Appropriate Use of Bifactor Analysis in Psychopathology Research: Appreciating Benefits and Limitations

Abstract

Co-occurrence of psychiatric disorders is welldocumented. Recent quantitative efforts have moved toward an understanding of this phenomenon, with the 'general psychopathology' or p-factor model emerging as the most prominent characterization. Over the past decade, bifactor model analysis has become increasingly popular as a statistical approach to describe common/shared and unique elements in psychopathology. However, recent work has highlighted potential problems with common approaches to evaluating and interpreting bifactor models. Here, we argue that, when properly applied and interpreted, bifactor models can be useful for answering some important questions in psychology and psychiatry research. We review problems with evaluating bifactor models based on global model fit statistics. We then describe more valid approaches to evaluating bifactor models and highlight three types of research questions for which bifactor models are wellsuited to answer. We also discuss the utility and limits of bifactor applications in genetic and neurobiological research. We close by comparing advantages and disadvantages of bifactor models to other analytic approaches and noting that no statistical model is a panacea to rectify limitations of the research design used to gather data. 1 Sometimes, group factors are called "specific factors." However, "specific factor" more correctly refers to an item's reliable (non-error) variance that is not shared with other items (5).

Heritability of overlapping impulsivity and compulsivity dimensional phenotypes

“…A criticism of bifactor models is their tendency to provide superior fit compared to competing models because they flexibly accommodate patterns of covariance regardless of population data structure 70,85,86 . However, results of model testing support use of the bifactor model as a theoretically consistent and empirically plausible representation of the current data.…”

The Psychometric Properties of the Frost Multidimensional Perfectionism Scale – Brief

“…However, a concern in bifactor modeling is "overfitting" due to capturing of unwanted noise and bifactor models' propensity to fit even random patterns (Bonifay et al, 2017). As a result, authors warn not to adopt models based primarily on which fit better (Murray and Johnson, 2013;Bornovalova et al, 2020). In addition, one would expect that pooling items from several different instruments together in one analysis increases the G factor relative to the subgroups.…”

Associations Between Resting State Functional Connectivity and a Hierarchical Dimensional Structure of Psychopathology in Middle Childhood

“…13 Alternatively, others suggested that p-factor may be too general and heterogeneous to reveal etiology, potentially representing functional consequences of psychopathology. 17,18 Hence, it remains unclear what level of specificity in phenotypes is most informative for understanding neural mechanisms.…”

“…13 Alternatively, others suggested that p-factor may be too general and heterogeneous to reveal etiology, potentially representing functional consequences of psychopathology. 17,18 Hence, it remains unclear what level of specificity in phenotypes is most informative for understanding neural mechanisms.At other levels of the hierarchy, 12 the 2-factor solution was comprised of broad internalizing and broad externalizing factors consistent with prior research. [19][20][21] In the 3factor structure, a neurodevelopmental factor (e.g.…”