Determining the Internal Structure of Responses to a Physical Activity Self-Efficacy (PASE) Scale Under a Latent Variable Approach: A Monte Carlo Study

The objective of this study was to improve the validity of the physical activity self-efficacy (PASE) measurement. To accomplish this objective, a latent variable approach commonly advocated, but infrequently used in practice, for determining the internal structure of response patterns to a PASE scale was systematically studied. Data were simulated using Monte Carlo experimental methods under conditions commonly observed in practice (e.g. N; exploratory/confirmatory latent variable models) to maximize potential implications for data analysis in practice. Three sets of outcomes were monitored: convergence issues, model-data fit, and parameter estimates. A one-dimensional model outperformed two-dimensional exploratory and confirmatory models when the population model was one-dimensional. A two-dimensional exploratory model outperformed a two-dimensional confirmatory model and a one-dimensional model when the population model was two-dimensional. These findings provide an initial empirical basis for fuller adherence to the long proposed, but rarely fully followed, validity guidelines for measuring PASE – though experimental-design related limitations exist.