Revisiting the Structure of DSM-5 Section II Personality Disorder Criteria Using Individual Participant Data Meta-Analysis

Abstract

The factor structure of Personality Disorder (PD) criteria has long been debated, but due to previous heterogenous findings, a common structure to represent covariation among DSM-IV/DSM-5 Section II PD criteria remains an open question. This study integrated a total of N = 30,545 PD assessments from 25 samples to conduct an individual participant meta-analytic factor analysis on the structure of PD criteria. Measurement invariance testing across gender, clinical status, and assessment methods indicated substantial structural differences between interview-based and self-reported measures. In interviews, a confirmatory ten-factor model with factors representing specific DSM-5 PDs showed misfit, which could be addressed by allowing for secondary loadings (using exploratory factor analysis with target rotation). In self-reports, a confirmatory ten-factor model showed stronger misfit than in interviews and exploratory solutions were less clear and suggested that a simpler model may be preferable. Factors showed some resemblance to maladaptive trait domains such as Negative Affectivity and Disinhibition when extracting five factors, but there were substantial differences in factor content between interviews and self-reports. In bifactor rotated models, a general factor showed higher explained variance in self-reports, while the content of general factors was similar across both assessment methods. Our results suggest that interview and self-reported measures of PD criteria are not structurally equivalent. To advance research on the structure of PD, it might be useful to consequently focus on the shared variance of multiple methods. For this purpose, future multimethod studies should combine interviews and self-reports with further assessment methods such as informant-reports.