Convergent and Discriminant Validity of Measures of Job, Leisure, Dyadic, and General Life Satisfaction by Causal Modeling Methodology

Abstract

This article demonstrates the usefulness and appropriateness of causal modeling methodology as an alternative approach in determining construct validity of measures used in leisure research. The present approach is less ambiguous and more rigorous than the widely used multitrait-multimethod analysis, ANOVA, and regression models. For the purpose of illustration, the Trafton-Tinsley data (1980) were reanalyzed using LISREL IV. The results from the present analysis are considerably different from those obtained by Trafton and Tinsley.