ABSTRACT
Multiple strategies are explored for comparing means and proportions across two or more time points with structural question modeling. The relation of these models to conventional tests such as repeated-measures ANOVA, MANOVA, McNemar's chi-square, and marginal homogeneity tests our outlined. Each model is illustrated with example analyses. The connections between simple models comparing dependent means and proportions across a few time points and more complex models, such as latent growth curve models and latent difference score models, are discussed.
