a longitudinal analysis of the relationship between multiple commitments and withdrawal cognitions

23
Management SCANDINAVIAN JOURNAL OF Scand. J. Mgmt. 21 (2005) 329–351 A longitudinal analysis of the relationship between multiple commitments and withdrawal cognitions Aaron Cohen , Anat Freund Department of Political Science, University of Haifa, Haifa 31905, Israel Received 1 October 2002; accepted 1 June 2004 Abstract This paper examines the relationship between multiple commitments (affective organiza- tional commitment, continuance organizational commitment, occupational commitment, and job involvement) and withdrawal cognitions, with three different time intervals between the two. One hundred and twenty-two community center employees (a 37% response rate) in Israel participated in the study. The findings showed that commitment forms were related to withdrawal cognitions, even when withdrawal cognitions measured earlier than or at the same time as commitment forms were controlled for. The results also showed that the timing of the measurement of the research variables had a strong effect on the findings. More specifically, the prediction of withdrawal is better, the shorter the interval between its measurement and the measurement of multiple commitments. The findings also showed that commitment and withdrawal are both dynamic concepts. That is, the effect of timing on the accuracy of the prediction can be a result of more immediate changes in commitment forms across time, or more immediate changes in withdrawal cognitions over time. Other implications of the findings for future research on commitment and withdrawal are also discussed. r 2005 Elsevier Ltd. All rights reserved. Keywords: Commitment; Withdrawal cognitions; Turnover ARTICLE IN PRESS www.elsevier.com/locate/scajman 0956-5221/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.scaman.2004.06.004 Corresponding author. Tel.: +972 4 824 0041; fax: +972 4 825 7785. E-mail address: [email protected] (A. Cohen).

Upload: aaron-cohen

Post on 27-Oct-2016

224 views

Category:

Documents


0 download

TRANSCRIPT

ARTICLE IN PRESS

ManagementS C A N D I N AV I A N J O U R N A L O F

Scand. J. Mgmt. 21 (2005) 329–351

0956-5221/$ -

doi:10.1016/j

�CorrespoE-mail ad

www.elsevier.com/locate/scajman

A longitudinal analysis of the relationshipbetween multiple commitmentsand withdrawal cognitions

Aaron Cohen�, Anat Freund

Department of Political Science, University of Haifa, Haifa 31905, Israel

Received 1 October 2002; accepted 1 June 2004

Abstract

This paper examines the relationship between multiple commitments (affective organiza-

tional commitment, continuance organizational commitment, occupational commitment, and

job involvement) and withdrawal cognitions, with three different time intervals between the

two. One hundred and twenty-two community center employees (a 37% response rate) in

Israel participated in the study. The findings showed that commitment forms were related to

withdrawal cognitions, even when withdrawal cognitions measured earlier than or at the same

time as commitment forms were controlled for. The results also showed that the timing of the

measurement of the research variables had a strong effect on the findings. More specifically,

the prediction of withdrawal is better, the shorter the interval between its measurement and the

measurement of multiple commitments. The findings also showed that commitment and

withdrawal are both dynamic concepts. That is, the effect of timing on the accuracy of the

prediction can be a result of more immediate changes in commitment forms across time, or

more immediate changes in withdrawal cognitions over time. Other implications of the

findings for future research on commitment and withdrawal are also discussed.

r 2005 Elsevier Ltd. All rights reserved.

Keywords: Commitment; Withdrawal cognitions; Turnover

see front matter r 2005 Elsevier Ltd. All rights reserved.

.scaman.2004.06.004

nding author. Tel.: +972 4 824 0041; fax: +972 4 825 7785.

dress: [email protected] (A. Cohen).

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351330

1. Introduction

Since the 1970s the concept of commitment in the workplace has been attractingthe attention of both academics and practitioners. There are many definitions ofcommitment, but the most common ones see it as an attitude that reflects feelingssuch as attachment, identification or loyalty to the object of the commitment(Morrow, 1993). Commitment was first proposed as a concept to replace orstrengthen other explanations that had produced somewhat disappointing findingsregarding the effects of job satisfaction on work behaviors such as turnoverand absenteeism (Mowday, Porter, & Steers, 1982). Lack of commitment or loyaltyhas been cited as an explanation for employee absenteeism, turnover, reducedeffort expenditure, theft, job dissatisfaction, and unwillingness to be relocated(Morrow, 1993). Today it is quite clear that neither of these phenomena has beenexplained in the way the researchers had anticipated. Commitment was foundto be only a modest predictor of turnover, certainly not of the magnitude expected.Some researchers, indeed, questioned the need for future research on this concept(Randall, 1990).Researchers naturally sought alternative ways of seeking to understand commit-

ment and to pinpoint its contribution in both theory and practice. One of theimportant alternative approaches was to characterize commitment as a multi-dimensional concept. Such an approach has been advanced by Morrow (1983, 1993)and in the works of Reichers (1985, 1986), Becker (1992), Cohen (1998, 1993a,1999a, b, 2000, 2003), Blau, Paul and St. John (1993) and others, and has come torepresent what can be termed the multiple-commitments approach.As a result, there has been a growing interest in the concept of multiple

commitments in recent years among researchers and practitioners, as evinced by thegreater attention to this concept in the literature. This increased interest includesspecific objects of commitment such as organization, work group, occupation, theunion, and one’s job (Blau, Paul, & St. John, 1993; Cohen, 1993b, 1999a, 2003;Randall & Cote, 1991). Perhaps most critical for an appreciation of the multiple-commitment approach is the extent to which it has ‘‘added value.’’ In other words, ifthe recognition that we live in a multiple-commitment world is to have eithertheoretical or practical importance, it must be shown that examining commitment tovarious constituencies and domains refines our understanding of work-relatedbehavior (Meyer & Allen, 1997; Morrow, 1993). Indeed, forms of commitment havebeen shown to predict important work outcomes such as withdrawal, performance,absenteeism, and tardiness (e.g., Blau, 1986; Cohen, 1993a, 1999b, 2000, 2003;Randall & Cote, 1991; Steers & Rhodes, 1978; Wiener & Vardi, 1980).Despite the growing interest in the concept of multiple commitments and its

relationship to outcomes, there is still a need for more research on this relationship.The relationship between commitment and outcomes is still somewhat unclear. First,assuming that commitment forms are related in different ways to outcomes, manymore findings are needed to clarify the nature of this relationship. We needto establish which commitment forms are related to which outcomes, and to lookat the magnitude of this relationship. There are some general hypotheses for

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351 331

this relationship, but few findings, and much more research is therefore needed toclarify it.Second and more importantly, most of the findings on the relationship between

multiple commitments and outcomes are based on cross-sectional designs. Naturallysuch designs are exposed to common methodological errors. Such errors are moreproblematic in the case of multiple commitments because the number ofintercorrelations among such commitments is relatively high, given that there aresome conceptual similarities among them (Morrow, 1983). Longitudinal designs aretherefore needed to provide more accurate findings on the nature of the relationshipbetween multiple commitments and outcomes. Beck and Wilson (2001) argued thatorganizations today are not static. Therefore, the identification of a cohort andmeasurements taken over a period of time may help direct the search for the factorsthat cause changes in work commitments.This study is aimed primarily at addressing the two needs outlined above. With the

help of a longitudinal design, the relationship is examined between multiplecommitments—namely affective and continuance organizational commitment,occupational commitment and job involvement—on the other hand, and threeforms of withdrawal cognitions on the other. Data on multiple commitments andwithdrawal cognitions was collected on three time frames. Longitudinal studies onmultiple commitments of this kind have rarely been performed. This data will enableus to test the relationship between multiple commitments and withdrawal cognitions,given different lengths of time between the two. We can also control for withdrawalcognitions at an earlier time. Such a design can provide a more accurate evaluationof the relationship between commitment and withdrawal cognitions and cancontribute to our understanding of the relationship between attitudes (commitmentin this present case) and turnover.

1.1. Conceptual framework and research hypotheses

1.1.1. Multiple commitments and outcomes

Proposed linkages between commitment and outcomes were initially driven by theproperties ascribed to each form of work-related commitment. One of thehypothesized benefits of being committed at work (job, career, and/or organization)is improved job performance (Somers & Birnbaum, 1998). The logic or theory as towhy commitment forms should be related to outcomes follows the exchangeapproach. Employees who experience positive exchanges with the organization, thejob, or the work group will reciprocate with higher levels of commitment, which willmotivate them to contribute to the organization in other ways, such as reducedturnover and absenteeism or better performance. This logic can be applied to therationale for examining multiple commitments.Another important aspect of the relationship between commitment forms and

outcomes concerns the nature of the relationship itself. How are different forms ofcommitment related to outcomes? One approach has argued that commitment formscan be expected to be related differently to various outcomes. It is contended that thetype of work-related commitment under consideration might have some bearing on

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351332

how commitment might direct behavior in organizations (Shore, Newton &Thornton, 1990; Siders, George, & Dharwadkar, 2001; Somers & Birnbaum,1998). As different objects of work commitment represent distinct attitudes,differential effects on behavioral outcomes can be expected. Researchers shouldtherefore attempt to match the focus of their independent variable with the focus oftheir work outcome variable (Becker, 1992).Given that the object of organizational commitment is the employing organiza-

tion, the most likely behavior to be affected by this commitment is organization-oriented behavior such as turnover intentions, actual turnover (Mowday et al.,1982), absenteeism, and organizational citizenship behavior, particularly thecompliance dimension. Similarly, the most likely behavior to be affected by jobinvolvement is task-oriented behavior such as organizational citizenship behaviorand performance. In keeping with previous empirical evidence (Blau, 1986), jobinvolvement is also expected to have a stronger relationship with absenteeism thanorganizational commitment. Behaviors that are expected to be related tooccupational commitment are occupation/career-oriented behavior such as with-drawal cognitions and actual withdrawal from the occupation. One should note inthis regard that there is some support in the literature for commitments affectingbehaviors that are not directly related to the focus of that given commitment. Aluttoand Belasco (1974), for example, found negative relationships between organiza-tional commitment and attitudinal militancy among teachers and nurses. Dalton andTodor (1982) found that grievance-filing behavior was more a function of a negativerelationship with organizational commitment than of a positive relationship withunion commitment.A second approach is based on the logic of proximity as presented by Cohen

(2003). This approach is based on Lawler’s (1992) theory of attachment attributedmainly to nested subgroups. This theory or principle of proximal rules explains why‘‘actors develop stronger affective ties to subgroups within a social system ratherthan to the social system, to local communities rather than to states, to workorganization, and so forth’’ (Lawler, 1992, p. 334). Interpersonal attachmentproduces a stronger commitment to subgroups than to the larger group, because thecredit for positive results from interpersonal bonds is likely to be attributed to theproximal subgroups, while the blame for negative effects is likely to be attributed tothe large group (Lawler, 1992). The above logic can explain why a person willdevelop a stronger personal attachment to their job than to their career. People candevelop a stronger attachment to the job, which is a proximal target in theirimmediate work unit, than to their career, which is a much more distant target.Gregersen’s (1993) and Mueller and Lawler’s (1999) argument is based primarily

on the idea that proximal variables exert the most significant influence on employees’actions because proximity provides more opportunities for exchange relationships.Mueller and Lawler (1999) focused on the nested nature of organizational units. Thegeneral principle they proposed is that, given the nesting of one unit within another,employees’ commitment to a particular organizational unit in the structure will beaffected primarily by the work conditions that are created and controlled by thatparticular unit. Commitment to the most proximate unit will be influenced especially

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351 333

by work conditions, because day-to-day experiences in the local unit exert thestrongest effects on positive emotions, and positive emotions produce commitmentprimarily to this more proximate unit.According to Gregersen, individuals can come to identify strongly with and

become significantly attached to proximal and potentially influential foci.Gregersen’s (1993) argument can usefully be applied to the interrelationships amongthe commitment forms, and seems to provide another explanation for the way thecommitment foci are positioned and related to outcomes. The organization, the job,and the occupation provide more proximal foci in terms of their relationship to theimmediate work setting. They are therefore expected to have stronger relationshipswith work outcomes. The Protestant work ethic and work involvement are lessproximal in terms of their relationships to the work setting. The Protestant workethic and work involvement should have some effect on work outcomes. People whobelieve in hard work and in the value of work would be expected to demonstrate thisbelief in their behavior at work. However, because the foci of these commitments areless proximal to the work setting than the organization, the job, and the occupationfoci, their effect is expected to be weaker than these commitments.An interesting and fruitful explanation, a third one, was advanced by Vandenberg

and Scarpello (1994) who argued that some settings might value an occupation morehighly than others, thereby affecting the strength of the relationship betweenoccupational commitment and organizational commitment and work outcomes. Ifan employee is more strongly committed to the occupation than to the organization,it can be expected that this commitment will have a stronger effect on behavioroutcomes than on organizational commitment. Where an employee is more firmlycommitted to the organization than to the occupation, we would expect thatorganizational commitment would affect her/his behavior (Somers & Birnbaum,1998) more strongly than occupational commitment.

1.1.2. Research hypotheses

In this paper withdrawal cognitions, and not actual turnover, will be thedependent variables. Krausz, Koslowsky, Shalom, and Elyakim (1995) asserted thatstudies have shown that intentions to leave are a better predictor of actual turnoverthan are attitudinal measures. They argued that studies using intentions as thecriterion have distinct advantages over those with behavior as the dependentvariable. First, behaviors are often influenced by a host of other variables (e.g.,company policies, economic conditions) that cannot be controlled by theinvestigator, but do influence the findings. Second, turnover research has shownthat a person’s self-expressed intentions are the best predictor of turnover.Therefore, turnover intentions can be used as the dependent variable, and anassumption can be made that some action on the part of the employee is likely tofollow. Moreover, Price and Mueller (1981) argued that expressed intentions ofleaving a job or an occupational field are a statement of an emotional response towork or the profession. Parasuraman (1989) in a study of turnover among staffnurses asserted that intentions play a key-mediating role between attitudes andturnover and are the immediate determinant of actual turnover. In the present study

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351334

three forms of withdrawal cognitions will be examined, namely, attraction of presentjob, availability of alternatives and turnover intentions.An important argument mentioned above is that the relationship between

commitment forms and work outcomes, turnover in particular, is not direct, but ismediated by behavioral intentions. While this assumption has its roots in theliterature that examined the relationship between commitment forms and outcomesseparately, it also seems to affect research on the multiple commitment–outcomesrelationship. This conjecture arises from strong evidence that the relationshipbetween attitudes (commitment foci in our case) and behaviors (turnover in our case)is not direct. Mobley (1977) and Mobley, Griffeth, Hand, and Meglino (1979)suggested a linear process model whereby an individual’s negative work attitudestrigger ideas about leaving an organization, leading to the intention to search forother employment, which precipitates a choice about staying or leaving, resulting ina final decision to quit (Farrell & Petersen, 1984; Parasuraman, 1982; Rosse, 1988).Four hypotheses will be advanced here two of them based on the above rationale.

All of them are based on the above argument that suggests the existence of awithdrawal process, namely thoughts of quitting, availability of alternativesbehavior, and turnover intentions. The first hypothesis predicts a differentiatedrelationship between the various forms of withdrawal cognitions and multiplecommitments. It is expected that commitment forms will be related in different waysto the three withdrawal attitudes. As organizational commitment is expected to berelated to organizationally oriented behaviors and attitudes, we expect that its twoforms, affective, and continuance, will be more strongly related to withdrawalcognitions than occupational commitment and job involvement. Occupationalcommitment and job involvement are expected to contribute above and beyond theeffect of organizational commitment.The second hypothesis is more specific in its prediction. In this hypothesis we

argue that commitment foci will be related in different ways to different forms ofwithdrawal cognitions. This hypothesis is based on the proximity rationale outlinedabove (Gregersen, 1993; Mueller & Lawler, 1999). The logic explains why a personwill develop a stronger personal attachment to their job, which is a proximal target intheir immediate work unit, than to their career, which is a much more distant target.As mentioned above, this argument is based primarily on the idea that proximalvariables exert the most significant influence on employees’ actions, becauseproximity provides more opportunities for exchange relationships.Consequently, the attraction of a present job and the availability of alternatives

constitute proximal factors relative to occupational commitment. We expect them tobe related less strongly to organizational commitment and more strongly tooccupational commitment. Turnover intention, on the other hand, is an attitude thatis more proximal to the organization. An individual develops intentions to leave aspecific organization. Turnover intention therefore is expected to be strongly relatedto organizational commitment and less strongly to occupational commitment. Theabove hypothesis is supported by earlier findings, demonstrating that commitmentfoci have differential relationships with various forms of withdrawal intentions(Cohen, 1993a, b; Krausz et al., 1995).

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351 335

Hypothesis 1. Affective and continuance organizational commitment will be morestrongly related to withdrawal cognitions than occupational commitment and jobinvolvement. Occupational commitment and job involvement will also contribute toexplaining the variance of withdrawal cognitions.

Hypothesis 2. Commitment forms will be related in different ways to differentforms of withdrawal. Turnover intentions will be more strongly related to affectiveand continuance organizational commitment. Attraction of present job andavailability of alternatives will be more strongly related to occupational commit-ment.

1.1.3. A longitudinal analysis of multiple commitments and withdrawal cognitions

The possibility that early withdrawal cognitions can mediate the relationshipbetween commitment forms and withdrawal cognitions measured at a later time hasrarely been studied in commitment research. In other words, employees at a giventime have not only a perception of their commitments to given objects, but also aperception of their withdrawal cognitions. This perception of withdrawal cognitionscould affect withdrawal cognitions in the future. If commitment forms are ameaningful determinant of withdrawal cognitions, then they should affect the latterabove and beyond the effect of these withdrawal cognitions measured at the sametime or after the measurement of commitment forms, but before the measurement ofwithdrawal cognitions as the dependent variable.This hypothesis is important particularly in light of the argument that there is a

conceptual similarity or even a concept redundancy between commitment forms andwithdrawal cognitions (Morrow, 1993; Mowday et al., 1982; O’Reilly & Chatmen,1986; Reichers, 1985). If this is true, perhaps there is no need to considercommitment when predicting turnover. However, if we are able to show thatcommitment at a given time can affect withdrawal cognitions measured later, evenwhen we control for withdrawal cognitions measured at the same time ascommitment or measured before the measurement of commitment, then the valueand importance of commitment as a unique and independent predictor of turnoverwill be demonstrated.

Hypothesis 3. Commitment forms will affect withdrawal cognitions above andbeyond the effect of withdrawal cognitions measured at the same time or before themeasurement of commitment forms and before the measurement of withdrawalcognitions as the dependent variable.

Finally Porter, Crampon, and Smith (1976) compared ‘‘stayers’’ and ‘‘leavers’’ atthree different points in time before the leavers actually quit their jobs and foundthat leavers who were 11/2 month or less short of quitting, reported significantly lesscommitment than stayers. When the leavers were 2–31/2 months short of quitting,they showed less commitment than equivalent stayers, but the difference was notsignificant. If eventual leavers were at least 6 months away from terminating theiremployment, their commitment was almost the same as that of equivalent stayers.These findings suggest that even though organizational commitment as an attitude is

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351336

perceived to be ‘‘somewhat more stable over time than job satisfaction,’’ (Mowdayet al., 1982, p. 28), the decline in commitment can be very rapid.If, for example, two people reported high levels of organizational commitment

at a given time, the prediction for both of them would be that they would remain inthe organization. Assuming that turnover data was collected one year later,and it was found that a decline in the level of organizational commitment hadoccurred for one of them two months after the commitment measurement and hadresulted in turnover, then this observation would contradict the original predictionand reduce its accuracy. The decline in organizational commitment was not capturedbecause of the long interval between the measurement of the attitude and thecollection of the turnover data. The longer this interval, the greater the probabilitythat additional events in the organization have caused more employees to changetheir level of commitment. This change may add to the errors in prediction andreduce the validity of the commitment–turnover relationship. This argument issupported by Cohen’s (1993a, b) meta-analysis as well as by the meta-analyses ofHom, Pruaaia, and Griffeth (1992) and Steel and Ovalle (1984). It is also supportedby Mitra, Jenkins, and Gupta (1992) and Morrow, McElroy, Laczniak, and Fenton(1999), who examined the absenteeism–turnover relationship. In the study athand we did not collect turnover data. However, we will advance a similarhypothesis based on the same logic as described above by Cohen (1993a, b),but using withdrawal cognitions as the dependent variable. Briefly, it is expected thatthe shorter the interval between the measurement of commitment forms andwithdrawal cognitions, the stronger will be the effect of commitment forms onwithdrawal cognitions.

Hypothesis 4. The relationship between commitment forms and withdrawalcognitions will be stronger, the shorter the intervals between the measurements ofthe two variables.

2. Method

2.1. Subjects and procedure

The target population of this study consisted of 327 employees, most of themcommunity center managers and all of them working for a national nonprofitorganization in Israel. All the employees of this organization were asked toparticipate in the study. Seventy-four of the employees (61%) who did participatewere community center managers, 14 (11%) were district managers, and 34 (28%)worked at the head office of the organization. The survey was administered on threeoccasions at intervals of 6 months. In the first round all 327 employees received thequestionnaire, and 203 (62%) of them returned usable questionnaires. In the secondround all 203 employees who had completed the first survey received thequestionnaire, and 153 (75.4%) of them returned usable questionnaires. In thethird and last round 153 employees received the questionnaire, and 122 (80%) of

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351 337

them returned usable questionnaires. Thus, the final sample includes 122 employeeswho completed the questionnaire in all three rounds. The response rate relative tothe total target population (327) is 37%.A comparison of the characteristics of the final sample (122 employees) with those

of the target population (327 employees) revealed similarities between the twogroups: 64% males in the final sample compared to 69% in the target population;95% Jewish in the final sample compared to 89% in the target population; anaverage age of 45.6 in the final sample compared to 42.3 in the target population; andan average tenure of 8.9 years in the final sample compared to 7.3 in the targetpopulation. We also compared the demographic characteristics of those whodropped out of the study with those who continued and found no meaningfuldifferences between the groups. The main conclusion from the above data is that thefinal sample represents the target population quite well. Analysis of some othercharacteristics of the final sample shows that 86% had a university degree, more than50% were in managerial positions, and 77% had tenure in the organization.

2.2. Measures

2.2.1. Control variables

Three variables were controlled for in the analysis. Education and tenure areknown predictors of work attitudes, such as organizational commitment, which inturn have been found to predict work withdrawal (Mathieu & Zajac, 1990).Education was measured on a scale of 1–5, the lowest indicating completion of highschool or less and the highest indicating completion of an advanced universitydegree. Tenure with the organization was measured in years. Gender was alsocontrolled for and was measured as a dichotomous variable, whereby 0 ¼ maleand 1 ¼ female.

2.2.2. Commitment foci

Affective and continuance organizational commitment were measured by the8-item scale for each one as developed by Meyer and Allen (1984), with a Cronbachalpha of .70 and .68, respectively. These alphas were found in the first round of thesurvey, and they are based on the larger sample (N-327). The alphas based on thefinal sample and each round of the survey are reported in Table 1. Careercommitment was measured by the 8-item measure developed by Blau (1985), and theresultant Cronbach alpha was .77. Job involvement (10 items) was measured by thescale developed by Kanungo (1979, 1982), with a Cronbach alpha of .83. Allcommitment constructs were measured on a 7-point scale (1 ¼ strongly disagree to7 ¼ strongly agree). All the commitment scales applied in this research were noted byMorrow (1993) as being the most commonly used as well as the most reliable andvalid. They were all also mentioned as having strong discriminant validity in theirrelationship to other commitment forms. Thus, the selection of the scales fit well withMorrow’s conceptualization. It should be noted that all the measures used here hadbeen translated from English to Hebrew in earlier works by one of the presentauthors (Cohen, 1999b; Cohen, 2000). The present study used these translations.

ARTIC

LEIN

PRES

S

Table 1

Intercorrelations among commitment forms and withdrawal cognitions in the three survey rounds (reliabilities in parentheses)

Affective

organizational

commitment

Continuance

organizational

commitment

Occupational

commitment

Job

involvement

Availability

of alternatives

Attraction

of present job

Turnover

intentions

Time

1

Time

2

Time

3

Time

1

Time

2

Time

3

Time

1

Time

2

Time

3

Time

1

Time

2

Time

3

Time

1

Time

2

Time

3

Time

1

Time

2

Time

3

Time

1

Time

2

Time

3

Affective

organizational

commitment

Time 1 (.66)

Time 2 .72** (.68)

Time 3 .59** .73** (.77)

Continuance

organizational

commitment

Time 1 .10 .10 .22* (.68)

Time 2 .08 .15 .22* .69** (.68)

Time 3 .16 .19* .30** .64** .63** (.67)

Occupational

commitment

Time 1 .62** .45** .39** �.06 �.12 .02 (.77)

Time 2 .50** .61** .49** �.15 �.19* �.02 .69** (.77)

Time 3 .49** .59** .75** .04 �.01 .07 .56** .70** (.79)

Job

involvement

Time 1 .59** .39** .36** �.05 �.12 �.04 .64** .47** .39** (.83)

Time 2 .42** .49** .39** �.08 �.11 �.03 .46** .63** .53** .68** (.81)

Time 3 .42** .53** .61** .08 �.03 .08 .43** .57** .69** .59** .81** (.86)

Availability

of alternatives

Time 1 .14 .05 .12 .56** .50** .40** �.09 �.16 .02 �.12 �.21* �.11 (.63)

Time 2 .10 .06 .15 .50** .67** .43** �.16 �.16 �.02 �.07 �.16 �.09 .66** (.67)

Time 3 .05 �.05 .08 .33** .48** .35** �.04 �.08 .03 �.16 �.13 �.12 .54** .67** (.61)

Attraction of

present job

Time 1 .13 �.04 �.03 �.08 .15 .09 .09 �.05 �.00 .07 �.02 �.06 .14 .15 .28** (.62)

Time 2 .13 .19* .19* .12 .19* .15 .10 .20* .17 .03 .04 .04 .08 .14 .10 .54** (.55)

Time 3 .08 .14 .22* .08 .19* .08 .07 .18 .19* �.09 �.01 .12 .13 .10 .17 .38** .47** (.65)

Turnover

intentions

Time 1 �.52** �.38** �.31** .00 .03 �.07 �.64** �.45** �.38** �.41** �.24* �.24** �.05 .05 �.11 �.29** �.26** �.14 (.74)

Time 2 �.44** �.60** �.48** �.07 �.10 �.13 �.41** �.62** �.48** �.27** �.38** �.35** �.10 �.06 �.09 �.13 �.39** �.36** .59** (.78)

Time 3 �.35** �.49** �.62** �.13 �.15 �.23* �.34** �.49** �.56** �.17 �.30** �.42** �.00 .03 �.12 �.09 �.32** �.44** .52** .65** (.86)

N ¼ 118–122.

*Pp.05; **Pp.01; ***Pp.001.

A.

Co

hen

,A

.F

reun

d/

Sca

nd

.J

.M

gm

t.2

1(

20

05

)3

29

–3

51

338

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351 339

2.2.3. Withdrawal cognitions

Three variables measured three aspects of the process of withdrawal, as describedby the model in Mobley et al. (1979). According to these authors intention to searchand search behavior could generally be expected to precede intention to quit andturnover. Following this logic, we measured the two above intentions. In addition,Mobley et al. pointed out that one of the primary determinants of intentions isthought to be the attraction or expected utility of the present job. This variable wasalso measured in our study.The variable, attraction of present job, was defined as an individual’s perception of

the attractiveness of her/his present job as demonstrated by her/his willingness ormotivation to continue working in it. Two items measured this variable: (1) I intendto ask for unpaid leave soon and (2) I intend to reduce the extent of my employmentsoon. The higher the score on this scale, the lower the attraction of the job. Theresultant Cronbach alpha based on the first round of the survey (N ¼ 203) was .61.The variable, availability of alternatives, was also based on the Mobley et al. modeland was defined as the individual’s intentions to search for alternatives to theircurrent employment. Three items measured this variable: (1) I can find a better job inmy profession, (2) There are a lot of employment opportunities in my occupation,and (3) If I wanted to, I could get a job in my occupation in a different organization.The Cronbach alpha here was .63. On the basis of the Mobley et al. model, turnover

intentions were defined as the magnitude of an employee’s intentions to leave his/hercurrent organization. This variable was measured by four items: (1) I see myselfworking in this organization three years from now, (2) I intend to leave theorganization, (3) I perceive the probability of staying in this organization as verygood, and (4) I would like to keep on working for this organization for a long time.The resultant Cronbach alpha was .80. All withdrawal cognitions constructs werealso measured on a 7-point scale (1 ¼ strongly disagree to 7 ¼ strongly agree).

3. Results

Table 1 presents the intercorrelations among commitment forms and withdrawalcognitions in the three administrations of the survey. It also shows the reliabilities ofthe measures at each administration. The findings indicated acceptable reliabilitylevels for the research variables. Only in one case, attraction of present job at time 2,was the reliability below .60 (.55). It should be noted that this scale includes twoitems. The correlations among commitment forms did not exceed .70 except for thecorrelation between affective organizational commitment and occupational commit-ment at time 3 (.75). None of the correlations between the commitment forms andthe withdrawal cognitions variables exceeded .70. Therefore, there seemed to be noproblem of multicollinearity.According to Hypotheses 1 and 2 commitment forms were expected to relate in

different ways to withdrawal cognitions. Tables 2 and 3 present the results ofmultiple hierarchical regressions between commitment forms and withdrawalcognitions. In each regression the control variables were introduced into the

ARTIC

LEIN

PRES

S

Table 2

Hierarchical regression analysis (standardized coefficients) of withdrawal cognitions (time 3) on control variables and multiple commitments (time 1)a

Attraction of present job (Time 3) Availability of alternatives (Time 3) Turnover intentions (Time 3)

Step 1 Step 2 Step 1 Step 2 Step 1 Step 2

Gender �.02 �.01 .08 .03 �.08 �.14

Education .02 .02 .02 .01 .10 .11

Tenure �.02 �.03 .04 �.02 �.08 �.02

Job involvement (Time 1) �.29* �.24* .15

Occupational commitment

(Time 1)

.20 .06 �.35**

Affective organizational

commitment (Time 1)

.10 .13 �.21

Continuance

organizational commitment

(Time 1)

.07 .32*** �.13

R2 (Adjusted) .00 (.00) .06 (.00) .01 (.00) .15 (.09) .02 (.00) .21 (.16)

F .05 .95 2.66* .84 4.09***

DR2 .06 .14 .19

F for DR2 1.62 4.44** 6.40***

N ¼ 118–122.

*Pp.05; **Pp.01; ***Pp.001.aIn this table commitment forms at time 1 are related to withdrawal cognitions at time 3. The control variables were introduced into the equation first in step

1, and commitment forms were entered next, in step 2.

A.

Co

hen

,A

.F

reun

d/

Sca

nd

.J

.M

gm

t.2

1(

20

05

)3

29

–3

51

340

ARTIC

LEIN

PRES

S

Table 3

Hierarchical regression analysis (standardized coefficients) of withdrawal cognitions (time 3) on control variables and multiple commitments (time 2)a

Attraction of present job (Time 3) Availability of alternatives (Time 3) Turnover intentions (Time 3)

Step 1 Step 2 Step 1 Step 2 Step 1 Step 2

Gender �.02 �.03 .08 .01 �.08 �.14

Education .02 �.02 .02 �.01 .10 .14

Tenure �.02 �.10 .04 �.06 �.08 .04

Job involvement (Time 2) �.18 �.17 .06

Occupational commitment (Time 2) .41** .31* �.40**

Affective organizational commitment (Time 2) �.03 �.18 �.28*

Continuance organizational commitment (Time 2) .28** .51*** �.17

R2 (Adjusted) .00 (.00) .12 (.06) .01 (.00) .23 (.18) .02 (.00) .35 (.30)

F .05 2.09* .26 4.45*** .84 8.07***

DR2 .12 .22 .33

F for DR2 3.63** 7.56*** 13.20***

N ¼ 118–122.

Pp.05; **Pp.01; ***Pp.001.aIn this table commitment forms at time 2 are related to withdrawal cognitions at time 3. The control variables were introduced into the equation first in step

1, and commitment forms were entered next, in step 2.

A.

Co

hen

,A

.F

reun

d/

Sca

nd

.J

.M

gm

t.2

1(

20

05

)3

29

–3

51

341

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351342

equation first in step 1, and commitment forms were entered next in step 2. Thedifference between the three tables lies in the time at which the measurement ofcommitment forms and withdrawal cognitions was made. In Table 2 commitmentforms at time 1 are related to withdrawal cognitions at time 3. In Table 3commitment forms at time 2 are related to withdrawal cognitions at time 3.Hypothesis 1 predicted that forms of commitment would be related in different

ways to different forms of withdrawal. The data provides some support for thishypothesis. Affective organizational commitment and occupational commitment arerelated to turnover intentions (see Table 3). Continuance organizational commitmentand occupational commitment are related to availability of alternatives (see Table 3).The effect of commitment forms on attraction of present job is unstable, in the sensethat it differs with each combination of time measurement. When it is measured attime 3, continuance organizational commitment and occupational commitment attime 2 are related to this effect (see Table 3), while job involvement only at time 1 isrelated to it (see Table 2).The first part of Hypothesis 2 predicted that affective and continuance

commitment would be more strongly related to withdrawal cognitions than jobinvolvement or occupational commitment. The findings in Tables 2 and 3 do notprovide strong support for Hypothesis 2. Continuance organizational commitmentseems to be the strongest predictor of availability of alternatives under any timecombination of the measurement of the research variables. However, continuanceorganizational commitment is not related to turnover intentions in any of the abovetables and is related to attraction of present job only when it was measured at time 2and attraction of present job at time 3. Affective organizational commitment isrelated only to turnover intentions, except when it was measured at time 1 andturnover intentions at time 3. Thus, the two organizational foci commitments seem tohave a more modest effect on withdrawal cognitions than we expected.The second part of Hypothesis 2 expected that occupational commitment and job

involvement would have some effect on withdrawal cognitions. The findings inTables 2 and 3 provide some support for this expectation. Occupational commitmentmeasured at time 2 is related to all three forms of withdrawal measured at time 3 (seeTable 3). The effect of occupational commitment was expected according toHypothesis 2. Job involvement seems to have a weaker effect on withdrawalcognitions than occupational commitment. Only when measured at time 1 is itrelated to attraction of present job and availability of alternatives measured at time 3(see Table 2).Hypothesis 3 predicted that commitment forms would be related to withdrawal

cognitions above and beyond the effect of withdrawal cognitions measured at thesame time or before the measurement of commitment forms. Tables 4 and 5 providethe results of the hierarchical regression analysis. In each table there are three steps.In all the tables the control variables were introduced in step 1. In Table 4withdrawal cognitions at time 1 were introduced in step 2 and commitment forms instep 3. In Table 5 the dependent variables (withdrawal cognitions) were measured attime 3, the mediating variables (withdrawal cognitions) were measured at time 1, andcommitment forms were measured at time 2. In general the findings support

ARTIC

LEIN

PRES

S

Table 4

Hierarchical regression analysis (standardized coefficients) of withdrawal cognitions (time 3) on control variables, withdrawal cognitions (time 1) and multiple

commitments (time 1)a

Attraction of present job (Time 3) Availability of alternatives (Time 3) Turnover intentions (Time 3)

Step 1 Step 2 Step 3 Step 1 Step 2 Step 3 Step 1 Step 2 Step 3

Gender �.07 �.02 .00 .18 �.03 �.05 �.19 �.33 �.34

Education .04 .05 �.04 .03 .01 .03 .11 �.02 .00

Tenure �.00 .00 �.01 .01 �.00 �.01 �.02 �.01 .00

Attraction of present job (Time 1) .44*** .46*** .22* .24* .09 .05

Availability of alternatives (Time 1) .08 .02 .52*** .45*** .02 .17

Turnover intentions (Time 1) �.05 .01 �.03 �.10 .72*** .62***

Job involvement (Time 1) �.42* �.22 .22

Occupational commitment (Time 1) .25 .05 �.09

Affective organizational commitment (Time 1) .09 �.05 �.25

Continuance organizational commitment (Time 1) .16 .14 �.27*

R2 (Adjusted) .00 (.00) .15 (.10) .20 (.12) .01 (.00) .34 (.30) .37 (.31) .02 (.00) .28 (.24) .33 (.27)

F .07 3.08** 2.51** .25 9.21*** 6.02*** .84 7.13*** 5.19***

DR2 .15 .05 .33 .03 .26 .05

F for DR2 6.09*** 1.56 18.06*** 1.16 13.14*** 1.93

N ¼ 118–122.

Pp.05; **Pp.01; ***Pp.001.aThe control variables were entered in step 1, withdrawal cognitions at time 1 were entered in step 2 and commitment forms were entered in step 3. The

dependent variables (withdrawal cognitions) were measured in time 3, the mediating variables (withdrawal cognitions) were measured in time 1, and

commitment forms were also measured in time 1.

A.

Co

hen

,A

.F

reun

d/

Sca

nd

.J

.M

gm

t.2

1(

20

05

)3

29

–3

51

343

ARTIC

LEIN

PRES

S

Table 5

Hierarchical regression analysis (standardized coefficients) of withdrawal cognitions (time 3) on control variables, withdrawal cognitions (time 1) and multiple

commitments (time 2)a

Attraction of present job (Time 3) Availability of alternatives (Time 3) Turnover intentions (Time 3)

Step 1 Step 2 Step 3 Step 1 Step 2 Step 3 Step 1 Step 2 Step 3

Gender �.07 �.02 .04 .18 �.03 �.07 �.19 �.33 �.39*

Education .04 .05 �.05 .03 .01 �.03 .11 �.02 .05

Tenure �.00 .00 �.01 .01 �.00 �.01 �.02 �.01 .00

Attraction of present job (Time 1) .44*** .47*** .22* .19* .09 .05

Availability of alternatives (Time 1) .08 �.01 .52*** .43*** .02 .12

Turnover intentions (Time 1) �.05 .19 �.03 �.03 .72*** .48***

Job involvement (Time 2) �.32 �.15 .05

Occupational commitment (Time 2) .55** .32* .28

Affective organizational commitment (Time 2) .05 �.27 �.32*

Continuance organizational commitment (Time 2) �.30 �.38** �.30*

R2 (Adjusted) .00 (.00) .15 (.10) .25 (.17) .01 (.00) .34 (.30) .40 (.34) .02 (.00) .28 (.24) .44 (.38)

F .07 3.08** 3.36*** .25 9.21*** 6.88*** .84 7.13*** 8.04***

DR2 .15 .10 .33 .06 .26 .15

F for DR2 6.09*** 3.36* 18.06*** 2.58* 13.14*** 7.02***

N ¼ 118–122.

Pp.05; **Pp.01; ***Pp.001.aThe control variables were entered in step 1, withdrawal cognitions at time 1 were entered in step 2 and commitment forms were entered in step 3. The

dependent variables (withdrawal cognitions) were measured in time 3, the mediating variables (withdrawal cognitions) were measured in time 1, and

commitment forms were measured in time 2.

A.

Co

hen

,A

.F

reun

d/

Sca

nd

.J

.M

gm

t.2

1(

20

05

)3

29

–3

51

344

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351 345

Hypothesis 3. The support for Hypothesis 3 was weak when commitment forms andwithdrawal cognitions were measured at time 1 and related to withdrawal cognitionsat time 3 (see Table 4). In this table commitment forms did not significantly relate towithdrawal cognitions at a later time above and beyond the effect of withdrawalcognitions measured at the same time as commitment.However, as Table 4 demonstrates, when withdrawal cognitions were measured at

time 1 and commitment forms at time 2 and the dependent variables at time 3,Hypothesis 3 was strongly supported. Occupational commitment (time 2)contributes to the explanation of attraction of present job (time 3) above andbeyond the effect of the three variables of withdrawal cognitions (time 1).Occupational commitment adds 10% to the variance already explained bywithdrawal cognitions. Continuance organizational commitment and occupationalcommitment (both measured at time 2) contribute to the explanation of availability

of alternatives (measured at time 3) above and beyond the effect of withdrawalcognitions (measured at time 1). Both commitments added 6% to the variancealready explained by withdrawal cognitions. Continuance organizational commit-ment and affective organizational commitment (time 2) contributed to theexplanation of turnover intentions (time 3) above and beyond the explanation ofwithdrawal cognitions (time 1). In this case the two commitments added 15% to thevariance already explained by withdrawal cognitions. The findings in Table 5provided strong support for Hypothesis 3 in comparison to the weak support inTable 4.Hypothesis 4 predicted that the relationship between commitment forms and

withdrawal cognitions would be stronger, the shorter the intervals between themeasurements of the two variables. The findings support this hypothesis.First, in Tables 2 and 3 it can be seen clearly that the amount of variance explainedby commitment forms is much higher when commitment forms were measured attime 2 and withdrawal cognitions at time 3 (see Table 3) than when commitmentforms were measured at time 1 and withdrawal cognitions at time 3 (see Table 2).This finding is confirmed in almost all the regression equations and is particularlystriking for two forms of withdrawal, availability of alternatives and turnover

intentions.The pattern of the findings is fairly similar when the equations are controlled by

withdrawal cognitions (Tables 4 and 5). The amount of variance explained bycommitment forms, above and beyond the variance explained by withdrawalcognitions, is much higher when commitment forms were measured at time 2 andwithdrawal cognitions at time 3 (see Table 5), than when commitment forms weremeasured at time 1 and withdrawal cognitions at time 3 (see Table 4). In general, thepattern of the findings provides strong support for this hypothesis.Finally, it should be noted that Hypotheses 1 and 2 receive additional support

from the findings in Table 5. In fact, the findings in Table 5 provide stronger supportfor Hypotheses 1 and 2 than do Tables 2 and 3. First, as expected in Hypothesis 2,the two forms of organizational commitment are related to turnover intentions. Aspredicted by Hypothesis 2, occupational commitment is related to attraction of

present job and availability of alternatives. The findings in Table 5 provide weaker

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351346

support for Hypothesis 1. Job involvement and affective organizational commitmenthave a weaker effect on withdrawal cognitions than predicted by Hypothesis 1.

4. Discussion

The goal of this study has been to examine the relationship between multiplecommitments and withdrawal cognitions. The study adopted a longitudinal designthat examined this relationship under three timing conditions. Multiple commit-ments have hardly ever been examined with the help of a longitudinal design, andnever at three different points in time. The need for a longitudinal analysis ofmultiple commitments and their correlates has been raised several times (Beck &Wilson, 2001; Cohen, 1993, 1998, 1999, 2003), but has rarely been actively taken up.Beck and Wilson (2001) pointed out the importance of studying changes incommitment over time in order to identify how commitment develops. Such data, itwas felt, would promote the development of reliable theoretical models of how andwhy different types of commitment develop. The present study responded to thisneed and has provided an important insight into the relationship between multiplecommitments and withdrawal cognitions in an attempt to increase our under-standing of turnover.The first two hypotheses in this study are similar to hypotheses previously

advanced in multiple commitments research (Cohen, 1998, 1993b), and the data hasgenerally supported them. Multiple commitments have different effects on with-drawal cognitions. In other words, different commitment forms are related towithdrawal cognitions. While some of the specific expectations advanced in thehypotheses were not supported by the data, for instance the expectation of a strongereffect of the two organization-oriented commitments, the relationship pattern foundhere does not differ greatly from that found in other research (Cohen, 1998, 1993b,1999; Krausz et al., 1995).More specifically, the present findings support the value of distinguishing different

points in the process of withdrawing from the organization, based on the finding thatcommitment foci have a differentiated relationship with the different constructs ofwithdrawal cognitions. This finding and this conclusion replicate similar resultspresented by Krausz et al. (1995). While these last researchers used differentdeterminants to predict the three turnover intentions, their results, together with ourfindings, confirm the value of distinguishing between the three types of withdrawalcognitions. Similar conclusions were reached by Cohen (1993a, b), in a study ofengineers and technicians in private manufacturing firms. However, it should benoted that the commitment forms tested here have explained the variance of each ofthe withdrawal cognitions much better than the model presented in Krausz et al.(1995), which used other determinants such as age, rank, education, burnout, andjob scope. Future research should look more closely at this diversity of effects ondifferent kinds of turnover, as well as investigating the effect of commitment formson actual turnover in addition to withdrawal cognitions. The complex relationship ofcommitment forms to withdrawal cognitions found here should be of particular

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351 347

interest. There is a need to replicate these findings before reaching any definiteconclusions about this relationship.The contribution of the present research lies in its use of a longitudinal data study

and in its results that have produced some important findings. First, the data hasdemonstrated the importance of commitment forms to our understanding of theturnover process. Commitment forms were found to be related to withdrawalcognitions, even when withdrawal cognitions measured at an earlier time or at thesame time as commitment forms were controlled for. This is an interesting findingthat has never been examined in prior research on multiple commitments. Whilesome doubts have been raised about the value of commitment in the prediction ofturnover because of its conceptual similarity to withdrawal cognitions (O’Reilly &Chatmen, 1986; Reichers, 1985), the present data has shown that commitment formsmake a notable contribution to the prediction of withdrawal cognitions. Whilewithdrawal cognitions at a given time are the main predictors of withdrawal at alater time, commitment forms explain the variance of withdrawal better thanprevious explanations have done. This finding strongly confirmed the importance ofcommitment forms for our understanding of the turnover process. In addition it alsodemonstrated the importance of withdrawal cognitions at a given time as predictorof the same construct at a later time. That is, regardless of the effects of other workattitudes, withdrawal cognitions are best predicted by the earlier measurement of theconstruct itself. Turnover research should take account of the possible effect of thisconstruct on actual turnover. Withdrawal cognitions measured at any given time willprobably have a consistent effect on turnover. The interval between the first and thesecond measurement of withdrawal cognitions in our case was a year. Futureresearch should examine whether the same effect occurs given a shorter or longerintervals between the measurements of withdrawal cognitions.Another important finding here emphasizes the significance of time and time lags

in our understanding of the commitment-withdrawal cognitions relationship, andperhaps of other similar relationships as well. The length of the intervals between themeasurements had a marked effect on the findings, even stronger than we hadexpected. Prior research had drawn our attention to the importance of time in therelationship between organizational commitment and actual turnover (Cohen,1993a). Many previous studies of the relationship between commitment andwithdrawal cognitions were cross-sectional and were criticized, quite rightly, fortheir design. Our study demonstrates that a more profound understanding of therelationship between commitment and withdrawal can be acquired only by way oflongitudinal designs. Some specific findings should be emphasized in this context.First, the findings showed that the prediction of withdrawal is better the shorter

the interval between its measurement and the measurement of multiple commit-ments. This finding is interesting because it replicates a similar effect found by Cohen(1993a). However, while Cohen’s research identified this effect only for organiza-tional commitment and for actual turnover as the dependent variable the presentstudy extended these findings to multiple commitments and withdrawal cognitions.The main implication of this is that accurate evaluation of the relationship betweencommitment and withdrawal can be achieved given shorter intervals between the

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351348

measurements of the two. The effect of time on the accuracy of the prediction can bea result of more immediate changes in commitment forms across time or of moreimmediate changes in withdrawal cognitions over time. Future research should seekto discover which of these two variables results in the need for shorter intervalsbetween measurements. Future research should also examine the possibility that therelationship between commitment and withdrawal cognitions might be even strongergiven intervals shorter than six-months interval used here.Another important finding is demonstrated in Table 5. The strongest equations

were found when withdrawal cognitions were measured or controlled for at time 1,commitment forms were measured at time 2, and the dependent variable, withdrawalcognitions, at time 3. This equation was much stronger than when commitment andwithdrawal were measured at the same time (time 1), regardless of the measurementof the dependent variable (see Table 4). One explanation for this finding is that theresearch design in Table 5 is less contaminated by common methodological errors,since each group of variables was measured at a different time. Here, too, there is aneed to replicate this finding that implies great strong potential for a betterunderstanding of the withdrawal process.The present findings support earlier research (Cohen, 1998; Gardner, 1992; Reilly

& Orsak, 1991), which found that occupational commitment has a strong effect onthe attitudes and behaviors of employees. Morrow and McElroy (1993) argued thatone of the limitations of multiple commitment research is that occupation/career focihave not been paid sufficient attention in such studies. The present findings showthat occupational commitment is an important form of commitment, and should begiven a place in future research. The strong relationship between occupationalcommitment and attraction of present job and availability of alternatives should alsobe noted. The present findings extended Cohen’s (1998) study in which it was arguedthat occupational commitment is more important for professional than fornonprofessional employees. The present study suggests that occupational commit-ment is an important determinant for nonprofessional employees as well. Inasmuchas the findings emphasize the importance of occupational commitment as animportant predictor of work outcome, they pose many questions about the relevanceof job involvement. Job involvement had a weak relationship with withdrawalcognitions in most of the equations examined here. This finding is consistent withearlier findings that demonstrated the weak effect of job involvement on workoutcomes (Cohen, 1998) and suggest that perhaps high job involvement is notnecessarily a desirable attitude in the workplace.Some recommendations for managers emerge from this study. The first is that

employee attitudes are important to an understanding of the turnover process. Usingsurveys in organizations as a diagnostic tool can thus help managers to understandand predict their employees’ behaviors. The second is that commitment foci areimportant factors in the turnover process. Managers should find ways to increasecommitment levels among their employees, if they want to reduce turnover rates intheir organization.Another recommendation is that mangers should concentrate on increasing

occupational commitment, a quality that seems to have become more important in

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351 349

the present-day workforce. Employees today feel a strong bond with theiroccupation and with their organizations, and managers have to consider this linkin their attempts to reduce turnover. Satisfying their employees’ occupational andprofessional needs is one means whereby they can achieve this goal. Such satisfactionwill increase their employees’ occupational commitment and will reduce turnoverrates. Managers should also take note that withdrawal cognitions at a given time willpredict withdrawal cognitions a year later. In other words, alert managers may detectan employee’s intention to quit as much as one year before they actually leave theorganization. Finally, Beck and Wilson (2001) maintained that longitudinal researchon commitment forms can be used to advise managers of appropriate interventionsfor the optimization of their employees’ commitment.To sum up, managers should be aware of the fact that commitment and

withdrawal are both dynamic concepts. Examining them with the help of static,linear, statistical manipulations using cross-sectional designs cannot give an accuratepicture of the relationship between them. The particular point in time when amanager collects data on employees’ work attitudes is an important factor in theability of that data to predict other work attitudes or behaviors at work. Moreempirical research is needed to validate the results and conclusions of this study.Longitudinal research designs allow for the measurement of multiple commitmentsand withdrawal at more than one point in time are essential for clarifying the natureof the commitment-withdrawal relationship. In this respect some of the argumentsand findings of the present study may well provide a good starting-point for suchresearch.

References

Alutto, J. A., & Belasco, J. A. (1974). Determinants of attitudinal militancy among nurses and teachers.

Industrial & Labor Relations Review, 27, 216–227.

Becker, T. E. (1992). Foci and bases of commitment: are they distinctions worth making? Academy of

Management Journal, 35, 232–244.

Beck, K., & Wilson, C. (2001). Have we studied, should we study, and can we study the development of

commitment? Methodological issues and the developmental study of work-related commitment.

Human Resource Management Review, 11, 257–278.

Blau, G. J. (1985). The measurement and prediction of career commitment. Journal of Occupational

Psychology, 58, 277–288.

Blau, G. J. (1986). Job involvement and organizational commitment as interactive predictors of tardiness

and absenteeism. Journal of Management, 12, 577–584.

Blau, G., Paul, A., & St. John, N. (1993). On developing a general index of work commitment. Journal of

Vocational Behavior, 42, 298–314.

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

Journal, 36, 1140–1157.

Cohen, A. (1993b). Work commitment in relation to withdrawal intentions and union effectiveness.

Journal of Business Research, 26, 75–90.

Cohen, A. (1998). An examination of the relationship between work commitment and work outcomes

among hospital nurses. Scandinavian Journal of Management, 14, 1–17.

Cohen, A. (1999a). Relationships among five forms of commitment: an empirical examination. Journal of

Organizational Behavior, 20, 285–308.

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351350

Cohen, A. (1999b). The relation between commitment forms and work outcomes in Jewish and Arab

culture. Journal of Vocational Behavior, 54, 371–391.

Cohen, A. (2000). The relationship between commitment forms and work outcomes: a comparison of three

models. Human Relations, 53, 387–417.

Cohen, A. (2003). Multiple commitments in the workplace: an integrative approach. Mahwah, NJ: Lawrence

Erlbaum Associates.

Dalton, D. R., & Todor, W. D. (1982). Antecedents of grievance filing behavior: attitude/behavior

constituency and the union steward. Academy of Management Journal, 25, 158–169.

Farrell, D., & Petersen, J. C. (1984). Commitment, absenteeism and turnover of new employees: a

longitudinal study. Human Relations, 37, 681–692.

Gardner, D. L. (1992). Career commitment in nursing. Journal of Professional Nursing, 8, 155–160.

Gregersen, H. B. (1993). Multiple commitments at work and extra–role behavior during three stages of

organizational tenure. Journal of Business Research, 26, 31–47.

Hom, P. W., Pruaaia, G. E., & Griffeth, R. W. (1992). A meta-analytical structural equations analysis of a

model of employee turnover. Journal of Applied Psychology, 77, 890–909.

Kanungo, R. N. (1979). The concept of alienation and involvement revisited. Psychological Bulletin, 86,

119–138.

Kanungo, R. N. (1982). Measurement of job and work involvement. Journal of Applied Psychology, 67,

341–349.

Krausz, M., Koslowsky, M., Shalom, N., & Elyakim, N. (1995). Predictors of intentions to leave the ward,

the hospital, and the nursing profession: a longitudinal study. Journal of Organizational Behavior, 16,

277–288.

Lawler, E. J. (1992). Affective attachment to nested groups: A choice process theory. American

Sociological Review, 57, 327–339.

Mathieu, J. E., & Zajac, D. M. (1990). A review and meta-analysis of the antecedents, correlates and

consequences of organizational commitment. Psychological Bulletin, 108, 171–194.

Meyer, P. J., & Allen, J. N. (1984). Testing the side bet theory of organizational commitment: some

methodological considerations. Journal of Applied Psychology, 69, 372–378.

Meyer, P. J., & Allen, J. N. (1997). Commitment in the workplace: theory, research, and application.

Thousand Oaks: Sage Publications.

Mitra, A., Jenkins, D., & Gupta, N. (1992). A meta-analytic review of the relationship between absence

and turnover. Journal of Applied Psychology, 77, 879–889.

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

Journal of Applied Psychology, 62, 237–240.

Mobley, W. H., Griffeth, R. H., Hand, H. H., &Meglino, B. M. (1979). Review and conceptual analysis of

the employee turnover process. Psychological Bulletin, 86, 493–522.

Morrow, P. C. (1983). Concept redundancy in organizational research: the case of work commitment.

Academy of Management Review, 8, 486–500.

Morrow, P. C. (1993). The theory and measurement of work commitment. Greenwich, CT: Jai Press Inc.

Morrow, P. C., & McElroy, J. C. (1993). Introduction: understanding and managing loyalty in a multi-

commitment world. Journal of Business Research, 26, 1–2.

Morrow, P. C., McElroy, J. C., Laczniak, K. S., & Fenton, J. B. (1999). Using absenteeism and

performance to predict employee turnover: early detection through employee records. Journal of

Vocational Behavior, 55, 358–374.

Mowday, R. T., Porter, L. M., & Steers, R. M. (1982). Employee-organizational linkage. New York:

Academic Press.

Mueller, C. W., & Lawler, E. J. (1999). Commitment to nested organizational units: some basic principles

and preliminary findings. Social Psychology Quarterly, 62, 325–346.

O’Reilly, C. A., & Chatmen, J. (1986). Organizational commitment and psychological attachment: the

effects of compliance, identification and internalization on prosocial behavior. Journal of Applied

Psychology, 71, 492–499.

Parasuraman, S. (1982). Predicting turnover intentions and turnover behavior: a multivariate analysis.

Journal of Vocational Behavior, 21, 111–121.

ARTICLE IN PRESS

A. Cohen, A. Freund / Scand. J. Mgmt. 21 (2005) 329–351 351

Parasuraman, S. (1989). Nursing turnover: an integrated model. Research in Nursing and Health, 12,

267–277.

Porter, L. W., Crampon, W. J., & Smith, F. J. (1976). Organizational commitment and managerial

turnover. Organizational Behavior and Human Performance, 15, 87–98.

Price, J. L., & Mueller, C. W. (1981). Professional turnover: the case of nurses. New York: SP Medical and

Scientific Books.

Randall, D. M. (1990). The consequences of organizational commitment: methodological investigation.

Journal of Organizational Behavior, 11, 361–378.

Randall, D. M., & Cote, J. A. (1991). Interrelationships of work commitment constructs. Work and

Occupation, 18, 194–211.

Reichers, A. E. (1985). A review and reconceptualization of organizational commitment. Academy of

Management Review, 10, 465–476.

Reichers, A. E. (1986). Conflict and organizational commitment. Journal of Applied Psychology, 71,

508–514.

Reilly, N. P., & Orsak, C. L. (1991). A career stage analysis of career and organizational commitment in

nursing. Journal of Vocational Behavior, 39, 311–330.

Rosse, J. G. (1988). Relations among lateness, absence, and turnover: is there a progression of withdrawal?

Human Relations, 41, 517–531.

Shore, L. M., Newton, L. A., & Thornton, G. C. (1990). Job and organizational attitudes in relation to

employee behavioral intentions. Journal of Organizational Behavior, 11, 57–67.

Siders, M. A., George, G., & Dharwadkar, R. (2001). The relationship of internal and external

commitment foci to objective performance measures. Academy of Management Journal, 44, 570–579.

Somers, M. J., & Birnbaum, D. (1998). Work-related commitment and job performance: it’s also the

nature of the performance that counts. Journal of Organizational Behavior, 19, 621–634.

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.

Steers, R. M., & Rhodes, S. R. (1978). Major influences on employee attendance: a process model. Journal

of Applied Psychology, 63, 391–407.

Vandenberg, R. G., & Scarpello, V. (1994). A longitudinal assessment of the determinant relationship

between employee commitments to the occupation and the organization. Journal of Organizational

Behavior, 15, 535–547.

Wiener, Y., & Vardi, Y. (1980). Relationships between job, organization and work outcomes: an

integrative approach. Organizational Behavior and Human Performance, 26, 81–96.