An empirical test of Mobley's model of employee turnover

Dissatisfied by the lack of integration in the turnover literature, many students of the discipline are turning to the development of theories and models of the processes underlying turnover. Mobley (1977) has proposed a model of the psychological processes underlying the decision to resign. The model depicts several steps in the individual's decision processes beginning with the evaluation of the existing job and job satisfaction, through the search for and evaluation of alternatives, culminating in the intention to stay/quit and the eventual decision to stay or leave. In the present study an attempt was made to test the causal structure suggested by Mobley's model, by predicting each succeeding step in the model from all preceding steps. Mobley's work suggests that the adjacent preceding step of the model provides the best prediction of each step. Additional tests were carried out to test for differences in the predictability of employee turnover from intrinsic and extrinsic satisfaction. An attempt was made to assess the generalization of model predictors from one to two year intervals for turnover measures.

There was no evidence of a predictive superiority for either intrinsic or extrinsic satisfaction in relation to one or two year turnover. The results indicated that the causal sequence of cognitive stages preceding employee turnover generally followed model predictions. In six of nine regression analyses, the best predictor of a step in the model was the immediately preceding step. However, results of this study failed to support the central position assigned in the model to turnover intentions. Turnover intentions were not well predicted by adjacent variables and failed to yield a substantial connection to one year turnover. The findings were mixed with respect to the generalization of turnover predictors across time intervals. Evaluation of alternatives was the best predictor of one year turnover. Intrinsic satisfaction explained the most variance in two year turnover. The intention to search for alternatives proved to be a useful predictor of turnover across both time intervals.

The results are discussed as providing partial support for the model. Reservations still remain particularly with respect to the role of turnover intentions. No firm conclusions could be drawn concerning the generalizations of turnover predictors over time. Directions for future model building were given.