Exploratory Bifactor Analysis of Symptoms and Traits Highlights Aspects of Psychopathology Inaccessible by Other Methods

Abstract

Bifactor analysis has the distinct ability to model general versus specific psychopathology liabilities and the extent to which these liabilities are involved in manifest symptoms. However, overreliance on confirmatory models of diagnostic covariance hampers substantive implications. We undertook an exploratory bifactor approach to symptom-level epidemiological data (N = 8,405) to critically examine known issues in bifactor analysis and contribute to the development of more finely grained psychopathology models. The resulting model included Distress, Harmful Alcohol Use, Antisocial Behavior, Attention Seeking, Social Alienation, and Psychosis specific factors. Within diagnostic categories, symptoms showed substantial heterogeneity regarding their degree of generality versus specificity and in their primary loadings on specific factors. Results clarify the extent to which shortcomings of bifactor models persist beyond the limitations of diagnosis-level data and/or confirmatory methods. These findings suggest additional layers of complexity and nuance to quantitative dimensional taxonomies and offer new insights into critical debates in the bifactor literature.