Latent State-Trait Models for Longitudinal Family Data

Abstract. Support is key to healthy family functioning. Using the family social relations model (SRM), it has already been shown that variability in perceived support is mostly attributed to individual perceiver effects. Little is known, however, as to whether those effects are stable or occasion-specific. Several methods have been proposed within the structural equation modeling (SEM) framework for the investigation of hypotheses on stable and occasion-specific aspects of such psychological attributes. In this paper, we explore the applicability of different models for determining the consistency of SRM effects of perceived support: the multistate model, the singletrait-multistate model, and the trait-state occasion model. We provide a detailed description of the model building process and assumption verification, as well as the supporting R-code. In addition to the methodological contribution on how to combine these models with the SRM, we also provide substantive insights into the consistency of perceived family support. We rely on round robin data on relational support from the Dutch RADAR-Y (Research on Adolescent Development and Relationships – Younger Cohort) study, a 6-year longitudinal study of 500 families with a 13-year-old target adolescent at the start of the study.