Cognitive, affective, and satisfaction variables as predictors of organizational behaviors: A structural equation modeling examination of alternative models

The role of cognitive, affective, and satisfaction variables in the prediction of organizational behaviors was studied via questionnaires administered to managers at a large Midwestern university at two time periods: a computerized questionnaire administered June through August, 1998 (N = 188) and a web-based questionnaire administered in September, 1998 ( N = 144). Models predicting turnover intentions, work withdrawal, organizational retaliation behaviors (ORBs), and organizational citizenship behaviors (OCBs) were proposed based on cognitive, affective, and satisfaction antecedents found in the theoretical and empirical literature. Two major models were proposed: Model I-R, a model based on the Modified Organizational Withdrawal/Adaptation Model (Hulin, 1991), and Model II, a model based on Affective Events Theory (Weiss & Cropanzano, 1996). Model I-R predicted overall job satisfaction directly influences the behaviors. Model II predicted overall job satisfaction influences turnover intentions and positive and negative affect at work directly influence the remaining three behavioral variables. Two alternative models, Models III and IV, were developed by adding paths to Models I-R and II. Results based on structural equation modeling indicated that Models I-R and II provided a better fit to the data than Models III and IV. Specifically, in Models I-R and II positive and negative affect at work were influenced by emotional dispositions and work events. Moreover, cognitive evaluation of work and negative and positive affect at work significantly predicted overall job satisfaction. In Model I-R, overall job satisfaction predicted all four behaviors at both time periods. In Model II, overall job satisfaction predicted turnover intentions and negative affect predicted ORBs at both time periods. A fifth model, referred to as the Hybrid Model, was created by adding a path from negative affect at work to ORBs to Model I-R; the fit of this model was similar to the fit obtained in Model I-R. Results indicated that the proposed relationships in Model I-R, II, and the Hybrid Model were invariant over time. Overall, results supported the majority of the relationships hypothesized in Affective Events Theory and the Modified Organizational Withdrawal/Adaptation Model. The theoretical and practical implications of these findings are discussed.