Why would workers prefer collective forms of representation? Evidence from the 2011
Workplace Employment Relations Study
Danat Valizade, Leeds University Business School (Work and Employment Relations
Division), e-mail: bndv@leeds.ac.uk
Abstract
Drawing on data from the 2011 Workplace Employment Relations Study the present article captures crucial factors that determine employee desire for collective representation mechanisms in the United Kingdom. I utilized the expectancy-value theoretical perspective to estimate the effects of employee-level outcomes such as job satisfaction, employment relations climate and perceived union instrumentality alongside the impact of the latest recession on workers’ propensity for collective voice. Uniquely, the current paper verifies the moderation effect of perceived union instrumentality in the relationship between job satisfaction and employee desire for collective voice where a detrimental influence of job satisfaction exponentially increases with corresponding levels of perceived union instrumentality. The findings of this study indicate a positive association between the certain implications of the economic crisis (notably restricted access to paid overtime work) and employee propensity for collective representation; this effect was shown to be moderated by perceived union instrumentality and membership status of employees.
Key words : employee representation, collective voice, trade unions, recession, employeelevel outcomes.
Introduction
It is been a long time since substantial union decline in the United Kingdom in conjunction with the dominance of neoliberal approach in economics caused the crisis of
collective employee representation (Heery, 1998, Heery et al., 2003, Visser, 2002). Even
nowadays, where the perception of trade unions as a major threat of parliamentary democracy and economic performance is not prevalent in the political landscape, the future of employee
representation is still blurred and rather pessimistic (Hyman, 1997). Given an increasing
uncertainty with regards to the prospects of collective voice this article is aimed at exploring the determinants of workers’ desire for collective representation alongside the changes in employee attitudes towards collective voice in the light of the latest recession.
The present article attempts to penetrate deeper into the complexity of factors that constitute employee propensity for collective representation at workplace level by deciphering the effects of employee-level outcomes on workers’ desire for collective voice.
Based upon the rationale for the expectancy-value theory where employee attitudes are
deemed to be a function of costs and payoffs (Klandermans, 1986, Klandermans, 1984,
Monnot et al., 2011) I followed prior research and investigated the extent to what job
satisfaction alongside employment relations climate and perceived union instrumentality
shape employee desire for collective voice (Bryson and Freeman, 2013, Charlwood, 2002,
Fiorito, 1987, Fiorito, 2001, Robinson, 1988). In addition, I analysed the ability of qualitative
variables such as employee membership status and occupational category of employees to
predict workers’ propensity for collective voice mechanisms (Campolieti et al., 2011,
Chacko, 1985, Chacko and Greer, 1982, Kochan, 1979, Prowse and Prowse, 2006). Given the
complexity of employee-level outcomes as predictors of workers propensity for collective voice (Kochan, 1979, Charlwood, 2002) I tested the interaction effect between perceived union instrumentality and job satisfaction as well as the proposition of a three way interaction
effect between the implications of the latest recession, perceived union instrumentality, and employee membership status on employee desire for collective voice.
I utilised data obtained from the 2011 Workplace Employment Relations Study (the
2011 WERS) to provide an insight into workers’ desire for collective voice through the following areas of employee representation: negotiations about pay increase, reduced hours or pay, training, dealing with employee complaints, and employers’ attempts to discipline employees. A distinctive feature of the 2011 WERS is the reflection of various outcomes of the latest recession which opens up a room for the speculations about its influence on employee desire for collective voice. Given the composition of variables in the 2011 WERS, particularly the fact that union representation was merged with other collective forms of employee representation, I focused on collective voice in general including trade unions and staff associations. This is not deemed to be an important issue since union representation is
acknowledge as a dominant form of collective voice (Givan et al., 2012, Heery, 2009); hence
prior research on employee attitudes towards trade unions is applicable for the purposes of the current article.
A number of factors underpin the relevance of the present study for industrial relations research. Firstly, there is lack of comprehensive research to reflect employee perspectives towards fundamental aspects of workplace representation; therefore, empirical studies are required to unravel the constituencies of employee desire for collective voice and
shed the light on its prospects in contemporary employment relations (Blanchflower et al.,
2007, Charlwood and Terry, 2007). Secondly, industrial relations were traditionally skewed
towards an institutional understanding of the processes that occur within the system of employee representation (Hyman, 1975); thereby, little attention has been paid to an individual experience of employees with collective forms of workplace representation
(Bryson and Freeman, 2013). The importance of the present study also increases in light of
the latest trends in the union movement towards focusing on employee level outcomes as means to provide active campaigning aimed at overcoming the consequences of membership
The remainder of this article is organised as follows. Firstly, I review prior research on employee perceptions of collective voice mechanisms. In this regard close attention has been paid to a theoretical explanation of employee propensity for union representation.
Secondly, the data, quantitative methods and hypotheses formulated on the basis of prior research are introduced. Thirdly, the outputs of the regression analysis are presented alongside the discussion upon the main findings. The paper concludes with nominating potential implications of the findings for industrial relations research.
Employee level outcomes and collective voice: theoretical background
The current state and prospects of employee collective voice have recently occupied
one of the central roles in employment relations literature (Bryson et al., 2004, Bryson et al.,
2010, Bryson and Freeman, 2013, Bryson et al., 2013b, Cotti et al., 2013, Hammer and
Avgar, 2005, Kochan, 2004, Simms and Holgate, 2010). The question as to why workers join
trade unions and their subsequent perceptions of the labour movement have become increasingly relevant due to dramatic rates of union decline in the Western countries followed
by the emergence of union revitalization agenda (Bryson and Freeman, 2013, Heery, 2009,
MacKenzie, 2010, Troy, 2001). The expectancy-value perspective has featured prominently
among psychological and employment relations studies as a preponderant approach to
explain employee desire for collective representation (Klandermans, 1986, Klandermans,
1984). The foregoing framework is rested on the assumption of balance between individual
costs and benefits of participation adjusted by employee expectations from collective representation; it has been extensively used to explain the dramatic figures of union decline
as well as the prospects of union revitalization agenda (Klandermans, 1984). Furthermore, the
expectancy-value theory asserts the relevance of one’s beliefs in a particular object or system
for predicting human decisions (Klandermans, 1984). Hence, as the theory of attitude, where
employees make decisions on the premise of expectations of their actions, potential benefits and downsides of these actions alongside personal preferences and beliefs, the expectancyvalue approach encompasses existing perspectives towards employee desire for collective voice. These perspectives were summarized by Charlwood (2002) as dissonance theories, where employee disappointment with working conditions encourages them to participate in the union movement, utility theories, where the combination of costs and effects of collective representation presupposes employee decision to join a trade union, and political/ideological beliefs theory, where employee political views and ideological tenets are most likely to shape their decisions in favour of collective voice mechanisms. Further down I review the most eminent determinants of employee desire for collective representation in light of the expectancy-value viewpoint.
Without any doubts, one of the most frequently studied determinants of employee
propensity for collective voice is job satisfaction (Bryson and Freeman, 2013, Charlwood,
2002, Robinson, 1988). Extant literature explicitly underlined its detrimental effect on
employee desire for collective representation demonstrating that higher levels of job satisfaction undermine the value of trade unions (or staff associations) for workers and thereby exert undesirable (for trade unions) employee perceptions of collective voice mechanisms (Charlwood, 2002, Bryson & Freeman, 2013). Job satisfaction as a predictor of employee desire for collective voice is rested on the variety of factors that designate
employee-level attitudes (Bryson and Freeman, 2013, Charlwood, 2002). Prior empirical
research pointed to the working conditions of employees and HRM practices employed in the organisation as crucial determinants of job satisfaction through which employees elicit their
attitudes towards collective representation mechanisms (Bryson & Freeman, 2013, Robinson,
1988). In this vein, Robinson’s (1988) study verified a positive association between hazardous conditions at workplace level and employee willingness to be unionised. The research conducted by Bryson and Freeman (2013) explored determinants of poor working conditions and showed their positive association with employee desire for union representation. Using representative national datasets in the UK and in the US they demonstrated similar effects of poor working conditions in both countries (Bryson &
Freeman, 2013). Furthermore, innovative human resource management (HRM) such as participative decision making strategies and information sharing activities were also shown to produce unfavourable for trade unions results through higher levels of job satisfaction and
improved relationship between employees and managers (Brown et al., 2008, Truss et al.,
2013, Yalabik et al., 2013). It is along these lines that HRM practices associated with greater
organizational commitment and higher levels of employee trust in managers exert a ‘union substitution effect’ and, thereby, deteriorate workers’ propensity for union representation
The expectancy-value approach provides an opportunity to conceptualise the effects of job satisfaction and positive HRM practices on employee desire for collective voice. From this viewpoint employee reliance on collective voice is rested on the expectancy of improved working conditions due to the higher effectiveness of collective representation in comparison with the situation where workers have to deal with employers on vis-à-vis
al., 2013, Fagerlind et al., 2013). Similarly to this logic, poor working conditions, work
intensification and higher levels of job strain cause workers’ frustration with their job and force them to seek support from employee representatives (Klandermans, 1986). On the contrary, wherever the levels of job satisfaction and employment relations climate indicate desirable outcomes for employees, their perceptions of costs and benefits of collective
representation stir into a negative zone which makes the idea of collective voice less
favourable for workers (Pyman et al., 2010, Truss et al., 2013). It should be noted though that
the isolation of job satisfaction and HRM practices from broader workplace characteristics and from the results of union activities makes these predictors incomprehensible and ambiguous. For these reasons prior research attempted to extend the catalogue of determinants of employee desire for collective voice in the lines with Kochan’s (1979) suggestion of the necessity to investigate the underlying factors that might presuppose the effects of job satisfaction and HRM practices on employee attitudes towards trade unions.
Extant literature has focused on a number of workplace issues in order to embrace the variety of motives behind employee propensity for collective representation. One of the most prominent predictors in this regard is perceived union instrumentality which reflects different
forms of union support at the workplace (Fullagar et al., 2004). From the angle of the
expectancy-value theory higher levels of union support at the workplace in conjunction with positive outcomes of trade unions (staff association) activities increase the value of collective representation in the eyes of workers and, therefore, foster employee desire for collective voice. Indeed, prior research associated various facets of union instrumentality with employee propensity for collective representation by demonstrating that union involvement in the workplace negotiations has a positive influence on workers’ desire for union membership
(Langford, 1994). A wider concept of union instrumentality that encompasses the political
agenda of trade unions has been shown to foster pro-union voting intents (Fiorito, 1987),
whereas the research in Australian context has revealed a negative association between employee dissatisfaction with union support and their attitudes towards collective voice
(Gahan, 2012). In the UK Charlwood (2002) discovered that apart from job dissatisfaction
perceived union instrumentality is positively associated with workers propensity for unionisation; in fact, he nominated this variable as a most powerful predictor of non-union
members’ desire for union representation (Charlwood, 2002). In light of the negative role of job satisfaction, further questions need to be asked as to whether the effect of union instrumentality overpowers the detrimental impact of job satisfaction and how these opposite predictors constitute employee propensity for collective representation when considered jointly. Prior research touched upon these issues only in a superficial manner; hence, the present paper, among other things, focuses on unravelling the foregoing conundrum.
Apart from workplace determinants, social characteristics of workers, such as their union membership status and occupational category to which they belong are deemed to be of high importance with regards to employee desire for collective voice. A significant amount of studies showed that union members are more positive towards collective voice than
employees who have not had prior union experience (Campolieti et al., 2011, Chacko, 1985,
Chacko and Greer, 1982, Kochan, 1979). Furthermore, trade unions have traditionally had a
strong support among low-skilled, most vulnerable groups of workers likewise among highly
educated professionals (Campolieti et al., 2011, Kelloway and Barling, 1993, Kochan, 1979,
McLlroy, 2012). In these lines, Kochan’s (1979) classic study reflected that union members
on average are more positive towards union instrumentality as well as highly-educated people and white collar workers. Chacko and Greer (1982) showed greater levels of employee perception of union instrumentality alongside stronger confidence in union leaders among union members in comparison with non-members; however, the latter demonstrated stronger beliefs in union power than union members. Prowse and Prowse (2012) studied workers’ perceptions of union effectiveness and demonstrated only narrow differences between current union members, former members and non-members. More importantly, they showed that former union members expressed a lower interest in renewing their membership than non-
members to join a trade union (Prowse and Prowse, 2006). The most recent empirical studies
verified the importance of the aforementioned issues in the light of employee propensity for representation mechanisms (Campoliety et al, 2013).
Further challenges to collective employee voice have been posed by the latest
recession (Széll, 2010, Bryson et al., 2013a, Bryson et al., 2011). The austerity measures
based upon neoclassical understanding of economic growth through higher profits of organizations rather than higher wages of employees have created a hostile atmosphere for
the union movement (Széll, 2010). However, it is evident that apart from threats the recession
provides opportunities for trade unions to elicit favourable employee attitudes and thereby to
increase union power at the workplace (Széll, 2010). From the expectancy-value perspective
poorer working conditions under the recession open up a room for higher levels of employee desire for collective voice and, hence, in broader terms, provide an opportunity for employee
representatives to reinforce their positions in employment relations (Kelly et al., 1987,
Waddington and Whitston, 1997). To what extent these opportunities have been delivered by
the latest recession is yet to be explored in industrial relations literature.
Hypotheses
Following prior research where job satisfaction is considered a main predictor of employee desire for collective voice I suggest that higher levels of job satisfaction are negatively associated with employee propensity for collective voice. It is highly anticipated to observe the aforementioned tendency across all major areas of employee representation. In addition, I utilise the proposition of a significant influence of positive HRM practices on workers’ desire for collective representation based upon the premise that these practices produce better employment relations climate which in turn deteriorates employee desire for
collective voice (Fiorito, 2001, Pyman et al., 2010). Hypotheses 1 and 2 are as follows.
Hypothesis 1 . Job satisfaction is negatively associated with employee desire for collective voice.
Hypothesis 2 . Employment relations climate at the workplace is negatively associated with employee desire for collective voice.
Following Charlwood’s (2002) conclusion about a direct relationship between perceived union instrumentality and non-union members’ propensity for union representation
I expect to observe a statistically significant positive effect of perceived union (staff association) instrumentality on employee desire for collective representation across all major areas of representation. Hypothesis 3 is as follows.
Hypothesis 3 . Perceived union (staff association) instrumentality is positively associated with employee desire for collective voice.
In spite of ongoing debates in the literature about the differences between union members, former union members and non-members in terms of their attitudes towards collective representation I support the idea of higher probability to prefer collective voice mechanisms for union members in comparison with former union members and nonmembers. I draw this proposition on a number of studies that showed higher levels of loyalty towards trade unions among current union members and former union members in
comparison with employees who have never had a union experience (Chacko, 1985, Chacko
and Greer, 1982, Kochan, 1979). Based upon prior research (Campolieti et al., 2011, Kochan,
1979, Martinez and Fiorito, 2009) I also expect to find stronger desire for collective voice
among technicians, low-skilled employees and clerical workers in comparison with professionals and administrative personnel likewise among professionals in comparison with managers. Hypotheses 4 and 5 are as follows.
Hypothesis 4 . Union members are more likely to prefer collective forms of employee representation than non-union members and former union members; former union (staff association) members are more likely to desire collective representation than non-members.
Hypothesis 5 . Technicians, clerical and low-skilled workers likewise professional are more likely to prefer collective voice mechanisms than managerial and administrative personnel.
As I mentioned in the literature review section it is unlikely that separate effects of job satisfaction and perceived union (staff association) instrumentality fully comprehend employee desire for collective voice. In the same vein as Glick et al (1977) and Charlwood’s
(2002) studies I put forward an interaction effect between job satisfaction and union instrumentality on employee propensity for collective representation. Given the lack of empirical and theoretical arguments to support a particular direction of this effect I stated the following hypothesis in quite a broad way.
Hypothesis 6 . The effect of job satisfaction on employee desire for collective voice is moderated by perceived union (staff association) instrumentality.
Uniquely, the 2011 WERS allows tracing the effects of various aspects of the economic crisis on employee desire for collective voice. In light of the expectance-value approach I anticipate finding a positive association between the outcomes of the recession and employee propensity for collective representation. I also suppose that this effect is moderated by perceived union instrumentality adjusted by union membership status of employees. Hypotheses 7 and 8 are as follows.
Hypothesis 7 . The outcomes of the recession are associated with higher levels of employee desire for collective voice.
Hypothesis 8 . The effect of the recession on workers’ desire for collective representation is moderated by perceived union (staff association) instrumentality at the workplace alongside union membership status of employees.
Methodology
Data
The present article is based upon the employee survey of the 2011 Workplace
Employment Relations Study (2011 WERS). The survey provides comprehensive information with regards to various aspects of employment relations and employees’ worklife quality across the UK and is considered suitable for the current study. The nested nature of the 2011 WERS alongside the large number of sampled workplaces provides a sufficient statistical power to examine relevant employee-level components and to provide a high validity of research findings. The employee sample, which is the largest and most reliable employment relations data in the UK, was drawn on 21,981 responses from 1,923 workplaces. The dataset required minor adaptations; in particular it was necessary to account for missing and non-applicable items among the variables used in this study. In particular I concentrated on the workplaces with recognised trade unions and/or staff associations and excluded the workplaces with less than five employees to ensure the higher validity of the data. After all manipulations, the dataset was reduced to 5600 employee responses, which, however, has not affected the statistical power of the data.
Dependent, independent and control variables
The descriptive statistics for all independent and dependent variables included in the present study can be found in Appendix 1. Following Bryson and Freeman (2013) I use a set of following categorical dependent variables to reflect employee desire for collective voice.
1.
Getting increased in employees’ pay.
2.
Getting hours or payments reduced.
3.
Getting training.
4.
Making a complaint about working.
5.
If managers wanted to discipline workers.
Respondents had the following options to choose from: ‘myself’, ‘trade union’,
‘employee representative (non-union)’, ‘line manager’, and ‘another employee’. I transformed an initial coding into a set of five dichotomous variables by coding ‘1’ as any form of collective representation, namely the answers ‘trade union’ and ‘employee representative (non-union)’, and ‘0’ as all other, non-collective forms of representation. Thus
I deliberately modified the approach employed by Bryson and Freeman (2013) in order to estimate employee propensity for collective voice in the main areas of representation separately.
Six independent variables that represent a mixture of categorical and continuous variables have been selected for the purposes of this study. These variables are as follows.
Firstly, job satisfaction . The following observed items were used to compute a composite variable for job satisfaction (respondents were asked to rate their satisfaction on five points Likert type scale from ‘very bad’ to ‘very good’; Cronbach’s Alpha (α) for this scale is satisfactory at the level equal to 0.861 and a confirmatory factor analysis verified the presence of one principle component):
1.
The senses of achievement employees get from their work.
2.
The scope of using employees own initiative.
3.
The amount of influence employees have over their job.
4.
The training workers receive.
5.
The opportunity to develop employees’ skills in their job.
6.
The amount of pay.
7.
The job security of employees.
8.
The work itself.
Secondly, ‘ perceived union (staff association) instrumentality
’. Employees were asked if they agree or disagree with the following statements based upon five points Likert type scale (α=0.776).
1.
Union/staff association here take notice of members’ problems.
2.
Union/staff association here are taken seriously by management.
3.
Union/staff association here make a difference to what it is like to work here.
Thirdly, employment relations climate . Based upon Pyman et al (2010) I selected one variable to reflect industrial relations climate at workplace. Respondents were asked how they would in general describe relations between managers and employees. The answers were recorded on five points Likert type scale from ‘very poor’ to ‘very good’.
Fourthly, occupational group and employees’ membership status
. These are categorical nominal variables with three categories each. Occupational group, initially divided into nine categories, was merged into three groups: ‘managerial and administrative personnel’, ‘professionals’ (including state workers), and ‘clerical/manual/unskilled workers’.
Membership status variable in turn divided the dataset into the following groups: union or staff association members, former union or staff association members and non-members.
Fifthly, the recession . The implications of the recession at workplace level were measured by eight categorical dichotomous variables. Respondents were asked whether the following aspects of the recession occurred at their workplace (answers ‘no’ were coded as
‘0’, answers ‘yes’ were coded as ‘1’).
1.
Workload increased.
2.
Workers were moved to another position.
3.
Wages were frozen.
4.
Non-wages benefits were reduced.
5.
Working hours were reduced.
6.
An access to overtime paid activities was restricted.
7.
Workers had to take an unpaid leave.
8.
Training was reduced.
The whole range of recession variables represent two broader groups. The first group reflects an intensification of labour process and includes such cases as increased workload or necessity to change the position in the organisation. The second group is characterised by direct impact on workers’ welfare and comprehends such factors as frozen wages, access to overtime paid arrangements and et cetera .
Following prior research (Bryson and Freeman, 2013, Charlwood, 2002, Fiorito,
2001) I controlled for a wide range of demographic and workplace characteristics, namely gender, age, ethnicity, income, educational level, basic contractual hours, tenure in the organisation, employers’ attitude towards trade unions (staff associations), and contract type.
Method
The pertinent approach used to test the hypotheses of the present study is a binary logistic regression which measures the relationship between a dichotomous dependent variable and independent variables (both, continuous and categorical) by using probability scores as predicted values of a dependent variable. Given this I used a set of binary logistic regressions to independently examine the relationships between dependent and independent variables in five main areas of employee representation.
With regards to the data employed in the present study common method variance (a systematic error variance based on biased responses that come from one group of respondents) is a point of possible concern (Podsakoff, MacKenzie, Lee and Podsakoff,
2003). I performed the Harman’s single-factor test (Podsakoff et al., 2003) to ensure that common method variance is not present in this study. I loaded all observed items into an exploratory factor analysis, fixed the number of factors to extract at one, and then examined
the factor solution. An emerging single factor from the factor analysis explained only 19% of the covariance among observed items. It points to the fact that common method variance is not likely to contaminate the outcomes of the regression models.
Results
The regression analysis was separately conducted for five models to estimate the association between independent variables and employee desire for collective voice with regards to the following dependent variables: pay increase (model 1), reduced working hours or pay (model 2), training at the workplace (model 3), employee complaints at the workplace
(model 4), and managers’ attempts to discipline workers (model 5). Tables 1 and 2 reflect the outcomes of the regression models (due to the large amount of independent variables and the number models which have been tested in the current paper I present the findings in two separate tables).
Table 1
Predictor Β
Model 1
S.E. Exp(
B)
β
Model 2
S.E. Exp(B)
Model 3
β S.E. Exp(B)
Occupational category (low skilled as a reference point)
Occupational category
(managers)
Occupational category
(professionals)
-.654***
-.100
.138
.083
.520
.905
-.636*** .139
-.129 .087
.529 -1.228*** .349
.879 -.616** .135
.293
.540
Occupational category (mangers as a reference point)
Occupational category
(professionals)
Occupational category (low skilled)
.555***
.653***
.126 1.741
.138 1.921
.504*** .125
.632*** .139
1.655
Recession workload increased
Recession (the absence of recession is a reference point)
.120 .082 1.128 .078 .086 1.081
.607* .345
1.882 1.222*** .349
Membership status (members as a reference point)
Membership status (former members)
Membership status (nonmembers)
Industrial relations climate
Union instrumentality
Job satisfaction
-1.201***
-1.638***
.090
.087
.301 -1.425*** .089
.194 -1.892*** .087
.240
.151
Membership status (members)
Membership status (former members)
Membership status (non-members as a reference point)
1.639*** .087 5.151 1.892*** .087 6.631
.437*** .108 1.548 .464*** .104 1.590
-.276 .171
-.607* .193
.607* .192
.325 .238
.759
.545
1.834
1.384
Continuous independent variables
-.325*** .044 .722 -.305*** .047 .737 -.322*** .066 .724
.501*** .059 1.650 .486*** .064 1.626 .378*** .093 1.460
-.543*** .058 .581 -.449*** .060 .638 -.622*** .089 .537
-.009 .135
1.836
3.393
.991
Recession workplace reorganised -.160 .089 .852 -.082 .093 .921 -.107 .141 .898
Recession moved to another position
Recession wages frozen
Recession non-wages benefits cut
.145
-.314***
-.054
.132 1.156
.081
.134
.731
.948
.017 .138
-.045 .085
.025 .140
1.017
.956
1.025
.410 .220
.271* .132
-.257 .202
1.506
1.311
.773
Recession hours were reduced
Recession overtime paid restricted
Recession unpaid leave
Recession training reduced
.178
-.352***
.202 1.195
.094 .703
-.020 .211
-.336** .100
.980
.715
.215 .260 1.239
.025 .094 1.025
Interaction effects
.048 .273 1.049
-.130 .102 .878
-.263 .265
-.188 .134
.163 .353
-.094 .156
.768
.829
1.177
.910
Job satisfaction * Union instrumentality
-.183** .060 .833 -.068 .062 .934 .012 .086 1.012
Recession overtime paid restricted * Union involvement *Membership status (union members as a reference points)
Recession overtime paid restricted * Union
-.486*** .130 .615 -.180 .127 .836 -.351 .237 .704 instrumentality*Membership status (former members)
Recession overtime paid restricted * Union instrumentality *Membership status (non-members)
-.013 .136 .987 -.179 .132 .836 -.171 .292 .843
Note : Nagelkerke R square for model 1 = 0.312; model 2 = 0.298; model 3= 0.178. Control variables in the equations: gender, age, occupational category, contractual hours, contract type, educational level, income, managers’ attitude towards unions, and ethnicity. *** p<0.001; ** p<0.01; * p<0.05. β – logistic regression coefficients; S.E. – standard error; Exp(B) – odds ratios.
Table 2
Predictor Β
Model 4
S.E. Exp(B) Β
Model 5
S.E. Exp(B)
Occupational category (low skilled as a reference point)
Occupational category (managers)
Occupational category (professionals)
-.595***
-.098
.154
.078
.552
.907
-.576***
-.237**
Occupational category (mangers as a reference point)
Occupational category (professionals)
.498** .145 1.645 .339**
Occupational category (low skilled)
.596*** .154 1.814 .573***
Membership status (members as a reference point)
Membership status (former members) -1.023*** .100 .360 -1.559***
Membership status (non-members)
Membership status (non-members as a reference point)
-1.382*** .105 .251 -2.009***
Membership status (members)
1.381*** .105 3.980 2.009***
.137
.083
.125
.137
.087
.088
.562
.789
1.403
1.774
.210
.134
.088 7.453
Membership status (former members)
Industrial relations climate
Union instrumentality
Job satisfaction
.359** .134 1.432
Continuous independent variables
-.313*** .040
.429*** .052
-.561*** .054
.731
1.536
.571
.448***
-.212***
.544***
-.474***
Recession (the presence of recession is a reference point)
Recession workload increased -.010 .077 .990 .042
Recession workplace reorganised
Recession moved to another position
Recession wages frozen
-.058
.077
.083
.083
.120
.076
.943
1.080
1.086
-.010
.064
.034
Recession non-wages benefits cut
Recession hours were reduced
Recession overtime paid restricted
Recession unpaid leave
Recession training reduced
-.054
-.032
-.127
.121
.191
.083
-.201 .226
-.129 .087
Interaction effects
.947
.969
.881
.818
.879
.076
.206
-.326**
.266
.002
.106
.043
.058
.057
.081
.087
.129
.080
.129
.199
.092
.245
.094
1.565
.809
1.723
.622
Job satisfaction * Union instrumentality
-.014 .055 .986 -.140* .058 .869
Recession overtime paid restricted * Union instrumentality *Membership status (union members as a reference points)
Recession overtime paid restricted *
Union instrumentality *Membership status (former members)
Recession overtime paid restricted *
-.027 .149 .974 -.139 .127 .870
Union instrumentality *Membership status (non-members)
.045 .168 1.046 -.047 .138 .955
Note : Nagelkerke R square for model 4 = 0.213; model 5 = 0.303. Control variables in the equations: gender, age, occupational category, contractual hours, contract type, educational level, income, managers’ attitude towards unions, and ethnicity. *** p<0.001; ** p<0.01; * p<0.05. β – logistic regression coefficients; S.E. – standard error;
Exp(B) – odds ratios.
1.042
.990
1.066
1.034
1.079
1.229
.722
1.304
1.002
The analysis revealed that models with independent variables are significantly better than block-0 models and models with control variables only. The predictive capacity of the full models is in the range of 75-78% whereas block-0 models account only for 53-55% of
the variance; Nagelkerke R-square also increases significantly from the models with control variables only to the full models likewise to the models with interaction effects; Hosmer-
Lemeshow test for goodness of fit is statistically insignificant across all considered models which demonstrated their statistical accuracy.
Regression analysis verified hypotheses 1 and 2 by demonstrating a negative association between job satisfaction, industrial relations climate and employee desire for collective voice across all the models (β-values for these predictors are significantly negative at ρ<0.001). On the contrary, in the lines with hypothesis 3, perceived union instrumentality produced a consistent positive effect on employee propensity for collective representation (βvalues are significantly positive across all models at ρ<0.001). A strong support was found for hypotheses 4 and 5. As it was suggested, the probability for union (staff association) members to prefer collective representation mechanisms is significantly higher in comparison with former union members and non-members (additional analysis revealed that former members are more likely to desire collective representation than non-members). The effect of membership status is persistent and equally significant across all considered areas of employee representation with the exception of model 3 (training) where statistically significant difference was observed only between union (staff association) members and employees who have never been union members at ρ<0.05. The impact of occupational group was shown to have a similar level of consistency. The regression analysis demonstrated higher levels of employee desire for collective voice across clerical, low-skilled workers, technicians and professionals in comparison with managerial staff. In addition, the results of the present article demonstrated higher levels of probability to prefer collective forms of representation among low-skilled employees in comparison with professionals in the models
3 and 5 (training and managers’ attempts to discipline employees) at ρ<0.001 and ρ<0.05 respectively.
The present study only partly verified hypothesis 6 and demonstrated statistical significance of the interaction effect between job satisfaction and perceived union (staff association) instrumentality in two models
1
out of five models involved in the analysis. The effect is negative and statistically significant at ρ<0.05 across models 1 (pay increase) and 5
(employers’ attempts to discipline workers). Arguably, the significance of the moderation effect across only two models in the present paper does not undermine its relevance since increase in pay and employers’ attempts to discipline employee have been frequently associated with traditionally union areas of expertise. Therefore I conclude that in certain cases the effect of job satisfaction on employee desire for collective voice is moderated by perceived union instrumentality; the direction of the moderation is negative which implies an exponential increase of detrimental impact of job satisfaction on employee desire for collective voice with corresponding increases in perceived union instrumentality. In order to probe and depict the interaction effect I used two independent techniques, particularly pick-apoint approach and Johnson-Neyman (J-N) technique. The latter estimates critical values of moderator to identify the borderline of a significant impact of focal predictor (job satisfaction) on a dependent variable whereas the former suggests to estimate the relationship between the dependent and independent variables on low, moderate, and high levels of a moderator (Hayes and Matthes, 2009). Johnson-Neyman (J-N) analysis showed that effect of focal predictor (job satisfaction) is significant across the entire range of moderator variable
(perceived union instrumentality). Pick-a-point approach revealed the significant effect of job satisfaction across all three categories of perceived union instrumentality (low, moderated,
1 In case of an interaction effect between two continuous variables I used mean centering approach for both focal predictor and moderator as it is suggested by Cohen et al (2003) and Hayes and Matthes (2009)
and high). Detailed statistical outcomes for these procedures are presented in Appendix 2
2
.
The graphical interpretation of the interaction effect is presented in Pictures 1 and 2.
Picture 1. Interaction effect between job satisfaction and perceived union instrumentality (dependent variable ‘pay increase’)
2 I used a template for probing an interaction effect designed by Hayes and Matthes (2009)
Picture 2. Interaction effect between job satisfaction and perceived union instrumentality (dependent variable ‘employers’ attempts to discipline employees’).
These graphs emphasise an increasingly negative effect of job satisfaction on the probability for workers to prefer collective forms of representation with higher levels of perceived union instrumentality. The effect is significant across all membership and occupational categories. However, the absence of an intersection point between the lines on the foregoing graphs points to the fact that higher level of perceived union instrumentality are still associated with greater probability for workers to be in favour of collective voice mechanisms.
Further analysis partly supported hypothesis 7 and verified a positive association between the peculiar outcomes of the latest recession and employee propensity for collective representation at the workplace. Regression analysis demonstrated that workers who have not experienced the influence of the recession are less likely to be inclined towards collective forms of representation; however, amongst all implications of the recession, only two
variables resulted in statistically significant influence on employee desire for collective voice, namely frozen wages, and restricted access to paid overtime work. The latter was proven the most persistent predictor by appearing in three models out of five (at ρ<0.05) with the exception of models 3 (dependent variable – getting training) and 4 (dependent variable – dealing with complaints).
Lastly, I estimated the interaction effect between the outcomes of the recession, perceived union (staff association) instrumentality, and membership status of respondents.
Among all recession variables I selected ‘restricted access to paid overtime work’ due to the fact that it demonstrated the most persistent results across the models involved in the present study. The interaction effect was proven to be statistically significant only in model 1 (pay increase) at ρ<0.01 for former union members in comparison with union members and nonmembers. However, given the importance of this model for employee representation, I state a moderate support for hypothesis 8 where the direction of the interaction effect between perceived union instrumentality and restricted access to paid overtime depends on union membership status of employees. To be more precise, the impact of perceived union instrumentality in conjunction with the restricted access to paid overtime work on employee desire for collective voice was equally positive across union (staff association) members and non-members (as it is shown in Picture 2, a dashed line indicates the presence of the recession outcome) whereas in case of former union (staff association) members this effect was shown to be opposite. I used a split group analysis to plot an interaction effect and its specific application for former union members (Picture 4, a dashed line indicates the presence of the recession outcome).
Picture 3. Interaction effect between recession outcomes (access to paid overtime was restricted), perceived union (staff association) instrumentality, and membership status (effect for union members, dependent variable ‘pay increase’)
Picture 4. Interaction effect between recession outcomes (access to paid overtime was restricted), perceived union (staff association) instrumentality, and membership status (effect for former union members, dependent variable ‘pay increase’)
Picture 4 emphasises that where the restricted access to paid overtime has not been reported by employees the effect of perceived union instrumentality across former union
(staff association) members on employee desire for collective voice has attenuated, whereas in case of the presence of foregoing outcome of the recession the effect of perceive union instrumentality on employee desire for collective voice has increased exponentially.
Discussion
The present study sought to capture crucial factors that shape employee propensity for collective voice. In this effect, I employed the expectancy-value theoretical framework to examine the direct effects of such employee-level outcomes as job satisfaction, employment relations climate, perceived union instrumentality alongside such qualitative indicators as membership status of employees, occupational category, and the implications of the latest recession on workers’ desire for collective representation. In the lines with the bulk of extant literature the present study verified a negative association between job satisfaction and industrial relations climate with employee desire for collective voice whereas perceived union instrumentality was demonstrated to provide an opportunity for eliciting desirable for
collective voice outcomes (Bryson and Freeman, 2013, Campolieti et al., 2011, Chacko,
1985, Chacko and Greer, 1982, Charlwood, 2002, Fagerlind et al., 2013, Fiorito, 1987,
Fiorito, 2001, Fullagar et al., 2004, Fuller and Hester, 1998, Kochan, 1979, Martinez and
Fiorito, 2009). Further analysis revealed greater levels of employee desire for collective voice
among union members and former union members in comparison with employees who have never had a trade union experience; in a similar vein, clerical, low-skilled employees and technicians were shown to more likely support collective representation mechanisms than professionals and administrative personnel. In addition, the present paper confirmed a direct positive association between the outcomes of the latest recession, namely frozen wages and restricted access to paid overtime work, and employee propensity for collective voice. These
findings fit the expectancy value framework where the value of collective voice for workers is expected to increase significantly with poorer working conditions of employees whereas higher levels of job satisfaction and industrial relations climate skew the balance of gains and costs that employees can elicit from collective representation mechanisms towards the
negative zone and thereby deteriorate employee desire for collective voice (Bryson and
Freeman, 2013, Robinson, 1988). In the same vein an expectancy-value rate of collective
voice for low-skilled employees and professionals was demonstrated to be significantly higher rather than the corresponding rates for managers and administrative personnel
(Chacko, 1985, Chacko and Greer, 1982, Kochan, 1979, Martinez and Fiorito, 2009).
However, the present study showed only moderate difference between low-skilled employees and professionals in their propensity for collective representation.
The crux of the current paper was in capturing the complexity of the impact of employee-level outcomes on workers’ desire for collective voice. In this vein, I demonstrated a statistically significant moderation effect of perceived union instrumentality in the relationship between job satisfaction and employee propensity for collective voice where the detrimental effect of job satisfaction exponentially increases with higher levels of perceived union instrumentality. On the flipside, this effect was shown to be statistically significant exclusively in two areas of employee representation, namely pay increase and dealing with managers’ attempts to discipline employees, which, however, are deemed to be of high relevance for union representation. A particular direction of the moderation effect contradicts
Charlwood’s (2002) suggestion that perceived union instrumentality will attenuate the negative impact of job satisfaction on employee propensity for collective voice; on the contrary, it performs in a similar vein with Chacko’s (1985) observation of pernicious impact of members’ satisfaction with a trade union on their intents to participate in union activities.
The rationale for this phenomenon is not straightforward; more focused studies are required
to unravel the reasons for perceived union instrumentality to strengthen the impact of job satisfaction on employee propensity for collective voice instead of producing desirable for collective representation outcomes. However, the expectancy-value approach opens up a room for general speculations on the aforementioned phenomenon. In particular, the principle of diminishing marginal returns (utility) might serve to shed the light on the moderation effect of perceived union instrumentality in the relationship between job satisfaction and employee propensity for collective voice. The rationale for marginal returns with regards to collective employee representation rests on the idea that each consecutive successful action of trade unions (or other employee representatives), adjusted by the growing job satisfaction,
becomes less valuable for workers (Brewer and Venaik, 2010); thereby, in spite of generally
positive perception of union instrumentality employees may have lower expectations from union activities and, hence, will less likely prefer collective forms of representation.
Surprisingly, but this approach has not been applied within the expectancy-value framework to explain employee behaviour with regards to their attitudes towards collective representation.
The current study also demonstrated a three way interaction effect between the most consistent outcome of the latest recession (restricted access to paid overtime work), perceived union instrumentality, and membership status of employees. The effect produced a significant result for former union (staff association) members and exclusively in such area of employee representation as pay increase. However, this seems natural due to the fact that reduces in overall earnings of employees will most likely switch their attention to the income at the main work; therefore, I believe there are convincing reasons to claim the importance of the aforementioned interaction effect for unravelling employee attitudes towards collective voice.
Regression analysis demonstrated that where employees experience a restricted access to overtime paid work, former union members’ desire for collective voice increases
exponentially in comparison with the workplaces untouched by the outcome of recession.
Inspired by the Rembrandt I call this effect ‘The Return of the Prodigal Son’ given the fact that former union members under the certain implications of the recession experience a significantly stronger desire for collective voice. Once again, the expectancy-value theory is deemed to be a pertinent approach to explain these relationships. In case of reductions in overtime paid work employees may rely on improvements with regards to their main work in order to partly reimburse the loses. In this light, prior positive union experience may foster higher expectations from union actions which in conjunction with potential gains and relatively low costs of representation might result in greater levels of former union member’ desire for collective voice.
Limitations and recommendations for future research
This paper attempted to estimate the effect of employee-level outcomes on workers’ propensity for collective voice by utilizing the survey of employees from the 2011 Workplace
Employment Relations Study. I was particularly interested in analysing a combined effect of such opposite employee attitudes as job satisfaction and perceived union instrumentality alongside the effect of the consequences of the latest recession on employee desire for collective representation. The findings of the present article might not revolutionise existing approaches to employee representation, but nevertheless they are deemed to throw light on future research on workers’ attitudes towards collective voice.
In spite of being based upon the most representative and reliable data in UK employment relations studies the current article has not escaped limitations. Firstly, this study is limited by its cross-sectional character which prevented more confident generalisations on the basis of the observed results. A longitudinal study may be considered a pertinent approach in future research to build on the findings of the present paper, especially in light of the impact of the recession on employee propensity for collective representation. Secondly, in
spite of all advantages of the 2011 WERS, it cannot fully comprehend a flamboyant picture of union activities and employees’ experience at the workplace. Thus, if the role of this study was to tell a ‘grand’ story about the association between employee-level outcomes and workers’ desire for collective voice, further studies may consider more specific measurements of employee attitudes in order to explore the effects shown in the present article at in-depth levels.
Perhaps the most promising findings of this research are two interaction effects: between job satisfaction and perceived union instrumentality alongside the interaction between the recession outcome (restricted access to paid overtime work), perceived union instrumentality, and union membership status of employees. The negative effect of the former on employee desire for collective voice is proposed to be related to the phenomenon on marginal returns where even positive union activities lose their value in time which leads to undesirable for collective voice outcomes at the workplaces where employees experience higher levels of job satisfaction. However, further empirical studies are required to verify this proposition and to unravel its application in contemporary industrial relations. The interaction between the recession outcome, perceived union instrumentality and membership status of employees, which I called ‘the return of the prodigal son’ effect, demonstrated an exponential increase in former union members’ desire for collective representation with the corresponding growth in perceived union instrumentation and under the restricted access to paid overtime work. This finding is deemed to stimulate the discussion on union organising strategies and the representation gap in the UK workplaces. A queue of former union members who desire to renew their membership, this is what trade unions most probably would like to see, but unfortunately for them do not observe in reality. Whether there is a potential for the union movement to foster the revitalisation agenda is yet to be studied.
Bibliography
BLANCHFLOWER, D. G., BRYSON, A. & FORTH, J. 2007. Workplace industrial relations in Britain, 1980–2004. Industrial Relations Journal, 38 , 285-302.
BRAUCHLI, R., SCHAUFELI, W. B., JENNY, G. J., FÜLLEMANN, D. & BAUER, G. F.
2013. Disentangling stability and change in job resources, job demands, and employee well-being — A three-wave study on the Job-Demands Resources model. Journal of
Vocational Behavior, 83 , 117-129.
BREWER, P. & VENAIK, S. 2010. GLOBE practices and values: A case of diminishing marginal utility? Journal of International Business Studies, 41 , 1316-1324.
BROWN, A., FORDE, C., SPENCER, D. & CHARLWOOD, A. 2008. Changes in HRM and job satisfaction, 1998–2004: evidence from the Workplace Employment Relations
Survey. Human Resource Management Journal, 18 , 237-256.
BRYSON, A., BARTH, E. & DALE-OLSEN, H. 2013a. THE EFFECTS OF
ORGANIZATIONAL CHANGE ON WORKER WELL-BEING AND THE
MODERATING ROLE OF TRADE UNIONS. Industrial & Labor Relations Review,
66 , 989-1011.
BRYSON, A., CAPPELLARI, L. & LUCIFORA, C. 2004. Does Union Membership Really
Reduce Job Satisfaction? British Journal of Industrial Relations, 42 , 439-459.
BRYSON, A., CAPPELLARI, L. & LUCIFORA, C. 2010. Why So Unhappy? The Effects of
Unionization on Job Satisfaction*. Oxford Bulletin of Economics and Statistics, 72 ,
357-380.
BRYSON, A., EBBINGHAUS, B. & VISSER, J. 2011. Introduction: Causes, consequences and cures of union decline. European Journal of Industrial Relations, 17 , 97-105.
BRYSON, A. & FREEMAN, R. 2013. Employee Perceptions of Working Conditions and the
Desire for Worker Representation in Britain and the US. Journal of Labor Research,
34 , 1-29.
BRYSON, A., WILLMAN, P., GOMEZ, R. & KRETSCHMER, T. 2013b. The Comparative
Advantage of Non-Union Voice in Britain, 1980–2004. Industrial Relations: A
Journal of Economy and Society, 52 , 194-220.
CAMPOLIETI, M., GOMEZ, R. & GUNDERSON, M. 2011. What Accounts for the
Representation Gap? Decomposing Canada—US Differences in the Desire for
Collective Voice. Journal of Industrial Relations, 53 , 425-449.
CHACKO, T. I. 1985. Member Participation in Union Activities: Perceptions of Union
Priorities, Performance, and Satisfaction. Journal of Labor Research, 6 , 363-373.
CHACKO, T. I. & GREER, C. R. 1982. Perceptions of Union Power, Service, and
Confidence in Labor Leaders: A Study of Member and Nonmember Differences.
Journal of Labor Research, 3 , 211-221.
CHARLWOOD, A. 2002. Why Do Non–union Employees Want to Unionize? Evidence from
Britain. British Journal of Industrial Relations, 40 , 463-491.
CHARLWOOD, A. & TERRY, M. 2007. 21st-century models of employee representation: structures, processes and outcomes. Industrial Relations Journal, 38 , 320-337.
COTTI, C. D., HALEY, M. R. & MILLER, L. A. 2013. Workplace Flexibilities, Job
Satisfaction and Union Membership in the US Workforce. British Journal of
Industrial Relations, n/a-n/a.
FAGERLIND, A.-C., GUSTAVSSON, M., JOHANSSON, G. & EKBERG, K. 2013.
Experience of work-related flow: Does high decision latitude enhance benefits gained from job resources? Journal of Vocational Behavior, 83 , 161.
FIORITO, J. 1987. Political Instrumentality Perceptions and Desires for Union
Representation. Journal of Labor Research, 8 , 271-289.
FIORITO, J. 2001. Human Resource Management Practices and Worker Desires for Union
Representation. Journal of Labor Research, 22 , 335-354.
FULLAGAR, C. J., CLARK, P. F., GALLAGHER, D. G. & CARROLL, A. E. 2004. Union
Commitment and Participation: A 10-Year Longitudinal Study. Journal of Applied
Psychology, 89 , 730-737.
FULLER, J. B. & HESTER, K. 1998. The Effect of Labor Relations Climate on the Union
Participation Process. Journal of Labor Research, 19 , 173-187.
GAHAN, P. 2012. 'Voice Within Voice': Members' Voice Responses to Dissatisfaction with
Their Union. Industrial Relations, 51 , 29-56.
GIVAN, R. K., HIPP & LENA 2012. Public Perceptions of Union Efficacy: A Twenty-Four
Country Study. Labor Studies Journal, 37 , 7-32.
HAMMER, T. H. & AVGAR, A. 2005. The Impact of Unions on Job Satisfaction,
Organizational Commitment, and Turnover. Journal of Labor Research, 26 , 241-266.
HEERY, E. 1998. Unions in a Changing World: Problems and Prospects in Selected
Industrialized Countries. International Journal of Human Resource Management, 9 ,
229-230.
HEERY, E. 2009. The representation gap and the future of worker representation. Industrial
Relations Journal, 40 , 324-336.
HEERY, E., KELLY, J. & WADDINGTON, J. 2003. Union Revitalization in Britain.
European Journal of Industrial Relations, 9 , 79.
HYMAN, R. 1997. The Future of Employee Representation. British Journal of Industrial
Relations, 35 , 309-336.
KELLOWAY, E. K. & BARLING, J. 1993. Members' Participation in Local Union
Activities: Measurement, Prediction, and Replication. Journal of Applied Psychology,
78 , 262-279.
KELLY, J., MILLWARD, N. & STEVENS, M. 1987. Trade Unions Through The Recession
1980-1984. British Journal of Industrial Relations, 25 , 275-282.
KLANDERMANS, B. 1986. Psychology and trade union participation: Joining, acting, quitting. Journal of Occupational Psychology, 59 , 189-204.
KLANDERMANS, P. G. 1984. Mobilization and participation in trade union action: An expectancy-value approach. Journal of Occupational Psychology, 57 , 107-120.
KOCHAN, T. A. 1979. How American workers view labor unions. Monthly Labor Review,
102 , 23.
KOCHAN, T. A. 2004. Collective Actors in Industrial Relations: What Future? Industrielle
Beziehungen, 2004 , 6-14.
LANGFORD, T. O. M. 1994. Involvement with Unions, Union Belief Perspectives, and
Desires for Union Membership. Journal of Labor Research, 15 , 257-270.
MACKENZIE, R. 2010. Why do contingent workers join a trade union? Evidence from the
Irish telecommunications sector. European Journal of Industrial Relations, 16 , 153-
168.
MARTINEZ, A. & FIORITO, J. 2009. General Feelings Toward Unions and Employers as
Predictors of Union Voting Intent. Journal of Labor Research, 30 , 120-134.
MCLLROY, J. 2012. Asa Briggs and the Emergence of Labour History in Post-War Britain.
Labour History Review (Liverpool University Press), 77 , 211-242.
MONNOT, M. J., WAGNER, S. & BEEHR, T. A. 2011. A contingency model of union commitment and participation: Meta-analysis of the antecedents of militant and nonmilitant activities. Journal of Organizational Behavior, 32 , 1127-1146.
PROWSE, P. J. & PROWSE, J. M. 2006. Are non-union workers different to their union colleagues? Evidence from the public services. Industrial Relations Journal, 37 , 222-
241.
PYMAN, A., HOLLAND, P., TEICHER, J. & COOPER, B. K. 2010. Industrial Relations
Climate, Employee Voice and Managerial Attitudes to Unions: An Australian Study.
British Journal of Industrial Relations, 48 , 460-480.
ROBINSON, J. C. 1988. Workplace Hazards and Workers' Desires for Union Representation.
Journal of Labor Research, 9 , 237-249.
SIMMS, M. & HOLGATE, J. 2010. Organising for what? Where is the debate on the politics of organising? Work, Employment & Society, 24 , 157-168.
SZÉLL, G. 2010. Trade Union Movement & Labour Policies in the EU after the Global
Financial Crisis. Indian Journal of Industrial Relations, 46 , 183-193.
TROY, L. 2001. Twilight for Organized Labor. Journal of Labor Research, 22 , 245-259.
TRUSS, C., SHANTZ, A., SOANE, E., ALFES, K. & DELBRIDGE, R. 2013. Employee engagement, organisational performance and individual well-being: exploring the evidence, developing the theory. The International Journal of Human Resource
Management, 24 , 2657-2669.
VISSER, J. 2002. Why Fewer Workers Join Unions in Europe: A Social Custom Explanation of Membership Trends. British Journal of Industrial Relations, 40 , 403-430.
WADDINGTON, J. & WHITSTON, C. 1997. Why Do People Join Unions in a Period of
Membership Decline? British Journal of Industrial Relations, 35 , 515-546.
YALABIK, Z. Y., POPAITOON, P., CHOWNE, J. A. & RAYTON, B. A. 2013. Work engagement as a mediator between employee attitudes and outcomes. The
International Journal of Human Resource Management, 24 , 2799-2823.
Appendix 1.
Descriptive statistics for dependent and independent variables
INDEPENDENT VARIABLES (CONTINIOUS)
Mean Median St. deviation
Job satisfaction
Industrial relations climate
Union/staff association involvement
3.5
3.49
3.57
3.65
4.00
3.66
0.73
0.99
0.71
INDEPENDENT VARIABLES (CATEGORICAL)
Membership status
Frequency Percent Valid
Percent
Yes
No, but have been in the past
No, have never been a member
Total
3774
858
968
5600
67.4
15.3
17.3
100.0
67.4
15.3
17.3
100.0
Managers/administrators
Professionals
Technicians/clerical/unskilled
Total not mentioned
Mentioned
Total not mentioned
Mentioned
Total not mentioned
Mentioned
Total not mentioned
Mentioned
Total not mentioned
Mentioned
Total
Occupational group
456
2818
2326
5600
8.1
50.3
41.5
100.0
Recession (my workload increased)
3647 65.1
1953
5600
34.9
100.0
Recession (my work was reorganised)
4176
1424
5600
74.6
25.4
100.0
Recession (I was moved to another job)
5169 92.3
431
5600
7.7
100.0
Recession (my wages were frozen or cut)
3209 57.3
2391
5600
42.7
100.0
Recession (my non-wage benefits were reduced)
5184 92.6
416
5600
7.4
100.0 not mentioned
Mentioned
Total not mentioned
Mentioned
Total not mentioned
Mentioned
Total not mentioned
Mentioned
Total
8.1
50.3
41.5
100.0
65.1
34.9
100.0
74.6
25.4
100.0
92.3
7.7
100.0
57.3
42.7
100.0
92.6
7.4
100.0
Recession (my contracted hours were reduced)
5440 97.1
160
5600
2.9
100.0
97.1
2.9
100.0
Recession (access to paid overtime was restricted)
4412 78.8
1188
5600
21.2
100.0
Recession (I was required to take unpaid leave)
78.8
21.2
100.0
5497
103
5600
98.2
1.8
100.0
98.2
1.8
100.0
Recession (access to training was restricted )
4639 82.8
961
5600
17.2
100.0
82.8
17.2
100.0
Skewness
-0.438
-0.562
-0.175
Cumulative
Percent
67.4
82.7
100.0
8.1
58.5
100.0
65.1
100.0
74.6
100.0
92.3
100.0
57.3
100.0
92.6
100.0
97.1
100.0
78.8
100.0
98.2
100.0
82.8
100.0
not mentioned
Mentioned
Total
Other forms
Collective forms
Total
Other forms
Collective forms
Total
Other forms
Collective forms
Total
Other forms
Collective forms
Total
Other forms
Collective forms
Total
Recession (other)
4112
1488
5600
73.4
26.6
100.0
DEPENDENT VARIABLES
Increase pay
1980
3620
5600
35.4
64.6
100.0
Reduced hours or pay
1636
3964
5600
29.2
70.8
100.0
Getting training
5156
444
5600
Complaint
3646
1954
5600
92.1
7.9
100.0
65.1
34.9
100.0
Discipline
1999
3601
5600
35.7
64.3
100.0
73.4
26.6
100.0
35.4
64.6
100.0
29.2
70.8
100.0
92.1
7.9
100.0
65.1
34.9
100.0
35.7
64.3
100.0
73.4
100.0
35.4
100.0
29.2
100.0
92.1
100.0
65.1
100.0
35.7
100.0
Appendix 2
1.
Pick-a-a point approach to probe moderation effect (dependent variable – increase in pay)
Focal Predictor Variable
Job satisfaction
Moderator Variable
Union/staff association involvement
Coding of binary Y for analysis:
Increase Analysis
.00 .00
1.00 1.00
Logistic Regression Summary
-2LL Model LL McFadden CoxSnell Nagelkrk n
6154.9951 1120.8553 .1541 .1814 .2494 5600.0000
================================================================
b se p Exp(B) Wald constant .5815 .1551 .0002 1.7887 14.0541
Occupati .4472 .0506 .0000 1.5639 78.2409
Membersh -.8989 .0398 .0000 .4070 509.5067
Industri -.3253 .0441 .0000 .7229 53.9930
Job_sati -.6578 .0534 .0000 .5180 151.9690
Union_in .3591 .0473 .0000 1.4320 57.5111 interact -.1844 .0577 .0014 .8316 10.2275
Interact is defined as:
Job satisfaction X Union/staff association instrumentality
=================================================================
Conditional Effect of Focal Predictor at Values of the Moderator Variable
Union_in b se Z p Wald
-.7142 -.5262 .0643 -8.1834 .0000 66.9683
.0000 -.6578 .0534 -12.3276 .0000 151.9690
.7142 -.7895 .0704 -11.2194 .0000 125.8758
Alpha level used for confidence intervals:
.05
Moderator values are the sample mean and plus/minus one SD from mean
2.
Johnson-Neyman technique to probe moderation effect (dependent variable – increase in pay)
Focal Predictor Variable
Job satisfaction
Moderator Variable
Union/staff association involvement
Coding of binary Y for analysis:
Increase Analysis
.00 .00
1.00 1.00
Logistic Regression Summary
-2LL Model LL McFadden CoxSnell Nagelkrk n
6154.9951 1120.8553 .1541 .1814 .2494 5600.0000
===================================================================
b se p Exp(B) Wald constant .5815 .1551 .0002 1.7887 14.0541
Membersh -.8989 .0398 .0000 .4070 509.5067
Industri -.3253 .0441 .0000 .7229 53.9930
Occupati .4472 .0506 .0000 1.5639 78.2409
Job_sati -.6578 .0534 .0000 .5180 151.9690
Union_in .3591 .0473 .0000 1.4320 57.5111 interact -.1844 .0577 .0014 .8316 10.2275
Interact is defined as:
Job satisfaction X Union/staff association instrumentality
=====================================================================
Conditional Effect of Focal Predictor at Values of Moderator Variable
Union_in b se Z p Wald
-1.9107 -.3056 .1178 -2.5929 .0095 6.7233
-1.7440 -.3363 .1094 -3.0753 .0021 9.4573
-1.5773 -.3670 .1011 -3.6316 .0003 13.1886
-1.4107 -.3977 .0930 -4.2757 .0000 18.2813
-1.2440 -.4285 .0853 -5.0223 .0000 25.2232
-1.0773 -.4592 .0780 -5.8852 .0000 34.6355
-.9107 -.4899 .0713 -6.8723 .0000 47.2284
-.7440 -.5207 .0653 -7.9760 .0000 63.6171
-.5773 -.5514 .0602 -9.1584 .0000 83.8764
-.4107 -.5821 .0563 -10.3344 .0000 106.7988
-.2440 -.6128 .0539 -11.3691 .0000 129.2566
-.0773 -.6436 .0531 -12.1124 .0000 146.7114
.0893 -.6743 .0541 -12.4673 .0000 155.4341
.2560 -.7050 .0567 -12.4400 .0000 154.7532
.4227 -.7358 .0607 -12.1230 .0000 146.9674
.5893 -.7665 .0659 -11.6355 .0000 135.3860
.7560 -.7972 .0720 -11.0765 .0000 122.6896
.9227 -.8279 .0788 -10.5102 .0000 110.4648
1.0893 -.8587 .0861 -9.9714 .0000 99.4291
1.2560 -.8894 .0939 -9.4756 .0000 89.7864
1.4227 -.9201 .1019 -9.0272 .0000 81.4900
Alpha level used for Johnson-Neyman method and confidence intervals:
.05
3.
Pick-a-a point approach to probe moderation effect (dependent variable – discipline)
Focal Predictor Variable
Job satisfaction
Moderator Variable
Union/staff association involvement
Coding of binary Y for analysis:
Discipline Analysis
.00 .00
1.00 1.00
Logistic Regression Summary
-2LL Model LL McFadden CoxSnell Nagelkrk n
6058.6306 1239.8665 .1699 .1986 .2727 5600.0000
===================================================================
b se p Exp(B) Wald constant 1.5267 .1560 .0000 4.6027 95.7414
Membersh -1.1140 .0409 .0000 .3282 740.5481
Industri -.2120 .0430 .0000 .8090 23.8131
Occupati .2584 .0510 .0000 1.2949 25.6704
Job_sati -.4970 .0526 .0000 .6083 89.2142
Union_in .5110 .0478 .0000 1.6669 114.4870 interact -.1423 .0567 .0120 .8673 6.3098
Interact is defined as:
Job satisfaction X Union/staff association instrumentality
=====================================================================
Conditional Effect of Focal Predictor at Values of the Moderator Variable
Union_in b se Z p Wald
-.7142 -.3954 .0624 -6.3364 .0000 40.1497
.0000 -.4970 .0526 -9.4453 .0000 89.2142
.7142 -.5987 .0701 -8.5347 .0000 72.8419
Alpha level used for confidence intervals:
.05
Moderator values are the sample mean and plus/minus one SD from mean
4.
Johnson-Neyman technique to probe moderation effect (dependent variable – discipline)
Focal Predictor Variable
Job satisfaction
Moderator Variable
Union/staff association involvement
Coding of binary Y for analysis:
Discipline Analysis
.00 .00
1.00 1.00
Logistic Regression Summary
-2LL Model LL McFadden CoxSnell Nagelkrk n
6058.6306 1239.8665 .1699 .1986 .2727 5600.0000
===================================================================
b se p Exp(B) Wald constant 1.5267 .1560 .0000 4.6027 95.7414
Membersh -1.1140 .0409 .0000 .3282 740.5481
Industri -.2120 .0430 .0000 .8090 23.8131
Occupati .2584 .0510 .0000 1.2949 25.6704
Job_sati -.4970 .0526 .0000 .6083 89.2142
Union_in .5110 .0478 .0000 1.6669 114.4870 interact -.1423 .0567 .0120 .8673 6.3098
Interact is defined as:
Job satisfaction X Union/staff association instrumentality
=====================================================================
Conditional Effect of Focal Predictor at Values of Moderator Variable
Union_in b se Z p Wald
-1.9107 -.2251 .1145 -1.9650 .0494 3.8612
-1.7440 -.2488 .1062 -2.3422 .0192 5.4860
-1.5773 -.2725 .0981 -2.7778 .0055 7.7160
-1.4107 -.2962 .0902 -3.2824 .0010 10.7741
-1.2440 -.3199 .0827 -3.8676 .0001 14.9582
-1.0773 -.3437 .0756 -4.5437 .0000 20.6448
-.9107 -.3674 .0691 -5.3157 .0000 28.2563
-.7440 -.3911 .0633 -6.1754 .0000 38.1360
-.5773 -.4148 .0585 -7.0895 .0000 50.2613
-.4107 -.4386 .0549 -7.9870 .0000 63.7922
-.2440 -.4623 .0528 -8.7606 .0000 76.7473
-.0773 -.4860 .0523 -9.2975 .0000 86.4436
.0893 -.5097 .0535 -9.5335 .0000 90.8876
.2560 -.5335 .0562 -9.4846 .0000 89.9567
.4227 -.5572 .0604 -9.2269 .0000 85.1349
.5893 -.5809 .0656 -8.8504 .0000 78.3300
.7560 -.6046 .0718 -8.4270 .0000 71.0140
.9227 -.6284 .0785 -8.0019 .0000 64.0306
1.0893 -.6521 .0858 -7.5993 .0000 57.7496
1.2560 -.6758 .0935 -7.2296 .0000 52.2669
1.4227 -.6995 .1014 -6.8955 .0000 47.5478
Alpha level used for Johnson-Neyman method and confidence intervals:
.05