Decreases in posttest variance and the measurement of change

A Monte Carlo study was used to evaluate the effects of reductions in posttest variance on several methods for detecting predictors of change in a two-wave design. When the predictor was dichotomous, the analysis of covariance approach was compared to the analysis of variance on difference scores. For a continuous predictor, partial correlations, difference score correlations with the predictor and latent change correlations with the predictor in structural equation growth models were used. When posttest variance decreased (e.g., ceiling effect) difference scores lost power, while the power of regression based methods (analysis of covariance and partial correlations) and structural equation measures of change were unaffected.