structural determinants of job satisfaction and organizational commitment in turnover models

STRUCTURAL DETERMINANTS OF JOBSATISFACTION AND ORGANIZATIONALCOMMITMENT IN TURNOVER MODELS

Stefan Gaertner

J. Mack Robinson College of Business

Georgia State University, Atlanta, GA, USA

Using Meta-Analytical Structural Equation Modeling (SEM), the relation-ships between structural determinants of job satisfaction and organiza-tional commitment were investigated in the context of contemporaryturnover models. Data for this study were obtained from nine empiricalstudies conducted under the direction of James L. Price and Charles W.Mueller at the University of Iowa. The results suggest that only threestructural determinants (distributive justice, promotional chances, andsupervisory support) are directly related to organizational commitment overand above their impact on job satisfaction. The effect size of these directlinks, however, is small in that they are not detectable in smaller samples(e.g., n = 244). On the other hand, most structural determinants are directlyrelated to job satisfaction. The notable exception is the amount of payemployees received which is not related to either job satisfaction ororganizational commitment. These findings and implications for futureturnover research are discussed.

Research in employee turnover is primarily concerned with voluntary turnoverdefined as ``individual movements across the membership boundary of a socialsystem which is initiated by the individual'' (Price, 1977). Starting from Marchand Simon (1958), this stream of research has generated several models ofdeterminants and processes underlying voluntary turnover (e.g., Hom &Griffeth, 1995; Lee & Mitchell, 1994; Mobley, 1977; Mowday, Porter, & Steers,1982; Price & Mueller, 1986). In recent models, job satisfaction (defined asdegree to which employees have a positive affective orientation towardsemployment by the organization; Price, 1997) and organizational commitment(defined as loyalty to a social unit; Price, 1997) are commonly viewed as

Direct all correspondence to: Stefan Gaertner, The W.T. Beebe Institute of Personnel and Employment

Relations, J. Mack Robinson College of Business, Georgia State University, Atlanta, GA 30303, USA. E-mail:

[email protected]

Human Resource Management Review, Copyright # 2000Volume 9, Number 4, 1999, pages 479±493 by Elsevier Science Inc.All rights of reproduction in any form reserved. ISSN:1053±4822

intervening variables in the turnover process (e.g., Hom & Griffeth, 1995;Mowday et al., 1982; Price & Mueller, 1986). Those two variables are viewed asan essential component of turnover models because their empirical relation-ship with voluntary turnover has been firmly established through numerousMeta-Analyses (e.g., Cohen, 1993; Cohen & Hudecek, 1993; Cotton & Tuttle,1986; Hom & Griffeth, 1995; Hom, Caranikas-Walker, Prussia, & Griffeth,1992; Steel & Ovalle, 1984; Tett & Meyer, 1993).

Turnover models differ, however, in how they view the relationship betweenthe structural determinants of turnover (defined as characteristics of the worksetting and patterns of social interaction within the work setting; Price &

Figure 1. Relationship of Structural Determinants and Organizational Commitment /JobSatisfaction as Suggested by Kim et al. (1996) (Error Terms and Paths among Exogenous

Constructs are Omitted for Clarity)

HUMAN RESOURCE MANAGEMENT REVIEW VOLUME 9, NUMBER 4, 1999480

Mueller, 1986) with job satisfaction and organizational commitment. Threealternative versions exist. First, all structural determinants are related to bothjob satisfaction and organizational commitment (Kim, Price, Mueller & Wat-son, 1996). This version is referred to in this paper as the Kim et al. (1996)model and is illustrated in Fig. 1. Second, all structural determinants arerelated to job satisfaction only (Price & Mueller, 1986; Williams & Hazer,1986). This version is referred to here as the Price and Mueller (1986) modeland is illustrated in Fig. 2. Third, different structural determinants havedifferent effects on job satisfaction and organizational commitment. Hom andGriffeth (1995), e.g., argue that some structural determinants have an impact

Figure 2. Relationship of Structural Determinants and Organizational Commitment /JobSatisfaction as Suggested by Price and Mueller (1986) (Error Terms and Paths among

Exogenous Constructs are Omitted for Clarity)

STRUCTURAL DETERMINANTS OF TURNOVER 481

on organizational commitment only while others have an impact on jobsatisfaction only.

The purpose of this article is to examine the empirical relationship betweenstructural determinants of turnover with job satisfaction and organizationalcommitment. Towards this end, the empirical fit of the three alternativepatterns of relationships presented above (i.e., all structural determinantsare directly related to job satisfaction and organizational commitment; allstructural determinants are directly related to job satisfaction only; differentstructural determinants differ in their effect on job satisfaction and organiza-tional commitment) will be tested using data obtained from nine empiricalstudies conducted under the direction of James L. Price and Charles W.Mueller at the University of Iowa.

METHOD

Meta-Analytical Structural Equation Modeling (SEM) (e.g., Hom & Griffeth,1995; Hom et al., 1992; Viswesvaran & Ones, 1995) was applied to test theempirical fit of the three alternative models. There are two basic steps inMeta-Analytical SEM. First, correlations from different studies are accumu-lated using Meta-Analytical procedures (e.g., Hunter & Schmidt, 1990) toobtain an overall matrix containing all correlations among all variables ofinterest. Second, this resulting correlation matrix is then used as data input ina SEM test (e.g., Bollen, 1989; Hayduk, 1987) of the a priori specified models.

In accumulating past research findings, it was first necessary to decidewhich studies to include. Only those studies that assessed all variables ofinterest, shown in Figs. 1 and 2 above, were included (Bollen, 1989; Hom et al.,1992). This procedure is equivalent to listwise deletion of missing values (Homet al., 1992; Viswesvaran & Ones, 1995). Pairwise deletion (i.e., to include allstudies that publish correlations between any two variables of interest in-cluded in the model) has been suggested as one possible alternative (Viswes-varan & Ones, 1995). However, pairwise deletion results in unequal samplesizes underlying the correlations in the cells of the resulting accumulatedcorrelation matrix. In line with Bollen (1989), I argue that this is a non-trivialproblem since unequal sample sizes among the cells of a correlation matrixrequire an ambiguous choice of the overall sample size underlying the entirecorrelation matrix. This ambiguous choice of an overall sample size, in turn,effects the resulting Chi-square statistics which is used to estimate the fit ofthe model and the standard errors which are used to test the statisticalsignificance of the paths linking the model constructs.

Nine studies were found that included all of the 12 variables of interest.Each of these studies was conducted in single organizations. As such thepresent study avoids the problems (i.e., non-interpretable results) that arisewhen accumulating research findings from different levels of analysis (Ostroff& Harrison, 1999). Almost all (eight) of the studies are unpublished doctoraldissertations which were completed under the guidance of Price, Mueller, or


both, at the University of Iowa. The remaining study was published by Price,Mueller, and two of their associates (Kim et al., 1996).1 The measures under-lying all variables, as well as the procedures to obtain the data in each study,are highly standardized across the studies. These studies reflect the work onmore recent versions of the Price and Mueller turnover model from 1990 untiltoday. The total sample size underlying the Meta-Analysis is 7,040. Thestudies included were conducted in a wide variety of settings from SouthKorea (three samples), Kenya (one sample) , and the USA (five samples). Threesamples consist of teachers, while the other samples consist of agriculturalprofessionals, auto workers, hospital employees, nurses, diverse white collarworkers, and physicians (one sample each). Approximately half of the overallsample is female (51%). All, except one study, (the study of auto workers) wereconducted with samples of white-collar workers.

Using a bare-bones Meta-Analysis (Hunter & Schmidt, 1990), the correla-tions obtained in all studies were averaged and weighted by sample size. Nocorrection for measurement error was conducted in the Meta-Analytical step ofthe data analysis. Instead, an average, weighted by sample size, of thecoefficient alphas was computed and unreliability of all variables was correctedfor in the SEM analysis using procedures first outlined by Kenny (1979). Theerror variance of each manifest indicator was set to be the quantity ``one minuscoefficient alpha.'' The factor loadings were set to be the square root ofcoefficient alpha.

LISREL 8.30 (JoÈreskog & SoÈrbom, 1999) was used to conduct the SEManalysis. The fit of the overall model was assessed by applying suggestionsfrom several authors (e.g., Bollen, 1989; Brannick, 1995; Hayduk, 1987; Hu &Bentler, 1999; Mueller, 1997). As a result, the fit of the models was assessedusing multiple indicators, including the variance explained in the dependentvariable, the reasonableness of the parameter estimates, the existence (ornon-existence) of improper solutions, and overall fit indexes. Based onrecommendations from Hu and Bentler (1999), two overall fit indexes werechosen, the Comparative Fit Index (CFI; Bentler, 1990) and the StandardizedRoot Mean Square Residual (SRMR; Bentler, 1995).

RESULTS

The bare-bones Meta-Analysis resulted in a correlation matrix containing theaverages, weighted by sample size, of the correlations obtained in all ninestudies. The correlation matrix and the sample size weighted coefficient alphasare shown in Table 1.

Using the correlation matrix shown in Table 1, I first tested the Kim et al.(1996) version of the Price and Mueller models. The model fit the dataperfectlyÐwhich, of course, did not come as a surprise since the estimationof this model required precisely all degrees of freedom that were available (i.e.,the model was saturated). The completely standardized parameter estimatesof this model are shown in the second and third columns of Table 2.


TA

BL

E1

Sam

ple

Siz

eW

eig

hte

dA

vera

ge

Corr

ela

tions

and

Alp

ha

Relia

bili

ties

Acro

ss

Nin

eE

mpiric

alT

ests

of

the

Price

and

Muelle

rT

urn

over

Models

Alp

ha

OC

JS

AT

PA

YP

RO

MD

JU

ST

PE

ER

SU

PLO

AD

RC

ON

FA

MB

IGA

UT

RO

UT

Org

aniz

atio

nal

Com

mitm

ent

0.8

11.0

0

Job S

atis

factio

n0.8

20.5

81.0

0

Pay

1.0

00.0

90.1

11.0

0P

rom

otio

nal

Chances

0.7

50.3

50.3

30.0

61.0

0

Dis

trib

utiv

eJustic

e0.8

60.4

20.3

70.0

90.3

61.0

0

Peer

Support

0.7

30.1

90.2

20.0

10.1

70.1

51.0

0

Superv

isory

Support

0.8

20.3

90.3

60.0

50.2

90.3

70.2

81.0

0

Work

load

0.7

0ÿ0

.16ÿ0

.17ÿ0

.03ÿ0

.12ÿ0

.23ÿ0

.07ÿ0

.13

1.0

0R

ole

Confli

ct

0.7

3ÿ0

.18ÿ0

.21ÿ0

.07ÿ0

.14ÿ0

.19ÿ0

.15ÿ0

.24

0.1

81.0

0R

ole

Am

big

uity

0.6

9ÿ0

.24ÿ0

.31ÿ0

.09ÿ0

.15ÿ0

.14ÿ0

.18ÿ0

.25

0.0

50.2

01.0

0A

uto

nom

y0.7

20.3

00.3

70.1

50.2

80.3

00.1

70.2

8ÿ0

.16ÿ0

.17ÿ0

.25

1.0

0R

outin

izatio

n0.7

4ÿ0

.33ÿ0

.44ÿ0

.11ÿ0

.28ÿ0

.26ÿ0

.15ÿ0

.25

0.0

70.0

70.1

5ÿ0

.38

1

Note

:N

=7,0

40.

All

corr

ela

tions

larg

er

than

0.0

2are

sig

nifi

cant

with

p<

0.0

5.


The signs of the parameter estimates are mostly consistent with theexpectations shown in Fig. 1. Some results, however, are inconsistent withthe Kim et al. (1996) model. First, pay has no significant effect on either jobsatisfaction or organizational commitment. Second, while most structuraldeterminants had significant links to job satisfaction, only three had additionalsignificant links with organizational commitment: distributive justice, super-visory support, and promotional chances. These results suggest that simplifi-cations to the Price and Mueller model are justifiable. These simplificationswould involve eliminating pay from the Price and Mueller model and deletingmost of the paths that directly link the structural determinants to organiza-tional commitment.

Next, I tested the empirical fit of the second model, which is referred to asthe Price and Mueller (1986) model (see Fig. 2). Since this alternative model isnot saturated, meaningful overall fit indexes can be calculated. According tothe standards suggested by Hu and Bentler (1999), this second alternativemodel fits the data well (SRMR = 0.022; CFI = 0.97). This good fit, along withthe fact that all except one path estimate (referring to the path between payand job satisfaction) were significant on a conventional level (p < 0.05),provides evidence that the Price and Mueller (1986) model gives a parsimo-nious but still accurate representation of the data (the completely standar-dized parameter estimates of this model are shown in the fourth and fifthcolumns of Table 2).

In the next step of this analysis, I incorporated the general idea proposedby Hom and Griffeth (1995) into the Price and Mueller model. Recall that

TABLE 2Completely Standardized Parameter Estimates for the Kim et al. (1996) and the

Price and Mueller (1986) Models

IndependentVariables

Kim et al. (1996) Price and Mueller (1986)

JobSatisfaction

OrganizationalCommitment

JobSatisfaction

OrganizationalCommitment

Pay 0.01 0.01 0.01 ±Promotional

Chances0.09* 0.11* 0.11* ±

DistributiveJustice

0.14* 0.16* 0.17* ±

Peer support 0.05* ÿ0.01 0.04* ±Supervisory

Support0.09* 0.14* 0.12* ±

Workload ÿ0.07* ÿ0.01 ÿ0.07* ±Role Conflict ÿ0.06* 0.02 ÿ0.05* ±Role Ambiguity ÿ0.21* ÿ0.01 ÿ0.20* ±Autonomy 0.06* ÿ0.02 0.05* ±Routinization ÿ0.36* 0.01 ÿ0.34* ±Job Satisfaction ± 0.55* ± 0.74*

*Statistically significant with p < 0.05.


Hom and Griffeth (1995) proposed that different structural determinantshave different effects on job satisfaction and organizational commitment.Since none of the specific suggestions from the Hom and Griffeth (1995)model was applicable to the relationships between the structural determi-nants and job satisfaction/organizational commitment specified in the Priceand Mueller models, I tested their general ideas through a series ofinductive exploratory tests. The objective of these tests was to determinewhether some structural determinants are directly related to organizationalcommitment over and above their impact on job satisfaction. Note that thesetests are at best a method for theory development. Consequently, theresulting model is not more than an empirically derived suggestion forfuture research. The exploratory test conducted in this section of the presentinvestigation is the equivalent to the stepwise procedure known in regres-sion analysis. Using Chi-square difference tests and modification indexes, Iconsecutively added direct paths from the structural determinants to orga-nizational commitment using the Price and Mueller (1986) model (illustratedin Fig. 2) as a starting point. This procedure was continued until the Chi-square statistics for the difference in fit between the more parsimoniousPrice and Mueller (1986) model and the saturated Kim et al. (1996) modelwas statistically insignificant.

The initial Chi-square difference test indicates that the Price and Mueller(1986) model fit the data significantly worse than the Kim et al. (1996) model(�x2 = 485.42; �df = 10; p < 0.01). The significant �x2 suggests that the fit ofthe Price and Mueller (1986) model can be improved by incorporating addi-tional paths into that model linking one or more structural determinantsdirectly to organizational commitment. The maximum modification indexsuggested to include a direct link between distributive justice and organiza-tional commitment. After this path was included, the Price and Mueller (1986)model still fit the data significantly worse than the Kim et al. (1996) model(�x2 = 158.95; �df = 9; p < 0.01), suggesting that additional direct pathsbetween the structural determinants and organizational commitment can beadded. The stepwise procedure was continued and led to the inclusion of twoadditional direct paths (supervisory support to organizational commitmentand promotional chances to commitment). After those paths were incorporatedinto the Price and Mueller model, it did not fit the data significantly worsethan the Kim et al. (1996) model (�x2 = 5.10; �df = 7; p = 0.65). These resultsare generally supportive of the Hom and Griffeth (1995) suggestions in thatdifferent structural determinant appear to have different effects on job satis-faction and organizational commitment. Most structural determinants arerelated to job satisfaction only while some structural determinants (distribu-tive justice, supervisory support, and promotional chances) are related to jobsatisfaction and organizational commitment.

The overall fit statistics for the resulting model (see Fig. 3) are excellent(CFI = 1; SRMR = 0.0019) but only marginally informative since the model wasderived inductively. The model explains 50% of the variance in job satisfactionand 57% of the variance in organizational commitment. This model, along with


the empirically derived completely standardized parameter estimates, isshown in Fig. 3.

In interpreting the results obtained, it is necessary to note that the samplesize underlying the Meta-Analytical SEM procedure is extremely large, muchlarger than commonly found in single studies of job satisfaction, organizationalcommitment, or turnover. As a consequence, the resulting Chi-square statisticsobtained is inflated (e.g., Bollen, 1989) relative to other empirical studies andeven negligible (i.e., practically insignificant) relationships are statisticallysignificant. To account for this potential problem, I conducted the samestepwise tests by specifying alternative hypothetical sample sizes in the

Figure 3. Relationship of Structural Determinants and Organizational Commitment /JobSatisfaction: Completely Standardized Parameter Estimates of an Inductively Derived

Model of Employee Turnover


LISREL 8.30 input syntax. By varying this specified sample size, it waspossible to investigate whether a given relationship found to be significantwith high sample sizes also holds for smaller samples. Three alternativesample sizes were specified: the smallest sample size of any of the studiesincluded in the meta-analysis (n = 244); the mean of the sample sizes across allstudies (n = 782); and the largest sample size of any of the studies included inthe meta-analysis (n = 1,773).

Assuming a sample size of n = 244, the x2 difference in fit between thePrice and Mueller (1986) model and the saturated Kim et al. (1996) modelwas insignificant on a conventional level , suggesting that no additionalpaths are needed to improve the fit of the Price and Mueller (1986) model(�x2 = 16.76; �df = 10; p = 0.08). Assuming a sample size of n = 782, thePrice and Mueller (1986) model fit the data significantly worse than the Kimet al. (1996) model (�x2 = 55.14; �df = 10; p < 0.01). The largest modificationindex shown in the LISREL output suggested an inclusion of a direct pathbetween distributive justice and organizational commitment to that model.After inclusion of this path, the revised model still fit the data significantly(albeit marginally so) worse than the Kim et al. (1996) model (�x2 = 17.64;�df = 9; p = 0.04). As a result a link between supervisory support andorganizational commitment was included. After the inclusion of this directpath the resulting model did not fit the data significantly worse than the Kimet al. (1996) model (�x2 = 5.55; �df = 8; p = 0.70). The results obtained fromspecifying n = 1,773 provided the same conclusion.2 The Price and Mueller(1986) model fit the data significantly worse than the Kim et al. (1996) modelunless direct paths were included that link distributive justice and super-visory support with organizational commitment. Once those paths wereincluded, the Price and Mueller (1986) did not fit the data significantlyworse than the Kim et al. (1996) model. These analyses demonstrate thatdifferences in fit between the Price and Mueller (1986) model and thesaturated Kim et al. (1996) model may not be detectable in smaller samplesizes (e.g., n = 244). These analyses further suggest that even studies withlarge (e.g., n = 782; 1,773) or very large (n = 7,040) sample sizes are unlikelyto detect many (i.e., more than three) direct linkages between structuraldeterminants and organizational commitment.

DISCUSSION

In the introduction, I asked the question whether the impact of structuraldeterminants on organizational commitment is completely mediated by jobsatisfaction (e.g., Price & Mueller, 1986; Williams & Hazer, 1986), incomple-tely mediated by job satisfaction (e.g., Kim et al., 1996), or whether differentstructural determinants have differential effects on job satisfaction and orga-nizational commitment as suggested by Hom and Griffeth (1995). The resultsof this article do not unambiguously confirm any of those patterns, i.e., each ofthose patterns seems to be both correct and incorrect to varying degrees.


The proposition of Kim et al. (1996) that all structural determinants arerelated to both job satisfaction, and organizational commitment appears to bethe pattern that obtained the least support in this study. The results indicatethat only few structural determinants (distributive justice, supervisory support,and promotional chances) are directly related to organizational commitmentover and above job satisfaction. Despite of a highly powerful empirical test usinga very large sample, 7 out of 10 of the direct linkages between the structuraldeterminants and organizational commitment proposed by Kim et al. (1996)could not be confirmed. Instead, a more parsimonious structural model, basedon suggestions by Price and Mueller (1986), provided a good fit for the data.

The suggestions of Price and Mueller (1986) received more support in thepresent study. The empirical fit of their parsimonious model to the data isgood, fulfilling the relatively conservative standards recently suggested by Huand Bentler (1999). The empirical fit of the Price and Mueller (1986) model isalso not significantly worse than the fit of the Kim et al. (1996) model in smallsamples. Even if the underlying sample size grows larger, only few additionaldirect paths between structural determinants and organizational commitmentare necessary to achieve a fit that is statistically equivalent to the fit of theKim et al. (1996) model. Thus the suggestion of Price and Mueller (1986), thatstructural determinants are directly related to job satisfaction only, appears toprovide a parsimonious yet accurate representation of the real relationships.On the other hand, the support for this pattern of relationships is notunequivocal. Exploratory analyses suggested that some structural determi-nants have in fact direct links with organizational commitment over and abovejob satisfaction which is consistent with the general idea put forward by Homand Griffeth (1995).

Therefore the idea indicated by Hom and Griffeth (1995) also received somesupport in this study since different structural determinants have differenteffects on job satisfaction and commitment; i.e., most structural determinantsare directly related to job satisfaction alone while others (i.e., distributivejustice, supervisory support, and promotional chances) are related to jobsatisfaction and organizational commitment. Recall, however, that the specificsuggestions from the Hom and Griffeth (1995) model regarding the differentialeffects of structural determinants on job satisfaction and organizationalcommitment were not tested in this study. For example, several variablesthat Hom and Griffeth (1995) suggested as direct antecedents of organiza-tional commitment only (e.g., procedural justice, employment security, jobinvestments) are not incorporated into the empirical tests. Consequently, theresults obtained cannot be taken as supporting the Hom and Griffeth (1995)model. Instead, these results only support their general idea that differentstructural determinants have different effects on job satisfaction and organi-zational commitment.

The results and shortcomings of this study lend themselves to six sugges-tions for future research. These suggestions are discussed below.

First, it is conceptually unclear why three structural determinants (super-visory support, promotional chances, and distributive justice) are directly


related to job satisfaction and organizational commitment whereas others areonly directly related to job satisfaction. Note that this finding was obtainedinductively and therefore requires empirical confirmation and conceptualexplanation from future research.

Second, the empirical results suggest that there is no statistically signifi-cant relationship between pay and job satisfaction/organizational commit-ment. These findings might be surprising; however, they confirm earlierqualitative reviews of the relationship between pay and the turnover processbased on studies conducted at the University of Iowa. Mueller and Price (1990),e.g., noted that almost 20 years of research on the Price and Mueller modelfound that pay is not appreciably related to job satisfaction or other turnoverprocess variables once other turnover determinants are included in the model.To be sure, these results do not suggest that pay is not important sincedistributive justiceÐwhich has a pay componentÐis found to be an importantdeterminant of both job satisfaction and organizational commitment. Rather,these findings suggest that the justice of the pay received (i.e., distributivejustice) rather than the amount of actual pay received has an impact on jobsatisfaction and organizational commitment.

Third, all of the studies that were included here were conducted essentiallyby one group of researchers. The advantage of this approach is that themeasures used to assess the variables of interest and the procedures used tocollect the data are highly standardized. On the other hand, however, alter-native operationalizations of the variables that would allow a triangulation ofthe findings are not considered in this study. As a result, replications of thefindings using alternative measures may increase the generalizability of thefindings obtained here.

Fourth, while the present study combined samples from a wide variety ofsettings, some employee groupsÐparticularly blue-collar workersÐare un-derrepresented in the overall sample. Future research should attempt toreplicate the pattern of results obtained here using blue-collar workers.

Fifth, the results of the present study outline an overall relationshipbetween the constructs of interest across several employees representingdifferent occupations and nations. However, moderator tests to compare, forinstance, the patterns of results obtained from samples of U.S. employeeswith the results from samples of South Korean employees were not conducted.The reason for this omission is that the Price and Mueller models in which allof the studies included in the present investigation were anchored, do notspecify such moderators. Also, past empirical research on the Price andMueller model does not suggest that any such moderators exist. Kim et al.(1996), e.g., tested empirically whether values (defined as employee concep-tions regarding the preferred courses of action; Kim et al., 1996) interact withthe structural determinants on their impact on job satisfaction and organiza-tional commitment. None of the interactions tested in that study reachedstatistical significance.

Sixth, this study lacks tests to probe the causal order of job satisfaction andorganizational commitment. The cross-sectional nature of the data obtained


here prohibited meaningful tests of that causal order (Bollen, 1989). Instead, Ispecified job satisfaction as a cause of organizational commitment which isconsistent with most current thinking in turnover research (e.g., Kim et al.,1996; Price & Mueller, 1986; Williams & Hazer, 1986) but by no meansundisputed (e.g., Farkas & Tetrick, 1989; Hom & Griffeth, 1995; Mathieu,1991). Future research should attempt to conduct longitudinal tests that allowmeaningful tests of the causal order of job satisfaction/organizational commit-ment in the context of models that allow for differential effects of structuraldeterminants on those two variables.

In sum, the present investigation tested several alternative ideas regardingthe relationship between structural determinants with job satisfaction andorganizational commitment originating from different turnover researchcamps. The results suggest that each of those ideas is both correct andincorrect to some degree. In that way, the findings summarized above empha-size the importance of collaboration and cross-fertilization among turnover(and other) researchers from different research camps.

NOTES

1. The following studies were included in the meta-analysis: Currivan, D. B. (1998).Unpublished Doctoral Dissertation, University of Iowa. Cyphert, S. T. (1990).Unpublished Doctoral Dissertation, University of Iowa. Han, N. C. (1992). UnpublishedDoctoral Dissertation, University of Iowa. Iverson, R. D. (1992). Unpublished DoctoralDissertation, University of Iowa. Jo, D. G. (1995). Unpublished Doctoral Dissertation,University of Iowa. Kim, S. W. (1996). Unpublished Doctoral Dissertation, University ofIowa. Kim, S. W., Price, J. L., Mueller, C. W. & Watson, T. W. (1996). Human Relations.Moorhead, S. A. (1993). Unpublished Doctoral Dissertation, University of Iowa.Mulinge, M. M. (1994). Unpublished Doctoral Dissertation, University of Iowa.

2. The Chi-square statistics can be requested from the author.

REFERENCES

Bentler, P. M. (1990). Comparative fit indexes in structural models. Psychological Bulletin,

107, 238±246.

Bentler, P. M. (1995). EQS structural equations program manual. Encino, CA: Multi-

variate Software.

Bollen, K. A. (1989). Structural equations with latent variables. New York: Wiley.

Brannick, M. T. (1995). Critical comments on applying covariance structure modeling.

Journal of Organizational Behavior, 16, 201±213.

Cohen, A. (1993). Organizational commitment and turnover: A meta-analysis. Academy of

Management Journal, 36, 1140±1157.

Cohen, A., & Hudecek, N. (1993). Organizational commitmentÐturnover relationship across

occupational groups. Group and Organization Management, 18, 188±213.

Cotton, J. L., & Tuttle, J. M. (1986). Employee turnover: A meta-analysis and review with

implications for research. Academy of Management Review, 11, 55±70.


Currivan, D. B. (1998). An analysis of causal relationships in a model of organizational

commitment. Iowa: University of Iowa (unpublished doctoral dissertation).

Cyphert, S. T. (1990). Employee absenteeism: An empirical test of a revision of the Brooke

model. Iowa: University of Iowa (unpublished doctoral dissertation).

Farkas, A. J., & Tetrick, L. E. (1989). A three-wave longitudinal analysis of the causal

ordering of satisfaction and commitment on turnover decisions. Journal of Applied

Psychology, 74, 855±868.

Han, N. C. (1992). An empirical study of an organizational commitment model in South

Korea. Iowa; University of Iowa (unpublished doctoral dissertation).

Hayduk, L. A. (1987). Structural equation modeling with LISREL: Essentials and advances.

Baltimore, MD: Johns Hopkins.

Hom, P. W., & Griffeth, R. W. (1995). Employee turnover. Cincinnati, OH: South Western.

Hom, P. W., Caranikas-Walker, F., Prussia, G. E., & Griffeth, R. W. (1992). A meta-analytical

structural equation analysis of a model of employee turnover. Journal of Applied


Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure

analysis: Conventional criteria versus new alternatives. Structural Equation Model-

ing, 6, 1±55.

Hunter, J. E., & Schmidt, F. L. (1990). Methods of meta-analysis. Newbury Park, CA: Sage.

Iverson, R. D. (1992). Employee intent to stay: An empirical test of a revision of the Price and

Mueller model. Iowa: University of Iowa (unpublished doctoral dissertation).

Jo, D. G. (1995). Employee attachment among teachers in South Korea: A comparative ana-

lysis with U.S. teachers. Iowa: University of Iowa (unpublished doctoral dissertation).

Kenny, D. A. (1979). Correlation and causality. New York: Wiley.

Kim, S. W. (1996). Employee intent to stay: The case of automobile workers in South Korea.

Iowa: University of Iowa (unpublished doctoral dissertation).

Kim, S. W., Price, J. L., Mueller, C. W., & Watson, T. W. (1996). The determinants of

career intent among physicians at a U.S. Air Force hospital. Human Relations, 49,

947±975.

Lee, T. W., & Mitchell, T. R. (1994). An alternative approach: The unfolding model of volun-

tary employee turnover. Academy of Management Review, 10, 51±89.

March, J. G., & Simon, H. A. (1958). Organizations. New York: Wiley.

Mathieu, J. E. (1991). A cross-level nonrecursive model of the antecedents of organizational

commitment and satisfaction. Journal of Applied Psychology, 76, 607±618.

Mobley, W. H. (1977). Intermediate linkages in the relationship between job satisfaction and

employee turnover. Journal of Applied Psychology, 62, 237±240.

Moorhead, S. A. (1993). Nurses' job satisfaction, commitment, search behavior and intent to

leave the Air Force: A test of a causal model. Iowa: University of Iowa (unpublished

doctoral dissertation).

Mowday, R. T., Porter, L. W., & Steers, R. M. (1982). Employee±organization linkages. New

York: Academic Press.

Mueller, C. W., & Price, J. L. (1990). Economic, psychological, and sociological determinants

of voluntary turnover. The Journal of Behavioral Economics, 19, 321±335.

Mueller, R. O. (1997). Structural equation modeling: Back to basics. Structural Equation

Modeling, 4, 353±369.

Mulinge, M. M. (1994). Job satisfaction and organizational attachment among agricultural

professionals in Kenya. Iowa: University of Iowa (unpublished doctoral dissertation).

Ostroff, C., & Harrison, D. A. (1999). Meta-analysis, level of analysis, and best estimates of

population correlations: Cautions for interpreting meta-analytic results in organiza-

tional behavior. Journal of Applied Psychology, in press.


Price, J. L. (1997). Handbook of organizational measurement. International Journal of

Manpower, 18, 303±558.

Price, J. L., & Mueller, C. W. (1986). Absenteeism and turnover of hospital employees. In

S. B. Bacharach (Ed.), Monographs in organizational behavior and industrial rela-

tions (Vol. 5). Greenwich, CON: JAI Press.

Steel, R. P., & Ovalle, N. K. (1984). A review and meta-analysis of research on the

relationship between behavioral intentions and employee turnover. Journal of

Applied Psychology, 69, 673±686.

Tett, R. P., & Meyer, J. P. (1993). Job satisfaction, organizational commitment, turnover

intention, and turnover: Path analyses based on meta-analytic findings. Personnel


Viswesvaran, C., & Ones, D. Z. (1995). Theory testing: Combining psychometric meta-ana-

lysis and structural equations modeling. Personnel Psychology, 48, 865±885.

Williams, L. J., & Hazer, J. T. (1986). Antecedents and consequences of job satisfaction and

commitment in turnover models: A reanalysis using latent variable structural equa-

tion methods. Journal of Applied Psychology, 71, 219±231.


structural determinants of job satisfaction and organizational commitment in turnover models

Documents