The General Factor of Psychopathology in the Adolescent Brain Cognitive Development (ABCD) Study: A Comparison of Alternative Modeling Approaches

• In Event: Substantive Developmental Research Questions Addressed with Applied Measurement Methods

Abstract

Many models of psychopathology include a single general factor of psychopathology (GFP) or “p factor” to account for covariation across symptoms. The Adolescent Brain Cognitive Development (ABCD) Study is a longitudinal study of over 11,000 youth across 21 sites that provides a seminal opportunity to investigate the development of the GFP. Taking full advantage of the opportunities afforded by ABCD to study the GFP, however, requires methodological rigor, with numerous questions about how to best model the GFP needing to be addressed. Indeed, the variety of approaches for modelling the GFP increases the “researcher degrees of freedom” for investigating the GFP in ABCD, which raises concerns about how these modeling choices impact scores on the GFP, as well the sub-factors. To address these concerns, we fit and compared 14 distinct GFP models that varied as a function of whether the GFP was modeled using item or scale level data, whether sub-factors were based on the standard a priori structure or preliminary IFAs/EFAs, and whether the model form followed a bifactor, a higher-order, or single-factor (all indicators loading on the GFP without sub-factors) structure.
The sample consisted of youth from the ABCD study (N=11,875), recruited from 21 research sites across the United States. The sample was roughly gender balanced (48.4% female) with a mean age of 9.92 years (SD=0.62 years). Around half (51.6%) of the sample was White, with the remaining participants identifying themselves as Hispanic (20.0%), Black (14.8%), Other/Multi-racial (10.0%), and Asian (2.10%) (data on ethnicity was missing for 1.50% of the sample). We used the parent-reported Child Behavior Checklist (CBCL) as the basis for GFP model construction. Several outcome variables tapping different domains of functioning were also selected from the ABCD protocol and correlated with GFP, internalizing, and externalizing factor scores. These variables were chosen to capture both alternative means of assessing psychopathology in youth (i.e., different raters and/or inventories), as well as nomologic correlates of psychopathology (e.g., school performance, general cognitive ability).
We then fit the 14 distinct GFP models. Overall, children’s rank-ordering on the GFP was stable across models (mean r=.92) with GFP scores similarly related to criterion variables. Similar patterns were observed for internalizing and externalizing sub-factors both within model type, and between model type when sub-factors were made more comparable (i.e., when lower order factors were residualized of the higher order GFP factor variance, making them more analogous to bifactor model specific factors). Results suggest that while theoretical debates about modeling the GFP continue, the practical implications of these choices for rank-ordering children and assessing external associations will often be modest. That is, results point to the substantial robustness of a “core structure” of psychopathology as measured by the CBCL—encompassing the GFP along with internalizing/externalizing dimensions—that emerges across a number of different models that are specified in different ways. By showcasing a variety of methods for calculating GFP scores in ABCD, these findings lay a foundation for future investigations into the development of psychopathology in early life and beyond.

Authors

• D. Angus Clark, University of Michigan - Ann Arbor
  Presenting Author

• Brian M Hicks, University of Michigan
  Non-Presenting Author

• Mike Angstadt, University of Michigan Health System
  Non-Presenting Author

• Saige Rutherford, University of Michigan Health System
  Non-Presenting Author

• Aman Taxali, University of Michigan
  Non-Presenting Author

• Luke W. Hyde, University of Michigan - Ann Arbor
  Non-Presenting Author

• Alexander Weigard, University of Michigan, Ann Arbor
  Non-Presenting Author

• Mary Heitzeg, University of Michigan Health System
  Non-Presenting Author

• Chandra Sripada
  Non-Presenting Author