Confirmatory Test of a Turnover Model Utilizing Multiple Data Sources

In recent years, there have been several confirmatory tests of turnover models, based almost exclusively on subject self-reports for all but the measurement of turnover itself. A test of a subset of the Mobley, Griffeth, Hand, and Meglino (1979) model was conducted using data from three sources. It was proposed that autonomy, workload, constraints, role ambiguity, and interper- sonal conflict (assessed by supervisor reports) would lead to job satisfaction (assessed by subject reports), which would lead to intention of quitting (as- sessed by subject reports), which would lead to turnover (from organization records). A trimmed version of the model (with role ambiguity and interper- sonal conflict deleted) yielded good overall fit, but job conditions accounted for a very small amount of job satisfaction variance, and workload related in a direction opposite to predictions.