Sei sulla pagina 1di 7

Causal ordering of stress, satisfaction and

commitment, and intention to quit: a structural


equations analysis

A.R. Elangovan
Faculty of Business, University of Victoria, British Columbia, Canada

Keywords Turnover has long been an important area of probably influence commitment more than
Staff turnover, Stress, research in several disciplines (e.g. would job characteristics. Meanwhile,
Job satisfaction, Commitment
psychology, sociology, economics, Williams and Hazer (1986) found that
Abstract organizational behaviour) (Williams and satisfaction causally affects commitment,
This study addresses the Hazer, 1986). The general trend has been while a study by Bateman and Strasser (1984)
confusion prevailing over the towards developing conceptual models of the showed that commitment is causally
nature of the relationship between antecedent to satisfaction. On the other hand,
turnover process and, later, subjecting them to
satisfaction and commitment in
empirical tests. Although researchers have Curry et al. (1986) found no support for either
regard to employee turnover, and
examines the causal pattern of postulated and identified a number of variables of the hypothesized causal linkages between
relationships among stress, as being associated with turnover, satisfaction, job satisfaction and commitment (i.e. neither
satisfaction, commitment, and commitment and intention to quit have been causally affected the other). To add to the
turnover intentions by employing a controversy, a study by Farkas and Tetrick
structural equations analysis
generally accepted as important antecedents to
turnover. There is considerable support for the (1989) suggested that the two variables may be
approach. The results indicate
that there are strong causal links notion that intention to quit is probably the either cyclically or reciprocally related. It is
between stress and satisfaction most important and immediate antecedent of important that we determine the causal
(higher stress leads to lower
turnover decisions (Mobley et al., 1979; relationships between these two work attitudes
satisfaction) and between for several reasons. From a theoretical
satisfaction and commitment Mitchel, 1981; Bluedorn, 1982). Intention to quit
represents an attitudinal orientation or a standpoint, a causal relationship between the
(lower satisfaction leads to lower
commitment), and a reciprocal cognitive manifestation of the behavioural two variables (either direction) would imply
relationship between commitment decision to quit. But what about the that studies which omit the relationship or the
and turnover intentions (lower
antecedents to intentions to quit? While few relevant variable have employed mis-specified
commitment leads to greater models, and their results are suspect (Curry et
intentions to quit which, in turn, would argue against the contention that
al., 1986). Also, correct specification of the roles
further lowers commitment). intention to quit is the immediate antecedent
of variables in absenteeism and turnover
to turnover, considerable diversity of opinion
models requires knowledge of correct causal
exists concerning the antecedents to intention
ordering. From a practical standpoint,
to quit. Several models have postulated job
knowledge of the causal ordering has
satisfaction and organizational commitment to
implications for intervention strategies by
be antecedents of turnover (e.g. Williams and
managers to affect commitment and employee
Hazer, 1986; Farkas and Tetrick, 1989; Arnold
turnover (Curry et al., 1986).
and Feldman, 1982). Clegg (1983) contends that
Second, the notion of job satisfaction and
satisfaction and commitment have been the
organizational commitment being causally
most frequently investigated components of
related has not been incorporated in most
affect with regard to turnover decisions.
turnover models (Farkas and Tetrick, 1989).
Although considerable research effort has
Studies have:
been invested in understanding job attitudes
1 failed to include both satisfaction and
and turnover processes, four major limitations
commitment in their models;
have stifled progress in this area. First, studies
2 neglected to specify causal relationships
Received: November 2000 examining the causal relationship between job
between them; or
Revised/accepted: satisfaction and organizational commitment
3 failed to adequately examine their
March 2001 have been sparse, and their results often
antecedents.
contradictory. For example, Porter et al. (1974)
suggested that satisfaction represents one For example, satisfaction played a critical
specific component of commitment. Later, role in affecting intention to quit in a study
Leadership & Organization Steers (1977) proposed that satisfaction would by Bluedorn (1979), but organizational
Development Journal
22/4 [2001] 159±165
The research register for this journal is available at The current issue and full text archive of this journal is available at
# MCB University Press
[ISSN 0143-7739] http://www.mcbup.com/research_registers http://www.emerald-library.com/ft

[ 159 ]
A.R. Elangovan commitment was not included. Studies by satisfaction and commitment. Another
Causal ordering of stress, Arnold and Feldman (1982) and Michaels explanation for the proposed link would be
satisfaction and commitment, and Spector (1982) included both variables
and intention to quit: a that the employee might rationalize or justify
structural equations analysis but their causal relationship was not his/her intentions to quit by ``discovering''
Leadership & Organization considered. Further causal analyses of the more negative aspects of the job/
Development Journal process leading to intentions to quit are organization, thus subsequently
22/4 [2001] 159±165 sparse (e.g. Bluedorn, 1979; Bedeian and experiencing lower satisfaction and
Armenakis, 1981). The result has been an commitment. In other words, initially job
inadequate understanding of the role attitudes affect intentions to quit, but these
played by satisfaction and commitment in intentions, in turn, might causally affect
affecting turnover intentions, and subsequent job attitudes (while not
confusion regarding the relationship
precluding the continued effect of job
between satisfaction and commitment.
attitudes on turnover intentions). This
Causal analyses of models that include both
implies that apart from the initial time
satisfaction and commitment, and examine
period, there will be a reciprocal relationship
causal relationships between them are
between job attitudes and turnover
required to enhance our comprehension of
the turnover process. intentions. This should be especially visible
Third, the research in this area has also in cross-sectional analyses such as those
been criticized for failing to examine the done by Williams and Hazer (1986) and
antecedents of satisfaction and commitment Bedeian and Armenakis (1981). This proposed
(point (3) above). One factor that has been reciprocal relationship between job attitudes
frequently associated with both work and turnover intentions needs to be tested
attitudes and turnover intentions is job using causal analysis.
stress. Lofquist and Dawis (1969) argued that In summary, the limited causal research
increasing levels of job tension or stress may suggests that employee's satisfaction,
lead to a decision to quit. A study by commitment, and stress are significantly
Sheridan and Abelson (1983) showed that related to turnover intentions. But
when individuals' perceived job tension researchers are yet to systematically explore
exceeded a threshold limit, they quit. these propositions and map the pattern of
Bedeian and Armenakis (1981) found that job- causal relationships among these variables.
induced tension causally affected satisfaction To address these issues, this study examines:
which, in turn, influenced propensity to . the simultaneous causal effects of
leave. There appears to be strong support for satisfaction, commitment and stress on
stress playing an important role in affecting turnover intentions;
intention to quit, but this needs to be . the causal ordering among job
examined in the context of satisfaction and satisfaction, organizational commitment
commitment. Does stress have a direct causal and job stress; and
effect on intention to quit or does it influence . the reciprocal relationship between
turnover intentions by affecting job satisfaction and commitment and
attitudes? If it affects job attitudes, does it intention to quit.
affect satisfaction or commitment or both?
What is the pattern of relationships among Reflecting the differing findings and
the three variables in affecting turnover theoretical viewpoints in prior research, a
intentions? series of models highlighting these
Finally, the role played by turnover differences are tested in this paper. The
intentions in the actual process needs to be intent here is to assess and evaluate different
reconsidered. As mentioned earlier, models, all of which are theoretically well-
intention to quit is probably the most grounded in prior research, rather than to
important antecedent to the turnover solely test one specific viewpoint.
decision. But intentions, once formed, might
also affect the way the individual perceives
the job and organization. There are two Method
possible explanations for this proposed
relationship. First, according to Bem's (1972) Sample
self-perception theory, employees might Data were collected from 155 graduate
perceive/modify their job attitudes based on business students enrolled in a large, public
an awareness of their intentions to quit. This university. Most subjects were part-time
suggests that an employee who becomes students and employed at the time of study.
aware of his/her intention to quit might The subjects had, on average, eight years of
attribute that to low satisfaction and/or full-time work experience and the mean age
commitment, thus, subsequently, reducing was 31.
[ 160 ]
A.R. Elangovan Procedure graphic representation of the model. The
Causal ordering of stress, Subjects were given the option of model consists of two parts:
satisfaction and commitment, the structural model:  ˆ  ‡  ‡ , (1)
and intention to quit: a participating or declining to participate
structural equations analysis before the survey questionnaires were the measurement model: x ˆ x  ‡ , (2)
Leadership & Organization administered to them. They were briefly y ˆ y  ‡ ", (3)
Development Journal informed at the beginning that the study
22/4 [2001] 159±165 Model specifications
pertained to how they felt about their jobs
and the organization they work for. The Measurement model. The x and y matrices
subjects were given the opportunity to clarify in the program contain the specification for
any doubts and ask questions. Finally, they the measurement parameters. Since the
were encouraged to answer all questions latent constructs, both endogenous and
candidly and were assured of complete exogenous ( and ), were measured using
anonymity. single indicators (X and Y variables), the
associated lambda parameters were assigned
Measures a value of 1 to fix the scale. The
Stress. Perceived stress at work was corresponding error terms (in  and "
measured by using the Parker and DeCotiis matrices) were assigned a conservative value
(1983) scale. The reliability analysis revealed of 0.2 which constitutes the upper end of
a coefficient alpha of 0.88 for the 13-item values derived by subtracting the square root
scale. of the reliability coefficients of the various
Satisfaction. A shortened version of the measurement scales from 1. This
Minnesota Satisfaction Questionnaire (MSQ) specification or variations of it are used
when a correlation matrix is analysed with
(Johnson and Weiss, 1971) was used. The
LISREL (e.g. see Curry et al., 1986). Using this
reliability analysis showed a coefficient
strategy provides an estimation of causal
alpha of 0.89 for the scale.
effects among the latent variables controlling
Commitment. The Mowday, Steers and
for errors in measurement.
Porter (1979) commitment scale was used to
Structural model. The and matrices in
measure commitment. The scale had a
the program contain the specifications for
coefficient alpha reliability of 0.92.
the structural parameters. As mentioned
Intention to quit. A modified version of
earlier, the various links or paths in the
Arnold and Feldman's (1982) measure was
model between exogenous constructs and
used. Five items using a seven-point scale
endogenous constructs ( ) and between
ranging from 1 (very low) to 7 (very high)
endogenous constructs ( ) are based on
measured both subjects' intention to change
findings from prior research (discussed
organizations as well search for alternatives.
earlier). None of the structural parameters
Analyses yielded a coefficient alpha
were fixed or constrained but were allowed to
reliability of 0.72 for the scale. be free for estimation.

Model development Model evaluation


The data were analysed using linear Goodness of fit of the estimated model is
structural relations modelling. The assessed by comparing the reproduced
structural equation model specifies the covariance matrix, based on the specified
phenomenon under study in terms of constraints, with the observed covariance
putative cause and effect variables and matrix. LISREL provides indices of fit such
indicators, and is especially useful when as the chi-square statistic, the root-mean-
analysing data from surveys, quasi- square-residual (RMSR), the goodness of fit
experimental designs and longitudinal index (GFI) and the adjusted goodness of fit
studies to identify causal relationships. For index (AGFI). The chi-square (2 ) and
survey data, it represents a statistical probability level (p) tests the likelihood that
approach for drawing out a pattern of causal the covariance matrix estimated is the same
relationships among a set of related as the observed sample covariance matrix
variables. Each equation represents a causal and that the variables are not arbitrarily
link and so parameters do not coincide with correlated. In effect, a small chi-square (non-
regression coefficients. The LISREL program significant chi-square value) is needed so
(Joreskog and Sorbom, 1985) was used to that the null hypothesis fails to be rejected
develop and test the model in this study. The (i.e. proposed model is accepted). The sample
program estimates unknown coefficients in size of 155 ensures that accepting the null
the set of linear structural equations by the hypothesis of no difference is not due to a
maximum likelihood method using the small sample size (small sample sizes of 50
correlation/ variance-covariance matrix of and below increase the possibility of the null
observed variables. Figure 1 shows the hypothesis being accepted) (Bentler and
[ 161 ]
A.R. Elangovan Bonett, 1980). A value for p greater than 0.10 poor fit thus suggests that the specifications
Causal ordering of stress, is usually taken to indicate that a model of Model A are not an accurate (or a good
satisfaction and commitment, adequately fits the data. The RMSR is a enough) representation of the relationships
and intention to quit: a
structural equations analysis measure of the average residuals between the among the variables.
Leadership & Organization observed covariance matrix and the Models B to E were subsequently tested to
Development Journal estimated matrix. The smaller the RMSR determine if they had a good fit to the data.
22/4 [2001] 159±165 The patterns of causal relationships in
relative to the average of the observed
variances and covariances, the better the fit these models appear in Figures 2-5. The
of the model. But GFI represents the goodness of fit estimates from the LISREL
proportion of the variance and covariance analyses are shown below each model. For
explained and so the closer this value is to Model B, although the goodness of fit
1.00, the better the fit. The AGFI adjusts the estimates are acceptable ( 2 = 0.52, df = 1, p
GFI for the degrees of freedom used to = 0.473, RMSR = 0.008, GFI = 0.998, AGFI =
estimate free parameters. T-values indicate 0.983), two major structural parameters
the significance of individual causal paths. (satisfaction ? turnover intentions ( 31)
and commitment ? turnover intentions
( 32)) are inappropriately large and non-
Results significant (indicated by dotted lines in
Figure 2). This is an indication that the
Several models specifying different patterns model has a poor fit to the data. The results
of relationships were tested using LISREL for Models C and D also indicate a poor fit of
analyses to determine the best fitting model. the models to the data (for Model C, 2 =
These models are shown in Figures 1 to 6. 11.34, df = 3, p = 0.010, RMSR = 0.027, GFI =
These models (A to F) are all based on prior 0.968, AGFI = 0.892; for Model D, 2 = 29.99,
research findings and represent the diverse df = 3, p = 0.000, RMSR = 0.121, GFI = 0.912,
standpoints taken by researchers regarding AGFI = 0.708). Further, the stress ?
the relationships among the variables. The satisfaction link ( 11) in Model D was not
pattern of relationships in the first model significant (indicated by a dotted line in
(Model A) is shown in Figure 1. Figure 4). The iterations for Model E failed
The parameter specifications for Model A to converge, indicating that the model
were based on the common notion that specifications do not fit the data (Figure 5).
commitment directly affects turnover The final model tested (Model F) was
intentions, and satisfaction affects turnover based on the discussion in the earlier part
intentions directly as well as through of the paper and on the results of the models
commitment. Stress was also specified as tested above. The specifications of Model F
independently affecting turnover are as follows: stress was hypothesized to
intentions. The results of the analysis for have a negative causal effect on satisfaction
Model A are: 2 = 10.59, df = 2, p = 0.005, ( 11); satisfaction was hypothesized to have
RMSR = 0.122, GFI = 0.968, and AGFI = 0.839. a positive causal effect on commitment
The probability level of 0.005 indicates that ( 21); and commitment and intention to quit
the model fits the data inadequately. In were hypothesized to have reciprocal
addition, the three of the four specified negative causal effects on each other ( 23
relationships failed to attain significance and 32). Figure 6 shows the pattern of
(based on T-values produced by LISREL; causal relationships in the model and
indicated by dotted lines in Figure 1). The results of the LISREL analysis are

Figure 1
Specifications for Model A

[ 162 ]
A.R. Elangovan presented in Table I. The goodness of fit
Causal ordering of stress, estimates indicate an extremely good fit of Discussion
satisfaction and commitment, the model to the data (2 = 0.67, df = 2, p =
and intention to quit: a This study focused on three key issues:
structural equations analysis 0.717, RMSR = 0.012, GFI = 0.998, AGFI = 1 the causal effects of satisfaction,
Leadership & Organization 0.989). All hypotheses were supported ± the commitment and stress on turnover
Development Journal relationships were significant (T-value intentions;
22/4 [2001] 159±165 greater than 2.000) and in the predicted 2 the causal ordering among job
direction. The implications of the findings satisfaction, organizational commitment
are discussed in the next section. and job stress; and
3 the proposed reciprocal relationship
between satisfaction and commitment and
Figure 2
intentions to quit.
Specifications for Model B
The results of LISREL analyses yielded
strong support for all the hypotheses (Model
F). The primary findings of the study are a
strong causal stress ? satisfaction ?
commitment ? turnover intentions link, and
a turnover intentions ? commitment link.
With regard to the role played by stress,
satisfaction and commitment in affecting
turnover, the results indicated that only
commitment directly affected turnover
intentions. Commitment had a very strong
negative effect on turnover intentions (±
0.756), i.e. the lower the commitment, the
higher the propensity for the employee to
leave. Satisfaction and stress did not directly
Figure 3 affect turnover intentions but only through
Specifications for Model C commitment. This finding concurs with that
of Williams and Hazer (1986), and is along the
lines of models proposed by Steers (1977) and
Porter et al. (1974). The finding that stress
does not directly affect turnover intentions
refutes suggestions by Lofquist and Dawis
(1969) regarding a causal relationship
between the variables. Further, the results
imply that models of turnover that do not
include commitment (e.g. Bluedorn, 1979;
Bedeian and Armenakis, 1981) are
misspecified, and that the satisfaction ?
turnover intentions causal link is spurious.
The effect of satisfaction operates through
commitment. From a practical standpoint,
interventions aimed at affecting turnover
need to focus more on influencing
Figure 4 commitment; influencing satisfaction by
Specifications for Model D changing job characteristics, pay, etc. will
not directly affect turnover but will only
serve as one indirect approach for affecting
commitment (among other antecedents of
commitment).
The LISREL results of the various models
also provided information on the causal
ordering among stress, satisfaction and
commitment. There was strong support for a
stress ? satisfaction ? commitment causal
link (Model F). Stress had a moderate
negative effect on satisfaction (±0.299),
implying that higher levels of stress lead to
lower job satisfaction. Stress did not have a
direct causal effect on commitment. Analyses
[ 163 ]
A.R. Elangovan on the causal links from satisfaction and Hazer (1986), models by Steers (1977) and
Causal ordering of stress, commitment to stress were also not Porter et al. (1974), and the commonly held
satisfaction and commitment, position, it refutes findings by Bateman and
and intention to quit: a supported. This implies that the effect of
structural equations analysis stress is primarily on job satisfaction. This Strasser (1984) and Curry et al. (1986). One
Leadership & Organization finding concurs with findings in prior explanation for the differences could be
Development Journal research (e.g. Bedeian and Armenakis, 1981). methodological problems in the Bateman and
22/4 [2001] 159±165 Strasser study (see Curry et al. (1986) for a
The results suggest that models of stress that
do not include job satisfaction as a dependant critique). More causal studies need to be done
variable, and models of job satisfaction that to gain further insight into these conflicting
do not include stress as an antecedent are findings.
misspecified or incomplete. Satisfaction had Perhaps the most interesting finding of the
a strong positive effect on commitment study is the reciprocal link between
(0.538), i.e. the higher the satisfaction, the commitment and turnover intentions. The
higher the commitment. Models specifying a results not only revealed a strong negative
commitment ? satisfaction link (Model D) or effect of commitment on turnover intentions
a reciprocal relationship (Model E) did not fit (discussed above), but also a strong negative
the data. This suggests that satisfaction is a effect of turnover intentions on commitment
causal antecedent to commitment. While this (±0.583). As predicted, the cross-sectional
is in accord with findings by Williams and analysis revealed that higher propensity to
leave further lowered organizational
Figure 5 commitment. But turnover intentions did not
Specifications for Model E have a causal effect on satisfaction. This
suggests that once the employee becomes
aware of his/her intentions to quit, merely
changing jobs within the organization (job
rotation, transfers, etc.) is not likely to stop
him/her, since it is the individual's attitude
towards the organization that is adversely
affected. This also implies that intervention
strategies aimed at retaining good employees
(especially those who are entertaining ideas
of quitting) should focus on enhancing
organizational commitment rather than
``selling'' job attributes or features.
It is important to note some of the
limitations of the current study, especially
the cross-sectional nature of the data and the
sample characteristics. Although structural
equations analyses are intended for and very
Figure 6 effective in establishing causal relationships
Specifications for Model F among a comprehensive set of variables, it
remains a statistical approach for
determining causality, not a temporal one.
Future research should employ dynamic
longitudinal designs to obtain a more
accurate picture of the causal relationships.
Longitudinal data for a newly recruited work
group will clearly reveal the initial
commitment ? turnover intentions link, and
the subsequent reciprocal relationship. Also,
this study used single indicators for the
latent constructs ( and ); using multiple
indicators for each latent variable would
increase the methodological rigour of the
analysis. Antecedents other than stress
Table I
should be included to obtain a more
Results of the LISREL analysis for Model F
comprehensive picture of the pattern of
11 ±0.299 21 0.538 relationships.
23 ±0.583 32 ±0.756 Lastly, although the nature of the sample
Notes: 2 = 0.67; df = 2; p = 0.717; RMSR = 0.012; in this study suggests good generalisability of
GFI = 0.998; AGFI = 0.989 the findings, it also highlights some
constraints. The respondents were primarily
[ 164 ]
A.R. Elangovan students in a part-time, graduate-level study'', Journal of Applied Psychology, Vol. 68,
Causal ordering of stress, business program, employed full-time in pp. 88-101.
satisfaction and commitment, various organizations at the time of the study Curry, J.P., Wakefield, D.S., Price, J.L. and
and intention to quit: a Mueller, C.W. (1986), ``On the causal ordering
structural equations analysis with an average of eight years of work
of job satisfaction and organizational
Leadership & Organization experience. While this sample is a suitable
commitment'', Academy of Management
Development Journal one for the current study, it would be useful
22/4 [2001] 159±165 Journal, Vol. 29 No. 4, pp. 847-58.
to further explore the role of age and Farkas, A.J. and Tetrick, L.E. (1989), ``A three-
organizational tenure in affecting job wave longitudinal analysis of the causal
attitudes and turnover intentions to improve ordering of satisfaction and commitment on
generalisability. Prior research has noted turnover decisions'', Journal of Applied
that older and longer-serving employees are Psychology, Vol. 74 No. 6, pp. 855-68.
often more satisfied and committed than Hinkin, T.R. and Schriesheim, C.A. (1989),
younger employees. Having a wider range of ``Development and application of new scales
age and tenure in the sample ± the average to measure the French and Raven (1959) bases
of social power'', Journal of Applied
age of the respondents in the current sample
Psychology, Vol. 74 No. 4, pp. 561-7.
was only 31 ± might reveal this relationship.
Johnson, D.A. and Weiss, D.J. (1971), Middle
Similarly, future research should include the Management, Decision-making and Job
impact of the nature of the industry and the Satisfaction, Educational Research and
economy to make the findings more development Council, St Paul, MN.
generalisable. The competitiveness and the Joreskog, K. and Sorbom, D. (1985), LISREL:
growth potential of any industry and the Analysis of Linear Structural Relations by the
general economic conditions significantly Method of Maximum Likelihood, National
affect alternate employment options and offer Educational Resources, Chicago, IL.
reference points for comparison which, in Lofquist, L. and Dawis, R. (1969), Adjustment to
turn, can influence job attitudes and Work, Appleton Century Crofts, New York,
NY.
turnover intentions. Overall, however, this
Michaels, E. and Spector, E. (1982), ``Causes of
study represents a useful step towards employee turnover: a test of the Mobley,
understanding the complex nature of the Griffeth, Hand, and Meglino model'', Journal
relationships among work attitudes and of Applied Psychology, Vol. 67, pp. 53-9.
employee turnover intentions. Mitchel, J.O. (1981), ``The effect of intentions,
tenure, personal, and organizational
References variables on managerial turnover'', Academy
Arnold, H. and Feldman, D. (1982), ``A of Management Journal, Vol. 24, pp. 742-51.
multivariate analysis of the determinants of Mobley, W.H., Griffeth, R.W., Hand, H.H. and
job turnover'', Journal of Applied Psychology, Meglino (1979), ``Review and conceptual
Vol. 67 No. 3, pp. 350-60. analysis of the employee turnover process'',
Bateman, T.S. and Strasser, S. (1984), ``A Psychological Bulletin, Vol. 86, pp. 493-522.
longitudinal analysis of the antecedents of Mowday, R.T., Steers, R.M. and Porter, L.W.
organizational commitment'', Academy of (1979), ``The measurement of organizational
Management Journal, Vol. 27, pp. 95-112. commitment'', Journal of Vocational
Bedeian, A.G. and Armenakis, A.A. (1981), ``A Behavior, Vol. 14, pp. 224-47.
path-analytic study of the consequences of Parker, D.F. and DeCotiis, T.A. (1983),
role conflict and ambiguity'', Academy of ``Organizational determinants of job stress'',
Management Journal, Vol. 24 No. 2, pp. 417-24. Organizational Behavior and Human
Bem, D. (1972), ``Self-perception theory'', in Performance, Vol. 32, pp. 160-77.
Berkowitz, L. (Ed.), Advances in Experimental Porter, W., Steers, M., Mowday, T. and Boulian,
Social Psychology, Vol. 6, Academic Press, V. (1974), ``Organizational commitment, job
New York, NY. satisfaction, and turnover among psychiatric
Bentler, P. and Bonett, D. (1980), ``Significance technicians'', Journal of Applied Psychology,
tests and the goodness of fit in the analysis of Vol. 59, pp. 603-9.
covariance structures'', Psychological Sheridan, J.E. and Abelson, M.A. (1983), ``Cusp
Bulletin, Vol. 88, pp. 588-606. catastrophe model of employee turnover'',
Bluedorn, A.C. (1979), ``Structure, environment, Academy of Management Journal, Vol. 26
and satisfaction: toward a causal model of No. 3, pp. 418-36.
turnover from military organizations'', Steers, M. (1977), ``Antecedents and outcomes of
Journal of Military and Political Sociology, organizational commitment'', Administrative
Vol. 7, pp. 181-207. Science Quarterly, Vol. 22, pp. 46-56.
Bluedorn, A.C. (1982), ``A unified model of Williams, L.J. and Hazer, J.T. (1986),
turnover from organizations'', Human ``Antecedents and consequences of
Relations, Vol. 35, pp. 135-53. satisfaction and commitment in turnover
Clegg, C.W. (1983), ``Psychology of employee models: a reanalysis using latent variable
lateness, absence, and turnover: a structural equation methods'', Journal of
methodological critique and an empirical Applied Psychology, Vol. 71 No. 2, pp. 219-31.

[ 165 ]

Potrebbero piacerti anche