+ All Categories
Home > Documents > e0b49527c9c6345d4a

e0b49527c9c6345d4a

Date post: 07-Oct-2015
Category:
Upload: atiq-ur-rehman
View: 212 times
Download: 0 times
Share this document with a friend
Description:
a
14
 A Four-Dimensional Model of Organizational Commitment among Belgian Employees* Nathalie Delobbe and Chr istian Vandenberg he Université Catholique de Louvain, Belgium Keywords:  Organizational commitment, internalization, compliance, affective and continuance commitment.  European Journal of Psychological Assessment, V ol. 16, Issue 2, pp. 1 25–138 Summary: We conducted a study to examine the reliability and validity of four dimensions of organizational commitment. Using two samples from various organizations in Belgium (  N s = 216 and 201), we investigated internalization, compliance, and affective and continuance commitment as dimensions of employee commit- ment to the organization. Confirmatory factor analysis by LISREL showed that the four factors were reasonably distinct. However, the scales are in need of further refinement because (a) internalization and affective commit- ment are strongly correlated with each other and display similar relationships with major criterion variables (e. g., met expectations, job satisfaction, and intent to leave); (b) the compliance scale shows weak reliability and needs to be expanded. The implications of these results for future research are discussed. Over the past decade, organizational commitment (OC) has become a topic of increasing importance in industri- al/organizational psychology. Generally defined as a psychological link between the employee and his/her or- ganization, OC has been found to be related to major work outcomes, namely, turnover intention and actual turnover (Mathieu & Zajac, 1990). Indeed, employees who are str ong ly commit ted to the ir org ani zat ion are les s likely to leave. Moreover, the understanding of employ- ee withdrawal has been enhanced by the emergence of multidimensional conceptualizations of commitment. That is, resea rchers have ident ified distincti ve dimen - sions within the construct of OC, which is now con- ceive d as a psychol ogical state su bsumi ng separa te com- ponents (Meyer & Allen, 1997). In subsequent sections, we provide definitions of commonly accepted commit- ment dimensions and examine how they can be integrat- ed into overarching models of employee attachment to organizations. Then, we propose an empirical test of these competing models of employee commitment with- in a French-speaking context. Because most commit- ment stu die s have to dat e bee n conduc ted inNorthAmer- ica, thi s stu dy shouldenhanceour und erstand ing of com- mitment processes in other cultures. Finally, we also ex- amine how commitment dimensions relate to a set of criterion variables. Four Dimensions of Commitment to the Organization Among current conceptualizations, two primary dimen- sions of OC can be distinguished. The first dimension has been labelled  attitudinal commitment , whereas the second is referred to as  calculative commitment (Mathieu & Zajac, 1990). This distinction is consistent with Etzioni’s (1961) contention that the bond between the individual and the organization is rooted in either moral or economic considerations. It also parallels March and Simon’s (1958) distinction between the mo- tivation to produce and the motivation to participate, which results respectively in value attachment and in- strumental commitment to the orga nizat ion (Mayer & Schoorman, 1992, 1998). EJP A 16 (2), © 2000 Hogrefe & Huber Publishers * The original data upon which thi s paper is based are available at http://www.hhpub.com/journals/ejpa
Transcript
  • N. Delobbe and C. Vandenberghe: Multiple Dimensions of CommitmentEJPA 16 (2), 2000 Hogrefe & Huber Publishers

    A Four-Dimensional Model ofOrganizational Commitment

    among Belgian Employees*Nathalie Delobbe and Christian Vandenberghe

    Universit Catholique de Louvain, Belgium

    Keywords: Organizational commitment, internalization, compliance, affective and continuance commitment.

    European Journal of Psychological Assessment, Vol. 16, Issue 2, pp. 125138

    Summary: We conducted a study to examine the reliability and validity of four dimensions of organizationalcommitment. Using two samples from various organizations in Belgium (Ns = 216 and 201), we investigatedinternalization, compliance, and affective and continuance commitment as dimensions of employee commit-ment to the organization. Confirmatory factor analysis by LISREL showed that the four factors were reasonablydistinct. However, the scales are in need of further refinement because (a) internalization and affective commit-ment are strongly correlated with each other and display similar relationships with major criterion variables(e. g., met expectations, job satisfaction, and intent to leave); (b) the compliance scale shows weak reliabilityand needs to be expanded. The implications of these results for future research are discussed.

    Over the past decade, organizational commitment (OC)has become a topic of increasing importance in industri-al/organizational psychology. Generally defined as apsychological link between the employee and his/her or-ganization, OC has been found to be related to majorwork outcomes, namely, turnover intention and actualturnover (Mathieu & Zajac, 1990). Indeed, employeeswho are strongly committed to their organization are lesslikely to leave. Moreover, the understanding of employ-ee withdrawal has been enhanced by the emergence ofmultidimensional conceptualizations of commitment.That is, researchers have identified distinctive dimen-sions within the construct of OC, which is now con-ceived as a psychological state subsuming separate com-ponents (Meyer & Allen, 1997). In subsequent sections,we provide definitions of commonly accepted commit-ment dimensions and examine how they can be integrat-ed into overarching models of employee attachment toorganizations. Then, we propose an empirical test ofthese competing models of employee commitment with-in a French-speaking context. Because most commit-ment studies have to date been conducted in North Amer-ica, this study should enhance our understanding of com-

    mitment processes in other cultures. Finally, we also ex-amine how commitment dimensions relate to a set ofcriterion variables.

    Four Dimensions of Commitmentto the Organization

    Among current conceptualizations, two primary dimen-sions of OC can be distinguished. The first dimensionhas been labelled attitudinal commitment, whereas thesecond is referred to as calculative commitment(Mathieu & Zajac, 1990). This distinction is consistentwith Etzionis (1961) contention that the bond betweenthe individual and the organization is rooted in eithermoral or economic considerations. It also parallelsMarch and Simons (1958) distinction between the mo-tivation to produce and the motivation to participate,which results respectively in value attachment and in-strumental commitment to the organization (Mayer &Schoorman, 1992, 1998).

    EJPA 16 (2), 2000 Hogrefe & Huber Publishers

    * The original data upon which this paper is based are available at http://www.hhpub.com/journals/ejpa

  • Internalization and Affective Commitment

    The concept of attitudinal commitment traces back to thework of Porter and colleagues (Mowday, Steers, & Porter,1979; Porter, Steers, Mowday, & Boulian, 1974). Theseauthors defined commitment as the relative strength of anindividuals identification with and involvement in a par-ticular organization, which is characterized by three fac-tors: a) a strong belief in and acceptance of the organi-zations goals and values; b) a willingness to exert con-siderable effort on behalf of the organization; and c) astrong desire to maintain membership in the organization(Mowday, Porter, & Steers, 1982, p. 27). For measuringthis construct, Porter et al. (1974) developed the Organi-zational Commitment Questionnaire (OCQ). In the mid-1980s, Meyer and Allen (1984) proposed a similar defi-nition of OC, which they conceived as affective innature. They defined it as the employees emotional at-tachment to the organization. Affectively committed em-ployees identify with, are involved in, and enjoy member-ship in the organization such that they stay with it becausethey want to (Allen & Meyer, 1990, 1996; Meyer,Bobocel, & Allen, 1991; Meyer & Allen, 1991). In a sim-ilar vein, Mayer and Schoorman (1992, 1998) definedvalue commitment as an affective orientation towardthe organization. This is consistent with Meyer and Al-lens view of affective commitment.

    Aside from the obvious similarities between these def-initions, Porter et al.s definition of commitment differsfrom Meyer and Allens and Mayer and Schoormans inthat it subsumes both the individuals identification to theorganization and his/her acceptance of organizationalvalues. In fact, the acceptance of organizational valuesrefers to the notion of internalization, a construct thatmight differ from identification. This view is emphasizedin Kelmans (1958) work on bases for attitudinal change.Kelman argues that identification results from a desire ofaffiliation with the organizations members and the ac-ceptance of its influence. On the other hand, internaliza-tion refers to the attachment to the organizations goalsbased on a perceived congruence between the individu-als values and those of the organization. As suggestedby OReilly and Chatman (1986), internalization mayrepresent a different form of commitment that might notunderlie Meyer and Allens affective commitment andMayer and Schoormans value commitment. The ques-tion that we raise here is whether internalization andidentification are distinguishable dimensions of commit-ment. In order to clarify this issue, we will test whetherOReilly and Chatmans (1986) measure of internaliza-tion is empirically distinguishable from Allen andMeyers (1990) measure of affective commitment. Thelatter basically reflects feelings of identification to theorganization.

    Compliance and ContinuanceCommitment

    The second primary dimension of commitment is con-ceived as calculative in nature. In this view, commitmentinvolves a rational comparison between the costs andbenefits associated with staying in or leaving the compa-ny (Lee & Mitchel, 1994). This dimension was popular-ized by Beckers side-bet theory, which defined commit-ment as a tendency to engage in consistent lines of ac-tivity (Becker, 1960, p. 33) based on the individualsrecognition of the costs associated with discontinuingthe activity. In this sense, individuals become tied to theorganization because they have voluntarily or involun-tarily invested in the organization (e. g., pension plans)so that they cannot afford to leave (Mathieu & Zajac,1990). This calculative orientation was labelled continu-ance commitment by Meyer and Allen (1984) and Mayerand Schoorman (1992, 1998).

    A somewhat different aspect of calculative commit-ment is illustrated by Kelmans (1958) notion of com-pliance. According to Kelman (1958, p. 496), compli-ance occurs when an individual adopts the induced be-havior [e. g., providing quality services to clients, stayingwith ones company] not because he believes in its con-tent but because he expects to achieve a favorable con-sequence. In other words, compliance results from theexpectation of rewards associated with the target behav-ior. However, contrary to continuance commitment,compliance does not imply increased investments in theorganization over time. Thus, the difference betweencontinuance commitment and compliance lies in the na-ture of the calculative process (i. e., accrued investmentsvs. expectations of rewards). We will test this possibilityby examining whether Allen and Meyers (1990) scale ofcontinuance commitment is distinguishable fromOReilly and Chatmans (1986) scale of compliance.

    Empirical EvidenceA primary purpose of this study is to examine the facto-rial structure of a four-factor model of commitment de-rived from the previous discussion. More specifically,we were interested in testing whether internalization, af-fective commitment (identification), continuance com-mitment (side-bets), and compliance represent distinc-tive dimensions of OC.

    A main concern is whether OC splits into two primarydimensions, attitudinal commitment (including internal-ization and affective commitment) vs. calculative com-mitment (merging continuance commitment and compli-ance), or whether it could be conceived as four-dimen-

    126 N. Delobbe and C. Vandenberghe: Multiple Dimensions of Commitment

    EJPA 16 (2), 2000 Hogrefe & Huber Publishers

  • sional. We are not aware of any published work providingevidence for the discriminant validity of the four OC di-mensions discussed above. Thus far, evidence suggestshowever that affective commitment and continuancecommitment are empirically distinguishable (Allen &Meyer, 1996; Dunham, Grube, & Castaneda, 1994; Hack-ett, Bycio, & Hausdorf, 1994; McGee & Ford, 1987;Somers, 1993) and correlate differentially with key orga-nizational outcomes such as intent to leave and turnover(Allen & Meyer, 1996; Cohen, 1993; Hackett et al., 1994;Jaros, 1997) and performance (Allen & Meyer, 1996;Meyer, Paunonen, Gellatly, Goffin, & Jackson, 1989).

    Some studies questioned the discriminant validity ofinternalization- and identification-based commitment.For example, using OReilly and Chatmans (1986) mea-sure of commitment, Caldwell, Chatman, and OReilly(1990) as well as Sutton and Harrison (1993) reportedidentification and internalization scales to fall down intoa single factor. In contrast, using confirmatory factor anal-ysis, Vandenberg, Self, and Seo (1994) showed that thesescales, albeit strongly correlated, can be represented byseparate factors. The issue of independence among inter-nalization and identification scales thus requires moreempirical work before firm conclusions can be reached.

    Finally, the distinction among compliance and contin-uance commitment remains a neglected issue in the lit-erature, though some authors tend to equate these con-structs (e. g., Lee & Mitchell, 1994). Thus far, evidencesuggests that OReilly and Chatmans (1986) scale ofcompliance is problematic in terms of reliability and fac-torial structure (e. g., Sutton & Harrison, 1993; Vanden-berg et al., 1994). A contribution of this study will be toassess the structure of compliance and continuance com-mitment, using confirmatory factor analysis.

    Correlates of CommitmentDimensions

    In order to test further the discriminant validity of thefour commitment scales, we will examine whether thefour OC dimensions correlate differentially with somespecific variables. Based on a review of the target litera-ture, we propose below a set of hypotheses concerningthese differential linkages.

    Tenure

    Organizational tenure is one of the most frequently citedcorrelates of OC. Several investigations (Dunham et al.,1994; Hackett et al., 1994; Meyer & Allen, 1991;OReilly & Chatman, 1986) and Mathieu and Zajacs

    meta-analytic study (1990) reported positive correlationsbetween tenure and affective commitment, and betweentenure and continuance commitment. However, the rea-sons for these positive correlations differ for affectiveand continuance commitment. Tenure is expected to in-crease continuance commitment through accrued invest-ments (e. g., skills specialization and pension plans). Incontrast, the positive correlation between tenure and af-fective commitment would reflect either the opportuni-ties received to better ones position over time (Hackettet al., 1994), or the fact that employees with low affectivecommitment choose to leave the organization in earlyemployment (Dunham et al., 1994).

    Thus far, the relationship between tenure and compli-ance- and internalization-based commitment has re-ceived little attention. However, OReilly and Chatman(1986) showed that compliance was negatively related totenure and that internalization was unrelated to it.

    Research findings thus suggest the following hypothesis:H1: Affective commitment (identification) and con-tinuance commitment are positively related to tenure,whereas compliance is negatively related to it.

    Employment Contract and Level ofEducation

    Specific antecedents of continuance commitment wouldbe the nature of the employment contract (short vs. longterm) and the level of education. Employees with a short-term contract would not be expected to have sunk costsas much as those with a long-term one. Likewise, em-ployees with a high level of education would have moreemployment opportunities because of a higher demandfor their skills (Allen & Meyer, 1996; Ko, Price, & Muel-ler, 1997; Mayer & Schoorman, 1998; Somers, 1993).So, employees with a low level of education should bemore committed on a continuance basis.

    This leads to the following hypotheses:H2: Employees with a long-term contract displayhigher levels of continuance commitmentH3: Level of education is negatively related to contin-uance commitment such that low-skilled employeesfeel more committed on a continuance basis.

    Met Expectations

    The extent to which employees expectations about theirwork and organization are met or confirmed is a positivepredictor of their subsequent affective commitment tothe organization (Ko et al., 1997; Meyer & Allen, 1988).

    N. Delobbe and C. Vandenberghe: Multiple Dimensions of Commitment 127

    EJPA 16 (2), 2000 Hogrefe & Huber Publishers

  • On the contrary, if their expectations are disconfirmed,individuals become less affectively committed to the or-ganization. This met-expectations hypothesis has re-ceived strong support in a recent meta-analysis (Wanous,Poland, Premack, & Davis, 1992). Note, however, thatthe notion of met expectations implies that it is the dis-crepancy between work experiences and what the indi-vidual expects that influences affective commitment(Meyer & Allen, 1997).

    Although no previous study has examined the linksbetween internalization and compliance, and met expec-tations, one can reasonably hypothesize that met expec-tations facilitate internalization. Indeed, the fact that ex-pectations are met can be operationalized as the relativesimilarity of values between the individual and the orga-nization (Meyer, Irving, & Allen, 1998), a phenomenonthat is critical in the development of internalization (Kel-man, 1958). In contrast, neither continuance commit-ment (Meyer et al., 1998) nor compliance should relateto met expectations because these psychological statesinvolve calculative processes, which do not imply a sim-ilarity of values nor emotional ties between the individ-ual and the organization.

    This leads to the following hypothesis:H4: Internalization and affective commitment (identi-fication) are positively related to met expectations.

    Job Satisfaction

    Job satisfaction is a strong attitudinal correlate of affec-tive commitment. This relationship has been observedusing both the OCQ (Mathieu & Zajac, 1990; Mowdayet al., 1979) and Meyer and Allens affective commit-ment scale (Allen & Meyer, 1996). In contrast, continu-ance commitment has been found to be either uncorre-lated or weakly correlated to job satisfaction (Allen &Meyer, 1996). On the other hand, Becker (1992) reportedcompliance to correlate negatively and internalizationpositively, with job satisfaction. As a whole, these resultssuggest that job satisfaction correlates positively withcommitment components that are emotionally oriented.

    Consequently, the following hypothesis can be drawn:H5: Job satisfaction is positively related to internal-ization and affective commitment (identification), andnegatively related to compliance.

    Intent to Leave

    Intent to leave and actual turnover are widely studiedconsequences of commitment (Cohen, 1993; Hackett et

    al., 1994; Jaros, 1997). Notably, negative relationshipsinvolving intent to quit were found for affective commit-ment (Allen & Meyer; 1996; Mathieu & Zajac, 1990)and continuance commitment (Hackett et al., 1994;Whitener & Walz, 1993). However, although both affec-tive and continuance commitment presumably increasethe likelihood that an individual will remain with an or-ganization, the reason for doing so is different. Employ-ees with a strong affective commitment stay becausethey want to, whereas those who have a strong continu-ance commitment remain because they have to (Meyer& Allen, 1991).

    Similarly, internalization has been found to correlatenegatively with intent to leave, while compliance hasbeen reported to relate positively to it (Becker, 1992;OReilly & Chatman; 1986). It is worth noting that con-tinuance commitment and compliance have oppositeconsequences in terms of intentions to quit. This may addto their discriminant validity.

    The previous discussion leads to the following hypothe-sis:

    H6: Internalization, affective commitment (identifica-tion), and continuance commitment correlate nega-tively, and compliance positively, with intent to leave.

    Purpose of the Study

    In this study, we assessed the factor structure of the fourOC dimensions using well-established measures of theseconstructs (see below). We used confirmatory factoranalysis to analyze the data from two independent sam-ples of subjects. We analyzed further the discriminantvalidity of the scales by examining their correlationswith organizational tenure, employment contract, levelof education, met expectations, job satisfaction, and in-tent to leave. These analyses allowed for testing our re-search hypotheses.

    MethodSamples and Procedure

    The study was conducted in Belgium. Data were collect-ed in the context of a larger survey. Demographic char-acteristics were substantially different for the two sam-ples of the study because data were collected for differentresearch purposes. However, these contrasted character-istics allowed us to test the generalizability of findingsacross the two samples. All questionnaires were mailed

    128 N. Delobbe and C. Vandenberghe: Multiple Dimensions of Commitment

    EJPA 16 (2), 2000 Hogrefe & Huber Publishers

  • directly to the respondents. Participants received the as-surance that data would be handled confidentially. Com-pleted questionnaires were sent back to the researchersoffice. A remainder was sent to nonrespondents after 2weeks.

    Sample 1

    The first sample included 216 employees newly hired in25 different firms from a variety of industries. The re-sponse rate was 57%. The tenure of respondents aver-aged 3.39 months (median = 3.00), with a minimum of1 month and a maximum of 6 months. The average ageof respondents was 26.15 years and 44% were female.Of the respondents, 55% held a master degree, 38% wereundergraduates, whereas 7% held a college degree.

    Sample 2

    The second sample included 201 new recruits (a 44%response rate) from four Belgian banks. They held com-mercial positions within their company. Their averagetenure was 17.02 months (median = 16.00), with a min-imum of 2 months and a maximum of 54 months. Theaverage age of respondents was 26.18 years. Thirty-eightpercent of respondents were female, 51% were graduatesin applied economics, 22% were graduates in other dis-ciplines, 14% were undergraduates in applied econom-ics, and 13% were undergraduates in other disciplines.

    Measures

    As we used mostly scales developed in English, we hadto translate them into French. In order to minimize cross-language discrepancies, we asked three French-Englishbilinguals to translate the items independently. This pro-cedure resulted in only a few discrepancies across trans-lations (less than 5%). These difficulties were solvedthrough a discussion among the translators.

    Commitment

    Affective and continuance commitment were measuredusing seven and six items, respectively, resulting from aFrench validation of Meyer and Allens scales (Vanden-berghe, 1996). Internalization (four items) and compli-ance scales (four items) were translated from OReillyand Chatmans (1986) scales.

    Met expectations were measured using a four-itemscale, two of them being taken from Feldman (1976), theothers being developed specifically for this study (e. g.,In some ways, I feel like this is not the right type of workfor me, or Im not the right type of person for this job reverse keyed). This measure emphasizes the extent to

    which employees perceived a fit between the character-istics of their job and their own expectations (cf. Meyer& Allen, 1997). Note that the met expectations measurewas collected from Sample 1 only.

    Job satisfaction was assessed through four items, twoof which were borrowed from OReilly and Caldwell(1981), the others being built for this study. Respondentsindicated, for instance, how satisfied they were with theirjob and the extent to which they would prefer another,more ideal job.

    Intent to leave was captured using two items. Respon-dents indicated the likelihood of their working for theorganization within the next three years and how fre-quently they thought of quitting their job. This variablewas measured for Sample 2 only.

    Demographic Data

    Tenure was a continuous variable and measured inmonths. Employment contract was categorized as long-or short-term (from 1 to 3), depending on the time spanof the contract. Note that this variable was reported forSample 1 only. Level of education was an ordinal vari-able, with a higher score indicating a higher level ofeducation or more marketable skills.

    Except for demographic data, a 6-point Likert scalewas used for all survey items.

    Analyses

    Commitment

    Confirmatory factor analysis by LISREL 8 (Jreskog &Srbom, 1993) was used for analyzing commitment da-ta. More specifically, we calculated a covariance matrixof observed indicators that was then used as input forestimating competing models of commitment (throughmaximum likelihood). First, we assessed a one-factormodel, with all indicators loading on a single commit-ment construct. This model corresponds to the idea thatcommitment is unidimensional or that data are affectedby a common method (data were collected from a singlesource and at a single point in time). The second modelwas a two-factor (oblique) structure, with internalizationand affective commitment capturing a single construct,and compliance and continuance commitment defininganother one. This model refers to the conceptualizationof commitment as either attitudinal or calculative. An-other two-factor oblique model was also tested, whichconstrained continuance commitment, compliance, andinternalization to form a single factor, and affective com-mitment to be independent. Next, two three-factor(oblique) solutions were analyzed: one in which affec-tive commitment and internalization collapsed and com-

    N. Delobbe and C. Vandenberghe: Multiple Dimensions of Commitment 129

    EJPA 16 (2), 2000 Hogrefe & Huber Publishers

  • pliance and continuance commitment were independent,and another one in which affective commitment and in-ternalization were independent but compliance and con-tinuance commitment were combined. Finally, the orig-inal four-factor (oblique) model was evaluated. A nullmodel that constrained all indicators to be orthogonalwas included as a baseline for comparison purposes.

    All models were assessed using a 2 test. A significantvalue for this test indicates that the model differs fromperfect fit. Several other fit indices were also calculated.These indices were the goodness-of-fit index (GFI), thenonnormed-fit index (NNFI; Bentler, 1990) and the com-parative fit index (CFI; Bentler, 1990). Normally, valuesabove .90 indicate good fit. In addition to these indices,we used Browne and Cudecks (1989) root mean squareerror of approximation (RMSEA). The RMSEA pro-vides a test of the null hypothesis of close fit and has beenrecommended as the best fit index, especially for testingmodels with numerous degrees of freedom (MacCallum,Browne, & Sugawara, 1996; MacCallum & Hong,1997), which was the case here. For the RMSEA, valuesof .05 and below signal excellent fit. Finally, we comput-ed the 2/df ratio for the models. It is admitted that a ratiobetween 1 and 3 is indicative of good fit (Carmines &McIver, 1981).

    We used a nested sequence approach for comparingthe relative fit of theoretical models (Bentler & Bonett,1980). According to this procedure, a series of nestedmodels was assessed (i. e., moving from one- to four-fac-tor solutions). The difference between two models withinthe sequence was evaluated using a 2 difference test.When this test is significant, the less constrained modelis considered to be an improvement over the more con-strained one. A look into the practical difference betweenthe models can also be achieved by comparing the valuesof fit indices. This comparison indicates how sizeableand meaningful this difference might be.

    Relationships with Criterion Measures

    Correlational analyses were used to measure how eachcommitment component related to criteria included inthe two surveys. When necessary, we compared the sizeof correlations between variables. This was achieved byusing Cohen and Cohens (1983, p. 56) formula for com-paring the size of dependent correlations.

    ResultsConfirmatory Factor Analyses

    Internal consistency coefficients, measured throughCronbachs , for commitment dimensions are reason-ably good for internalization (.80 and .80), affectivecommitment (.82 and .86), and continuance commitment(.70 and .72). In contrast, compliance displays very weakreliabilities (.36 and .47). The latter may indicate that thecompliance scale is a poor operationalization of the con-struct. Note that a similar finding has been reported inprevious research (Sutton & Harrison, 1994; Vandenberget al., 1994).

    We first fitted the four-factor model of commitment ineach sample, using the 21-item covariance matrix. Inboth cases, model fit was moderate: 2 (183, N = 185) =327.25, GFI = .85, NNFI = .86, CFI = .87, RMSEA = .07for Sample 1, and 2 (183, N = 184) = 330.71, GFI = .85,NNFI = .88, CFI = .89, RMSEA = .07 for Sample 2. Theinspection of factor loadings revealed that two itemsfrom the compliance scale displayed a weak loading inat least one sample. Item COMP2 had loadings of .25 and.03 in Sample 1 and 2 respectively, whereas the loadingsfor COMP4 were .04 and .40. This might be one reasonwhy -coefficients were so poor for this scale. Becausethese items were a source of unreliability, we deletedthem and reconducted the analyses (using a 19-item co-

    Table 1. Overall fit indices for the four commitment scales.

    Sample 1 Sample 2Model 2 df GFI NNFI CFI RMSEA 2 df GFI NNFI CFI RMSEA

    Null model 1303.89 171 .40 NA NA .19 1524.64 171 .35 NA NA .21One-factor 444.62 152 .77 .71 .74 .10 507.42 152 .73 .70 .74 .11Two-factor obliquea 375.45 151 .82 .78 .80 .09 403.48 151 .80 .79 .81 .10Two-factor obliqueb 437.91 151 .77 .71 .75 .10 488.50 151 .74 .72 .75 .11Three-factor obliquec 298.98 149 .85 .85 .87 .07 311.62 149 .84 .86 .88 .08Three-factor obliqued 355.52 149 .83 .79 .82 .09 357.55 149 .82 .82 .85 .09Four-factor oblique 275.95 146 .86 .87 .89 .07 255.67 146 .87 .91 .92 .06

    Notes. Sample 1: N = 191, sample 2: N = 185, based on listwise deletion of missing data. GFI = goodness-of-fit index; NNFI = nonnormedfit index; CFI = comparative fit index; RMSEA = root mean square error of approximation; NA = not applicable. aAffective commitmentand internalization versus continuance commitment and compliance, bAffective commitment versus internalization, continuance com-mitment and compliance, cAffective commitment and internalization, continuance commitment, and compliance, dContinuance com-mitment and compliance, affective commitment, and internalization.

    130 N. Delobbe and C. Vandenberghe: Multiple Dimensions of Commitment

    EJPA 16 (2), 2000 Hogrefe & Huber Publishers

  • variance matrix). The -coefficients for this shortenedcompliance scale improved to .53 in Sample 1 and .56 inSample 2. The 19-item covariance matrices that we usedfor analyzing commitment models are provided in Ap-pendixes A and B.

    The results of confirmatory factor analyses are report-ed in Table 1. Results are quite comparable across thetwo samples, which adds to their generalizability. In anabsolute sense, the four-factor oblique model receivesthe strongest support from the analyses. This model out-performs all other models in the nested sequence, and

    notably the three-factor model in which internalizationand affective commitment are combined [ 2 (3, N =191) = 23.03, p < .001 for Sample 1; 2 (3, N = 185) =55.45, p < .001 for Sample 2], and the three-factor modelwith compliance and continuance commitment com-bined [ 2 (3, N = 191) = 79.57, p < .001 for Sample 1; 2 (3, N = 185) = 101.88, p < .001 for Sample 2]. Thedifferences among fit indices between the four-factormodel and the three-factor representations are also size-able from a practical point of view (i. e., from .01 to .08in Sample 1 and from .03 to .09 in Sample 2). Differences

    Table 2. Factor loadings of the commitment items for the four-factor oblique model.

    Item Affective Continuance Internalization ComplianceCommitment CommitmentS1 S2 S1 S2 S1 S2 S1 S2

    Identification (Affective Commitment)ID1. I feel a strong sense of belonging to my organization. .80 .87ID2. I feel emotionally attached to this organization. .81 .80ID3. I feel like part of the family at my organization. .70 .77ID4. I enjoy discussing my organization with people outside it. .35 .69ID5. I really feel as if this organizations problems are my own. .58 .64ID6. This organization doesnt have a great deal of personal meaning

    for me (R). .63 .65ID7. I would be unhappy to spend the rest of my career with this

    organization (R). .55 .47

    Continuance CommitmentCC1. I feel that I have enough options to consider leaving this

    organization (R). .50 .73CC2. One of the few consequences of leaving this organization would

    be the scarcity of available alternatives. .49 .70CC3. I am afraid of what might happen if I quit my job without having

    another one lined up. .64 .52CC4. One of the major reasons I continue to work for this organization

    is that leaving would require considerable personal sacrifice another organization may not match the overall benefits I have here. .44 .55

    CC5. Right now, staying with my organization is a matter of necessityas much as desire. .44 .43

    CC6. It would be very hard for me to leave my organization right now,even if I wanted to. .63 .43

    InternalizationINT1. My attachment to this organization is primarily based on the

    similarity of my values and those represented by the organization. .72 .86INT2. The reason I prefer this organization to others is because of what it

    stands for, its values. .74 .75INT3. Since joining this organization, my personal values and those of the

    organization have become more similar. .71 .68INT4. If the values of this organization were different, I would not be as

    attached to this organization. .68 .59

    ComplianceCOMP1. Unless Im rewarded for it in some way, I see no reason to expend

    extra effort on behalf of this organization. .54 .67COMP3. My private views about this organization are different than those

    I express publicly. .65 .61

    Notes. S1: sample1, S2: sample2. Factor loadings are based on completely standardized solution results.

    N. Delobbe and C. Vandenberghe: Multiple Dimensions of Commitment 131

    EJPA 16 (2), 2000 Hogrefe & Huber Publishers

  • of this size are generally considered to be important (Wi-daman, 1985). Although, for the four-factor model, theGFI and NNFI values in Sample 1 and that of the GFI inSample 2 are below the .90 benchmark, the correspond-ing CFI values are either very close or above the .90level. Moreover, the RMSEA for the four-factor modelyields acceptable values in both Sample 1 (.07) and Sam-ple 2 (.06). Finally, the 2/df ratio for this model is alsogood in both Sample 1 (1.89) and Sample 2 (1.75).

    Table 2 provides the factor loadings for the four-factormodel within the two samples. A close inspection ofthese loadings reveals that most of them are significantand above .40. An exception is the affective commitmentitem 4, which has a loading of .35 in Sample 1 (cf. Table2). It is worth noting that the compliance items also havegood loadings. In fact, the average variance that the com-pliance construct explains in its indicators averages 36%in Sample 1 and 41% in Sample 2, which is acceptablecompared to the proportions of variance accounted forby internalization (51% and 53%), affective commit-ment (42% and 50%), and continuance commitment(28% and 33%) in their own indicators.

    Table 3 displays the intercorrelations (estimated byLISREL) among the four commitment factors across thetwo samples. Although all correlations are significantlydifferent from unity (cf. Anderson & Gerbing, 1988),some of these are quite high. Indeed, internalization andaffective commitment are correlated between .86 and .89whereas compliance is also strongly related (though neg-atively) to affective commitment (.74 and .76). Thesefindings suggest that affective commitment and internal-

    ization partly tap the same content domain and that com-pliance, as it is operationalized, partly reflects the reverseof affective commitment. Further examination of the dis-criminant validity of these scales is provided belowthrough the inspection of their relationships with criterionvariables.

    Relationships with Criterion Variables

    Table 4 provides the correlations between commitmentdimensions and criterion measures. As can be seen, ten-ure is positively correlated with continuance commit-ment in Sample 1 but uncorrelated with it in Sample 2.Affective commitment and compliance were unrelated totenure in either sample. On the whole, these results lendonly partial support for Hypothesis 1.

    Results from Table 4 also indicate that continuancecommitment is negatively related to employment status(data from Sample 1). That is, those employees with along-term labor contract are more committed on a con-tinuance basis than those with a short-term one. In otherwords, long-term employment increases the costs ofleaving. Hypothesis 2 is thus confirmed. Continuancecommitment is also negatively associated with level ofeducation but only in Sample 2. The reason for the dis-crepancy between Sample 1 and 2 may be that the natureof education levels was different across the two samples,which may result in differences in skills transferability.Hypothesis 3 is thus partly supported.

    Affective commitment and internalization are posi-tively associated with met expectations (cf. Table 4,Sample 1). This provides support for Hypothesis 4. Notethat the correlation involving affective commitment andmet expectations is significantly higher than that involv-ing internalization and met expectations (t = 2.19, p