Examination of Alternative Models of Job Satisfaction

Researchers examining efficacy of alternative measures of job facet satisfaction have generally assumed overall job satisfaction to be an additive function of weighted facet scores. This paper considers the linear compensatory model implied by such an assumption as well as two nonlinear alternatives. It is found that conjunctive and disjunctive models predict overall job satisfaction about as well as does the linear compensatory model. Further, simulated general satisfaction scores are used to examine the extent to which aggregation of data across individuals using differing combinatory models may lead to systematic bias in favor of a particular model. The results suggest that while use of the linear compensatory model may be justifiable on the basis of mathematical simplicity, conclusions drawn by researchers using that model should be tempered by the recognition that alternative models with sharply differing theoretical bases and practical implications may have equivalent descriptive ability.