The general factor of psychopathology (p): Choosing among competing models and interpreting p

Abstract

Over the past 10 years, the general factor of psychopathology, p, has attracted interest and scrutiny. We review the history of the idea that all mental disorders share something in common, p; how we arrived at this idea; and how it became conflated with a statistical representation, the Bi-Factor Model. We then leverage the Environmental Risk (E-Risk) longitudinal twin study to examine the properties and nomological network of different statistical representations of p. We find that p performed similarly regardless of how it was modelled, suggesting that if the sample and content are the same the resulting p factor will be similar. We suggest that the meaning of p is not to be found by dueling over statistical models but by conducting well-specified criterion-validation studies and developing new measurement approaches. We outline new directions to refresh research efforts to uncover what all mental disorders have in common.

Fig. 1.

The nomological network of psychopathology. Each panel shows associations between the first-order factors from the correlated-factors model (in green), different versions of the p-factor (in yellow), and adjusted p-factor scores (in brown) with (a) family history of psychopathology, (b) social class origins, (c) childhood IQ, (d) childhood self-control problems, (e) childhood maltreatment, (f) adolescent victimization, and (g) elevated inflammation. Values shown are βs (error bars indicate 95% confidence intervals). Ext = Externalizing; Int = Internalizing; ThD = Thought Disorders.

Fig. 2.

What do the p-free factors from bifactor models mean? The figure shows associations between the three first-order factors from the correlated-factors model, each of the three first-order factors from the correlated-factors model adjusted for the other two first-order factors, and p-free factors from the bifactor models with (a) family history of psychopathology, (b) social class origins, (c) childhood IQ, (d) childhood self-control problems, (e) childhood maltreatment, (f) adolescent victimization, and (g) elevated inflammation. Values shown are βs (error bars indicate 95% confidence intervals). Ext = Externalizing; Int = Internalizing; ThD = Thought Disorders.

Fig. 3.

p is a developmental life-course phenomenon. Higher p—derived from a factor analysis of all symptoms reported by participants over the first half of their lives—is associated with early onset, persistence, and cumulative diversity of mental disorders, from age 11 to age 45 (data from Caspi et al., 2020). y = years.