ALFRED
P.
WORKING PAPER
SLOAN SCHOOL OF MANAGEMENT
A FIELD STUDY EXAMINATION OF BUDGETARY PARTICIPATION
AND LOCUS OF CONTROL
by
Peter Brownell
WP 1169-80
November 1?80
(Revised, February 19 82
Forthcoming, Accounting Review, October 1982
MASSACHUSETTS
INSTITUTE OF TECHNOLOGY
50 MEMORIAL DRIVE
CAMBRIDGE, MASSACHUSETTS 02139
A FIELD STUDY EXAMINATION OF BUDGETARY PARTICIPATION
AND LOCUS OF CONTROL
by
Peter Brownell
WP 1169-80
November 1?80
(Revised, February 19 82)
Forthcoming, Accounting Review, October 1982
A Field Study Examination of Budgetary Participation
and Locus of Control
ABSTRACT
The results of this field study are generally consistent with those
from a previous laboratory experiment which showed that the relationship
between budgetary participation and performance is moderated by the
personality variable internal-external locus of control
.
Internals
(individuals who feel that they are in control of their own destinies)
appear more job satisfied and perform better under conditions of high
externals (individuals who attribute the
By contrast,
participation.
results of their actions to chance, luck, or fate) are more job satisfied
The
and perform better under conditions of low participation.
complimentary findings of the two studies are considered and the
implications of the results in the areas of personnel management and
control system design are discussed.
>^u4i~\^(^^
A Field Study Examination of Budgetary Participation
and Locus of Control
question
The
will
leEd
of
conditions
the
under
outcomes
desirable organizational
to
is
matter of concern to
a
researchers in managerial accounting for two reasons;
widely believed
of
provide
to
organizational
literature
is
motivation
for
and
the
(2)
evidence
which
This
question
conflict.
considerable
laboratory study reported
a
elsewhere
the role of internal-external locus of control,
a
participation is
(1)
managerial approach to improving performance
a
personnel
in
participation
budgetary
which
in
the
provided
the
exists
[Brownell,
of
1981]
personality variable, as
a
moderator of the relationship between budgetary participation and
present study is to validate the results
The purpose of the
performance.
of the laboratory experiment using a field-study approach.
the
previous
will
study
participation
locus
and
also
of
be
extended
control
on
job
by
The results of
considering the
satisfaction,
effects
well
as
of
on
as
performance
are based
The hypothesized moderating effects of locus of control
the
notion
of
chsracteristics
Several
studies
congruence
between
of
situation
a
in
task
the
an
in
which
the
individual
indicate
literature
psychology
personality
individual's
that
is
and
on
the
placed.
"internals"
(individuals who generally feel they are in personal control of their own
destiny)
prefer,
and
perform better
in,
task
situations characterized by
"self-control", while "externals" (individuals who attribute the outcome of
their actions to chance,
fate,
or
luck)
prefer,
control is in the hands of others [Cromwell et al
and
.
,
perform better,
1961;
when
Rotter and Mulry,
Baumal
and
V/atson
19fi5;
"source of control
derives
from
in
that
budgetary participation
conditions
under
between
parallelism
The
1972].
particular situation" and
a
view
the
Houston,
1967;
.
budgetary
high
of
participation, an individual has substantial control over the budget, which
basis
the
forms
often
for
evaluation,
performance
budgetary
low
while
participation denies the individual such control.
The
test
results of
laboratory experiment
a
question,
this
provided
[Brownell,
confirmation
strong
designed to
1981],
predicted
the
of
interactive effects of participation and locus of control for independent
samples
of
subjects
accounting
drawn
students,
organizations.
and
highest
The
participation condition,
managers
middle-level
internals
performing
while the
manufacturing
in
were
those
in
who
were
"more
in
assigned
enjoyment"
finding
Brov/nell
to
their
of the
suggests
that
[1981,
f tn
.
congruent
experimental
job
participation
task
satisfaction
because
satisfaction
job
has
often
high
from
was
It
that subjects from both samples
11]
employed
also
may
ir,
be
examined
been
indicated
condition
interaction between participation and locus of control.
and
a
highest performing externals were
the experimental group experiencing a low participation treatment.
further noted
—
populations
different
quite
two
from
that
This
study.
affected
by
the
Because of this,
as
the
criterion
variable in previous studies of participative budgeting (e.g., see Hofstede
[1967], and Cherrington and Cherrington
[1973])
this variable was included
t
in the present study.
The major
Brownell
strength of experimental designs,
[1981],
is
in
their ability to generate valid causal
However, their most typical weakness involves
external validity.
such as that employed
In other words,
a
by
inferences.
lack of general izability or
while an experimenter may be confident
that the effects of differential experimental treatments have been causally
there cannot be absolute assurance tha^
linked to some dependent variable,
this
treatment
•intended
contrast,
generate
a
J
setting,
while
statements,
Together,
can
we
cannot
weaker
in
terms
field
and
is
it
absolutely
be
Brov.-r)ell
Ey
counterpart.
world
real
effectively address
experimental
process
world
of participation used in
its
far
real
the
context,
present
embraces
effectively
field
causal
issue.
the
In
of
the operational definition
849
p.
replica
faithful
a
represent.
to
certain that
[1981,
is
of
the
ability
its
to
generalizability
approaches
are
mutually
reinforcing.
HYPOTHESES
The
hypothesis
tested
in
my
previous
experimental
study
was
thaz
budgetary participation would interact with locus of control in affecting
performance.
This hypothesis also
tested in this study.
is
In
addition,
further evidence will be sought concerning the role of this interaction in
its
effects
on
job
satisfaction,
results discussed above.
In
their
a
dependent
null
forms,
variable
suggested
by
the
the two hypotheses of the
present study can be stated as follows:
Hj
There will be no significant interaction between participation and
locus of control affecting performance.
H^
There will be no significant interaction between participation and
locus of control affecting job satisfaction.
METHOD
1
The method used in this study was
a
survey questionnaire administered
to 48 middle level cost-center managers employed by a 'large San Francisco
This sample of managers was the same group
Bay Area manufacturing company.
which
formed
were
managers
experimental
second
the
from
drav/n
sample
in
The
the
of
divisions
functional
separate
eight
study.
previous
my
•organization, each of which was involved in one phase of the production or
Preliminary meetings with plant
distribution activities of the business.
management indicated that technological and environmental characteristics,
while
critical
not
resulting in the use of
that
a
characteristic
this
varied
themselves,
in
var'iety of managerial
of
would
sample
the
across
somewhat
approaches.
ensure
divisions,
the
It
felt
of
range
wide
a
was
responses on the participation measures.
The
sample
involvement
since
company
to
The
managers
in
the
chosen
was
project required
the
in
part
take
questionnaire.
responding
of
a
experimental
one-half day
study
and
questionnaire was administered
the conclusion of the laboratory sessions.
complete
rather
the
managers at
M8
This procedure eliminated nonand assured
response (together with the biases usually associated with it)
that- all
from the
leave
to
the
to
management,
plant
by
questionnaires were completed by the individuals in the sample,
than
by
disadvantage
assistants,
staff
of
this
for
approach
example.
concerns
the
potential
one
However,
possibility
that
the
treatments in the laboratory sessions sensitized respondents
experimental
The results of some tests
to the purpose of the questionnaire.
possibility are reported later.
possibility
of
completing
both
fatigue
the
A
resulting
experimental
fui'ther
from
the
session
potential
the
this
problem concerns the
lengthy
somewhat
and
for
process
of
in
one
performance,
job
questionnaire
sitting.
Three
variables
satisfaction,
and
were
measured
by
questionnaire:
budgetary participation.
Locus of control
was measured
experimental sessions.
at the beginning of the
see Brownell
For details,
[1981, p. 848].
Performance
selection
the
In
consideration:
the
of
performance
a
measure,
two
source of the performance ratings,
ality of the rating measure.
the first
On
issue,
the dimension-
and
was
it
required
issues
self-
felt that
ratings because of evidence
ratings would be preferable to superiors'
in
the literature which suggests that the latter are subject to "halo" error
when
used
Miner,
conjunction with multi-dimensional taxonomies [Lawler,
in
Thornton,
1968;
1968;
Nealey and Owen,
1967;
With respect to the
1970].
second issue, the need exists to recognize the multi-dimensional nature of
performance,
but
taxonomies
involving
excessive
an
number
dimensions
of
should also be avoided.'
the
On
Mahoney,
basis of these
Jerdee and Carroll [1963;
dimensions together
Independent
Penfield
for
a
considerations,
vjith a
assessments
[1974],
rating
1965]
self-rating
a
1
the
global rating, was chosen for the present study.
of
(low
of
taxonomy, using eight performance
the
Mahoney
measure,
Heneman
by
support its reliability and validity.
from
version
performance)
to
(high
9
[197^]
and
The measure calls
on
each
The eight
sub-
performance)
dimension, as well as for the global, or overall, rating.
dimensions of performance are planning, investigating, coordinating,
evaluating, supervising, staffing, negotiating, and representing.
Job Satisfaction
The Minnesota
Satisfaction
Questionnaire
was chosen to measure job satisfaction.
questionnaire which calls for
a
(MSQ)
[V/eiss
et
al
.
,
1967]
The MSQ is a 100-item Likert-type
response on
a
five-point,
fully anchored
scale
(the
satisfied") to
a
"very
descriptors of which are
polar
dissatisfied"
"very
and
The measure provides satisfaction
set of job attributes.
The
scores on each of 20 sub-scales, five items addressing each sub-scale.
measure also provides for an overall rating of job satisfaction which is
derived
as
each
For
follov.'s.
item among the
representative"'*
of
the
20
sub-scales,
five
for
each
sub-scale
the
is
"most
one
chosen.
The
scores on these 20 "most representative" items are then summed giving the
overall rating.
Independent assessment of the validity of the MSQ is due to Dunham,
Smith and Blackburn [1977] who reported that the MSQ outperformed two other
satisfaction measures in terms of both convergent validity (the ability of
different
questionnaire
validity
discriminant
items
(the
measure
to
ability
of
same
the
different
construct)
items
questionnaire
and
to
measure different constructs).
Participation
Several direct (as opposed to factor-analytic) attempts at measurement
of budgetary participation are
found
in
the
Examples include
literature.
those of Vroom [I960], Likert [1961], Hofstede [1967], Heller [1971], Vroom
and
Yetton
[1973]
Milani
and
[1975].
Of
these,
measures were selected for the present study.
developed
for
use
in
a
context
identical
to
Hofstede
the
and
Milani
Only these two measures were
present context,
the
and
the
use of either one would permit an integration of the results of the present
study into the most relevant body of previous literature.
For purposes of
cross-validation, both measures were employed in this study,
Milani measure,
testing because,
permitting
a
a
seven-point Likert-type- scale
unlike
the
Hofstede measure,
reliability assessment.
,
although the
will be used in hypothesis
it
is
a
multi-item measure
RELIABILITY AND VALIDITY CHECKS
Performance
statistical
A
variations
the
in
rating
global
explained by ratings on the
check,
global
the
separate dimensions
dimensions explained
This
result
is
ratings
in
a
conducted
was
check
from
eight
were
to
the
separate
regressed
assess
dimensions.
on
the
60.8 percent of the variance
with
Jerdee
Mahoney,
could
in
conduct
To
ratings
eight
eight
sepai-ate
global
ratings.
Carroll's
and
be
this
the
on
The
the
which
to
instrument
Mahoney
single multiple regression.
consistent
extent
the
[1963]
developmental work where it was found that approximately 55 percent of the
functions critical to effective managerial performance were common
^^52
managerial assignments in
different companies studied,
13
approximately ^5 percent were job specific [pp.
also consistent with Heneman's [197^^]
presents
simple
each
the
of
correlations
eight
separate
betv;een
global
dimensions
for
both
performance
the
the
while
The results are
106-10?].
use of the Mahoney measure.
the
to
Table
1
ratings
and
sample
and
present
Heneman's study.
INSERT TABLE
The results of this comparison reveal
dimensions 2 (investigating) and
similarly
important
in
sample and the present
4
explaining
sample.
1
HERE
that,
with the possible exception of
(evaluation), the set of dimensions are
overall
This
performance
similarity of results
evidence that no difficulties were encountered
measure in this study.
in
in
the
both
Heneman's
provides
some
use of the Mahoney
Also
provided
Table
in
is
1
matrix of the
a
the eight sub-dimensions of performance.
It
intercorrelations among
is noted
that several of these
coefficients are sufficiently large to warrant some caution
the
results
statistics
Descriptive
tests
hypothesis
of
on
sub-dimension.
particular
any
Mahoney measure responses are presented
the
for
based
interpreting
in
in
Table 2.'
INSERT TABLE
HERE
2
Job Satisfaction
A reliability
assessment of the MSQ was made by
reliability coefficients computed
a
comparison of the
each of the 20 sub-scales,
for
plus the
single overall rating, with the coefficients for 27 norm groups reported by
et
V.'eiss
al
and
[1967]
.
managers,
also
variance
technique,
reported
with
also
Weiss
by
used
et
al
Weiss
by
coefficients
the
et
al
.
coefficients from the present administration.
for
overall
satisfaction
was
0.72,
which,
was
,
lower than the coefficients reported by V/eiss et al
and
(0.88)
across the
for
20
of managers
group
the
sub-scales in
present
the
(0.85).
analysis
Hoyt
used
135
derive
to
of
the
reliability coefficient
The
while
sample of
a
The
[1967].
.
for
satisfactory
.
per
se
,
is
for the 27 norm groups
The
coefficient
average
administration
was
(in
0.71
the
range of 0.50 to 0.81), compared with 0.86 (0.67 to 0.95) and 0.84 (0.78 to
0.93)
f
respectively, for the two comparison groups reported by Weiss et al
Inter-item correlations
among
the
from
Given
that
+0.01
to
+0.85.
sufficiently large
to
reject
a
20 sub-scales averaged +0.38
null
this. average
correlation
of
and
ranged
+0.38
is
hypothesis of no relationship at the
five percent level of confidence, it will be necessary to exercise caution
in
interpreting
results
the
involving
hypotheses
of
scales, due to the lack of independence among
this
concern,
last
results
the
this
of
of
any
sub-scales.
the
testing
can
these
sub-
Aside
from
considered
be
satisfactory, particularly the sub-scale reliabilities, which, while
lower
generally
those
than
Weiss
by
reported
conclusion
that
et
al
quite
are
[1967],
.
MSQ administration
was
adequate to
support
successful.
Descriptive statistics for the job satisfaction responses are
presented in Table
the
the
2.
Participation
The responses to
correlated
measures
provide
An
To
the
statistical
measure
as
a
factor.
To
analyzed.
six
of
load
the
explained
61
the second
conducted
rating-scale
the measure
issue,
on
a
single
should
the
six
items
in
The
factor.
two eigenvalues greater than unity,
Five
variance,
common
does
it
the
on
tap one
measure.
did)
Milani
(as
Milani
Milani
implies
single dimension
were
responses
factor-
factor-anal ^liical terms, one would hope to discover that all
In
items
this
examine
was
test
summated
that each of the six items in
or
percent
55
the two
implies that
Although this result
were
reasonably satisfactory test of convergent validity.
additional
treat
Hofstede participation measures
and
approximately
only
share
a
(p<0.01).
+0.74
at
Milani
the
the
analysis led
to
the emergence of
two factors therefore being
measure
loaded
on
a
single
extracted.
percent of the combined variance across the measure.
factor barely qualified
which
factor,
In
fact,
for consideration with an eigenvalue of
1.037 and an additional explanatory power of only 17 percent.'
These
results
can
be
viewed
additivity assumption implicit
results,
it
was decided
to
in
as
fairly
satisfactory
Milani 's measure.
On
evidence
of
the
the basis of these
construct participation scores additively,
the
1C
approach
by Milani.
'jsed
alternative approach of constructing
The
factc-
scores was decided against in order to ensure comparability of the results
of this
with
study
those of nilani.
further in the study.
Hofstede measure
The
not
was
usec
Table 2 also presents descriptive statistics for the
Milani measure.
RESULTS
Questionnaires
responses
were
omitted
were
collected
due
to
from
improper
all
48
managers.
completion
of
sets
Two
both
cf
Hahonej
the
measure and the MSQ, while an additional six sets of responses were omittei
due to evidence of response set bias in completion of the MSQ only.
Tr.e
final sample size for hypothesis testing was reduced to N=UO as a result cT
these procedures.
Performance:
The
H
,
performance
hypothesis
approach fitting the model presented
Y = e,
+ e,x + B3Z + B.
performance,
Mahoney measure
where Y is
tested,
was
in
|(X - Z)j
The
a
regressic-
multiple
Equation (1).
|(x - z)|
as
(1)
by
n:easured
X is standardized locus of control
Z
using
the
global
rating
score ([X. - X]/o
is standardized budgetary participation ([Z.
on
the
)
- Z]/o
)
is thie interaction between locus of control and participatior.
absolute value of difference between X and Z.
rationale
for
the
interaction terms
is
as
follows:
scores
high
,
cr.
locus of control (external), combined with low scores on participation (Icv
participation)
,
v.-ill
produce large absolute difference terms,
as
will
lev
locus of control scores (internals) combined with high participation scores
(high participation).
These two cr-binations are expected,
as
per
Hj,
t;
11
be
associated
performance
high
with
scores.
other
The
combinations
two
(low participation internals and high participation externals) will produce
scores
difference
lower
absolute
lower
performance.
In
other
expected
and
are
words,
for
consistent with the expectations,
rejection
positive
a
to
associated
be
cf
K,
in
a
with
fashion
is required.
8^
The results of the regression are presented in Table 3.
INSERT TABLE
e^
the interaction coeeff icient
significance only at the
was highly significant
.10
,
3
HERE
is found to be positive, although reaching
level.
«
the
q,,
however,
(p<0.01),
participation
coefficient,
suggesting that while
the
interaction contributes to an explanation of performance differences across
the sample,
participation alone exerts
performance.
In
positively associated
with
performance,
among internals than externals.
substantial positive influence on
participation
words,
other
a
Bj,
but
appears
this
to
generally
be
asscciation
is
stronger
the coefficient :zt locus of control,
is also positive but fails to reach significance at any reasonable level.
Job Satisfaction:
To test
Hj,
H
,
the overall measure of job
as the dependent
used
Table
4
variable in
a
regression identical to Equation (1).
presents the results of this regression.
INSERT TABLE
Q^,
the
satisfacticn from the MSQ was
coefficient
again consistent
for
the
interaction
with expectations.
^
HERE
term,
is
Participation
significant
(5,)
(p<0.025),
apparently exerts
very
influence
direct
little
satisfaction,
job
on
depending on locus of control.
This
is
in
contrast
strictly
effects
its
results
the
to
for
performance pi'esented above.
Further A nalysis
The tests
^
of Hj
and
reported
Hj
above were based
hypotheses
identical
were
(1)
performed
sub-dimensions
performance
eight
the
for
Equation
to
or
Two further sets of
overall, ratings for performance and job satisfaction.
regressions
global,
the
on
explore
to
the
for
and
the
job
20
satisfaction sub-scales.
the
In
staffing
positive,
area
performance,
of
significant
produced
expected.
as
The
regression
a
Together,
suggest
exists
that
association
is
involving
planning
of
coefficients,
and
both
sub-dimension
the
of
significant coefficient for participation.
participation
between
the
results
interaction
(pj<0.10)
planning also produced
these
sub-dimensions
the
and
that
a
this
sub-dimension
stronger
positive
association
internals
among
of
generally
performance,
but
externals.
The
than
results also reveal significant positive coefficients for participation in
the regressions for three other
sub-dimensions - evaluating,
negotiating,
and representing.
Of the
regressions involving the job
20
satisfaction
sub-scales,
15
produced significant (p£0.10) interaction coefficients, all in the expected
direction.
The
creativity,
recognition,
responsibl ity
,
strongest
and
results
variety,
security.
uere
moral
and recognition.
values,
Seven
activity,
for
supervision
positively associated with participation..
creativity,
found
sub-scales
-
the
sub-scales
human
were
of
relations,
significantly
These were ability utilization,
responsiblity
,
supervision-technical,
13
cases where the interaction term, but not the participation term,
In
is
significant,
results
the
imply
that
satisfaction
with
particular
the
job-aspect is higher for internals in high participation conditions and for
externals in low participation conditions.
The most
significant
(p£0.05)
results of this type were found for the sub-scales of company policies and
practice,
supervision,
security,
interaction
term
In
cases
where
significant,
participation term are
the
and
variety.
and
both
the
the
results
imply that satisfaction is positively associated with participation
overall, but that these associations are stronger for intet-nals under high
participation and externals under low participation.
The most
significant
(p£0.05) results of this type were found for the sub-scales of creativity,
recognition, and responsibility.
As discussed earlier,
sub-dimensions,
performance
eight
suggest
the
section,
due
need
to
the patterns of intercorrelation among both
to
carefully
evidence
of
and
interpret
lack
results
the
independence,
of
sub-scales,
satisfaction
20
the
the
discussed
particularly
in
this
in
the
latter case.
DISCUSSION AND LIMITATIONS
Generally speaking,
the
results of the survey phase of the research
program can be viewed as consistent with those of the previously reported
experimental
control
research,
support
phase.
affecting
although
the
performance
only
conclusion
that
individuals.
from
is
marginally
expectation
internally-oriented
tentative
interaction
The
the
between
participation
significant
in
the
survey
participation
In
the
previous
in
area
study
is
of
both
and
of
phases of the
The
phase.
results
effective
for
satisfaction,
the
most
job
locus
concerning
the
interactive
m
effects
participntion
of
and
locus
satisfaction
affecting
control
of
is
strongly confirmed in the present study.
One important difference in the results of the
importance of
experimental
on
performance.
experimental
participation,
phase,
taken
alone,
participation.
In
insignificant
had
the
effects
should be noted that care was taken tc ensure that the
It
treatments
high
for
ensure that, ceteris pa ribus
particular
effects of
unmoderated
direct,
the
phases concerns the
tv/o
advantage
participation
low
and
to
subjects in one experimental condition had no
,
terms of performance)
(in
designed
v.'ere
over
those
in
other.
the
The results of the experimental phase confirm the success of this endeavor.
By contrast, the results of the field study indicate that participation may
exert
this
a
direct positive influence on performance, (although the nature of
influence
strengthening
appears
it,
and
be
to
conditioned
externals
locus
by
weakening
of
result
This
it).
internals
control,
has
some
intuitive appeal when considered in the light of the characteristics of the
organization
from
which
sample
the
earlier, respondents were drawn
the organization.
and,
hence,
organization
participation
a
Notably,
higher
studied.
can
for
Previous
in
from
a
subjects
a
literature
this
areas across
high level of interdependence
cooperation
achieving
As mentioned
was dravm.
variety of functional
there existed
need
play
of
among
has
these
suggested
cooperation
areas
the
across
in
role
the
which
different
functional units (e.g., Waterhouse and Tiessen [1978] and Thompson [196?]).
Considering the survey data for performance and job satisfaction together,
the results suggest that while participation is not necessarily satisfying,
it is viewed
independent
each.
as essential
functional
to achieve the aoordination of activities across
units
necessary
for
effective
performance
within
15
contrast can be drawn between the results of this study and
A useful
those of Milani
only
find
evidence
weak
performance.
who,
[1975],
of
v/orthy
is
It
respondents were all drawn
Hayes
has
[1977]
organization
factors
v/hich
using
of
from
indicated
effects
positive
note
that
on
of
questionnaire
area of one
organization.
production
that
participation
of
sample
his
production
the
could
participation measure,
same
the
area
one
is
the
within
often enjoys some buffering from external environmental
(including
factors internal
other
to
functional
areas
the
within
organization), especially when its own activities are more or less routine.
Hage
and
Aiken
suggested
that,
participative
results
planning
the
for
structures
could
can
be
Honour
and
of
such
have
centralized,
non-
pp.
other
In
the
to
25-27]
[1975,
activities,
successful.
attributable
been
have
Swieringa
and
[1959]
of
nature
Milani 's
words,
the
activities
engaged in by the organizational sub-unit from which his sample was drawn.
Participation might not be expected
performance
in
such
circumstances
to
contribute
substantially
non-participative
where
schemes
to
are
equally effective.
The
above
limitations
of
discussion
should
study
this
and,
be
tempered
by
research
entire
the
indeed,
consideration
a
of
the
program.
Several deserve elaboration.
In
study,
The
connection
it should
measurement
with
the
measurement
instruments
used
be recognized that these are subject to
of
performance
remains
problematic,
in
the
survey
some limitations.
and,
despite
the
advantages of self-ratings alluded to earlier, it should be noted that any
rating
of
performance,
from
whatever
source,
is
highly
subjective.
The
possibility of attr ibutional biases in these ratings cannot be ruled out.
For
example,
it
is
not
inconceivable
that
highly
participating
managers
16
rationalize
will
investment
thei'-
should
also
That
is,
in
linkage in thi^
The causal
High
by the organization
performance
as
on
th£
prerequisite
a
tc
linkage could be the reverse of thst
causal
the
perceivir.r
process by
this
carefully assessed.
be
part of managers might be viewed
participation.
time
pcformance higher.
and hence ratinsi, their ovm
relationship
of
conventionally proposed.
potential
A
concerns the
at
earlier,
is
questionnaire
benefits
of
survey
the
conclusion
the
of
of
phase
progran
research
the
administering the questionnaire
the
experimental
sessions.
note;
As
possible that the differential experimental treatments (higr.
participation)
low
and
of
unknown consequences of
immediately
it
limitation
responses
achieving
gave
two
percent
100
differences
systematic
some
to
these
from
a
rise
groups
response
of
rate
iz
respondents.'"
The
must
off
traded
be
against the potential costs associated with the p'-ocedure used here.
role of variables omitted
The
not
also
overlooked.
be
previously [Brov/nell,
suggest
Reward
in
the
present research design
structure
which
for
1981]
importance
its
from
evidence
is
exists
present context.
the
one
mentionec
variable
[Gregory,
Future research
1978]
tc
would
cc
well to examine this and other types of variables likely to exert an
on
the
style)
relationship.
and
Variables
from
organizational levels (e.g.,
the
inte'-personal
structu'-e)
(e.g.,
shoulr
impac-
leadership
represent two classes cf
influences on the participation/critc-ion relationship which remain largely
unstudied,
and
the
inclusion
of
which
would
likely
have
improved the overall explanatory power of the models tested
Finally,
the
focus
on
a
single
organizational
in
sample
substantially
this study.
constitutes
further grounds for caution in interpreting the "esults too liberally.
17
these
Notwithstanding
research
this
of
phases
two
the
limitations,
program have provided mutually supportive evidence of the need to recognize
the existence of important moderating effects of the personality variable,
locus
of
control,
relationship
the
on
participation
between
and
both
performance and job satisfaction.
The potential implications of these results (and others confirming the
role
of
personality
different
possibly
participation)
are
selection
placement,
and
organizations that
participation
traits
most
important
members,
its
afforded
level
For
budgetary
of
that
conceivable
is
it
managerial
of
system design.
the
over
discretion
of
effect
the
of
areas
the
in
control
perhaps,
and,
permitted
are
moderators
as
the
role
specifications could be modified (as far as participation is concerned)
suit
the
personality
characteristics of the
role
Even where this is not feasible, management can
need
to
consider
organizational
a
roles
mismatch
possible
and
their
as
least
at
between
occupants
occupant
be
incumbent.
aware of the
characteristics
the
a
or
cause
of
to
of
sub-standard
performance and/or job satisfaction.
More
likely
than
the
scenario
above
is
the
possibility
that
environmental and technological circuastances confronting the organization
necessitate
example,
a
particular level of participation in budgetary matters.
readily
a
programmable,
effectively controlled via
involving
procedure
Crozier [196^^]).
a
activity
centralized,
specification
By contrast,
routine
and
often
can
non-participative
rule-setting
(see,
suggested
in
task
[Burns
complexity
and
Stalker,
most
process
example,
other activities in the organization may be
poorly specifiable and subject to environmental uncertainty.
increase
for
be
For
associated
1961;
with
Thompson
these
1967]
The resulting
activities
to
be
best
has
been
managed
lb
through
a
participative
highly
decentralized,
process.
In
these
circumstances, it will often be the case that the best information base to
support managerial
than
with
management
level
of
top
is
decisions will
management
constrained
budgetary
be
located
[Galbraith,
by
these
participation
1977 3.
types
into
at
of
role
the
In
action
point
circumstances
factors to build
descriptions,
the
a
rather
where
specific
burden
of
achieving the right fit between individuals and roles shifts to personnel
selection and placement.
Further research is obviously needed to expand our
understanding
and
awareness of other individual characteristics that complement design
features of budgetary control systems.
Indeed, the more basic question of
what control system features are appropriate to manage activities that face
differing
environmental
largely unanswered
and
technological
circumstances
still
remains
19
FOOTNOTES
to space limitations, the discussion of the measures chosen for
performance, job satisfaction and participation is kept relatively brief.
However, complete copies of these questionnaire items can be obtained,
upon request, from the author.
'Due
error is the tendency to evaluate "globally," or on a single
High intercorrelations among separate performance
cognitive dimension.
dimensions is evidence of halo error
^Halo
3
Kavanagh, MaciCinney and Wolins [1971] provide a good example of the need
The authors of that study employed a 20-dimension
to avoid this problem.
rating scale and obtained disappointing results on a discriminant
validity test.
The one "most representative" item among the five for each of the 20 subThe chosen item for each
scales i:; prescribed by V/eiss et al
[19671.
sub-scale is that which was found, in developmental work en the measure,
to be most highly correlated with its respective total sub-scale score
(the sum of five items).
.
5
complete tabulation of data, including descriptive statistics for
each of the eight performance sub-dimensions, and the 20 satisfaction
sub-scales, can be obtained, upon request, from the author.
A more
6
including the Hoyt analysis of
for each sub-scale, descriptive
statistics for each sub-scale, and the matrix of intercorrelations among
the sub-scales, can be obtained, upon request, from the author.
A
more
complete
tabulation
of
data,
variance reliability coefficients
7
Complete tabulation of the results of this analysis can be obtained, upon
request, from the author.
8
The three independent variables are virtually uncorrelated
the strongest
association being r=+0. 10 (n.s.) between locus of control and the
interaction term.
Therefore, the square of the t-statistic i's an
accurate approximation of the F-statistic conventionally used to test the
significance of the added explanatory power of a model which includes the
It is possible to
interaction term, over a model which excludes it.
conclude, therefore, that, in this case, the added explanatory power of
the model which includes the interaction term is also approximately
significant at the .10 level.
,
9
Complete tabulation of the results suiimarized
obtained, upon request, from the author.
in
this
section
can
be
1
the present context, however, no such differences could be uncovered.
test for this possibility, the scores for each of the variables
measured in the questionnaire phase (performance, job satisfaction, and
participation) were compared across the two experimental treatment
conditions.
None of the differences between means across the two
conditions reached significance at any reasonable level.
In
To
20
TABLE
1
TABLE OF INTERCORRELATIONS AMONG PERFORMANCE DIMENSIONS
1
2
3
5
'4
6
7
1.
Planning
2.
Investigating
37
3.
Coordinating
28
25
1
4.
Evaluating
3i
39
32
5.
Supervising
3^
27
28
23
1
6.
Staffing
41
03
29
50
13
7.
Negotiating
31
10
20
29
-19
32
1
8.
Representing
32
16
02
47
14
35
51
8
,
1
Note:
1
1
1
1
Decimal point suppressed
r
>_
39 significant at p
r
>
30 significant at p < 0.05
_<
0.01
CORRELATIONS BETOEEN OVERALL PERFORMANCE AND THE SEPARATE
DIMENSIONS OF MAHONEY MEASURE
21
TABLE 2
DESCRIPTIVE STATISTICS FOR MEASURED VARIABLES
Variable
Performance
Mean
Std
.
Dev'n
Possible Range
6.85
0.97
Job Satisfaction
77-95
7.27
20-100
Participation
25.13
7-36
6-^42
1-9
Actual
Range
N
5-8
^46
58-93
^0
9-39
^^6
22
23
TABLE
4
REGRESSION RESULTS:
Coefficient
H
2
2H
REFERENCES
"Participation in Budgeting, Locus of Control arBrov/nell,
P.,
Organizational Effectiveness," The Accounting Review (October 1981), ?t .
"Budgetary Control ar.d Organizatirr
Bruns, W.J., Jr. and J,H. Wsterhouse
Research
(Autunn 1975), ??. 177-203.
Accounting
Journal
of
Structure,"
,
Burns,
T.
and G.M. Stalker, Th e Management of Innovation (Tavistock,
I96I'-
"Appropriate Reinforceme-D.J. and J,0. Cherrington
Cherrington
Contingencies in the Budgeting Process,"
Empirical Research i~,
A ccounting:
Research
( 1
,
Sel ect ed Stud ies,
supplement
the
to
Jourr;al of Accouni:i-~c
T'sTT" PpTT? b -2 5 3
"Reaction Tine,
Cromwell, R.L., D. Rosenthal, D.K. Shakow and T.P. Zahn
Locus of Control, Choice Behavior and Descriptions of Pa-ental Behavi:in Schizophrenic and Normal Subjects," Journal of Personality (Dececbe1961)
pp. 363-380.
,
,
Crozier,
1
96iJ
)
M.
The Bureaucratic Phenomenon
,
(University
of
Chicago
Pre35,
.
Dunham, R.B., F.J. Smith and R.S. Blackburn, "Validation of the Index rf
Organizational Reactions with the JDI, the MSQ and Faces Scales," Acacc-^y
of Management Journal (September 1977), pp. 120-432.
Galbraith, J. Des i gning Complex Organizations
,
(
Addison-Wesley,
1977).
Gregory, W.L., "Locus of Control for Positive and Negative Outcomes."^
Journal of Personali t y and Social Psychology (August 1978), pp. 8^0-8^?.
Hage, J. and M. Aiken,
Goals,"
Organizational
1969), pp.
"Routine Technology, Social Structure a'
(Septer.b^Administrative Science Quarterly
,
366-376.
Hayes, D.C., "The Contingency Theory of Managerial
Accountiing Review, (January 1977), PP. 22-39.
Heller, F.A., Manager Decision Ma k ing (Tavistock,
Accounting,"
T"-
1971).
Heneman, H.G. Ill, "Comparisons of Self and Superior Ratings of Manage'-iaL
Performance," Journal of Applied Psychology (October 197^), pp. 638-6^2.
Hofstede, G.H., The Game of B'ldget Contro l
(Van Gorcum,
1967).
Houston, B.K., "Control Over Stress, Locus of Control, and Response z.zStress," Journal of Personality and Social Psychology (February
1972 ,
pp. 2^9-255.
Kavanagh, M.J., A.C. MacKinney and
"Issues in ManageriaJL
V/olins,
L.
Performance:
Multitrait-J'.ultimeth.od Analyses of Ratings," Psychologic aJ,
Bulletin (January 1971), pp. 34-49.
25
Lawler, E.E., "The Multitrait-Hultirater Approach to Measuring Managerial
(October
1967),
pp.
Journal of Applied Psychology
Performance,"
Job
369-381.
Likert,
R.
,
New Patterns of Management (McGraw-Hill,
Mahoney. T.A.,
Performance:
1963).
1961).
Jerdee and S.J. Carroll, Development of Managerial
T.H.
Company,
Publishing
(South-V/estern
A Research Approach
T.H. Jerdee and S.J. Carroll, "The Job(s)
Industrial Relations (February 1965), PP. 97-110.
,
"The Relationship of Participation
Milani, K.W.
Performance and Attitudes:
Supervisor
Industrial
Accounting Re view (April 1975), pp. 27^1-285.
,
in
A
of
Management,"
Budget-Setting to
Study," The
Field
A Capsule Review
Miner, J.B., "Management Appraisal:
(October
1968), pp. 83-96.
References," Business Horizons
and
Current
Nealey, S.M. and T.W. Owen, "A Mul titrait-Multimethod Analysis of
Predictors and Criteria of Nursing Performance," Organizational Behavior
and Human Performance (July 1970). PP. 3^^8-365.
Penfield, R.V., "Time Allocation Patterns and Effectiveness of Managers,"
Personn el Psychology (Summer 197^^). PP. 2^5-255.
Rotter, J.B. and R.C. Mulry, "Internal Versus External Control of
Journal of Personality and Social
Time,"
Decision
and
Reinforcement
Psychology (October 1965), pp. 598--60n.
Swieringa, R.J. and R.H. Moncur Some Effects of Participative Budgeting on
Managerial Behavior (National Association of Accountants, 1975).
,
Thompson, J.D.
Or ganizations in Action
.
(McGraw-Hill,
1967).
Thornton, G.C., "The Relationship Between Supervisory- and Self-Appraisals
of Executive Performance," Personnel Psychology (Winter 1968), PP
Vroom, V.H., and P.W. Yetton, Leadership and Decision Making (University of
Pittsburgh Press, 1973)
Tiessen, "A Contingency Framework for Management
Accounting, Organizations and Society
Research,"
No. 1), pp. 65-76.
Waterhouse, J.H. and
Systems
Accounting
(1978, Vol
3
P.
,
Watson, D. and E. Baumal
"Effects of Locus of Control and Expectation of
Performance," Journal of Personality and
Present
Control
Upon
Future
Social Psychology (June 1967), pp. 212-215.
,
Weiss, D.J., R.V.Dawis, G.W. England
Minnesota Satisfaction Questionnaire
Rehabilitation, 1967, vol XXII).
and
L.H.
(Minnesota
Lofquist,
Studies
Manual for the
Vocational
in
Date Due
;iV*'
ft
^y ^?
MAR
5 19^5
SEPL
iQP
1989
J
SE U3'89
%
'•3b
IDG2i199"
oc
%7l9'
•JAN
W^§
OCT
\
Q
1986
imsi
Lib-26-67
B03K''::J..,\'3 CO., INC.
DEC 8
1983
100 CAME:?nGE STREET
CHARLESTOWN, MASS.
I
HD28-IV1414 no.ll74- 80
Brownell, Pete/Leadership
HD28.M414 no.ll65- 80
continuity
WlcKersie, Robe/Change and
741433
D»BKS
_
740979
001_3;3Z9
001 "na 630
TOflD
_
MBKS
TOflO
_
_.
00
r.Tt^rn%V|c.ti?e.nW^^^^^^
OUlJ.Wi.!?..,,
,P*BKS
741452
002 0M2
TOflO
3
fi74
MIT LIBRARIES
,1^-7
i
TDflO DDM
3
-^O
193
/
bm
557
HD28.M414 no.ll68- 80
Pete/Participation
Brownell
D»BKS
740958
"
002 001
TOflO
3
budget
in
.B"'"""'"
07fl
HD28,M414 no,1169- 80
Brownf
Brownell,
Pete/Participation
D»BKs
740971
m
budget
qnir.05/
002 001
TOflO
52
Mil LibRAftJES
e^q-?0
3
TOfiO
DD3 0^1 T53
HD28.M414 no.ll70- 80
Brownell, Pete/The effects of personal
D»BKS
740974
001 TT2 bbS
TOflO
3
00.1323""
HD28.M414 no.ll71- 80
accounting
OQl.3.2371
Brownell, Pete/Tfie role of
D»BKS
740972
HD28.M414
Brownell,
_
1172-
80
Pete/Participation
D*BKS
740970
3
no.
.
001 TT2 bTT
TOflO
3
.
TOflO
in
001 TT2 715
HD28.IV1414 no.ll73- 80
Brownell, Pete/Improving the
8
740968.
D*
.D*BKJ
lllliiilillill,
3
bu
the
00132372
informati
pqi32638
ilil
TOflO
001 TT5
Tfll
\<^=^
behavior,
bu
00.132629
T^5 ao^