The Relative Contribution of Perceived Skill-Utilization and Other Perceived Job Attributes to the Prediction of Job Satisfaction: A Cross-Validation Study

Abstract

With a representative sample of 1383 Adelaide employees it was found, using simple multiple regression, that job satisfaction was significantly predicted by the perceived job attributes of skill-utilization, influence, variety, pressure, and interaction. For the total sample, skill-utilization was the strongest predictor of job satisfaction. In order to measure the degree of over-estimation of R², a double cross-validation of the data was conducted. The R² varied from .47 to .51 and the beta coefficient for skill-utilization remained the highest and varied from .48 to .50. The effects of multi-collinearity were measured using principle components analysis and measures of the variance inflation factor for each beta coefficient. It was concluded that multicollinearity was not a significant problem for these data. The degree to which the observed association between job satisfaction and skill-utilization was due to content similarity of the scale was assessed by deleting ability items from the job-satisfaction scale. There was only a slight reduction in the magnitude of the skill-utilization beta coefficient. Skill-utilization remained the strongest predictor of job satisfaction when the JDI (work itself) was used as the criterion with a separate sample. There was a small reduction in R² and beta coefficients but not sufficient to suggest that the original associations were due to content similarity. Finally, the generality of the regression results was examined by repeating the regression within occupational groupings.