Longitudinal bifactor modeling of anxiety, depression and schizotypy - The role of rumination as a shared mechanism

Abstract

A bifactor model with a general (p) factor reflecting shared variance and specific factors reflecting additional variance in individual symptoms has been introduced to explain common co-occurrence among anxiety, depression and schizotypy. However, longitudinal evidence is lacking and the validity of bifactor modeling is debatable. The current study aimed to examine the presence of the p factor together with specific factors in accounting for relationships between anxiety, depression and schizotypy both cross-sectionally and longitudinally, and to investigate the relationship between these factors and rumination. A validated sample of university students were surveyed on levels of anxiety, depression, schizotypy and rumination at baseline (N = 2291), one year (N = 1833) and two years (N = 1656). Models were estimated using exploratory structural equation modeling (ESEM) and compared at each time point. Longitudinal invariance of the best-fitting model was examined and all potential within- and between-factor stability pathways were tested in an SEM framework. A bifactor model with a p factor and four specific factors (representing residual information of composite anxiety and depression, cognitive-perceptual, interpersonal and disorganized schizotypy respectively) consistently outperformed a correlated-factors model. The bifactor structure appeared longitudinally stable. Within-factor stabilities were moderate, and between-factor pathways reflected a few significant interactions, mostly involving the p factor. Rumination was independently associated with p and four specific factors at each time point. Therefore, there is a p factor accounting for concurrent and sequential co-occurrence of anxiety, depression and schizotypy. Rumination explained partly the p and specific factors. Transdiagnostic interventions should target rumination.