Job Satisfaction a Meta-Analysis of Stabilities

Job Satisfaction: A Meta-Analysis of StabilitiesAuthor(s): Christian Dormann and Dieter ZapfSource: Journal of Organizational Behavior, Vol. 22, No. 5 (Aug., 2001), pp. 483-504Published by: John Wiley & SonsStable URL: http://www.jstor.org/stable/3649554Accessed: 21/06/2010 04:23

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available athttp://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and youmay use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained athttp://www.jstor.org/action/showPublisher?publisherCode=jwiley.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

John Wiley & Sons is collaborating with JSTOR to digitize, preserve and extend access to Journal ofOrganizational Behavior.

http://www.jstor.org

http://www.jstor.org/stable/3649554?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp

http://www.jstor.org/action/showPublisher?publisherCode=jwiley

Journal of Organizational Behavior J. Organiz. Behav. 22, 483-504 (2001) DOI: 10.1002/job.098

Job satisfaction: a meta-analysis of stabilities

CHRISTIAN DORMANN* AND DIETER ZAPF Johann Wolfgang Goethe-University of Frankfurt, Institute of Psychology, Mertonstr. 17, D-60054 Frankfurt/Main, Germany

Summary Evidence suggesting that job satisfaction is caused by individual dispositions is reviewed, and stability coefficients for job satisfaction in previous studies are analysed with a meta-analytic procedure. Previous longitudinal studies analysing job changer samples imply an upper limit estimate of 0.51 for direct dispositional influences on job satisfaction. A study of job changers considering the stability of working conditions suggests that this estimate has to be considerably corrected downwards. At present, it is concluded that it is more likely that dispositions indirectly affect job satisfaction via selection and self-selection processes. Implications for job satisfaction as a tool for organizational assessment are discussed. Copyright ? 2001 John Wiley & Sons, Ltd.

Introduction

Job satisfaction is one of the best-researched concepts in work and organizational psychology for at least two reasons. Job satisfaction is relevant for all those who are interested in the subjective evaluation of working conditions such as responsibility, task variety, or communication requirements (e.g., Hackman and Oldham, 1980) because job satisfaction is supposed to be strongly caused by such conditions. Job satisfaction is also of major concern whenever outcome variables such as absenteeism (e.g., Breaugh, 1981; Keller, 1983; Tharenou, 1993), fluctuation (e.g., Farkas and Tetrick, 1989; Rusbult and Farrell, 1983), or organizational inefficiency such as counterproductive behavior (e.g., Gottfredson and Holland, 1990) or sabotage (Chen and Spector, 1992), are dealt with because job satisfaction is supposed to be a major cause of such problems. By integrating these two perspectives job satisfaction is placed as a central concept in work and organizational psychology, which mediates the relation between working conditions on the one hand and organizational and individual outcomes on the other hand.

Although intuitively convincing, working conditions as the major cause of job satisfaction have been challenged. One counterargument emphasizes the role of personality variables. Staw and colleagues (Staw et al., 1986; Staw and Ross, 1985) suggested that job satisfaction might reflect stable underlying

* Correspondence to: Christian Dormann, Johann Wolfgang Goethe-University of Frankfurt, Institute of Psychology, Mertonstr. 17, D-60054 Frankfurt/Main, Germany. E-mail: [email protected]

Received 1 June 1999 Accepted 30 August 2000

Copyright ? 2001 John Wiley & Sons, Ltd. Published online 26 June 2001

484 C. DORMANN AND D. ZAPF

dispositions which might be genetically determined. Obviously, this supposition challenges the use of job satisfaction for the assessment of work and organization.

Despite the diverse plausible mechanisms which have been suggested, the significance of personality dispositions in the development of job satisfaction has been subjected to much criticism and inspired an intensive discussion in the literature during recent years (e.g., Arvey et al., 1989, 1993; Bouchard et al., 1992, 1990; Cropanzano and James, 1990; Davis-Blake and Pfeffer, 1989; Gerhart, 1987; Gutek and Winter, 1992; Judge and Hulin, 1993; Keller et al., 1992; Levin and Stokes, 1989; Newton and Keenan, 1991; Staw et al., 1986; Staw and Ross, 1985; Watson and Slack, 1993). The central issue raised with the assumption of dispositional influences on job satisfaction is concerned with the extent to which individuals' job satisfaction can be changed by organizational measures. If the malleable parts of job satisfaction were only small and stable personality traits were the major causes, a variety of conclusions would have to be drawn. Measuring job satisfaction for organizational assessment, for example, work design or organizational climate, would be questionable. Instead, based on the trait-like character of job satisfaction, one would tend to follow suggestions to use individuals' job satisfaction in personnel selection procedures (Staw and Ross, 1985) because highly satisfied people would be also satisfied in the future. Turnover, absenteeism, and other factors of organizational inefficiency might be affected by this (Carsten and Spector, 1987; Brayfield and Crockett, 1955; Herzberg et al., 1957; Mobley et al., 1978; Nicholson et al., 1976; Vroom, 1964). Thus, it is of great importance to know how strong dispositional influences are in comparison to situational determinants such as working conditions when it comes to job satisfaction.

Several explanations have been given for the relation of job satisfaction and personality traits. Job satisfaction may be affected by emotion-related personality traits because job satisfaction has been equated with a pleasurable emotional state (e.g., Locke, 1969, 1976). Recent theorizing on the dispositional influences on job satisfaction has mainly focused on negative affectivity (NA) and to a lesser extent on positive affectivity (PA) (e.g., Brief et al., 1988; Brief and Roberson, 1989; Munz et al., 1996). Negative affectivity has been sometimes equated with neuroticism (Burke et al., 1993; Watson and Clark, 1984), and it has been interpreted as a general dimension which lowers the threshold to experience negative emotions. Similarly, PA has been defined as a dimension which increases the likelihood to experience positive emotions. Opposed to such internal psychological processes, it can also be assumed that personality has an impact on job satisfaction via influencing the objective working conditions (e.g., Hulin, 1991). Personality traits are relevant for job choice and for being selected and promoted by the organization (Hogan, 1991). Mental ability and personality tests are frequently used in personnel selection. A study by Spector et al. (1999), for example, showed that NA is more strongly correlated with non-incumbent (job analyst and supervisor) measures of job characteristics than with incumbent measures. In a laboratory and in field studies Cook et al. (1995 - paper presented at the SIOP convention, Lake Buena Vista, FL) showed that people scoring high in NA are less likely to successfully complete selection interviews. These studies suggest that NA, PA, and other personality dispositions such as extraversion, openness or intelligence affect which job a person gets and, by this, affect the working conditions. The working conditions in turn affect job satisfaction. In other words, the effect of personality dispositions on job satisfaction is mediated by working conditions. This may either take place via self-selection and career decisions made by the individual or by selection and promotion by the organization. In contrast to other mechanisms, the usefulness of job satisfaction for evaluation purposes is not threatened if selection due to personality dispositions applies because job satisfaction is a reaction to working conditions. Even if individuals with certain dispositions are exposed to bad working conditions, working conditions could be improved independently of these dispositions leading to higher levels of job satisfaction.

In this article, we first provide a brief overview of results obtained from different approaches used in studies on dispositional causes of job satisfaction to provide an estimate of the effect of personality

Copyright ? 2001 John Wiley & Sons, Ltd. J. Organiz. Behav. 22, 483-504 (2001)

STABILITY OF JOB SATISFACTION 485

dispositions on job satisfaction. Second, we present a meta-analysis of stabilities obtained from longitudinal studies. This analysis comprises a great variety of different occupations and organizations, so the results may apply to the overall population. We distinguish between job changers and job stayers and argue that stability coefficients of job changers represent evidence for the upper-limit of the dispositional influence on job satisfaction. Third, we investigate the indirect effects of personality dispositions on job satisfaction via working conditions. Our aim is to demonstrate that stability of job satisfaction may arise because of the stability in working conditions even when job changes occur. Therefore, we apply a partial correlation approach to a longitudinal sample of job changers. It is hypothesized that a substantial part of dispositional influence on job satisfaction is indirect through the influence on job conditions caused by selection. The reduction in job satisfaction stability after partialling job conditions should reflect the indirect effects of personality in contrast to its direct effects. The general aim of the present article is to provide a basis to decide whether job satisfaction should be either used for subjective evaluation of the organization or as a personality variable which might be useful elsewhere, for example, in employee selection, or both.

Approaches to the Dispositional Basis of Job Satisfaction

Research on the role of personality variables as determinants of job satisfaction can be distinguished according to their use of a direct or indirect approach. Whereas direct approaches tried to explicitly identify certain dispositions as determinants of job satisfaction, indirect approaches were used to demonstrate that some unspecified disposition to be satisfied or dissatisfied with the job is likely to exist.

Indirect approaches

In a longitudinal study, Schneider and Dachler (1978) reported an average test-retest correlation of the job description index (JDI; Smith et al., 1969) satisfaction-measure of 0.57 over a time interval of 16 months. Assuming a reliability of 0.84 and correcting this estimate for attenuation (which was the average reliability obtained from the meta-analysis of longitudinal studies presented below) implies a corrected stability of 0.68. This result tempted some researchers to draw the conclusion that the reason for the high stability of job satisfaction cannot be found in working conditions (e.g., Staw and Ross, 1985) because the latter should be expected to change considerably. Taking into account that American employees change jobs frequently (Wegemann, 1991), such a high test-retest correlation may undercut the idea that job satisfaction is primarily caused by working conditions. Rather, it may suggest that traits or stable individual dispositions are responsible for satisfaction with the job.

Although a particular disposition that may cause job satisfaction cannot be derived from our research, test-retest correlations allow us to estimate the maximum dispositional effect on job satisfaction. Since dispositions are supposed to be stable at least across short time periods, their influence should be almost constant over time. Consequently, dispositional variance inherent in job satisfaction should covary across time. As Gerhart has noted '... an observed relation between previous and current job satisfaction should perhaps be viewed as an upper bound on the total effect of traits on job satisfaction.' (1987, p. 371). The test-retest correlation thus provides an upper limit estimate of the variance attributable to those causal factors that are stable over time. It should be noted that the



correlation directly corresponds to the maximum effect of an underlying stable factor and thus should not be squared (Bouchard et al., 1990; Bouchard, 1997).

The test-retest correlation usually does not provide an unbiased estimate of the actual dispositional effect because there are other variables which also remain constant over time to some degree. Only if every non-dispositional cause of job satisfaction had completely changed between two measurements, test-retest correlations would be unbiased estimates of the absolute dispositional effect. This condition is partly met if individuals change their occupation and/or their employer, as suggested by Staw and Ross (1985). Moreover, long time lags between subsequent measurements also contribute to this condition because the probability of changes in work and non-work conditions increases with time. How- ever, it can be supposed that people will not be spread over jobs at random after a job change. Rather, a variety of working conditions will remain similar to those before the change because people remain in their occupation. Nevertheless, job conditions should be less stable in samples of job changers. The test-retest correlations for job satisfaction obtained under such conditions (i.e., change of job and employer) reported by Staw and Ross (1985) was 0.19 across five years. Assuming that the one item-measure applied has a reliability of 0.57 (which was the average estimate of the minimum reliability obtained by Wanous et al. in their recent meta-analysis, 1997), and correcting the test-retest correlation for attenuation led to an estimate of 0.33 for the maximum effect for dispositional influences. This estimate is only about one half of the corresponding measure (0.65) which was obtained for subjects who neither changed occupation nor employer during the same time.

Given the widespread use of job satisfaction measures and the increased availability of longitudinal studies, the bulk of evidence relevant for the discussion of dispositional influences on job satisfaction can certainly be found in this line of inquiry. Collecting all available information on the stability of job satisfaction in order to provide a more accurate estimate of the maximum dispositional effect thus seemed to be a promising avenue to pursue. We present a comprehensive meta-analysis of stabilities in Study 1.

Another kind of indirect evidence for dispositional influences emerges from twin studies. In 1989, Arvey and his colleagues demonstrated that monozygotic twins who were reared apart for most of their lives shared about 31 per cent of their variance in general and intrinsic job satisfaction. Intrinsic aspects of the job are directly related to the tasks (e.g., skill variety), whereas extrinsic aspects are related to external circumstances (e.g., promotion opportunities). For extrinsic satisfaction, no common variance among twins was found. This result was replicated partly by Arvey et al. (1993), although correlations in job satisfaction scores for dizygotic twins were lacking in this study. Nevertheless, these results suggest a search for traits that are known to be genetically determined.

Direct approaches

Based on the definition of job satisfaction as a pleasurable emotional state resulting from the appraisal of one's job (e.g., Locke, 1969, 1976), many authors considered dispositions related to the experience of positive (PA) and negative emotions (NA) as good candidates to affect job satisfaction and dissatisfaction.

First, there are several cross-sectional studies estimating the relation between affectivity and job satisfaction. The strongest effect for NA was found in the study of Munz et al. (1996), where the corrected common variance between general job satisfaction and NA was 21 per cent. In the same study, the effect for PA was 30 per cent. Brief and Roberson (1989), after correcting for measurement error, reported the common variance of NA with two measures of job satisfaction to be 14 per cent on average. For PA the average corrected common variance was 34 per cent. Levin and Stokes (1989) reported an intermediate corrected effect of 12 per cent for NA. Somewhat lower associations were found by Watson and Slack (1993) who reported 14 per cent corrected common variance for PA



with job satisfaction and 8 per cent for NA. Similarly, approximately 8 per cent corrected common variance was obtained by Staw et al. (1986).

Second, there are longitudinal studies that tested whether job satisfaction can be predicted by affect variables measured long before. Although such designs do not actually prove the causal role of affective dispositions, evidence obtained from such studies is more trustworthy than cross-sectional ones. Longitudinal studies were carried out by Gustavsson et al. (1997), Staw et al. (1986), and Watson and Slack (1993). In the study of Staw et al. the 50-year lagged corrected common variance of affective disposition and job satisfaction was 20 per cent, assuming that their overall satisfaction measure had a reliability of 0.80. However, in this study, the synchronous corrected common variance was only 8 per cent. It can be supposed that the most important selection processes take place in early adulthood when the individual him- or herself and others decide on the individual's future career. Therefore, the fact that the common variance of prior (childhood) affective disposition and job satisfaction was larger when compared to the concurrent (adult) affective disposition supports the selection hypothesis. Gustavsson et al. used neuroticism factor-scores to predict job satisfaction nine years later. Assuming a reliability of 0.73 for neuroticism, which was the average reliability of the nine scales loading on this factor, 12 per cent corrected common variance emerged for two cross-validation samples, on average. In the two-year prospective study of Watson and Slack, the squared time-lagged correlation (corrected for attenuation) in job satisfaction explained by negative emotionality was 0.01 and 0.13 for positive emotionality.

Even stronger evidence for a possible causal role of affective disposition comes from longitudinal analyses which test whether the stability of job satisfaction breaks down when affectivity is controlled, as suggested by Judge (1992). This was tested by Schaubroeck et al. (1996) in a seven-year follow-up. For employees who did not change their job, the average stability of various facet satisfaction scales was reduced from 0.34 to 0.33 when NA and PA were controlled, and for employees who changed their job the average reduction was from 0.25 down to 0.23. Obviously, the reduction obtained was so small that the results question the role of NA and PA as underlying dispositions of job satisfaction.

Although most of the direct approaches dealt with NA or PA, there have been claims to include other traits in dispositional research (e.g., House et al., 1996). Recently, Judge et al. (1998) suggested that NA, and possibly PA, represent aspects of a core self-evaluation factor which also comprises self- esteem, generalized self-efficacy, and locus of control. In two samples, the authors showed that the core self-evaluation factor was indeed related to job satisfaction. A scale by scale analysis revealed that the strongest associations existed for self-esteem (the corrected common variance was 26 per cent), which the authors think of as '... the source (or an important source) of positive affectivity' (Judge et al., 1998; p. 19). Since PA was not directly measured, the results of Judge et al. are thus still in line with the notion that affective dispositions seem to be most relevant. Noteworthy, the weakest association among the four variables studied by Judge et al. was found for NA, for which the corrected common variance was 14 per cent (in line with the selection hypothesis, Judge et al. (2000) recently showed that at least part of the relation between core self-evaluations and job satisfaction is mediated by job characteristics).

Summary of previous findings

The indirect approaches of Staw and Ross (1985) and Arvey et al. (1989) indicate that the maximum dispositional effect can be expected to be some 30 per cent. This approach, which combines the effects of all possible personality traits, shows higher estimates than those typically found in more direct approaches which test the influence of certain personality variables. In most instances, direct



approaches explain between 10 and 20 per cent of variance in job satisfaction, and NA seems to be less important than PA.

In the following section, we first present the results of a meta-analysis (Study 1). This study repre- sents an indirect approach because it summarizes stabilities of job satisfaction reported in previously published articles. If stable organizational variables contribute to the stability of job satisfaction, then

changes of these organizational variables should lower the stability of job satisfaction. Such changes are more likely to occur if employees change their jobs. Thus, we supposed that test-retest correlations reported in the literature should be smaller for job changers than for job stayers.

The foregoing supposition rests on the assumption that job changers actually experience changes in

working characteristics, which in turn affect their job satisfaction level. However, the stabilities of some of the situational factors before and after a job change are often underestimated. Of course, there are factors such as organizational climate or leadership, which may fundamentally change in a given sample. For example, persons being badly treated by their supervisors may succeed in finding a new job under brilliant supervision, thus moving from the very bottom to the very top of this dimension. However, this is unlikely for job content and job stressors. Job content is closely related to job skills. Usually, people acquire skills for a specific occupation and skills within an occupation are typically more similar than skills between occupations. In most cases, people remain in their occupation after a job change. If not, they try to get a job with a similar or a somewhat better job content. It should also be noted that job content variables usually show the highest relations with job satisfaction compared with other organizational variables such as leadership or payment (Dawis et al., 1974; Zapf, 1991). As is the case with job content, almost the same is true for job stressors, although these may show more variation in comparison with job content. An example is a firefighter who changes his or her job and moves to another fire brigade. Time pressure or job hazards will probably not change to a large extent. That is, some of the organizational variables are sufficiently controlled when using a job changer sample. How- ever, other variables, in particular job content and job stressors, may remain stable even after a job change. These factors may be responsible for similar levels of job satisfaction before and after the job change. To our knowledge, studies controlling for organizational variables in job changer samples have not yet been carried out. Therefore, we decided to test in a second study (Study 2) whether partialling job characteristics would result in reduced test-retest correlations of job satisfaction for job changers. This analysis was based on data from a German multi-wave study.

Study 1: A Meta-Analysis of Stabilities of Job Satisfaction

Method

A computer-based literature search on the stability of job satisfaction was conducted in order to identify as much empirical information as possible. This search was applied to the electronic database PSYCLIT up to September 1997. Search terms were 'satisfaction and (stability or retest or panel or longitudinal or follow up) and (organization or job or work or management or business)'. References given in these articles were also taken into account, leading to 45 articles providing information on job satisfaction over the course of time.

Recorded information We intended to investigate overall job satisfaction, which is often measured using single item measures or particularly designed scales. On the other hand, there are measures that assess different facets of job satisfaction, such as the JDI (Smith et al., 1969). However, '... JDI total does have meaning and can



be used as a global measure of satisfaction' (Wanous, 1974, p. 140). Regarding the Minnesota satisfaction questionnaire (MSQ; Weiss et al., 1967), or other facet-based measures, a similar argument can be made. Therefore, we either collected information on global job satisfaction or aggregated scales of facet satisfaction.

Test-retest correlations, r-square increments from hierarchical regression analysis where Time 1 job satisfaction was exclusively entered in the first step predicting Time 2 satisfaction, and mean levels at two points in time together with the corresponding standard deviations and the test statistic (e.g., t-values, F-ratios, or p-values with at least two decimals) were recorded. The latter two kinds of information are useful because r-square increments and differences in mean levels over time can be transformed into a correlation.

Twelve articles dealt with issues which required the formation of subsamples, for example, Type A versus Type B behavior, job stayers versus leavers, etc. Sometimes information about the stability of job satisfaction was reported for both the sample as a whole and the subsamples. In such cases we recorded the information about the overall sample to avoid the problem of dependent samples, with the exception of studies where subsamples for stayers and leavers were reported. Thus, our data basis consists of 60 samples taken from 42 studies.

Some of the studies found were multi-wave studies comprising more than two measurements. Since we were interested in computing an upper limit estimate of dispositional influences on job satisfaction, it was reasonable to choose the test-retest correlation based on the longest time lag available. Besides information about stability, we also recorded time lags and reliabilities.

Calculations All available information was algebraically transformed into test-retest correlations. Not all of the studies reported information on the reliability of job satisfaction scales. When no information about reliability was reported, we estimated the reliability using the approach implemented in the meta-analytic procedure of Raju et al. (1991). In other words, the approach of Raju et al. allows for correction of test-retest correlations for unreliability in the job satisfaction measure (at both measurement periods) by taking sampling error in the sample-based reliability estimates into account. We did not correct for range restriction because no information about this was available, which may have led to slightly conservative estimates. We used the computer program MAIN by Raju and Fleer (1997). The corrected test-retest correlations include the amount of dispositional variance plus the amount of variance due to lack of change in environmental conditions. Thus, the test-retest correlations are actually upper-limit estimates of dispositional variance because other sources of stability are also included.

Results

In Figure 1 the corrected test-retest correlations of job satisfaction across time are shown. In the top part of Figure 1 all samples identified in the literature are included (overall N= 14 944), whereas the middle part of Figure 1 comprises only stayer samples (N= 3512), and the bottom part of Figure 1 shows only samples where employees changed their job, the organization, or both (N= 4033). Unfor- tunately, the number of samples where employees changed either employer (n = 2) or their job (n = 6) was rather small so we did not consider this distinction further.

Table 1 shows the meta-analytic results. The mean correlation (Mr) for all samples was 0.42. The sample-weighted corrected test-retest correlation (Md) was 0.50 (for an average sample-weighted time-lag of 35.89 months). The standard error of this mean was 0.01, resulting in a 95%-confidence interval of 0.44 up to 0.48. The 95% upper credibility value, below which 95% of all true stabilities lie, was 0.85.



The top part of Figure 1 shows that stabilities decrease when time lags increase (r = - 0.41). The amount of unexplained variance remaining (83 per cent) suggests that there might be differences in the development of the various satisfaction scales over time. We could not analyse this possibility because with the exception of the MSQ (4 times) and the JDI (14 times), most other satisfaction measures were only rarely used. Although global job satisfaction measured by one item was used in most cases (18 times), equivalence among the 1-item measures was unclear to a great extent because the verbatim wording was often not reported. Another possibility might be that the samples analysed comprised jobs with varying degrees of stability. For example, working conditions for foresters might be more stable across time than working conditions in the high-tech industry. Again, there are so many diversities in the samples analysed that we were unable to test for this possibility. The next analysis dealt with differences in the stability of job satisfaction between job stayers and job leavers. Since not all studies found in the literature reported whether or not subjects changed, the number of stayer samples was only 19. The mean sample-weighted corrected test-retest correlation (Md) for stayers was 0.48. For changers, M? was 0.35 (see Table 1). In line with Finkelstein et al. (1995), we used standard normal deviate comparisons to test the differences between the mean corrected test-retest correlations using the formula given in Quifiones et al. (1995). The difference between these two estimates was significant (z = 5.36, p < 0.01).

A look at the middle and bottom part of Figure 1 reveals that for job stayers and for job changers only moderate relations between time lag and stability emerged. For job stayers the relationship was slightly negative as expected (r = - 0.20), whereas, contrary to expectation, the relation was slightly positive when job changers were considered (r = 0.22). However, because of the very small sample size, we will not interpret this result.

All Samples (N= 60) 1,1

1.0 a

S.3 Time Lag (Months)

~ 1

o

Time Lag (Months)

Figure 1. Corrected test-retest correlations (ps) of job satisfaction for varying time lags separately for all samples, samples where participants stayed within their jobs, and samples where participants changed their jobs or their employer between two measurements



Pure Stayer Samples (N = 19) 1,1

1,0

,9 _

o ,8

c ,7

o S Q 4

4W,

43

-,3

Time Lag (Months)

Changer Samples (N= 10) 1,1

-.3

1,1

1,0 26a a o

.8

u ,13

-.2 -,3

0 12 24 3 40 60 84 96 108 120 132 144

Time Lag (Months)

Figure 1. Continued.



Table 1. Meta-analytic results for the stability of job satisfaction

Observed statistics Estimated population parameters

Condition N No. of Mr SDr M0 SD, SEMi 95% 95% CI effects upper CV around M6

All samples 14.944 60 0.42 0.22 0.50 0.21 0.01 0.85 0.44 to 0.48 Stayer samples 3.512 19 0.42 0.17 0.48 0.23 0.02 0.86 0.45 to 0.51 Changer samples 4.033 10 0.18 0.15 0.35 0.14 0.02 0.58 0.31 to 0.38

Note: SEM, is an asymptotic standard error of the estimated mean of p. CV = Credibility value. CI = Confidence interval.

As already noted, the sample-weighted corrected ps for stayers and leavers were 0.48 and 0.35, respectively. Note that these values are conservative. They may even be higher if range restriction is taken into account, supporting the position that one should not be too optimistic when assuming that job satisfaction could be easily changed. However, this estimate is based on stable environmental factors as well, and the question arises as to what degree test-retest correlations are lowered when working conditions are partialled. Using a direct longitudinal approach, this issue was addressed in Study 2.

Study 2: Partialling Work Characteristics From Test-retest Correlations of Job Satisfaction

It can be assumed that even for those employees who changed their job or their employer, working conditions of the new job or in the new organization are not completely different from those before. People are likely to search for jobs that fit their qualification, their needs, and their job expectations. Therefore, they often remain in the same branches and hold a job in the same occupation. For example, high-qualified jobs give more control to employees. Control at work should thus be correlated even if people change their jobs, and, if control affects job satisfaction, a stability of job satisfaction should emerge because control as one underlying cause does not completely change over time. If causes of job satisfaction are partialled from the stability of job satisfaction, this stability should be reduced. This argument is similar to partialling personality from the stability of job satisfaction, which should result in a significant decrease. However, as already noted, previous results (Schaubroeck et al., 1996) failed to demonstrate that partialling NA and PA as personality characteristics reduced the stability of job satisfaction. Partialling job characteristics takes the opposite point of view and was applied in the present study.

Method

The data analysed were collected in the AHUS-project. AHUS is a German acronym for 'active behavior in a radical-change situation.' The AHUS-project has been carried out in former East Germany and comprises six waves of data collection between 1990 and 1995. The general purpose of the study was to analyse how working conditions change as a consequence of the unification of West and East Germany in 1990. In this context, job satisfaction was also measured. Other parts of the project, which did not focus on job satisfaction, were published in Dormann and Zapf (1999), Frese et al. (1997), Frese et al. (submitted), and Frese et al. (1996).



The six waves took place in July/August 1990 (when West German currency was introduced into East Germany), in October/November 1990 (a few weeks after the unification of East and West Germany), in July/August 1991, July/August 1992, July/August 1993, and July/August 1995.

Contextual Sidebar

Dresden and the unification of East and West Germany

Dresden until 1989 Dresden was one of the biggest cities in the German Democratic Republic (GDR). It was founded about 1200. It rapidly grew and had its maximum of about 630 000 inhabitants in 1939. In 1945, shortly before the end of World War II, Dresden was heavily bombed by British and U.S. American air forces resulting in about 350 000 victims. In October 1949, the GDR was founded on the grounds of the former Soviet zone of occupation. In 1968, a socialistic constitution was established, which stressed the exclusive leadership of a single party termed Sozialistische Einheitspartei Deutschlands (SED). After foundation of the GDR, Dresden became capital of the district of Dresden, one out of 15 districts in the GDR. In 1990, Dresden had about 500 000 inhabitants and it was the third biggest city in the GDR.

In the period up to 1989 the largest enterprises in Dresden were a power plant, a transformer and x-ray enterprise, enterprises in the electrical engineering branch and the electronics and microelectronics sectors, typewriter works, Pentacon (cameras), Nagema (a firm producing machinery for the foods and related industries), a high-vacuum plant, pharmaceuticals, air-conditioning and refrigera- tion plants.

The unification Stimulated by the perestroika politics in the USSR, the civil rights movement in the GDR received more and more word of encouragement in the population, resulting in mass demonstrations in 1989. The people forced the opening of the frontiers. In November 1989, the socialistic government col- lapsed and Berlin's wall breached. The elections in March 1990 were won by parties that were look- ing forward to an unification of the GDR and the Federal Republic of Germany (FRG), which has had a social market economy and a political system that has been strongly devoted to democratic principles. The unification process began in July 1990 with the introduction of a common currency system. The unification process was finalized in October 1990, when the newly founded Federal States of the GDR joined the FRG.

After an initial period of enthusiasm following the unification, East Germans (i.e., the former GDR) became more and more disillusioned. Before the unification, the work in the GDR was poorly organized; supplies were often missing, tools were of poor quality and qualifications often counted less than political correctness. Supervisors usually prescribed in detail how the work had to be done, but they had little sanction power. After the unification, a lot of East German firms were acquired by West German and international companies. East German supervisors were often replaced by West Germans, western business rules were quickly adopted, and a lot of employees were fired. In the GDR there was virtually no unemployment, but as early as in 1991 there were mass demonstrations against unemployment in Leipzig, the second biggest city in the former GDR.

Copyright ( 2001 John Wiley & Sons, Ltd. J. Organiz. Behav. 22, 483-504 (2001)


Dresden since 1990 About 45 billion 4 have been invested in Dresden since 1990. The number of businesses has risen from 12000 to 34600. The industrial diversity has facilitated the transformation of economic structures since the unification, but nevertheless this was difficult. Even those industries that had the potential to compete in a world market, such as some of the electrical and electronic industries, were vastly overstaffed in terms of West German standards. However, for some branches the transformation was successful. Today the most important branches are the high-tech sectors with microelectronics and associated applications, telecommunications and information technology. There are currently around 400 companies with more than 15 000 employees operating directly in microelectronics or in affiliated branches (e.g., Infineon Technologies and Advanced Micro Devices). In addition, there are 15 independent research institutes and more than 150 software offices. Also very important are the pharmaceutical sector, environmental engineering and bio- technologies, precision engineering and apparatus, opitcal engineering and cameras, aircraft engineering, medical technology, fine chemicals, printing and publishing, food and tobacco, construction and the service sector. Three quarters of all employees in the city are employed in the service sector: more than 10000 service enterprises have been newly established since 1990. The 64 banking and insurance branches in the city employ some 7000 people, and there are over 700 lawyers, countless tax, business and personnel consultants, more than 800 estate agents and accommodation bureaus, and a great number of other business-related service providers. The productivity of Dresden's enterprises is today more than four times higher than it was in 1990. The population of Dresden now achieves the highest purchasing power in Eastern Germany.

More than 30 institutes and scientific centres, including those of the Leibnitz Society, the Fraunhofer Society and the Max Planck Society, are located in Dredsen. The Technical University of Dresden is a large and important university in Germany. It employs more than 9000 people, including 800 professors and senior lecturers, and there are about 24 000 registered students.

Participants Participants were sampled using a random route method: streets were selected at random, and within each selected street every third house was visited. In larger blocks of flats, the inhabitants of every fourth apartment (in smaller houses every third) were contacted. All participants were assured of con- fidentiality. The refusal rate was 33 per cent.

The data were collected in the district of Dresden, one of the three big cities of former East Germany. During the first wave of data collection, 365 subjects participated in the study. At Time 2, data from 202 additional subjects were collected. Since there are relatively few participants available for Time 1-Time 6 analyses, we decided to omit Time 1 data and analysed data from Time 2 and Time 6.

With respect to age, social class, and male/female workers, the participants were representative of the working population of Dresden (Frese et al., 1996). All participants were paid for their participa- tion in the study. Forty-nine per cent of the participants were male and 51 per cent were female. Age ranged between 16 and 63 years (M = 39, SD = 11.42). Most subjects worked in public or private ser- vices (35.9 per cent). Trade or manufacturing enterprises employed 30.9 per cent of the participants. There were 18.9 per cent office workers holding jobs which require little qualification. Managers or professionals with high qualification requirements formed 27.4 per cent of the sample. There were 12.5



per cent higher-level public servants mostly employed in schools and universities, and 16.5 per cent skilled and 14.9 per cent unskilled blue-collar workers, respectively.

To determine the degree to which the stability of job satisfaction is due to stable environmental factors, work-related variables were controlled in a job changer sample. In line with the meta-analysis, we did not differentiate between those who changed either their job or their employer. Participants unemployed at Time 2 or at Time 6 were excluded, as well as participants with missing values for job satisfaction at either time. The sample analysed consisted of 184 subjects. For these subjects, there was an overall rate of missing values of about 3 per cent. Missing data were accounted for by application of the EMCOV computer program (see Graham and Donaldson, 1993; Graham et al., 1996). We tested whether the sample analysed differed from the whole sample with respect to gender, age, socioeconomic status, job satisfaction, job content (see below), and job stressors (see below). Those participants who changed their job were 2.65 years younger on average compared to those subjects not analysed (t - - 2.58; df- 566; p = 0.01). We further tested whether the reason for excluding certain participants was responsible for differences between participants included in the study and excluded participants. Subjects who were employed at Time 2, but who did not report their job satisfaction and were thus excluded from analyses, had significantly (t= 2.01, df= 197, p < 0.05) lower satisfaction scores at Time 6 (M = 3.12, SD= -0.68) than those participants analysed (M= 3.48, SD = 0.66). For subjects excluded due to unemployment, no differences were found.

Measures Job satisfaction. The job satisfaction measure used was adopted from Warr et al. (1979). Eight items asked participants how satisfied they were with respect to several aspects of their work. These aspects were 'Possibilities to develop new skills and knowledge,' 'Possibilities to carry out work as it is most suitable to oneself,' 'Availability and condition of working resources which facilitate task accomplishment (properties, devices etc.),' 'Social recognition,' 'Environmental conditions at work (noise, light, temperature etc.),' 'Pay and social benefits,' 'Trust received by supervisors,' and 'Promotion opportunities.' The items required a response on a 5-point scale that ranged from 1 (very dissatisfied) to 5 (very satisfied). Job content. Measures for job contents and job stressors were taken from the Instrument of Stress- oriented Job Analysis (ISTA; Semmer et al., 1995). A composite scale was used to assess job content. This scale comprised items used to assess control at work and complexity. A sample item for control was 'Can you decide on your own how to fulfill your tasks?' A sample item for complexity was 'Is it possible to use your skills and knowledge at work?' Responses were given on a 5-point scale ranging from 1 (very little) to 5 (very much). The resulting composite scale of job content is similar to the measure of job control used by Karasek (Karasek and Theorell, 1990). For a discussion on how job control and complexity are related, see Frese (1987), or Frese et al. (submitted). Job stressors. A composite scale similar to Frese (1985) was used to assess job stressors. This scale comprised scales for organizational problems (e.g., 'How good is your working equipment?'), concentration requirements and time pressure (e.g., 'How often do you have to work under time pressure?'), and uncertainty (e.g., 'How often do you receive unclear instructions?'). Responses were given on a 5-point scale ranging from 1 (very seldom) to 5 (very often). Psychometric information of the variables used in this study can be found in Table 2.

Analytical procedure Since there is no basic statistical test available to evaluate whether test-retest correlations are statis- tically reduced when job conditions (i.e., job content and job stressors) are controlled, we followed the procedure described by Williams et al. (1996). In brief, we used two series (with and without



Table 2. Descriptive statistics of study 2- Variables (only job changers; n = 184)

No. of M SD 1 2 3 4 5 6 items

1. Job satisfaction t2 8 2.99 0.66 0.77 2. Job satisfaction t6 8 3.50 0.66 0.26 0.81 3. Job content t2 8 3.33 0.71 0.38 0.26 0.80 4. Job content t6 8 3.52 0.71 0.16 0.38 0.61 0.82 5. Job stressors t2 15 2.70 0.56 -0.38 -0.05 0.18 0.31 0.76 6. Job stressors t6 15 2.56 0.50 -0.40 -0.38 0.05 0.24 0.51 0.79

Note: Correlations exceeding 0.17 in absolute value are significant forp < 0.01; correlations exceeding 0.12 in absolute value are significant with p < 0.05 (one-sided). Correlations appearing in the table were corrected for missing data. Cronbach's alpha appears in the diagonal.

measurement models) of three different structural equation models. The first two models were satu- rated structural models. The first model estimated the zero-order correlation (i.e., stability) of the job satisfaction scale without any other variable involved. The second model estimated the residual correlation after the four job condition-scales (job content in 1990 and 1995 and job stressors in 1990 and 1995) were controlled. Finally, the third model was identical to the second model but the restriction was imposed that the residual correlation between the two job satisfaction scales in 1990 and 1995 equals the zero-order correlation obtained in the first model. Model 2 and Model 3 can then be compared using Chi-square difference tests (Bentler and Bonnett, 1980). A significant difference between the two Chi-square values indicates that the reduction in stability after controlling for job conditions is significant.

Results

Descriptive statistics of Study 2-variables are presented in Table 2. Interestingly, the stability of job satisfaction (0.26) was lower than the stabilities of job content (0.61) and job stressors (0.51). It was also lower than the respective average correlation obtained from the meta-analysis of stabilities.

In the next step, the test-retest correlation of job satisfaction after partialling job content and job stressors was analysed. The partial correlation was 0.01. A Chi-square difference test revealed that it was significantly lower than the zero-order correlation of 0.26 (Ax2 = 23.24, df= 1, p < 0.01).

A similar series of analyses was conducted with latent variables in order to account for measurement error. In this instance, the available indicators were randomly distributed across three item parcels for each latent factor (job satisfaction, job content, and job stressors) at both points in time. The zero-order correlation between latent job satisfaction in 1990 and 1995 was 0.29, which was reduced to - 0.04 after controlling for job content and job stressors (AX2 = 23.60, df= 1, p < 0.01). The results suggest that even in samples of job changers, several characteristics of their work remained stable, and it cannot be excluded that these stable characteristics accounted for the stability in changers' job satisfaction scores.

General Discussion

In this article we have presented an overview of studies concerned with dispositional determinants of job satisfaction. Indirect approaches suggest that unspecified dispositional determinants explain about

Copyright 0 2001 John Wiley & Sons, Ltd. J. Organiz. Behav. 22, 483-504 (2001)


30 per cent of variance in job satisfaction; corresponding effects obtained from direct approaches explain 10 to 20 per cent of variance in most instances.

Direct approaches typically analyse the role of affectivity. Positive affectivity seems to be more important than NA, which often shows only small effects. Most indirect evidence for the dispositional argument emerges from longitudinal studies analysing the stability of job satisfaction across time. An overview of such studies revealed substantial stabilities for job satisfaction both for employees staying with their job and for job changer samples. Interestingly, the meta-analytic comparison of stabilities of job changers and job stayers resulted in a difference as small as 0.13. For each group the relationship between time lag and size of correlation was weak: it was slightly negative for stayers and slightly positive for changers.

From the point of view that organizational conditions primarily determine the level of job satisfaction, this is certainly a surprise and the results seem to support the dispositional argument, or at least weaken the argument that organizational conditions should be the main causes of job satisfaction. However, these analyses do not take into consideration that levels of job satisfaction may be maintained by stable environmental conditions even under the condition of job change. Job changers are not randomly exposed to organizational conditions after the change. Therefore, stability in job satisfaction is likely to be only partly due to dispositions. Rather, it is substantially maintained by environmental characteristics that are malleable in principle but nevertheless remain constant. Of course, people try to improve their working conditions, but, all in all, they either stay in their occupation or find a job in a related occupation, and thus may keep certain organizational conditions constant, in particular job content and job stressors. Therefore, we analysed a sample of East Germans who underwent a societal change, where more drastic organizational changes should occur. Nevertheless, the corrected stability of job satisfaction of 0.33 in this study is similar to the average sample-weighted and corrected test-retest correlation of the job changer samples analysed in previously published studies (0.35). This suggests that the between-organization variance in working conditions affecting job satisfaction was not affected by societal changes and, thus, that our results may generalize beyond the present study.

Partialling relatively stable variables that are typically major determinants of job satisfaction should further reduce this correlation. When job content variables (complexity, control at work) and job stressors (organizational problems, concentration requirements and time pressure, and uncertainty) are controlled, the stability of job satisfaction decreases to - 0.04. This result clearly challenges the assumption of a direct effect of an underlying dispositional factor on job satisfaction, but may still be compatible with the idea that selection effects are responsible for personality-job satisfaction relations mediated by working conditions.

The mediating effect of working conditions triggered by selection effects can be tested by partialling work content and stressors. In the study of Arvey et al. (1989) this did not reduce the job satisfaction correlations of twins very much, although the twins shared a substantial amount of variance in work content and stressors. Obviously, these variables were unrelated with job satisfaction in the study of Arvey et al., which is contrary to many findings in the job satisfaction literature. The problem may be that Arvey et al. derived work content and stressors by using the Dictionary of Occupational Titles (DOT; U.S. Department of Labor, 1977). Although variables derived from the DOT are less prone to several sorts of biases when compared to self-reports, they may be too fuzzy. Comparing DOT-based and self-reported measures of complexity as predictors of job satisfaction, Gerhart (1987) found that DOT-based measures were much weaker predictors of job satisfaction. Indeed, concurrent self- reported complexity was the strongest predictor in his analyses, and it was even a stronger predictor than past satisfaction.

The correlation of - 0.04 obtained after partialling job content variables and job stressors should not be taken as a proof that there is no dispositional influence at all. Rather, this finding could be inter-



preted as a support of the selection hypothesis: people with certain personality traits obtain their jobs through self-selection and selection by the organization. For example, intelligence may lead to a high level of qualification which determines the level of complexity (e.g., Kohn and Schooler, 1982) and, in turn, increases satisfaction with the job. Extraversion might be associated with social competencies, which may be helpful in swaying a supervisor's decision to promote a person to a job with better work content, which again leads to higher satisfaction. Such reasoning leads to the conclusion that when stable environmental factors are partialled from job satisfaction, dispositional causes of job satisfaction through selection effects are being partialled, too.

The results may strengthen the selection hypothesis in several respects. First, although our results are in contrast to the finding of Arvey et al. (1989), we partialled different job content variables and job stressors, which may explain the divergence across studies. Second, since partialling NA and PA from the stability of job satisfaction had virtually no effect in the study of Schaubroeck et al. (1996), our results suggest that selection processes are more likely to be triggered by other personality variables which are more strongly related to qualification issues.

In sum, our results suggest that personality factors play an important role leading to high stabilities of job satisfaction. However, it seems likely that a substantial part of this effect operates indirectly through selection mechanisms: personality affects job conditions and these affect job satisfaction. These kinds of personality effects, however, do not question the usability of job satisfaction for the subjective assessment of the organization. Even some part of the direct effects of PA and NA on job satisfaction may be a result of the selection effects, an assumption which is supported by the already mentioned studies of Cook et al. (1995 - paper presented at the SIOP convention, Lake Buena Vista, FL), which found that individuals scoring high in NA were less successful in selection interviews.

A careful interpretation of the present data must allow for the fact that not all effects of personality variables are mediated by working conditions. Rather, it has to be taken into account that a small part of variance in job satisfaction remains that is directly determined by personality variables and which may be considered as a contaminating factor in the measurement of job satisfaction. This could limit the use of job satisfaction measures as a tool for work and organizational assessment. The problem can be circumvented to some degree if longitudinal designs are applied. For example, when prior satisfaction is used as the first predictor in hierarchical regression with subsequent satisfaction as the outcome, it can be assumed that the major part of dispositional variance is also partialled because of the relation between prior satisfaction and dispositions (Zapf et al., 1996). This is based on the assumption that dispositional factors affect organizational variables similarly at different points in time and that there is no interaction between personality traits and the factors responsible for change. Since job satisfaction is highly stable, the residual variance in current job satisfaction would only be small in such instances, and it is not easy to find further substantial predictors. Thus, even small effects of working characteristics could then be of enormous importance. Moreover, if means of satisfaction before and after an intervention are compared, the difference can be assumed to be due to differences in the environment apart from random errors, if personality effects are very stable.

Limitations and directions for research

An issue which was not dealt with in the present article has to do with the measurement of job satisfaction. We did not differentiate between global satisfaction measures and compounds of several facets. This echoes the view of other authors (e.g., Wanous, 1974) who equate both ways to measure global satisfaction. But our main reasons for doing so were practical. The number of studies in the meta-analysis was too small to allow for a break-down of different satisfaction measures. In addition,

Copyright ) 2001 John Wiley & Sons, Ltd. J. Organiz. Behav. 22, 483-504 (2001)


there was no global satisfaction measure available in our own empirical study. However, the effects of personality dispositions on all facets and global satisfaction items are not necessarily uniform. Some facets may be more susceptible to trait influences than others (cf, Arvey et al., 1989). This may apply to trait effects on emotional information processing as well as selection effects that are triggered by traits. Future research should take the possibility into consideration that there are either differential or general effects.

In addition to the issue of differential versus general effects another issue is worth attention. Williams et al. (1996) distinguished between confounding measurement models and confounding congeneric models and suggested testing the models using structural equations. Using the confounding measurement model the authors tested the hypothesis that personality biases, among others, job satisfaction items (indicators). The confounding congeneric model was applied to test the hypothesis that personality affects, among others, the true score of (latent) job satisfaction. Although we think that the issue of item nuisance versus true score-effects is as important as the issue of differential versus general effects, we did not pursue it in the present study. The reason was two-fold. First, there was no way to consider this distinction in the meta-analysis because of the lack of appropriate information (i.e., stabilities of items) in the literature. Second, it would have been possible to analyse confounding measurement models and congeneric models using our own data, but this implies at least two problems: too many parameters would have then to be estimated relative to our sample size and the interpretation would be far from clear. The interpretation problem lies in the fact that we could not partial personality directly but only indirectly inasmuch as their effects are included in job conditions and job stressors. However, we should once again refer to the study by Spector et al. (1999). Their study did not apply different structural equation modelling techniques to test these competing models against each other, but provided a more direct test by correlating objective (non-incumbent) and subjective (incumbent) measures. As we have already mentioned, their results clearly favoured the selection model (i.e., the confounding congeneric model).

One problem with the selection hypothesis is that people are selected to good as well as to poor working conditions, but that a typical finding of job satisfaction research is that more than 80 per cent of a sample report being satisfied. This result has also proved stable over time (Weaver, 1980). Most job satisfaction researchers, however, are convinced that the organizational reality does not justify this high number of satisfied people. If the result cannot be explained by organizational factors, must not the selection hypothesis be discarded and direct effects of personality factors be the explanation? Diener and Diener (1996) argue that people typically report positive satisfaction data, which seems to be a necessity for their well-being and that one should not conclude that 'just because people report positive levels of satisfaction with their work or with a consumer product, for example, this does not mean inevitably that the work or product is highly desirable' (p. 185). On the basis of the presented studies it can be hypothesized that reporting satisfaction rather than dissatisfaction is a process affecting every individual: it leads to increased levels of satisfaction but to little change of the rank order of the individuals.

A related issue is that job satisfaction is dependent on individuals' aspiration levels (Bruggemann et al., 1975; Biissing, 1992). Again, the question is whether aspiration levels affect job satisfaction like a personality trait. Bruggemann and colleagues have suggested in their theoretical model that there are various 'types' of job satisfaction, one of them called 'resigned job satisfaction'. If there is a discrepancy between the perception of one's work and of one's expectations, individuals start to cope with this discrepancy. One strategy is to lower one's aspiration level, resulting in resigned job satisfaction. This may contribute to the result that most employees report to be satisfied with their jobs. The processes of resigned job satisfaction are probably the results of an interaction between person and situation. There may be a personality component to the distinction between employees who resign under dissatisfying working conditions and those who do not. Of course, there is also a strong situational



component because dissatisfying working conditions are a prerequisite to elicit the attenuation of one's aspiration level. In any case, these are issues that need to be pursued in future research.

In this article we intended to show that the effects of personality dispositions do not impose a strong limitation on the use of job satisfaction for work and organizational assessment. Rather, our results support the hypothesis that a strong part of personality influence on job satisfaction is due to selection mechanisms, which lead to differences in working conditions. However, we do not believe that measures of job satisfaction are without problems. There is still a lack of theory, but process models such as the one developed by Bruggemann et al. (1975) are promising. Their consideration should lead to an improvement of job satisfaction measures. The present paper has shown that such improve- ments are not limited by the dispositional determination of job satisfaction.

Acknowledgements

We are grateful to Michael Frese who gave us the data from the project AHUS. The project AHUS (Aktives Handeln in einer Umbruchsituation - active actions in a radical change situation) was supported by the Deutsche Forschungsgemeinschaft (DFG, No Fr 638/6-6) (principal investigator: Michael Frese).

We are also grateful to Hanspeter Irmer for his helpful comments on an earlier draft of this paper.

References

Arvey RD, Bouchard TJ Jr, Segal NL, Abraham LM. 1989. Job satisfaction: environmental and genetic components. Journal of Applied Psychology 74: 187-192.

Arvey RD, McCall BP, Bouchard TJ, Taubman P, Cavanaugh MA. 1993. Genetic influences on job satisfaction and work value. Personality and Individual Differences 17: 21-33.

Bentler PM, Bonett DG. 1980. Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin 88: 588-606.

Bouchard TJ Jr. 1997. Genetic influence on mental abilities, personality, vocational interests and work attitudes. In International Review of Industrial and Organizational Psychology 1997, Vol. 12, Cooper CL, Robertson IT (eds). Wiley: Chichester; 373-395.

Bouchard TJ Jr, Lykken DT, McGue M, Segal NL, Tellegen A. 1990. Sources of human psychological differences: the Minnesota study of twins reared apart. Science, 250: 223-228.

Bouchard TJ, Arvey RD, Keller LM, Segal NL. 1992. Genetic influences on job satisfaction: a reply to Cropanzano and James. Journal of Applied Psychology 77: 89-93.

Brayfield AH, Crockett WH. 1955. Employee attitudes and employee performance. Psychological Bulletin 52: 396-424.

Breaugh JA. 1981. Predicting absenteeism from prior absenteeism and work attitudes. Journal of Applied Psychology 66: 555-560.

Brief AP, Roberson L. 1989. Job attitude organization: an exploratory study. Journal ofApplied Social Psychology 19: 717-727.

Brief AP, Burke MJ, George JM, Robinson B, Webster J. 1988. Should negative affectivity remain an unmeasured variable in the study of job stress. Journal of Applied Psychology 73: 193-198.

Bruggemann A, Groskurth P, Ulich E. 1975. Arbeitszufriedenheit [Job satisfaction]. Huber: Bern, Switzerland. Burke MJ, Brief AP, George JM. 1993. The role of negative affectivity in understanding relations between self-

reports of stressors and strains: a comment on the applied psychology literature. Journal of Applied Psychology 73: 402-412.

Biissing A. 1992. A dynamic view of job satisfaction in psychiatric nurses in Germany. Work and Stress 6: 239-259.



Carsten JM, Spector PE. 1987. Unemployment, job satisfaction, and employee turnover: a meta-analytic test of the Muchinsky model. Journal of Applied Psychology 72: 374-381.

Chen PY, Spector PE. 1992. Relationships of work stressors with aggression, withdrawal, theft, and substance use: an exploratory study. Journal of Occupational and Organizational Psychology 65: 177-184.

Cropanzano R, James K. 1990. Some methodological considerations for the behavioral genetic analysis of work attitudes. Journal of Applied Psychology 75: 433-439.

Davis-Blake A, Pfeffer J. 1989. Just a mirage: the search for dispositional effects in organizational management. Academy of Management Review 14: 385-400.

Dawis RV, Pinto PP, Weitzel W, Nezzer M. 1974. Describing organizations as reinforcer systems: a new use for job satisfaction and employee attitude surveys. Journal of Vocational Behavior 4: 55-66.

Diener E, Diener C. 1996. Most people are happy. Psychological Science 7: 181-185. Dormann C, Zapf D. 1999. Social support, social stressors at work and depression: testing for moderating effects

with structural equations in a 3-wave longitudinal study. Journal of Applied Psychology 84: 874-884. Finkelstein LM, Burke MJ, Raju MS. 1995. Age discrimination in simulated employment contexts: an integrative

analysis. Journal of Applied Psychology 80: 652-663. Frese M. 1985. Stress at work and psychosomatic complaints: a causal interpretation. Journal of Applied

Psychology 70: 314-328. Frese M. 1987. A theory of control and complexity: implications for software design and integration of computer

system into the work place. In Psychological Issues of Human Computer Interaction at the Work Place. Frese M, Ulich E, Dzida W (eds). North-Holland: Amsterdam; 313-337.

Frese M, Kring W, Soose A, Zempel J. 1996. Personal initiative at work: differences between East and West Germany. Academy of Management Journal 39: 37-63.

Frese M, Fay D, Leng K, Hilburger T, Tag A. 1997. The concept of personal initiative: operationalization, reliability, and validity in two German samples. Journal of Occupational and Organizational Psychology 70: 139-161.

Gottfredson GD, Holland JL. 1990. A longitudinal test of the influence of congruence: job satisfaction, competency utilization, and counterproductive behavior. Journal of Counseling Psychology 37: 389-398.

Graham JW, Donaldson SI. 1993. Evaluating interventions with differential attrition: the importance of nonresponse mechanisms and use of follow-up data. Journal of Applied Psychology 78: 119-128.

Graham JW, Hofer SM, MacKinnon DP. 1996. Maximizing the usefulness of data obtained with planned missing value patterns: an application of maximum likelihood procedures. Multivariate Behavioral Research 31: 197-218.

Gustavsson JP, Weinryb RM, Goransson S, Pederson NL, Asberg M. 1997. Stability and predictive ability of personality traits across 9 years. Personality and Individual Differences 22: 783-791.

Hackman JR, Oldham GR. 1980. Work Redesign. Addison-Wesley: Reading, Mass. Herzberg F, Mausner B, Peterson RO, Capwell DE 1957. Job Attitudes: Review of Research and Opinion.

Psychological Service of Pittsburgh: Pittsburgh, PA. Hogan RT. 1991. Personality and personality measurement. In Handbook of Industrial and Organizational

Psychology Vol. 2, Dunnette MD, Hough LM (eds). Consulting Psychologists Press: Palo Alto, CA; 873-919.

House RJ, Shane SA, Herold DM. 1996. Rumors of the death of dispositional research are vastly exaggerated. Academy of Management Review 21: 203-224.

Hulin CL. 1991. Adaptation, persistence, and commitment in organizations. In Handbook of Industrial and Organizational Psychology, Vol. 2, Dunnette MD, Hough LM (eds). Consulting Psychologists Press; Palo Alto, CA; 445-505.

Judge TA. 1992. The dispositional perspective in human resources research. Research in Personnel and Human Resources Management 10: 31-72.

Judge TA, Hulin CL. 1993. Job satisfaction as a reflection of disposition: a multiple source causal analysis. Organizational Behavior and Human Decision Processes 56: 388-421.

Judge TA, Locke EA, Durham CC, Kluger AN. 1998. Dispositional effects on job and life satisfaction: the role of core evaluations. Journal of Applied Psychology 83: 17-34.

Judge TA, Bono JE, Locke EA. 2000. Personality and job satisfaction: the mediating role of job characteristics. Journal of Applied Psychology 85: 237-249.

Karasek RA, Theorell T. 1990. Healthy Work. Basic Books: New York. Keller LM, Bouchard TJ, Arvey RD, Segal NL, Dawis RV. 1992. Work values: genetic and environmental

influences. Journal of Applied Psychology 77: 79-88. Keller RT. 1983. Predicting absenteeism from prior absenteeism, attitudinal factors, and nonattitudinal factors.

Journal of Applied Psychology 68: 536-540.



Kohn ML, Schooler C. 1982. Job conditions and personality: a longitudinal assessment of their reciprocal effects. American Journal of Sociology 87: 1257-1286.

Levin I, Stokes JP. 1989. Dispositional approach to job satisfaction: role of negative affectivity. Journal of Applied Psychology 74: 752-758.

Locke EA. 1969. What is job satisfaction? Organizational Behavior and Human Performance 4: 309-336. Locke EA. 1976. The nature and causes of job satisfaction. In Handbook of Industrial and Organizational

psychology, Dunnette MD (ed). Rand McNally: Chicago, IL; 1297-1349. Mobley WH, Homer SO, Hollingsworth AT. 1978. An evaluation of precursors of hospital employee turnover.

Journal of Applied Psychology 63: 408-414. Munz DC, Huelsman TJ, Konold TR, McKinney JJ. 1996. Are there methodological and substantive roles for

affectivity in job diagnostic survey relationships. Journal of Applied Psychology 81: 795-805. Newton T, Keenan T. 1991. Further analyses of the dispositional argument in organizational behavior. Journal of

Applied Psychology 76: 781-787. Nicholson N, Brown CA, Chadwick-Jones JK. 1976. Absence from work and job satisfaction. Journal of Applied

Psychology 61: 728-737.

Quifiones MA, Ford JK, Teachout MS. 1995. The relationship between work experience and job performance: a conceptual and meta-analytic review. Personnel Psychology 48: 887-910.

Raju NS, Fleer PE 1997. MAIN: A Computer Program for Meta-Analysis. Illinois Institute of Technology: Chicago, IL.

Raju NS, Burke MJ, Normand J, Langlois GM. 1991. A new meta-analytic approach. Journal of Applied Psychology 76: 432-446.

Rusbult CE, Farrell D. 1983. A longitudinal test of the investment model: the impact on job satisfaction, job commitment, and turnover of variations in rewards, costs, alternatives, and investments. Journal of Applied Psychology 68: 429-438.

Semmer NK, Zapf D, Dunckel H. 1995. Assessing stress at work: a framework and an instrument. In Work and Health - Scientific Basis of Progress in the Working Environment, Svane O, Johansen C (eds). Office for Official Publications of the European Communities: Luxembourg; 105-113.

Smith PC, Kendall LM, Hulin CL. 1969. The Measurement of Satisfaction in Work and Retirement: A Strategy for the Study of Attitudes. Rand McNally: Chicago, IL.

Spector PE, Fox S, Van Katwyk, PT. 1999. The role of negative affectivity in employee reactions to job characteristics: bias effect or substantive effects. Journal of Organizational and Occupational Psychology 72: 205- 218.

Staw BM, Bell NE, Clausen JA. 1986. The dispositional approach to job satisfaction: a lifetime longitudinal test. Administrative Science Quarterly 31: 56-77.

U.S. Department of Labor. 1977. Dictionary of Occupational Titles, 4th edn. Government Printing Office: Washington, DC.

Vroom VH. 1964. Work and Motivation. Wiley: New York. Wanous JP. 1974. Individual differences and reactions to job characteristics. Journal of Applied Psychology 59:

616-622. Wanous JP, Reichers AE, Hudy MJ. 1997. Overall job satisfaction: how good are single item measures. Journal of

Applied Psychology 82: 247-252. Warr P, Cook J, Wall T. 1979. Scales for the measurement of some work attitudes and aspects of psychological

well-being. Journal of Occupational Psychology 52: 129-148. Watson D, Clark LA. 1984. Negative affectivity. The disposition to experience aversive emotional states.

Psychological Bulletin 96: 465-490. Watson D, Slack AK. 1993. General factors of affective temperament and their relation to job satisfaction over

time. Organizational Behavior and Human Decision Processes 54: 181-202. Weaver CN. 1980. Job satisfaction in the United States in the 1970s. Journal of Applied Psychology 65: 364-367. Wegemann RG. 1991. From job to job. Journal of Employment Counseling 28: 8-12. Weiss DJ, Dawis RV, England GW, Lofquist LH. 1967. Manual for the Minnesota Satisfaction Questionnaire.

Minnesota Studies in Vocational Rehabilitation: Minnesota; XXII. Williams LJ, Gavin MB, Williams ML. 1996. Measurement and nonmeasurement processes with negative

affectivity and employee attitudes. Journal of Applied Psychology 81: 88-101. Zapf D. 1991. Arbeit und Wohlbefinden [Work and well-being]. In Wohlbefinden: Theorie, Empirie, Diagnostik,

Abele A, Becker P (eds). Juventa: Weinheim, Germany; 227-244. Zapf D, Dormann C, Frese M. 1996. Longitudinal studies in organizational stress research: a review of the

literature with reference to methodological issues. Journal of Occupational Health Psychology 1: 145-169.



Further Reading

Adkins CL. 1995. Previous work experience and organizational socialization: a longitudinal examination. Academy of Management Journal 38: 839-862.

Agho AO, Mueller CW, Price JL. 1993. Determinants of employee job satisfaction: an empirical test of a causal model. Human Relations 46: 1007-1027.

Bateman TS, Strasser S. 1983. A cross-lagged regression test of the relationships between job tension and employee satisfaction. Journal of Applied Psychology 68: 439-445.

Begley TM, Czajka JM. 1993. Panel analysis of the moderating effects of commitment on job satisfaction, intent to quit, and health following organizational change. Journal of Applied Psychology 78: 552-556.

Blegen MA, Mueller CW. 1987. Nurses' job satisfaction: a longitudinal analysis. Research in Nursing and Health 10: 227-237.

Breeden SA. 1993. Job and occupational change as a function of occupational correspondence and job satisfaction. Journal of Vocational Behavior 43: 30-45.

Cramer D. 1995. Life and job satisfaction: a two-wave panel study. Journal of Psychology 129: 261-267. Downey HK, Sheridan JE, Slocum JW. 1976. The path-goal theory of leadership: a longitudinal analysis.

Organizational Behavior and Human Performance 16: 156-176. Farkas AJ, Tetrick LE. 1989. A three-wave longitudinal analysis of the causal ordering of satisfaction and

commitment on turnover decisions. Journal of Applied Psychology 74: 855-868. Gerhart B. 1987. How important are dispositional factors as determinants of job satisfaction? Implication for job

design and other personnel programs. Journal of Applied Psychology 72: 366-373. Griffin RW. 1981. A longitudinal investigation of task characteristics relationships. Academy of Management

Journal 24: 99-113. Griffin RW. 1991. Effects of work redesign on employee perceptions, attitudes, and behaviors: a long-term

investigation. Academy of Management Journal 34: 425-435. Gutek BA, Winter SJ. 1992. Consistency of job satisfaction across situations: fact or framing artifact? Journal of

Vocational Behavior 41: 61-78. Hardin E. 1965. Perceived and actual change in job satisfaction. Journal of Applied Psychology 49: 363-367. Heaney CA, Israel BA, House JS. 1994. Chronic job insecurity among automobile workers: effects on job

satisfaction and health. Social Science and Medicine 38: 1431-1437. Howard JH, Cunningham DA, Rechnitzer PA. 1986. The effects of personal interaction on triglyceride and uric acid

levels, and coronary risk in a managerial population: a longitudinal study. Journal of Human Stress 12: 53-63. Judge TA, Watanabe S. 1993. Another look at the job satisfaction-life satisfaction relationship. Journal of Applied

Psychology 78: 939-948. Keller RT, Szilagyi AD. 1978. A longitudinal study of leader reward behavior, subordinate expectancies, and

satisfaction. Personnel Psychology. 31: 119-129. Kopelman RE. 1977. Psychological stages of careers in engineering: an expectancy theory taxonomy. Journal of

Vocational Behavior 10: 270-286. Koslowsky M. 1991. A longitudinal analysis of job satisfaction, commitment, and intention to leave. Applied

Psychology: An International Review 40: 405-415. LaRocco JM. 1983. Job attitudes, intentions, and turnover: an analysis of effects using latent variables. Human

Relations 36: 813-825. Long BC. 1993. Coping strategies of male managers: a prospective analysis of predictors of psychosomatic

symptoms and job satisfaction. Journal of Vocational Behavior 42: 184-199. Lyons TF, Dickinson TL. 1973. A comparison of perceived- and computed-change measures over a three-year

period. Journal of Applied Psychology 58: 318-321. Nathan BR, Mohrman AM, Milliman JF. 1991. Interpersonal relations as a context for the effects of appraisal

interviews on performance and satisfaction: a longitudinal study. Academy of Management Journal 34: 352-369.

Near JP. 1984. Relationships between job satisfaction and life satisfaction: test of a causal model. Social Indicators Research 15: 351-367.

Newcomb MD. 1995. Prospective dynamics of intoxication in the workplace: personal and job-related predictors and consequences. Experimental and Clinical Psychopharmacology 3: 56-74.

Newman JM, Krzystofiak FJ. 1993. Changes in employee attitudes after an acquisition: a longitudinal analysis. Group and Organization Management 18: 390-410.



O'Driscoll M. 1987. Attitudes to the job and the organisation among new recruits: influence of perceived job characteristics and organisational structure. Applied Psychology: An International Review 36: 133-145.

O'Driscoll M, Thomas D. 1987. Life experiences and job satisfaction among mobile and stable personnel on large- scale construction projects. New Zealand Journal of Psychology 16: 84-92.

Saks AM, Ashforth BE. 1997. A longitudinal investigation of the relationships between job information sources, applicant perceptions of fit, and work outcomes. Personnel Psychology 50: 395-426.

Schaubroeck J, Green SG. 1989. Confirmatory factor analytic procedures for assessing change during organizational entry. Journal of Applied Psychology 74: 892-900.

Schaubroeck J, Ganster DC, Kemmerer B. 1996. Does trait affect promote job attitude stability? Journal of Organizational Behaviour 17: 191-196.

Schmitt N, Mellon PM. 1980. Life and job satisfaction: is the job central? Journal of Vocational Behavior 16: 51-58.

Schneider B, Dachler PH. 1978. A note on the stability of the job description index. Journal of Applied Psychology 63: 650-653.

Snizek WE, Bullard JH. 1983. Perception of bureaucracy and changing job satisfaction: a longitudinal analysis. Organizational Behavior and Human Performance 32: 275-287.

Staw BM, Ross J. 1985. Stability in the midst of change: a dispositional approach to job attitudes. Journal of Applied Psychology 70: 469-480.

Steel RP, Rentsch JR. 1997. The dispositional model of job attitudes revisited: findings of a 10-year study. Journal of Applied Psychology 82: 873-879.

Tharenou P. 1993. A test of reciprocal causality for absenteeism. Journal of Organizational Behaviour 14: 169-190.

van der Velde MEG, Feij JA. 1995. Change of work perceptions and work outcomes as a result of voluntary and involuntary job change. Journal of Occupational and Organizational Psychology 68: 273-290.

Vandenberg RJ, Lance CE. 1992. Examining the causal order of job satisfaction and organizational commitment. Journal of Management 18: 153-167.

Wanberg CR. 1995. A longitudinal study of the effects of unemployment and quality of reemployment. Journal of Vocational Behavior 46: 40-54.

Wanous JP. 1975. A causal-correlational analysis of the job satisfaction and performance relationship. Journal of Applied Psychology 59: 139-144.

Copyright 0 2001 John Wiley & Sons, Ltd. J. Organiz. Behav. 22, 483-504 (2001)

Documents

Job Satisfaction a Meta-Analysis of Stabilities