Bifactor analyses have become a popular way to model a general psychopathology factor (or p factor) that accounts for correlations among higher order internalizing, externalizing, and thought-disorder dimensions. However, to advance dimensional nosology and construct validation of the p factor, this body of research needs to be synthesized. This study focuses on the p factor as a substantive construct by testing its consistency across samples and models. Bifactor analyses were run for three large epidemiological and clinical data sets: National Comorbidity Survey, Collaborative Psychiatric Epidemiology Surveys, and Methods to Improve Diagnostic Assessment and Services. Despite generally good model fit and strong factor congruence, interpretation of the p factor varied across samples and methods. This limited consistency suggests a lack of clear, generalizable operationalization of the p factor for applications such as measure development. Results are interpreted in terms of implications for the p factor as a substantive construct.
Keywords: assessment; bifactor analysis; dimensional models; measurement; p factor; psychopathology; structure.