Multitrait-Multimethod-Multioccasion Modeling

The author reviews structural equation approaches to modeling multitrait-multimethod (MTMM) longitudinal data and defines new models based on stochastic measurement theory. The new models can be used to analyze latent states and latent change in an MTMM context. The author provides a detailed psychometric analysis of the new models and illustrates their use in an empirical application. He also presents a Monte Carlo simulation study, in which the applicability of the models is scrutinized for different sample sizes. The author discusses various practical modeling issues in detail, such as model selection, testing of measurement invariance, dealing with indicator-specific effects over time, and assessment of convergent and discriminant validity of change. In the final section, the author compares the new models to alternative approaches, discusses advantages and limitations, and provides detailed guidelines for applied researchers.