The Structure of Psychopathology in a Sample of Clinically Referred, Emotionally Dysregulated Early Adolescents.

Abstract

This investigation answers and amplifies calls to model the transdiagnostic structure of psychopathology in clinical samples of early adolescents and using stringent psychometric criteria. In 162 clinically referred, clinically evaluated 11-13-year-olds, we compared a correlated two-factor model, containing latent internalizing and externalizing factors, to a bifactor model, which added a transdiagnostic general factor. We also evaluated the bifactor model psychometrically, including criterion validity with broad indicators of psychosocial functioning. In doing so, we compared alternative approaches to defining and interpreting criterion validity: a recently proposed incremental definition based on amounts of variance in criterion factors explained, and the more typical definition based on the presence of conceptually meaningful relationships. While traditional fit statistics favored the bifactor model as expected, psychometric analyses added important nuance. Despite moderate reliability, the general factor was not fully transdiagnostic (i.e., was not informed by several externalizing scores), and was partially redundant with internalizing scores. Approaches to criterion validity yielded opposing results. Compared to the correlated two-factor model, the bifactor model redistributed, without incrementally increasing, the total variance explained in criterion indicators of psychosocial functioning. Yet, the bifactor model did improve the precision of clinically important relationships to psychosocial functioning, raising questions about meaningful tests of bifactor psychopathology models.