META-ANALYSES

The most comprehensive and widely cited meta-analytic review of the organizational commitment literature was that conducted by Mathieu and Zajac (1990). Mathieu and Zajac reported mean sample-weighted correlations, corrected for unreliability, between organizational commitment and 48 other work-related variables divided into three categories: antecedents (26), consequences (8), and correlates (14). I will organize my review of meta-analytic findings around these same three categories. Within each category, I will begin by summarizing the findings obtained by Mathieu and Zajac and then present the results of more recent analyses. The latter generally had a narrower focus and included more detailed assessments of moderator effects. Unless otherwise stated, all correlations reported in this section of the article are sample- weighted and corrected for attenuation due to unreliability.

Antecedents of Commitment

Among the variables Mathieu and Zajac ¢1990) classified as antecedents of commitment were those reflecting personal characteristics, job characteristics, group-leader relations, organizational characteristics, and role states. Of these, the strongest correlations were obtained for job characteristics, particularly job scope (enrichment), and group-leader relations (e.g. leader communication, participative leadership, task interdependence). Personal characteristics (with the exception of perceived personal competence) were generally found to have weak correlations with commitment. Sex was one of the personal characteristics found not to be related to commitment. Aven, Parker, and McEvoy (1993) confirmed this result in a more recent meta-analysis based on a larger set of studies. Few studies examined relations between commitment and organizational characteristics and those that did reported relatively weak correlations. Role states (ambiguity, conflict, overload) were not found to have meaningful links to commitment, although there was sufficient between-study variance to suggest that moderators might be operating.

Cohen and his colleagues conducted a series of meta-analyses to address more specific antecedent-commitment relations and potential moderating effects. For example, Cohen and Lowenberg (1990) reviewed the findings of studies conducted to test Becker's side-bet theory of commitment. According to Becker, commitment increases as employees make side-bets, or investments, that would be lost if they were to leave the organization. Cohen and Lowenberg found that the correlations between commitment and 11 side-bet variables (e.g. age, tenure, gender) were generally weak, and concluded that there is little evidence to support side-bet theory. They cautioned, however, that the side-bet and commitment measures used in the studies included in their analyses might not have been appropriate for testing side-bet theory (cf. Meyer & Allen, 1984).

Cohen (1992) tested for the moderating effect of occupation on the correlations between commitment and various personal and organizational antecedent variables. He found that commitment was more strongly related to personal characteristics (e.g. tenure, education, marital status, gender, motivation) for employees in blue-collar and non-professional white-collar occupations than for professionals. Correlations between commitment and organizational characteristics also varied across occupational groups, but the pattern was less consistent. For example, role ambiguity correlated more strongly with commitment among non-professionals, whereas autonomy and communication correlated more strongly with commitment among professionals. These latter findings suggest that, in general, the needs and preferences of members of these broad occupational groups differ, and that the commitment they experience varies as a function of whether these needs and preferences are satisfied at work.

In a related analysis, Cohen (1993a) examined the correlations between commitment and age and tenure within different career stages. He found that the correlation between age and commitment was stronger among younger employees (i.e. those under 30) than it was for the other age groups. In contrast, the correlation between tenure and commitment was greater among the more senior employees (i.e. those with more than nine years of experience). These findings suggest that the correlations involving age and tenure may not be linear, and might help to account for the relatively weak correlations reported by Mathieu and Zajac (1990).

Cohen and Gattiker (1994) used meta-analysis to examine the link between organizational commitment and rewards, operationalized as actual income and pay satisfaction. Across all studies, commitment was more strongly related to pay satisfaction than to actual income. These relations were moderated to some extent, however, by structural characteristics. Specifically, pay satisfaction correlated more strongly with commitment in the private versus public sector, and the correlation between actual income and commitment was greater for professional than for clerical employees.