-analytic comparisons of the Job Diagnostic Survey
and Job Characteristics Inventory as correlates of work satisfaction and performance.
Journal of Applied Psychology, 76, 690-697.
Fried, Y., & Ferris, G.R. (1987). The validity of the Job Characteristics
Model: A review and meta-analysis. Personnel Psychology, 40, 287-322.
Fried, Y., & Ferris, G.R. (1991). The dimensionality of job characteristics:
Some neglected issues. Journal of Applied Psychology, 71, 419-426.
Friedman, L.C., Nelson, D.V., Baer, P.E., Lane, M., Smith, F.E., & Dworkin,
R.J. (1992). The relationship of dispositional optimism, daily life stress, and
domestic environment to coping methods used by cancer patients. Journal of
Behavioral Medicine, 15, 127-141.
Friesen, D., Holdaway, E.A., & Rice, A.W. (1983). Satisfaction of school
principals with their work. Educational Administration Quarterly, 19, 35-58.
373
Frone, M.R., Russell, M., & Cooper, M.L. (1994). Relationship between job
and family satisfaction: Causal or noncausal covariation? Journal of Management,
20, 565-579.
Fung-kam, L. (1998). Job satisfaction and autonomy of Hong Kong
registered nurses. Journal of Advanced Nursing, 27, 355-364.
Ganster, D.C. (1989). Worker control and well-being: A review of research
in the workplace. In S. Sauter., J. Hurrell., & C. Cooper (Eds.), Job Control and
Worker Health. (pp. 3-24). Chichester, U.K: Wiley.
Ganster, D.C., Dwyer, D.J., & Fox, M.L. (2001). Explaining employees’
health care costs: A prospective examination of stressful job demands, personal
control, and physiological reactivity. Journal of Applied Psychology, 86, 954-964.
Ganster, D.C., & Fusilier, M.R. (1989). Control in the workplace. In C.L.
Cooper., & I. Robertson (Eds.), International Review of Industrial and
Organizational Psychology. (pp. 235-280). London: John Wiley
Gardner, G. (1977). Is there a valid test of Herzberg’s two-factor theory?
Journal of Occupational Psychology, 50, 197-204.
Gaziel, H.H. (1989). Determinants of perceived deficiency of autonomy
among elementary school administrators. Social Behavior and Personality, 17, 5766.
Geyer, P.D., & Daly, J.P. (1998). Predicting job satisfaction for relocated
workers: Interaction of relocating consequences and employee age. The Journal of
Psychology, 132, 417-427.
374
Gillet, B., & Schwab, D.P. (1975). Convergent and discriminant validities of
corresponding Job Descriptive Index and Minnesota Satisfaction Questionnaire
Scales. Journal of Applied Psychology, 60, 313-317.
Graham, W.K., & Balloun, J. (1973). An empirical test of Maslow’s need
hierarchy theory. Journal of Humanistic Psychology, 13, 97-108.
Grant, G.P., & Murray, C.E. (1999). Teaching in America: The Slow
Revolution. Cambridge, Mass: Harvard University Press.
Gunthert, K.C., Cohen, L.H., & Armeli, S. (1999). The role of neuroticism in
daily stress and coping. Journal of Personality and Social Psychology, 5, 1087-1100.
Hackman, J.R., & Oldham, G.R. (1975). Development of the Job Diagnostic
Survey. Journal of Applied Psychology, 60, 159-170.
Hackman, J.R., & Oldham, G.R. (1976). Motivation through the design of
work: Test of a theory. Organizational Behavior and Human Performance, 16, 250279.
Hackman, J.R., & Oldham, G.R. (1980). Work Redesign. Reading, Mass:
Addison-Wesley.
Hall, D.T., & Nougaim, K.E. (1968). An examination of Maslow’s need
hierarchy theory in an organizational setting. Organizational Behavior and Human
Performance, 3, 12-35.
Halliday, C.A., & Graham, S. (2000). “If I get locked up, I get locked up”:
Secondary control and adjustment among juvenile offenders. Personality and Social
Psychology Bulletin, 26, 548-559.
375
Hallqvist, J., Diderichsen, F., Theorell, T., Reuterwall, C., & Ahlbom, A.
(1998). Is the effect of job strain on myocardial infarction risk due to interaction
between high psychological demands and low decision latitude? Results from
Stockholm Heart Epidemiology Program (SHEEP). Social Science and Medicine,
46, 1405-1415.
Hart, P.M. (1999). Predicting employee life satisfaction: A coherent model
of personality, work and non-work experiences, and domain satisfaction. Journal of
Applied Psychology, 84, 564-584.
Harwood, M.K., & Rice, R.W. (1992). An examination of the referent
selection processes underlying job satisfaction. Social Indicators Research, 27, 1-39.
Hatfield, J.D., Robinson, R.B., & Huseman, R.C. (1985). An empirical
evaluation of a test for assessing job satisfaction. Psychological Reports, 56, 39-45.
Hawkins, M.J., Hawkins, W.E., & Ryan, E.R. (1989). Self-actualisation as
related to age of faculty members at a large Midwestern University. Psychological
Reports, 65, 1120-1122.
Heckhausen, J., & Schulz, R. (1995). A life span theory of control.
Psychological Review, 102, 284-304.
Heckhausen, J., Schulz, R., & Wrosch, C. (1997). Technical Report:
Optimization in primary and secondary control: OPS Scales, a measurement
instrument for general and domain specific strategies of control and development
regulation. Berlin: Max Planck Institute for Human Development.
Heeps, L., Croft, C., & Cummins, R.A. (2000). The Primary and Secondary
Control Scale (2nd ed). Melbourne: Deakin University.
376
Herzberg, F. (1966). Work and the Nature of Man. Cleveland: World Publ
Co.
Herzberg, F., Mausner, B., & Snyderman, B.B. (1959). The Motivation to
Work. New York: Wiley.
Herzberg, F., Mausner, B., & Snyderman, B.B. (1993). The Motivation to
Work. New Brunswick, N.J: Transaction Publishers.
Highhouse, S., & Becker, A.S. (1993). Facet measures and global job
satisfaction. Journal of Business and Psychology, 8, 117-127.
Hill, M.D. (1986). A theoretical analysis of faculty job
satisfaction/dissatisfaction. Educational Research Quarterly, 10, 36-44.
Himle, D.P., & Jayaratne, S. (1991). Buffering effects of four social support
types on burnout among social workers. Social Work Research and Abstracts, 27,
22-28.
Hirschfeld, R.R. (2000). Does revising the intrinsic and extrinsic subscales of
the Minnesota Satisfaction Questionnaire short form make a difference? Educational
and Psychological Measurement, 60, 255-271.
Hoppock, R. (1935). Job Satisfaction. New York: Harper.
Howard, J.L., & Frink, D.D. (1996). The effects of organizational restructure
on employee satisfaction. Group and Organization Management, 21, 278-303.
Iaffaldano, M.T., & Muchinsky, P.M. (1985). Job satisfaction and job
performance: A meta-analysis. Human Relations, 44, 287-307.
377
Ironson, G.H., Smith, P.C., Brannick, M.T., Gibson, W.M., & Paul, K.B.
(1989). Construction of a job in general scale: A comparison of global, composite,
and specific measures. Journal of Applied Psychology, 74, 193-200.
Iverson, R.D., & Maguire, C. (2000). The relationship between job and life
satisfaction: Evidence from a remote mining community. Human Relations, 53, 807813.
Jansen, P.G.M., Kerkstra, A., Abud-Saad, H.H., & Van der Zee, J. (1996).
The effects of job characteristics and individual characteristics of job satisfaction and
burnout in community nursing. International Journal of Nursing Studies, 33, 407421.
Johnson, J.V., & Hall, E.M. (1988). Job strain, work place social support,
and cardiovascular disease: a cross-sectional study of a random sample of the
Swedish working population. American Journal of Public Health, 78, 1336-1342.
Johnson, J.V., & Hall, E.M. (1994). Social support in the work environment
and cardiovascular disease. In S.A. Schumaker., & S.M. Czajkowski (Eds.), Social
Support and Cardiovascular Disease. Plenum Series in Behavioral
Psychophysiology and Medicine. (pp. 145-166.) New York: Plenum Press.
Johnson, J.V., Hall, E.M., & Theorell, T. (1989). Combined effects of job
strain and social isolation on cardiovascular disease morbidity and mortality in a
random sample of the Swedish male working population. Scandinavian Journal of
Work, Environment and Health, 15, 271-279.
378
Johnson, S.M., Smith, P.C., & Tucker, S.M. (1982). Response format of the
Job Descriptive Index: Assessment of reliability and validity by the multitraitmultimethod matrix. Journal of Applied Psychology, 67, 500-505.
Judge, T.A., & Bono, J.E. (2001). Relationship of core self-evaluations
traits- Self-esteem, generalised self-efficacy, locus of control and emotional stability
with job satisfaction and job performance: A meta-analysis. Journal of Applied
Psychology, 86, 80-93.
Judge, T.A., Bono, J.E., & Locke, E.A. (2000). Personality and job
satisfaction: The mediating role of job characteristics. Journal of Applied
Psychology, 85, 237-249.
Judge, T.A., & Locke, E.A. (1993). Effect of dysfunctional thought
processes on subjective well-being and job satisfaction. Journal of Applied
Psychology, 78, 475-490.
Judge, T.A., Locke, E.A., Durham, C.C., & Kluger, A.N. (1998).
Dispositional effects on job and life satisfaction: The role of cognitive evaluations.
Journal of Applied Psychology, 83, 17-34.
Judge, T.A., & Watanabe, S. (1994). Individual differences in the nature of
the relationship between job and life satisfaction. Journal of Occupational and
Organizational Psychology, 67, 101-107.
Kandel, D.B., Davies, M., & Raveis, V.H. (1985). The stressfulness of daily
social roles for women: Marital, occupational and household roles. Journal of Health
and Social Behavior, 26, 64-78.
379
Karasek, R.A. Jr. (1979). Job demands, job decision latitude, and mental
strain: Implications for job redesign. Administrative Science Quarterly, 24, 335-357.
Karasek, R.A., Brisson, C., Kawakami, N., Houtman, I., Bongers, P., &
Amick, B. (1998). The Job Content Questionnaire (JCQ): An instrument for
internationally comparative assessments of psychosocial job characteristics. Journal
of Occupational Health Psychology, 3, 322-355.
Karasek, R.A., & Theorell, T. (1990). Healthy work: Stress, productivity
and the Reconstruction of Working Life. New York: Basic Books.
Karasek, R.A., Triantis, K.P., & Chaudry, S.S. (1982). Co-worker and
supervisor support as moderators of associations between task characteristics and
mental strain. Journal of Occupational Behaviour, 3, 181-200.
Kelly, J. (1992). Does job re-design theory explain job re-design outcomes?
Human Relations, 45, 753-775.
Keppel, G. (1991). Design and analysis: A Researchers Handbook (3rd ed.).
New Jersey: Prentice Hall.
King, N. (1970). Clarification and evaluation of the two factor theory of job
satisfaction. Psychological Bulletin, 74, 18-31.
Klecker, B.M., & Loadman, W.E. (1999). Male elementary school teachers
ratings of job satisfaction by years of teaching experience. Education, 119, 504-604
Koeske, G.F., Kirk, S.A., & Koeske, R.D. (1993). Coping with job stress:
Which strategies work best? Journal of Occupational and Organizational
Psychology, 66, 319-335.
380
Kohn, P.M., Hay, B.D., & Legere, J.J. (1994). Hassles, coping styles, and
negative well-being. Personality and Individual Differences, 17, 169-179.
Kunin, T. (1955). The construction of a new type of attitude measure.
Personnel Psychology, 8, 65-77.
Lahey, K.E., & Vihtelic, J.L. (2000). Finance faculty demographics, career
history, diversity, and job satisfaction. Financial Practice and Education, 10, 111123.
Landeweerd, J.A., & Boumans, N.P.G. (1994). The effect of work
dimensions and need for autonomy on nurses’ work satisfaction and health. Journal
of Occupational and Organizational Psychology, 67, 207-218.
Landry, M.B. (2000). The effects of life satisfaction and job satisfaction on
reference librarians and their work. Reference and user Services Quarterly, 40, 166178.
Landsbergis, P.A., Schnall, P.L., Deitz, D., Friedman, R., & Pickering, T.
(1992). The patterning of psychological attributes and distress by ‘job strain’ and
social support in a sample of working men. Journal of Behavioral Medicine, 15,
379-405.
LaRocco, J.M., House, J.S., & French, J.R.P. (1980). Social support,
occupational stress, and helath. Journal of Health and Social Behavior, 21, 202.
Laschinger, H.K.S., Finegan, J., & Shamian, J. (2001). Promoting nurses’
health: Effect of empowerment on job strain and work satisfaction. Nursing
Economics, 19, 42-52.
381
Latack, J.C. (1986). Coping with job stress: Measures and future direction
for scale development. Journal of Applied Psychology, 71, 377-385.
Lawler, E.E., & Suttle, J.L. (1972). A causal correlational test of the need
hierarchy concept. Organizational Behavior and Human Performance, 7, 265-287.
Lazarus, R.S., & Folkman, S. (1984). Stress, Appraisal and Coping. New
York: Springer.
Leong, F.T.L., & Dollinger, S.J. (1991). NEO Personality Inventory. In D.J.
Keyser., & R.C. Sweetland (Eds.), Test Critiques, Vol VIII. (p.527-539). Texas:
ProEd
Leung, K. (1997). Relationships among satisfaction, commitment and
performance: A group level analysis. Applied Psychology: An International Review,
46, 199-205.
Leung, T., Siu, O., & Spector, P.E. (2000). Faculty stressors, job satisfaction
and psychological distress among university teachers in Hong Kong: The role of
locus of control. International Journal of Stress Management, 7, 121-138
Locke, E.A. (1969). What is job satisfaction? Organizational Behavior and
Human Performance, 4, 309-336.
Locke, E.A. (1976). Nature and causes of job satisfaction. In M. Dunnette
(Ed.), Handbook of Industrial and Organizational Psychology. (pp. 1297-1350).
Chicago, IL: Rand McNally.
Loher, B.T., Noe, R.A., Moeller, N.L., & Fitzgerald, M.P. (1985). A metaanalysis of the relation of job characteristics to job satisfaction. Journal of Applied
Psychology, 70, 280-289.
382
Lu, L. (1999). Work motivation, job stress and employees’ well-being.
Journal of Applied Management Studies, 8, 61-69.
Lykken, D., & Tellegen, A. (1996). Happiness is a stochastic phenomenon.
Psychological Science, 7, 186-189.
Ma, X., & MacMillan, R.B. (1999). Influences of workplace conditions on
teachers job satisfaction. Journal of Educational Research, 93, 39-47.
Maher, E., & Cummins, R.A. (2001). Subjective quality of life, perceived
control, and dispositional optimism among older people. Australasian Journal on
Aging, 20, 139-146.
Maher, E., & Cummins, R.A. (2002). Situation Specific Primary and
Secondary Control Scale. Melbourne: Deakin University.
Maher, E., Misajon, R., Heeps, L., & Cummins, R.A. (2001). Primary
Control and Secondary Control Scale (4th ed.). Melbourne: Deakin University.
Mannheim, B., Baruch, Y., & Tal, J. (1997). Alternative models for
antecedents and outcomes of work centrality and job satisfaction of high-tech
personnel. Human Relations, 50, 1537-1562.
Marriott, A., & Sexton, L. (1994). Components of job satisfaction in
psychiatric social workers. Health and Social Work, 19, 199-206.
Maslow, A.H. (1954). Motivation and Personality. New York: Harper.
Maslow, A.H. (1970). Motivation and Personality. New York: Harper &
Row.
McConatha, J.T., & Huba, H.M. (1999). Primary, secondary and emotional
control across adulthood. Current Psychology, 18, 164-170.
383
McCrae, R.R., & Costa, P.T. (1991). Adding Liebe und Arbeit: The full
Five-Factor Model and well-being. Personality and Social Psychology Bulletin, 17,
227-232.
McCrae, R.R., & Costa, P.T. Jr. (1992). Discriminant validity of NEO-PIR
Facet Scale. Educational Psychological Measurement, 52, 229-237.
Melamed, S., Kushnir, T., & Meir, E.I. (1991). Attentuating the impact of
job demands: Additive and interactive effects of perceived control and social
support. Journal of Vocational Behavior, 39, 40-53.
Michalos, A.C. (1985). Multiple Discrepancies Theory (MDT). Social
Indicators Research, 16, 347-413.
Miles, E.W., Patrick, S.L., & King, W.C. (1996). Job level as a systemic
variable in predicting the relationship between supervisory communication and job
satisfaction. Journal of Occupational and Organizational Psychology, 69, 277-285
Misajon, R. (2002). The Homeostatic Mechanism: Subjective Quality of Life
and Chronic Pain. Unpublished Ph.D. Thesis, Deakin University, Melbourne.
Misajon, R., & Cummins, R.A. (in press). Subjective quality of life, control
and the impact of chronic illness. Arthritis Care and Research.
Mitchell, T.R. (1974). Expectancy models of job satisfaction, occupational
preference and effort. Psychological Bulletin, 81, 1053-1077.
Moorman, R.H. (1993). The influence of cognitive and affective based job
satisfaction measures on the relationship between satisfaction and organizational
citizenship behavior. Human Relations, 46, 759-776.
384
Munro, L., Rodwell, J., & Harding, L. (1998). Assessing occupational stress
in psychiatric nurses using the full Job Strain Model: The value of social support to
nurses. International Journal of Nursing Studies, 35, 339-345.
Narayanan, L., Menon, S., & Spector, P.E. (1999). Stress in the workplace: a
comparison of gender and occupation. Journal of Organizational Behavior, 20, 6373.
Neher, A. (1991). Maslow’s theory of motivation: A critique. Journal of
Humanistic Psychology, 31, 89-112.
Nicolle, P.G. (1994). The effect of work dimensions and need for autonomy
on nurses’ work satisfaction and health. Journal of Occupational and Organizational
Psychology, 64, 207-218.
Niemann, Y.F., & Dovidio, J.F. (1998). Relationship of solo status, academic
rank, and perceived distinctiveness to job satisfaction of racial/ethnic minorities.
Journal of Applied Psychology, 83, 55-71.
Norman, P., Collins, S., Conner, M., Martin, R., & Rance, J. (1995).
Attributions, cognitions, and coping styles: Teleworkers’ reactions to work-related
problems. Journal of Applied Social Psychology, 25, 117-128.
O’Driscoll, M.P., & Beehr, T.A. (2000). Moderating effects of perceived
control and need for clarity on the relationship between role stressors and employee
affective reactions. The Journal of Social Psychology, 140, 151-157.
Olsen, D. (1993). Work satisfaction and stress in the first and third year of
academics appointment. Journal of Higher Education, 64, 453-462.
385
Osipow, S.H., Doty, R.E., & Spokane, A.R. (1985). Occupational stress,
strain and coping across the life span. Journal of Vocational Behavior, 27, 98-108.
Osowiecki, D., & Compas, B.E. (1998). Psychological adjustment to cancer:
Control beliefs and coping in adult cancer patients. Cognitive Therapy and
Research, 22, 483-499.
Osowiecki, D.M., & Compas, B.E. (1999). A prospective study of coping,
perceived control and psychological adjustment to breast cancer. Cognitive Therapy
and Research, 23, 169-180.
Parahoo, K., & Barr, O. (1994). Job satisfaction of community nurses
working with people with a mental handicap. Journal of Advanced Nursing, 20,
1046-1055.
Park, C.L., Folkman, S., & Bostrom, A. (2001). Appraisals of controllability
and coping in caregivers and HIV+ men: Testing the goodness-of-fit hypothesis.
Journal of Consulting and Clinical Psychology, 69, 481-488.
Parker, S.K., & Sprigg, C.A. (1999). Minimising strain and maximising
learning: The role of job demands, job control and proactive personality. Journal of
Applied Psychology, 84, 925-939.
Parkes, K.R. (1984). Locus of control, cognitive appraisal and coping in
stressful episodes. Journal of Personality and Social Psychology, 46, 655-668.
Parkes, K.R. (1989). Personal control in an occupational context. In A.
Steptoe., & A. Appels (Eds.), Stress, Personal Control, and Health (pp. 21-47).
Chichester: Wiley.
386
Parkes, K.R. (1990). Coping, negative affectivity and the work environment:
Additive and interactive predictors of mental health. Journal of Applied Psychology,
75, 399-409.
Parkes, K.R. (1994). Personality and coping as moderators of work stress
process: Models, methods and measures. Work and Stress, 8, 110-129.
Parkes, K.R., & von Rabenau, C. (1993). Work characteristics and wellbeing
among psychiatric health care staff. Journal of Community and Applied Social
Psychology, 3, 243-260.
Parsons, M.B. (1998). A review of procedural acceptability in organizational
behavior management. Journal of Organizational Behavior Management, 18, 173190.
Payne, R., & Fletcher, B.C. (1983). Job demands, supports, and constraints
as predictors of psychological strain among schoolteachers. Journal of Vocational
Behavior, 22, 136-147.
Pearson, C.A.L., & Chong, J. (1997). Contribution of job content and social
information in organizational commitment and job satisfaction: an exploration in a
Malaysian nursing context. Journal of Occupational and Organizational Psychology,
70, 357Pelfrene, E., Vlerick, P., Mak, R.P., De Smets, P., Kornitzers, M., & De
Backer, G. (2001). Scale reliability and validity of the Karasek ‘Job Demand
Control Support’ model in the Belstress study. Work and Stress, 15, 297-313.
387
Perrewe, P.L., & Zellars, K.I. (1999). An examination of attributions and
emotions in the transactional approach to the organizational stress process. Journal
of Organizational Behavior, 20, 739-752.
Porter, L.W. (1961). A study of perceived need satisfaction in bottom and
middle management jobs. Journal of Applied Psychology, 45, 1-10.
Porter, L.W. (1963). Job attitudes in management: II. Perceived importance
of needs as a function of job level. Journal of Applied Psychology, 47, 141-148.
Pritchard, R.D., & Sanders, M.S. (1973). The influence of valence,
instrumentality, and expectancy on effort and performance. Journal of Applied
Psychology, 57, 55-60.
Pulakos, E.D., & Schmitt, N. (1983). A longitudinal study of a valence
model approach for the prediction of job satisfaction of new employees. Journal of
Applied Psychology, 68, 307-312.
Quinn, R.O., & Staines, G.L. (1979). The 1977 Quality of Employment
Survey: Descriptive statistics with comparison data from the 1960-1970 and the
1972-73 surveys. University of Michigan: Ann Arbor.
Rain, J.S., Lane, I.M., & Steiner, D.D. (1991). A current look at the job
satisfaction/life satisfaction relationship: Review and future considerations. Human
Relations, 44, 287-307.
Reinharth, L., & Wahba, R.A. (1976). A test of alternative models of
expectancy theory. Human Relations, 29, 257-272.
388
Renn, R.W., & Vandenberg, R.J. (1995). The critical psychological states:
An underrepresented component on Job Characteristics Model research. Journal of
Management, 21, 279-303.
Rice, R.W., McFarlin, D.B., & Bennett, D.,E. (1989). Standards of
comparison and job satisfaction. Journal of Applied Psychology, 74, 591-598.
Roberts, K.H., & Glick, W. (1981). The job characteristics approach to task
design: A critical review. Journal of Applied Psychology, 66, 193-217.
Roberts, K.H., Walter, G.A., & Miles, R.E. (1971), A factor analytic study of
job satisfaction items designed to measure Maslow need categories. Personnel
Psychology, 24, 205-220.
Roberts, S.M. (1995). Applicability of the goodness of fit hypothesis to
coping with daily hassles. Psychological Reports, 77, 943-954.
Roberts, T.B. (1982). “Comments on Mathes’s article”. Journal of
Humanistic Psychology, 22, 97-98.
Rodriguez, I., Bravo, M.J., Peiro, J.M., & Schaufeli, W. (2001). The
demands-control-support model, locus of control and job dissatisfaction: a
longitudinal study. Work and Stress, 15, 97-114.
Rothbaum, F., Weisz, J.R., & Synder, S.S. (1982). Changing the world and
changing the self: A two process model of perceived control. Journal of Personality
and Social Psychology Bulletin, 42, 5-37.
Roznowski, M. (1989). Examination of the measurement properties of the
Job Descriptive Index with experimental items. Journal of Applied Psychology, 74,
805-814.
389
Russell, D.W., Altmaier, E., & Van Velzen, D. (1987). Job-related stress,
social support, and burnout among classroom teachers. Journal of Applied
Psychology, 72, 269-274.
Ryan, E.M., & Deci, E.L. (2001). On happiness and human potentials: A
review of research on hedonic and eudaimonic well-being. Annual Review of
Psychology, 141-155.
Saklofske, D.H., & Kelly, I.W. (1995). Coping and personality.
Psychological Reports, 77, 481-482.
Salancik, G.R., & Pfeffer, J. (1977). An examination of need satisfaction
models of job attitudes. Administrative Science Quarterly, 22, 427-456.
Schappe, S.P. (1998). The influence of job satisfaction, organizational
commitment, and fairness perceptions on organizational citizenship behavior. The
Journal of Psychology, 132, 277-291.
Schaubroeck, J., & Merritt, D.E. (1997). Divergent effects of job control on
coping with work stressors: The key role of self-efficacy. Academy of Management
Journal, 40, 738-754.
Scheier, M.F., Weintraub, J.K., & Carver, C.S. (1986). Coping with stress:
Divergent strategies of optimists and pessimists. Journal of Personality and Social
Psychology, 51, 1257-1264.
Schmidt, G.L. (1976). Job satisfaction among secondary school
administrators. Educational Administration Quarterly, 12, 68-86.
390
Schonfeld, I.S. (2000). An updated look at depressive symptoms and job
satisfaction in first year women teachers. Journal of Occupational and
Organizational Psychology, 73, 363-371.
Schwab, D.P., Olian-Gottlieb, J.D., & Heneman, H.G. (1979). Betweensubjects expectancy theory research: A statistical review of studies predicting effort
and performance. Psychological Bulletin, 86, 139-147.
Schulz, R., & Heckhausen, J. (1996). A life-span model of successful aging.
American Psychologist, 51, 702-714.
Schulz, R., & Heckhausen, J. (1999). Aging, culture and control. Journal of
Gerontology: Psychological Sciences, 54, 139-145.
Schwartz, J.E., Neale, J., Marco, C., Shiffman, S.S., & Stone, A.A. (1999).
Does trait coping help? A momentary assessment approach to the evaluation of
traits. Journal of Personality and Social Psychology, 77, 360-369.
Secret, M., & Green, R.G. (1998). Occupational status differences among
three groups of married mothers. Journal of Women and Social Work, 13, 47-70.
Senate Employment, Education and Training Referees Committee. (1998). A
Class Act: Inquiry into the Status of the Teaching Profession. Canberra.
Sergiovanni, T. (1967). Factors which affect satisfaction and dissatisfaction
of teachers. Journal of Educational Administration, 5, 66-82
Shoura, M.M., & Singh, A. (1999). Motivation parameters for engineering
managers using Maslow’s theory. Journal of Management in Engineering, 15, 44-55.
Silver, P.F. (1987). Job satisfaction and dissatisfaction revisited. Educational
and Psychological Research, 7, 1-20.
391
Smith, C.A., Organ, D.W., & Near, J.P. (1983). Organizational citizenship
behavior: Its nature and antecedents. Journal of Applied Psychology, 68, 653-663.
Smith, C.S., Tisak, J., Hahn, S.E., & Schmeider, R.A. (1997). The
measurement of job control. Journal of Organizational Behavior, 18, 225-237.
Smith, P.C., Kendall, L.M., & Hulin, C.L. (1969). The Job Descriptive
Index. Bowling Green, OH: Department of Psychology, Bowling Green State
University.
Sobel, R.S. (1971). Tests of preperformance and postperformance models of
satisfaction with outcomes. Journal of Personality and Social Psychology, 19, 213321.
Spector, P.E. (1986). Perceived control by employees: A meta-analysis of
studies concerning autonomy and participation at work. Human Relations, 39, 10051017.
Spector, P.E. (1987). Interactive effects of perceived control and job
stressors on affective reaction and health outcomes for clerical workers. Work and
Stress, 1, 155-162.
Spector, P.E. (1997). Job Satisfaction: Applications, Assessment, Causes and
Consequences. London: Sage Publications.
Spector, P.E., Dwyer, D.J., & Jex, S.M. (1988). Relation of job stressors to
affective, health and performance outcomes: A comparison of multiple data sources.
Journal of Applied Psychology, 73, 11-19.
Spector, P.E., & O'Connell, B.J. (1994). The contribution of personality
traits, negative affectivity, locus of control and Type A to the subsequent reports of
392
job stressors and job strains. Journal of Occupational and Organizational
Psychology, 67, 1-11.
Spokes, A. (1998). Subjective quality of life of people with schizophrenia:
Examination of perceived control and positive illusions. Unpublished Honours
Thesis, Deakin University, Melbourne, Australia.
Staw, B.M., Bell, N.E., & Clausen, J.A. (1986). The dispositional approach
to job attitudes: A lifetime longitudinal test. Administrative Science Quarterly, 31,
56-77.
Staw, B.M., & Ross, J. (1985). Stability in the midst of change: A
dispositional approach to job attitudes. Journal of Applied Psychology, 70, 469-480.
Stone, A.A., & Neale, J.M. (1984). New measure of daily coping:
Development and preliminary results. Journal of Personality and Social Psychology,
46, 892-906.
Tabachnick, B.G., & Fidell, L.S. (1996). Using Multivariate Statistics (3rd
ed.). Northridge: Harper Collins.
Taber, T.D., & Taylor, E. (1990). A review and evaluation of the
psychometric properties of Job Diagnostic Survey. Personnel Psychology, 43, 467512.
Tait, M., Padgett, MY., & Baldwin, T.T. (1989). Job and life satisfaction: A
reevaluation of the strength of the relationship and gender effects as a function of the
date of the study. Journal of Applied Psychology, 74, 502-507.
Teacher Stress in Victoria: A survey of teachers’ views: The Report (1990).
Victoria: Ministry of Education.
393
Teas, R.K. (1981). A within-subject analysis of valence models of job
preference and anticipated satisfaction. Journal of Occupational Psychology, 54,
109-124.
Terry, D.J., & Hynes, G.J. (1998). Adjustment to a low-control situation:
Reexamining the role of coping responses. Journal of Personality and Social
Psychology, 74, 1078-1092.
Terry, D.J., Nielsen, M., & Perchard, L. (1999). Effects of work stress on
psychological well-being and job satisfaction: The stress-buffering role of social
support. Australian Journal of Psychology, 45, 168-175.
Thurstone, L.L., & Jones, L.V. (1957). The rational origin for measuring
subjective values. Journal of the American Statistical Association, 52, 458-471.
Tiegs, R.B., Tetrick L.E., & Fried, Y. (1992). Growth need strength and
context satisfactions as moderators of the relations of the Job Characteristics Model.
Journal of Management, 18, 575-593.
Tinsley, H.E.A. (1994). NEO Personality Inventory-Revised. In D.J.
Keyser., & R.C. Sweetland (Eds.), Test Critiques, Vol X. (p. 443-456). Texas:
ProEd.
Thompson, S.C., Collins, M.A., Newcomb, M.D., & Hunt, W. (1996). On
fighting versus accepting stressful circumstances: Primary and secondary control
among HIV-positive men in prison. Journal of Personality and Social Psychology,
70, 1307-1317.
394
Thompson, S.C., Nanni, C., & Levine, A. (1994). Primary versus secondary
and central versus consequence-related control in HIV-Positive men. Journal of
Personality and Social Psychology, 67, 540-547.
Thompson, S.C., Sobolew-Shubin, A., Galbraith, M.E., Schwankovsky, L., &
Cruzen, D. (1993). Maintaining perceptions of control: Finding perceived control in
low-control circumstances. Journal of Personality and Social Psychology Bulletin,
64, 293-304.
Thompson, S.C., Thomas, C., Rickabaugh, C.A., Tantamjarik, P., Otsuki, T.,
Pan, D., Garcia, B.F., & Sinar, E. (1998). Primary and secondary control over agerelated changes in physical appearance. Journal of Personality, 66, 583-605.
Thurber, C.A., & Weisz, J.R. (1997). “You can try or you can just give up”:
The impact of perceived control and coping style on childhood homesickness.
Developmental Psychology, 33, 508-517.
Thurstone, L.L., & Jones, L.V. (1957). The rational origin for measuring
subjective values. Journal of the American Statistical Association, 52, 458-471.
Tokar, D.M., & Subich, L.M. (1997). Relative contributions of congruence
and personality dimensions to job satisfaction. Journal of Vocational Behavior, 50,
482-491.
Unden, A. (1996). Social support at work and its relationship to absenteeism.
Work and Stress, 10, 46-61.
Valentiner, D.P., Holahan, C.J., & Moos, R.H. (1994). Social support,
appraisals of event controllability and coping: An integrative model. Journal of
Personality and Social Psychology, 66, 1094-1102.
395
Van Der Doef, M., & Maes, S. (1999). The job demand-control (-support)
model and psychological well-being: a review of 20 years of empirical research.
Work and Stress, 13, 87-114.
Van Eerde, W., & Thierry, H. (1996). Vroom’s expectancy models and
work-related criteria: A meta-analysis. Journal of Applied Psychology, 81, 575-586.
Vitaliano, P.P., DeWolfe, D., Maiuro, R.D., Russo, J., & Katon, W. (1990).
Appraised changeability of a stressor as a modifier of the relationship between
coping and depression: A test of the hypothesis of fit. Journal of Personality and
Social Psychology, 59, 582-592.
Vitaliano, P.P., Russo, J., Carr, J.E., Maiuro, R.D., & Becker, J. (1985). The
Ways of Coping Checklist: Revision and psychometric properties. Multivariate
Behavioral Research, 20, 3-26.
Vroom, V.H. (1964). Work and motivation. New York: Wiley.
Wahba, M.A., & House, R.J. (1976). Expectancy theory in work and
motivation: Some logical and methodological issues. Human Relations, 27, 121-147.
Wahba, M.A., & Bridwell, L.G. (1976). Maslow reconsidered: A review of
the research on the need hierarchy theory. Organizational Behavior and Human
Performance, 15, 212-240.
Wall, T.D., Clegg, C.W., & Jackson, P.R. (1978). An evaluation of the Job
Characteristics Model. Journal of Occupational Psychology, 51, 183-196.
Wall, T.D., Jackson, P.R., Mullarkey, S., & Parker, S.K. (1996). The
demands-control model of job strain: A more specific test. Journal of Occupational
and Organizational Psychology, 69, 153-166.
396
Wanous, J.P., Reichers, A.E., & Hudy, M.J. (1997). Overall job satisfaction:
How good are single-item measures? Journal of Applied Psychology, 82, 247-252.
Warr, P., Cook, J., & Wall, T. (1979). Scales for the measurement of some
work attitudes and aspects of psychological well-being. Journal of Occupational
Psychology, 52, 129-148.
Warr, P.B. (1990). Decision latitude, job demands, and employee well-being.
Work and Stress, 4, 285-294.
Waters, L.K., & Waters, C.W. (1969). Correlates of job satisfaction and job
dissatisfaction among female clerical workers. Journal of Applied Psychology, 53,
388-391.
Weiss, D.J., Dawis, R.V., England,G.W., & Lofquist, L.H. (1967). Manual
for the Minnesota Satisfaction Questionnaire. Minneapolis: Industrial Relations
Centre, University of Minnesota.
Weiss, H.M., Nicholas, J.P., & Daus, C.S. (1999). An examination of the
joint effect of affective experienced and job beliefs on job satisfaction and variations
in affective experiences over time. Organizational Behavior and Human Decision
Processes, 78, 1-24.
Wernimont, P.F. (1966). Intrinsic and extrinsic factors in job satisfaction.
Journal of Applied Psychology, 50, 41-50.
White, R.W. (1959). Motivation reconsidered: The concept of competence.
Psychological Review, 66, 297-333.
Whitsett, D.A., & Winslow, E.K. (1967). An analysis of studies critical of
the motivation-hygiene theory. Personnel Psychology, 391-414.
397
Wicker, F.W., Brown, G., Wiehe, J.A., Hagen, A.S., & Reed, J.L. (1993). On
reconsidering Maslow: An examination of the Deprivation/Domination proposition.
Journal of Research in Personality, 27, 118-133.
Wicker, F.W., & Wiehe, J.A. (1999). An experimental study of Maslow’s
deprivation-domination proposition. Perceptual and Motor Skills, 88, 1356-1358.
Wilensky, H.L. (1960). Work, careers and social integration. International
Social Science Journal, 12, 543-560.
Williams, S., & Cooper, C.L. (1996). Occupational Stress Indicator, Version
2. Harrogate, North Yorkshire: RAD.
Winefield, A.H. (2000). Stress in academe: Some recent research findings.
In D.T. Kenny., & J.G. Carlson., F.J. McGuigan., J.L. Sheppard (Eds.), Stress and
Health: Research and Clinical Applications. (pp. 437-446). Amsterdam: Harwood
Academic.
Witt, L.A., Andrews, M.C., & Kacmar, K.M. (2000). The role of
participation in decision-making in the organisational politics-job satisfaction
relationship. Human Relations, 53, 341- 350.
Wong, K.S., & Cheuk, W.H., & Rosen, S. (2000). The influence of job stress
and supervisor support on negative affects and job satisfaction in kindergarten
principals. Journal of Social Behavior and Personality, 15, 85-99.
Wright, T.A., Bennett, K.K., & Dun, T. (1999). Life and job satisfaction.
Psychological Reports, 84, 1025-1028.
398
Wyrwich, K.W., Nienaber, N.A., Tierney, W.M., & Wolinsky, F.D. (1999).
Linking clinical relevance and statistical significance in evaluating individual
changes in health-related quality of life. Medical Care, 37, 469-478.
399
5.7
Appendices
400
Appendix A- Plain Language Statement for Study One
Dear Sir/Madam,
My name is Elise Maher, and I am completing my Ph.D. in Psychology at Deakin
University. As part of my studies, I am undertaking a research project under the
supervision of Professor Robert Cummins, a researcher in the School of Psychology.
This study is investigating job satisfaction and control. The study aims to provide
useful information about how the amount of choice that an employee has influences
job satisfaction. The results will provide information that will enhance programs that
increase job satisfaction.
You are invited to participate in this research. If you agree, you will be asked to
complete the enclosed questionnaire. Any information you provide will be
anonymous and confidential. Only group results will be reported and no individuals
will be identified. Upon completion of the study, data will be secured in a locked
cabinet in the School of Psychology, Deakin University, for a minimum period of six
years from the date of publication.
The questionnaire should take around 30 minutes to complete and your participation
would be greatly appreciated. Examples of questions are: "My work is boring", "In
my job, I can choose the amount I earn", "I am not a worrier" and "How satisfied are
you with your close relationships with family or friends". You are free to withdraw
at any time during the study in which event your participation in the research study
will immediately cease and any information obtained will not be used. You are free
to refuse to answer any questions.
Following the completion of the study, I will provide your employer with a summary
of the results. If you would like a copy of the summary sent directly to you, please
contact Elise Maher.
If you have any further questions regarding the study, please contact:
Elise Maher on 9251 7153 or Email: ecmaher@deakin.edu.au
Or you can contact Professor Robert Cummins on 9244-6845 or Email:
robert.cummins@deakin.edu.au.
If you are happy to be involved in this study, please complete the enclosed
questionnaire and return it in the reply-paid envelope supplied (i.e., NO
STAMP NEEDED).
________________________________________________________________________
Should you have any concerns about the conduct of this research project, please contact the
Secretary, Deakin University Ethics Committee, Research Services, Deakin University, 221
Burwood Highway, BURWOOD, VIC, 3125, Tel (03) 9251 7123
401
Appendix B- Job Autonomy Scale used in Study One (Revision of Ganster,
1989, cited in Dwyer & Ganster, 1991)
Indicate your agreement with the following 13 statements by ticking () a number
ranging from 1 to 10, where 1= Do not agree at all, and 10= Agree completely. All
of the statements begin with “In my job, I can choose….”
In my job, I can choose:
1) In my job, I can choose among a variety of tasks or projects to do.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
2) In my job, I can choose the order in which I do my work.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
3) In my job, I can choose how quickly I work.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
4) In my job, I can choose how I schedule my rest breaks.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
5) In my job, I can choose the physical conditions of my workstation.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
6) In my job, I can choose when I interact with others.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
7) In my job, I can choose the amount I earn.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
8) In my job, I can choose the number of times I am interrupted at work.
0
Do not
agree at all
1
2
3
4
5
6
7
8
9
10
Agree
Completely
402
9) In my job, I can choose how my work is evaluated.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
10) In my job, I can choose the quality of my work.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
11) In my job, I can choose the policies and procedures in my work unit.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
12) In my job, I can choose among a variety of methods to complete my work.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
13) In my job, I can choose how much work I get done.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
14) In general, how much are you able to influence work and work-related
matters?
0
Very Little
1
2
3
4
5
6
7
8
9
10
Very Much
403
Appendix C- Primary and Secondary Control Scale used in Study One
(Revision of Heeps et al., 2000)
Indicate your agreement with the following statements by selecting a number ranging
from 1 to 10, where 1=Do not agree at all, and 10= Agree Completely
1) When a goal that I have at work is difficult to reach, I think about different ways to
achieve it.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
2) When I want something at work to change, I think I can make it happen.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
3) When a work task really matter to me, I think about it a lot.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
4) When I really want to reach a goal at work, I believe I can achieve it.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
5) When faced with a difficult work situation, I believe I can overcome it.
0
Do not
agree at all
1
2
3
4
5
6
7
8
9
10
Agree
Completely
404
Indicate your agreement with the following statements by selecting a number ranging
from 1 to 10 where 1=Do not agree at all, and 10=Agree completely. All the
statements begin with “When something bad happens that I cannot change…”
When something bad happens at work that I cannot change:
When something bad happens at work that I cannot change
1) I can see that something good will come of it.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
When something bad happens at work that I cannot change
2) I remember you can’t always get what you want.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
When something bad happens at work that I cannot change
3) I know things will work out OK in the end.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
When something bad happens at work that I cannot change
4) I remember I am better off than many other people.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
When something bad happens at work that I cannot change
5) I remember I have already accomplished a lot in life.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
When something bad happens at work that I cannot change
6) I remember the success of my family or friends.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
When something bad happens at work that I cannot change
7) I think nice thoughts to take my mind off it.
0
Do not
agree at all
1
2
3
4
5
6
7
8
9
10
Agree
Completely
405
When something bad happens at work that I cannot change
8) I remind myself the situation will change if I am just patient.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
When something bad happens at work that I cannot change
9) I tell myself it doesn’t matter.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
When something bad happens at work that I cannot change
10) I think about my success in other areas.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
When something bad happens at work that I cannot change
11) I don’t feel disappointed because I knew it might happen.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
When something bad happens at work that I cannot change
12) I can see it was not my fault.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
When something bad happens at work that I cannot change
13) I ignore it by thinking about other things.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
When something bad happens at work that I cannot change
14) I realise I didn’t need to control it anyway.
0
Do not
agree at all
1
2
3
4
5
6
7
8
9
10
Agree
Completely
406
Appendix D- Job Satisfaction Scale used in Study One (Revision of Roznowski,
1989)
Indicate your agreement with the following 15 statements by ticking () a number
ranging from 1 to 10, where 1= Do not agree at all, and 10= Agree completely.
1) My work is boring.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
2) My co-workers are stupid.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
3) My pay is bad.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
4) My supervisors know how to supervise.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
5) There is a good chance for promotion in my job.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
6) My co-workers are responsible.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
7)
10
Agree
Completely
I am well-paid.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
8) My work is dull.
0
Do not
agree at all
1
2
3
4
5
6
7
8
9
10
Agree
Completely
407
9) There is a fairly good chance for promotion in my job.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
10) My supervisors are bad.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
11) My work is interesting.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
12) My pay is unfair.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
13) My supervisors are annoying.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
14) My co-workers are a waste of time.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
15) There are good opportunities for advancement in my job.
0
Do not
agree at all
1
2
3
4
5
6
7
8
9
10
Agree
Completely
408
Appendix E- Life Satisfaction Scale used in Study One (Cummins, 1997)
Please tick () the box that best describes how SATISFIED you are with each
area. Do not spend too much time on any one question. There are no right or wrong
answers.
1) How Satisfied are you with the THINGS YOU OWN ?
0
1
2
3
4
5
6
7
8
9
Completely
dissatisfied
10
Completely
satisfied
2) How Satisfied are you with your HEALTH?
0
1
2
3
4
5
6
7
8
9
Completely
dissatisfied
10
Completely
satisfied
3) How Satisfied are you with what you ACHIEVE IN LIFE ?
0
1
2
3
4
5
6
7
8
9
Completely
dissatisfied
10
Completely
satisfied
4) How Satisfied are you with your CLOSE RELATIONSHIPS with FAMILY
or FRIENDS ?
0
1
2
3
4
5
6
7
8
9
Completely
dissatisfied
10
Completely
satisfied
5) How Satisfied are you with HOW SAFE YOU FEEL ?
0
1
2
3
4
5
6
7
8
9
Completely
dissatisfied
10
Completely
satisfied
6) How Satisfied are you with feeling part of your COMMUNITY?
0
1
2
3
4
5
6
7
8
9
Completely
dissatisfied
10
Completely
satisfied
7) How Satisfied are you with YOUR OWN HAPPINESS ?
0
Completely
dissatisfied
1
2
3
4
5
6
7
8
9
10
Completely
satisfied
409
Appendix F- Personality Scale used in Study One (Costa & McCrae, 1992)
This questionnaire contains 24 statements. Read each statement carefully. For each
statement tick () the box with the response that best represents your opinion.
1) I am not a worrier.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
2) I like to have a lot of people around me.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
3) I often feel inferior to others.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
4) I laugh easily.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
5) When I'm under a great deal of stress, sometimes I feel like I'm going to
pieces.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
6) I don't consider myself especially light hearted.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
7) I rarely feel lonely or blue.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
8) I really enjoy talking to people.
0
Do not
agree at all
1
2
3
4
5
6
7
8
9
10
Agree
Completely
410
9) I often feel tense and jittery.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
10) I like to be where the action is.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
11) Sometimes I feel completely worthless.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
12) I usually prefer to do things alone.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
13) I rarely feel fearful or anxious.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
14) I often feel as if I am bursting with energy.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
16) I often get angry at the way people treat me.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
16) I am a cheerful, high-spirited person.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
17) Too often, when things go wrong, I get discouraged and feel like giving up.
0
Do not
agree at all
1
2
3
4
5
6
7
8
9
10
Agree
Completely
411
18) I am not a cheerful optimist.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
19) I am seldom sad or depressed.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
20) My life is fast-paced.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
21) I often feel helpless and want someone else to solve my problems.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
22) I am a very active person.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
23) At times I have been so ashamed I just want to hide.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
Completely
24) I would rather go my own way than be a leader of others.
0
Do not
agree at all
1
2
3
4
5
6
7
8
9
10
Agree
Completely
412
Appendix G-Levels of Job Satisfaction reported by various occupational groups
No
Author
Occupation
Scale
N
1
Leung et al.,
(2000)
Academics
106
2
Judge et al.,
(1998)
Judge et al.,
(2000)
Renn &
Vandenberg
(1995)
Hackman &
Oldham
(1975)
Dollard et al.,
(2000)
Physicians
Job Satisfaction
Scale (OSI-2;
Williams &
Cooper, 1996)
Brayfield & Rothe
(1951)
Brayfield & Rothe
(1951)
Job Diagnostic
Survey (Hackman
& Oldham, 1975)
Job Diagnostic
Survey (Hackman
& Oldham, 1975)
Global measure
Job
Satisfaction
(%SM)
57.6
164
72.67
384
107
188
69.6,
76.53
64.8
658
60.3
786
64
Warr, Cook & Wall
(1979) intrinsic
satisfaction
1451
51.33
Accounting
firms
Manual and
non-manual
workers
Facet measure (12
facets)
Global measure
236
68
60
58
Nurses
Faces scale (Kunin,
1955)
Job Diagnostic
Survey (Hackman
& Oldham, 1975)
Global measure (4
items)
501
(manual)
788 (nonmanual)
136
39
70
177
52.5
550
62.04
3
4
5
6
7
8
9
10
11
12
13
Wall,
Jackson,
Mullarkey &
Parker (1996)
O’Driscoll &
Beehr (2000)
Fletcher &
Jones (1993)
Fox et al.,
1993
Mannheim,
Baruch & Tal
(1997)
Beutell &
WittigBerman
(1999)
Agho, Price
& Mueller
(1992)
General
Management,
counseling,
administration
General
Public sector
welfare
workers
Manufacturing
employees
Managerial
personnel
MBA students
Employees of
medical centre
Brayfield & Rothe
(1951)
76.33
413
14
Fisher (2000)
General
Faces scale (Kunin,
1955)
124
87.0
15
Howard &
Frink (1996)
Job Diagnostic
Survey (Hackman
& Oldham, 1975)
248
65.83
16
Managers,
administrators
police,
firefighters,
labourers
Nurses
Jansen,
Kerkstra,
Abud-Saad &
Van der Zee
(1996)
Laschinger et Nurses
al., (2001)
Facet measure
(Algera, 1980)
355
nurses, 92
nurse
auxiliaries
63.75
(nurses) &
68.0 (nurse
auxiliaries)
Job Diagnostic
Survey (Hackman
& Oldham, 1975)
Graduate
Quinnes & Staines
teachers
(1979)-Global
Administrators Hoppock (1935)
600
44.75
184
80
169
78.5
Teachers
Global items (4)
2,202
80.25
Private sector
organisation
Insurance
company
workers
Global items (3)
174
57
Minnesota
Satisfaction
Questionnaire
(Weiss et al., 1967)
Brayfield & Rothe
(1951)
Job Diagnostic
Survey (Hackman
& Oldham, 1975)
Job perception
scale (Hatfield,
Robinson &
Huseman, 1985)
Hoppock (1935)
150
71.50
47
71.36
286
68.88
713
57.5
1251
63.75
17
18
19
20
21
22
23
24
Schonfeld
(2000)
Finlay,
Martin,
Romas &
Blum (1995)
Ma &
Macmillan
(1999)
Geyer &
Daly (1998)
Schappe
(1998)
Parsons
Nurses
(1998)
Pearson &
Nurses
Chong (1997)
25
Miles, Patrick Manufacturing
& King
employees
(1996)
26
Witt,
Andrews &
Kacmar
(2000)
Public sector
organisation
414
27
Bogg &
Copper
(1995)
Weiss,
Nicholas &
Daus (1999)
Moorman
(1993)
Parahoo &
Barr (1994)
Spector,
Dwyer & Jex
(1988)
Civil servants,
executives
OSI (Cooper, Sloan
& Williams, 1988)
1051
1056
55.6
62.0
Managers
24
82
35
62.25
71
75
155
78
32
Spector &
O’Connell
(1994)
University
Graduates
33
Wong et al.,
(2000)
Klecker &
Loadman
(1999)
Kindergarten
principles
Teachers
Faces Scale (Kunin,
1955), and global
scale
Brayfield & Rothe
(1951)
Global measure (1
item)
Michigan
Orgnizational
Assessment
(Cammann,
Fichman, Jenkins &
Klesh, 1979)
Michigan
Orgnizational
Assessment
(Cammann et al.,
1979)
Global item
108
54
1874
68.16
Marriott &
Sexton
(1994)
Frone,
Russell &
Cooper
(1994)
Social workers
National follow-up
survey of teachers
education
graduates- 7 facets
Global measure- 1
item
188
66.9
Global Scale
(Kandel, Davies &
Raveis, 1985)
631
73.33
MEAN
(N= 41)
66.75
28
29
30
31
34
35
36
Manufacturers
Nurses
Secretaries
Random
66.66
415
Appendix H- Primary and Secondary Control Scale for Study Two (Maher et
al., 2001)
The following items assess the difficulties that you have at work.
Please tick () the areas in which you experience difficulties in your work.







Time management (making time to do everything)
Motivation
Interpersonal relationships (colleagues, or supervisors)
Nature of the Work
Promotions
Pay
Other
b) How often do you have difficulty doing something at work? (ie., think of the
examples given above, or other difficulties you may have had at work)
0
Never
1
2
3
4
5
6
7
8
9
10
All the time
416
Here are ways people deal with difficult situations at work.
How often have you had these thoughts when facing a difficulty at work OVER
THE PAST WEEK?
I thought………
1) It will work out okay in the end.
0
1
2
3
4
5
6
7
8
9
Never
10
Every time
2) I knew it would happen.
0
1
2
3
4
5
6
7
8
9
Never
10
Every time
3) I can't always get what I want.
0
1
2
3
4
5
6
7
8
9
Never
10
Every time
4) It doesn’t matter.
0
1
2
3
4
5
6
7
8
9
Never
10
Every time
5) I am better off than many other people.
0
1
2
3
4
5
6
7
8
9
Never
10
Every time
6) It was not my fault.
0
Never
1
2
3
4
5
6
7
8
9
10
Every time
417
Here are other ways people deal with difficult situations at work.
How often have you done these things when facing a difficulty at work
OVER THE PAST WEEK?
7) I looked for different ways to overcome it.
0
1
2
3
4
5
6
7
8
9
Never
10
Every time
8) I kept trying.
0
1
2
3
4
5
6
7
8
9
Never
10
Every time
9) I told someone about it.
0
1
2
3
4
5
6
7
8
9
Never
10
Every time
10) I worked to overcome it.
0
1
2
3
4
5
6
7
8
9
Never
10
Every time
11) I thought of the success of my family or friends.
0
1
2
3
4
5
6
7
8
9
Never
10
Every time
12) I thought about my success in other areas.
0
1
2
3
4
5
6
7
8
9
Never
10
Every time
13) I did something different, like going for a walk.
0
1
2
3
4
5
6
7
8
9
Never
10
Every time
14) I ignored it.
0
Never
1
2
3
4
5
6
7
8
9
10
Every time
418
15) I worked out how to remove obstacles.
0
1
2
3
4
5
6
7
8
9
Never
10
Every time
16) I looked for something else that was positive in the situation.
0
Never
1
2
3
4
5
6
7
8
9
10
Every time
419
Appendix I- Plain Language Statement for Study Two
Dear Sir/Madam,
My name is Elise Maher, and I am completing my Ph.D. in Psychology at Deakin
University. As part of my studies, I am undertaking a research project under the
supervision of Professor Robert Cummins, a researcher in the School of Psychology.
This study is investigating job satisfaction and control. The study aims to provide
useful information about how the amount of choice that an employee has influences
job satisfaction. The results will provide information that will enhance programs that
increase job satisfaction.
You are invited to participate in this research. If you agree, you will be asked to
complete the enclosed questionnaire. Any information you provide will be
anonymous and confidential. Only group results will be reported and no individuals
will be identified. Upon completion of the study, data will be secured in a locked
cabinet in the School of Psychology, Deakin University, for a minimum period of six
years from the date of publication.
The questionnaire should take around 30 minutes to complete and your participation
would be greatly appreciated. Examples of questions are: "I am satisfied with the
praise I get for doing a good job", "How much can you choose the amount that you
earn", "I am not a worrier", "How satisfied are you with your close relationships with
family or friends", and “Which management style do you prefer”. You are free to
withdraw at any time during the study in which event your participation in the
research study will immediately cease and any information obtained will not be used.
You are free to refuse to answer any questions.
Following the completion of the study, I will provide your employer with a summary
of the best coping strategies. If you would like a copy of the summary sent directly
to you, please contact Elise Maher. If you have any further questions regarding the
study, please contact: Elise Maher on 9251 7153 or Email:
elisem@deakin.edu.au, or you can contact Professor Robert Cummins on 92446845 or Email: cummins@deakin.edu.au
If you are happy to be involved in this study, please complete the enclosed
questionnaire and return it in the reply-paid envelope supplied (i.e., NO STAMP
NEEDED).
Thank you very much for your time.
_________________________________________________________________
Should you have any concerns about the conduct of this research project, please contact the Secretary,
Deakin University Ethics Committee, Research Services, Deakin University, 221 Burwood Highway,
BURWOOD, VIC, 3125, Tel (03) 9251 7123
420
Appendix J- Job Autonomy Scale for Study Two (Hackman & Oldham, 1975)
The following 3 items assess how much freedom you have at your work. For each
item, please tick () a number ranging from 1 to 10, where 1= Do not agree at all,
and 10= Agree completely.
1) In my job, I can decide on my own how to go about doing my work.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
completely
2) In my job, I have the chance to use my personal initiative and judgement in
carrying out the work.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
completely
3) In my job, I have considerable opportunity for independence and freedom.
0
Do not
agree at all
1
2
3
4
5
6
7
8
9
10
Agree
completely
421
Appendix K- Need for Autonomy Scale for Study Two (de Rijk et al., 1998)
The following 4 items assess how important it is for you to do certain things at work.
Please tick a box ranging from 1=Not important at all to 10=Could not be more
important.
1) How important is it for you to set the pace of your tasks at work?
0
1
2
3
4
5
6
7
8
9
Not
important
at all
10
Could not
be more
important
2) How important is it for you to have control over what you do at work and the
way that you do it?
0
1
2
3
4
5
6
7
8
9
Not
important
at all
10
Could not
be more
important
3) How important is it for you to do your own planning at work?
0
1
2
3
4
5
6
7
8
9
Not
important
at all
10
Could not
be more
important
4) How important is it for you to give orders at work instead of receiving them?
0
Not
important
at all
1
2
3
4
5
6
7
8
9
10
Could not
be more
important
422
Appendix L- Job Satisfaction Scale for Study Two (Weiss et al., 1967)
Please indicate how satisfied you are with the following aspects of your work.
Please tick a box ranging from (1) Very dissatisfied to (10) Very satisfied.
On my present job, this is how I feel about…..
1) Being able to keep busy all the time.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
2) The chance to work alone on the job.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
3) The chance to do different things from time to time.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
4) The chance to be “somebody” in the community.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
5) The way my boss handles his/her work.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
6) The competence of my supervisor in making decisions.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
7) Being able to do things that don’t go against my conscience.
0
Very
dissatisfied
1
2
3
4
5
6
7
8
9
10
Very
satisfied
423
8) The way my job provides for steady employment.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
9) The chance to do things for other people.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
10) The chance to tell people what to do.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
11) The chance to do something that makes use of my abilities.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
12) The way company politics are put into practice.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
13) My pay and the amount of work I do.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
14) The chances for advancement on my job.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
15) The freedom to use my own judgement.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
16) The chance to try my own methods of doing the job.
0
Very
dissatisfied
1
2
3
4
5
6
7
8
9
10
Very
satisfied
424
17) The working conditions.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
18) The way my co-workers get along with each other.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
19) The praise I get for doing a good job.
0
1
2
3
4
5
6
7
8
9
Very
dissatisfied
10
Very
satisfied
20) The feeling of accomplishment I get from the job.
0
Very
dissatisfied
1
2
3
4
5
6
7
8
9
10
Very
satisfied
425
Appendix M- Social Support Scale for Study Two (Revision of Karasek &
Theorell, 1990)
The following 8 questions ask about your supervisor and your co-workers. Please
circle a number 1= Not true at all to 10= Could not be more true.
1) My supervisor shows concern for me.
0
1
2
3
4
5
6
7
8
9
Not true at
all
10
Could not
be more true
2) My supervisor pays attention to me.
0
1
2
3
4
5
6
7
8
9
Not true at
all
10
Could not
be more true
3) My supervisor is helpful getting work done.
0
1
2
3
4
5
6
7
8
9
Not true at
all
10
Could not
be more true
4) My supervisor creates a good teamwork environment for me.
0
1
2
3
4
5
6
7
8
9
Not true at
all
10
Could not
be more true
5) My co-workers are friendly to me.
0
1
2
3
4
5
6
7
8
9
Not true at
all
10
Could not
be more true
6) My co-workers are helpful to me.
0
1
2
3
4
5
6
7
8
9
Not true at
all
10
Could not
be more true
7) My co-workers are personally interested in me.
0
1
2
3
4
5
6
7
8
9
Not true at
all
10
Could not
be more true
8) My co-workers are competent.
0
Not true at
all
1
2
3
4
5
6
7
8
9
10
Could not
be more true
426
Appendix N-Plain Language Statement used in Study Three
Dear Sir/Madam,
My name is Elise Maher, and I am completing my Ph.D. in Psychology at
Deakin University. As part of my studies, I am undertaking a research project under
the supervision of Professor Robert Cummins, a researcher in the School of
Psychology. This study is investigating job satisfaction and coping. The study aims
to provide useful information about the best type of coping strategies that workers
should use. The results will provide information that will enhance programs that
increase job satisfaction.
You are invited to participate in this research. If you agree, you will be asked
to complete the enclosed questionnaire. Any information you provide will be
anonymous and confidential. Only group results will be reported and no individuals
will be identified. Upon completion of the study, data will be secured in a locked
cabinet in the School of Psychology, Deakin University, for a minimum period of six
years from the date of publication.
The questionnaire should take around 20 minutes to complete and your
participation would be greatly appreciated. Examples of questions are: "I can decide
on my own about how to go about doing my work", "Your co-workers really care
about you", "What type of difficulties do you face at work?" and "How satisfied are
you with your close relationships with family or friends." You are free to withdraw
up until you have returned the survey, in which event your participation in the
research study will immediately cease and any information obtained will not be used.
You are free to refuse to answer any questions.
Following the completion of the study, I am happy to provide you a summary
of the best coping strategies. If you would like a copy of the summary or if you have
any further questions regarding the study, please contact:
Elise Maher on (03) 9251 7153 or Email: ecmaher@deakin.edu.au, or
Professor Robert Cummins on (03) 9244-6845 or Email: cummins@deakin.edu.au.
If you are happy to be involved in this study, please complete the enclosed
questionnaire and return it in the reply-paid envelope supplied (i.e., NO STAMP
NEEDED).
Thank you very much for your time.
__________________________________________________________________
Should you have any concerns about the conduct of this research project, please contact the Secretary,
Deakin University Ethics Committee, Research Services, Deakin University, 221 Burwood Highway,
BURWOOD, VIC, 3125, Tel (03) 9251 7123
427
Appendix O- Primary and Secondary Control Scale for Study Three (Maher &
Cummins, 2002)
People may experience several kinds of difficulties in their work. They can control
some of them, but not others.
For example, Worker ‘A’, a teacher, can control difficulties involving students,
parents and time management. They cannot however control difficulties involving
school policies and work times.
Another example, Worker ‘B’ a supermarket operator, can control difficulties
involving customers and co-workers. They cannot control difficulties involving pay,
promotion and holiday leave.
1) Tick the difficulties you experience at work that you CAN CONTROL.











Difficulties with Supervisor(s)
Difficulties with Promotion
Difficulties with Co-worker(s)
Difficulties with Time Management
Difficulties with Kind of work you do
Difficulties with Motivation
Difficulties with Pay
Difficulties with Work Times
Difficulties with Work-place rules
Difficulties with Amount of Work
Other………………………………………………………………………
2) Consider the difficulty that you experience MOST OFTEN, and which you
CAN CONTROL.
How often do you experience this difficulty?
1
2
3
Rarely
Sometimes
Often
4
Always
428
3) When you face this difficulty that you CAN CONTROL, how often do you
do the following?
a) Discuss solutions with the people involved
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
b) Think that the difficulty doesn’t matter
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
c) Think that this difficulty will work out okay in the end
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
d) Choose a solution and act on it
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
e) Think that I knew this difficulty would happen
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
f) Think that I can’t always get what I want
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
g) Work harder
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
h) Think that I am better off than many other people
0
Never
i)
1
2
3
Rarely
Sometimes
Often
4
Always
Think that this difficulty is not my fault
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
j) Keep trying
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
429
k) Tell someone about this difficulty to make me feel better
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
l) Think of the success of my family/friends
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
m) Think about my success in other areas
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
n) Do something different, like going for a walk
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
o) Ignore this difficulty
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
p) Look for something else that is positive in the situation
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
q) Other …………………………..(please specify)
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
430
The following questions examine difficulties that you CANNOT CONTROL
4) Tick the difficulties you experience at work that you CANNOT CONTROL.











Difficulties with Supervisor(s)
Difficulties with Promotion
Difficulties with Co-worker(s)
Difficulties with Time Management
Difficulties with Kind of work you do
Difficulties with Motivation
Difficulties with Pay
Difficulties with Work Times
Difficulties with Work-place rules
Difficulties with Amount of Work
Other………………………………………………………………………
5) Consider the difficulty that you experience MOST OFTEN, and which you
CANNOT CONTROL. How often do you experience this difficulty?
1
Rarely
2
3
Sometimes
Often
4
Always
6) When you face this difficulty that you CANNOT CONTROL, how often do
you do the following?
a) Discuss solutions with the people involved
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
b) Think that the difficulty doesn’t matter
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
c) Think that this difficulty will work out okay in the end
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
d) Choose a solution and act on it
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
431
e) Think that I knew this difficulty would happen
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
f) Think that I can’t always get what I want
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
g) Work harder
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
h) Think that I am better off than many other people
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
Think that this difficulty is not my fault
i)
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
j) Keep trying
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
k) Tell someone about this difficulty to make me feel better
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
Think of the success of my family/friends
l)
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
m) Think about my success in other areas
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
n) Do something different, like going for a walk
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
432
o) Ignore this difficulty
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
p) Look for something else that is positive in the situation
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
q) Other ………………(please specify)
0
Never
1
2
3
Rarely
Sometimes
Often
4
Always
433
Appendix P- Life Satisfaction Scale for Study 3 (Cummins et al., 2001)
1) How satisfied are you with your standard of living?
0
1
2
3
4
5
6
7
8
9
Completely
dissatisfied
10
Completely
satisfied
2) How satisfied are you with your health?
0
1
2
3
4
5
6
7
8
9
Completely
dissatisfied
10
Completely
satisfied
3) How satisfied are you with what you achieve in life?
0
1
2
3
4
5
6
7
8
9
Completely
dissatisfied
10
Completely
satisfied
4) How satisfied are you with your personal relationships?
0
1
2
3
4
5
6
7
8
9
Completely
dissatisfied
10
Completely
satisfied
5) How satisfied are you with how safe you feel?
0
1
2
3
4
5
6
7
8
9
Completely
dissatisfied
10
Completely
satisfied
6) How satisfied are you with feeling part of your community?
0
1
2
3
4
5
6
7
8
9
Completely
dissatisfied
10
Completely
satisfied
7) How satisfied are you with your future security?
0
Completely
dissatisfied
1
2
3
4
5
6
7
8
9
10
Completely
satisfied
434
Appendix Q- Social Support Scale for Study 3 (Ducharme & Martin, 2000)
The following 6 questions ask about your co-workers. Please circle a number
ranging from 0 to 10, where 0= Do not agree at all and 10= Agree completely.
1) My co-workers really care about me.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
completely
2) I feel close to my co-workers.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
completely
3) My co-workers take a personal interest in me.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
completely
4) My co-workers assist with unusual work problems.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
completely
5) My co-workers are helpful in getting the job done.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
completely
6) My co-workers give useful advice on job problems.
0
Do not
agree at all
1
2
3
4
5
6
7
8
9
10
Agree
completely
435
The following 6 questions ask about your supervisor(s). For each item, please circle
a number ranging from 0 to 10, where 0= Do not agree at all, and 10= Agree
completely
1) My supervisor really cares about me.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
completely
2) I feel close to my supervisor.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
completely
3) My supervisor takes a personal interest in me.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
completely
4) My supervisor assists with unusual work problems.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
completely
5) My supervisor is helpful in getting the job done.
0
1
2
3
4
5
6
7
8
9
Do not
agree at all
10
Agree
completely
6) My supervisor gives useful advice on job problems.
0
Do not
agree at all
1
2
3
4
5
6
7
8
9
10
Agree
completely