A multitrait‐multioccasion generalization of the latent trait‐state model: Description and application

Abstract

Single‐trait, multioccasion (STMO) measurement models have been developed to decompose variance into the constituent components of common trait, state, item‐specific, and random error. In this article, a generalization of the STMO model was provided via a confirmatory measurement model for multitrait‐multioccasion (MTMO) data. A brief description of several commonly employed confirmatory measurement models (e.g., congeneric factor analytic model) was provided and contrasted with the MTMO. An illustration of the MTMO model was provided with a large, four‐wave data set involving family support and alcohol involvement. Conceptual issues regarding the trait‐state distinction were discussed, as were statistical issues about the estimation of parameters corresponding to item‐specific and measurement error components.