Longitudinal analysis of dyads using latent variable models: Current practices and constraints

Abstract

Interdependencies between dyads have long been recognized and taken into account in the analysis of partnership and marital data. However, most of the research that has examined dyadic influences is based on cross-sectional data or basic longitudinal models. When more complex longitudinal models are examined, several limitations and barriers arise. In this chapter, some of the practical issues with dyadic analyses of multi-time point samples will be discussed. In particular, we discuss (1) applications of latent growth curve mixture modeling trajectories of intimate partner relationship adjustment and (2) latent difference score modeling associations between relationship adjustment and depressive symptoms over time. A 4-year longitudinal sample of 237 families assessed over six time points will be used to illustrate these practical issues.