اعتبار مدل سه مولفه تعهد سازمانی در یک محیط چین

This study examined the construct validity of Meyer and Allen’s (1991) three-component model of organizational commitment in a Chinese context and compared levels of OC between the Chinese sample and previously published data from Canada and South Korea. In a study of 226 Chinese employees in six foreign-invested companies, a five-factor oblique model, which included both substantive and method factors, fit the data best. However, the three-component model had a reasonably good fit. The proposed antecedents of each of the three principal OC dimensions tended to associate most strongly with their respective scales. Affective commitment and normative commitment significantly predicted job satisfaction and all three components predicted turnover intention, although associations between continuance commitment and these outcomes were moderated by normative commitment. Finally, we found that normative and affective commitment were significantly higher in the Chinese sample than in previously published samples from Canada and South Korea. Continuance commitment in the Chinese sample was lower than the Canadian and Korean samples. Implications of these findings are discussed in more detail.

A variety of studies have been conducted to explore the concept of organizational commitment (OC) (e.g., Becker, 1960; Mowday, Porter, & Steers, 1982), to examine the validity of different measures of organizational commitment (e.g., Allen & Meyer, 1996; Hackett, Bycio, & Hausdorf, 1994), and to investigate the relations between organizational commitment and its antecedents and outcomes (e.g., Allen & Meyer, 1990; Buchko, Weinzimmer, & Sergeyev, 1998). Although there have been several definitions and measures of OC (e.g., Becker, 1960; Meyer and Allen, 1984 and Meyer and Allen, 1991; Porter, Steers, Mowday, & Boulian, 1974), Meyer and Allen’s (1991) three-component model of OC has been the dominant framework for OC research in the past decade because it is based on a more comprehensive understanding of OC. The three-component model consists of: (a) affective commitment [AC, an emotional attachment to one’s organization, similar to Porter et al.’s (1974) conceptualization]; (b) continuance commitment (CC, attachment based on the accumulation of valued side bets such as pension, skill transferability, relocation, and self-investment that co-vary with organizational membership); and, (c) normative commitment (NC, attachment that is based on motivation to conform to social norms regarding attachment). Common to these three components is the view that commitment is a psychological state that characterizes the employee’s relation with the organization, and has implications for the decision to continue or discontinue membership in the organization (Meyer & Allen, 1991). In addition, cross-cultural studies of organizational commitment have been conducted in many countries such as Canada, Great Britain, Belgium, Australia, Russia, South Korea, Japan, Singapore (e.g., Ko, Price, & Mueller, 1997; Lee, Allen, Meyer, & Rhee, 2001; Vandenberghe, 1996). With increasing globalization and awareness of cultural issues that bear on social, cognitive, and attitudinal issues in the workplace (e.g., Hofstede, 1980a), extending research on organizational commitment to other borders is important. The current research examines the construct validity of Meyer and Allen’s (1991) three-component model of OC in China, the world’s largest population center.

The means, SDs, internal consistency estimates (Cronbach’s α), and scale inter-correlations for the Chinese sample as well as for the Canadian and South Korean comparison samples are presented in Table 1. The internal consistency reliabilities of the three scales were not as high in the Chinese sample as they were in the Canadian or South Korean samples. The reliability coefficients of AC and NC were acceptable (α>.75), but the reliability of CC was not acceptable (α=.52). The internal consistencies of CC in the South Korean samples were also below acceptable levels.For normative commitment, the Chinese sample has a significantly higher level than the Canadian sample (z=14.29, p<.001) and than the South Korean sample (z=4.74, p<.001). The South Korean sample also has a higher level of normative commitment than the Canadian sample (z=8.77, p<.001). Therefore, Hypothesis 1 was supported. However, contrary to Hypothesis 2, the Chinese sample had a significantly higher affective commitment than the Canadian sample (z=8.51, p<.001) and the South Korean sample (z=4.38, p<.001). The South Korean sample also has a higher affective commitment than the Canadian sample (z=2.89, p<.01). Therefore, Hypothesis 2, that AC would be lower in the Asian samples than the Canadian sample, was not supported. On the other hand, the Chinese sample had significantly lower continuance commitment than the Canadian sample (z=4.21, p<.001) and than the South Korean sample (z=5.43, p<.001). No significant difference was found between the South Korean sample and the Canadian sample on continuance commitment. Therefore, Hypothesis 3 is partially supported. Finally, two of the three inter-correlations among three scales were significantly higher in the Chinese sample than in the Canadian sample. Specifically, in the Chinese sample, the correlation between AC and CC was .36, whereas it was −.05 in the Canadian sample (z=6.01, p<.0001). The correlation between CC and NC was .48 in the Chinese sample and .11 in the Canadian sample (z=5.81, p<.0001). However, the correlation between AC and NC was .64 in the Chinese sample and .63 in the Canadian sample (z=.24, n.s.), and there is no significant difference between two samples. Similarly, the correlation between AC and CC was higher in the Chinese sample (r=.36) than in the South Korean sample (r=−.03) (z=4.59, p<.0001); the correlation between CC and NC was also higher in the Chinese sample (r=.48) than in the South Korean sample (r=.17) (z=3.97, p<.0001); and there is no significant difference between the Chinese sample (r=.64) and the South Korean sample (r=.58) (z=1.08, n.s.) on the correlation between AC and NC. On the other hand, the Canadian sample and the South Korean sample are quite similar with respect to the intercorrelations among three subscales; it was found that two samples do not have significant differences on any of the intercorrelations (zrac=−.26, n.s.; zrcn=−.78, n.s.; zran=1.01, n.s.).Table 2 provides the results of the confirmatory factor analyses. The oblique models (Models 2 through 6) fit the data better than did the orthogonal models (Models 7 through 11). Among the oblique models, the five-factor model (Model 2) produced the lowest χ2 value and lowest χ2/df ratio (see Table 2). Although the remaining models were significantly different from Model 2, the fit indices suggested that the other oblique models, including the three-component model, were not appreciably worse than the 5-factor oblique model. Even the one-factor model (model 12) was not appreciably worse than any of the multi-factor, oblique models. The fit indices for any of these models were only marginally good, but they are similar to values reported in the literature using Canadian samples (Hackett et al., 1994).All of the structural coefficients, or loadings, between the latent factors and the OC items in the five-factor oblique model were significant. However, the path between CC-High Sacrifice and Item 11 (If I had not already put so much of myself into this organization, I might consider working elsewhere) was negative (standardized coefficient=−.27), the path between CC-Low Alternatives and Item 12 (One of the few negative consequences of leaving this organization would be the scarcity of available alternatives) was relatively low (.15) as was the path between CC-Low Alternatives and Item 10 (I feel that I have too few options to consider leaving this organization) (.23). All other paths/loadings were above .34. In addition the correlations among the latent factors were all significant and quite high. The weakest correlation was between AC-negative wording and CC-Low Alternatives (r=.57). Thus, the three-component model of OC among Chinese workers (Hypothesis 4) appears to have modest support. The three primary dimensions are much more highly inter-correlated in the Chinese sample than they are in the Canadian and South Korean samples. Furthermore, the CFA shows that the 5-factor solution, which contains substantive sub-factors for the CC dimension and method factors for the AC dimension is the best-fitting, albeit mediocre, solution. Further inspection of this model reveals less-than-optimal loadings on several of the CC items.To further examine the construct validity of the three-component model of OC in a Chinese context, we examined the associations between antecedents and outcomes of OC that have been well researched in other cultural contexts, primarily Canadian.Hierarchical multiple regression analyses were conducted to examine the relative strength of each OC dimension’s proposed antecedents compared to the other antecedents in predicting each of the OC subscales (the three-component model was utilized). Note that for the remaining analyses, item 11, the CC item that produced a negative loading in the CFA analyses, was eliminated from the scale. The resulting 5-item scale had a internal consistency estimate of α=.64. The results of the hierarchical regression analyses are provided in Table 3. This table shows that each OC dimension’s proposed antecedent variables significantly predicted the dimension. Adjusted R2s for the proposed antecedent variables on its dimension were .47 for AC, .30 for CC, and .33 for NC. However, antecedents of other dimensions explained portions of the variance over and above the proposed antecedents.AC was significantly predicted by Perceived Equity, Peer Group Cohesion, Personal Importance, and Role Clarity (negatively) which constituted 4 of the 8 proposed antecedents for AC. Receipt of Benefits and Socialization Toward Loyalty, NC antecedents, also significantly predicted AC. None of the CC antecedents predicted AC. Three of the four CC antecedents significantly predicted CC: Alternatives (negatively), pension, and costs of leaving. Peer Group Cohesion, an AC antecedent, negatively predicted CC, and Role Clarity, another AC antecedent, positively predicted CC. Neither NC antecedent predicted CC. Both NC antecedents, Receipt of Benefits and Socialization Toward Loyalty, were the strongest predictors of NC. Costs of leaving, a CC antecedent, and Goal Difficulty, an AC antecedent, also significantly predicted NC (see Table 3). Contrary to previous research, none of the demographic or organizational structure variables predicted any of the OC dimensions with the exception of gender. Men reported higher levels of NC (M=4.43, SD=1.09) than did women (M=4.06, SD=1.08). Given the relatively high inter-correlations between AC, CC, and NC in the Chinese sample, it is not surprising that many of the respective antecedent variables significantly predicted other OC dimensions. The pattern of these results only provides moderate support for convergent and divergent validity among the antecedent variables. Nonetheless, with two exceptions (goal difficulty and role clarity), antecedent variables were more strongly associated with their respective OC dimensions than with opposing dimensions. Therefore, we found moderate support for Hypothesis 5, Hypothesis 6 and Hypothesis 7.Controlling for the demographic and structural variables, AC and NC significantly predicted job satisfaction. Of these, AC was the stronger predictor, thus supporting Hypothesis 8. There was, however, a significant 2-way interaction between CC and NC. Simple slopes (Aiken & West, 1991) were computed to diagnose the interaction. When NC was low (−1 SD below the mean) the simple slope of job satisfaction on CC was .12 (unstandardized). When NC was high (+1 SD above the mean), the simple slope of job satisfaction on CC was .00. This indicates that, among Chinese workers, strong normative commitment nullified the positive association between continuance commitment and job satisfaction. All three OC dimensions significantly predicted turnover intentions. Stronger levels of each of these OC dimensions were associated with lower turnover intentions. AC and CC were the strongest predictors. These findings partially support Hypothesis 9, but we did not expect CC to be a stronger predictor of turnover intentions than NC. Again, however, there was a significant CC × NC interaction (see Table 4). Analysis of simple slopes revealed that at low levels of NC (−1 SD), the slope of turnover intentions on CC was −.39. At high levels of NC (+1 SD), the slope was −.13. Similar to the interaction between these terms on job satisfaction, high levels of NC appear to reduce the association between CC and Turnover Intentions.There was no support for the hypothesis that NC would negate associations between AC and job satisfaction and turnover intentions. The AC × NC interactions were nonsignificant in both equations (see Table 4). Therefore Hypothesis 9 was not supported.