Assessing the Psychometric Properties of the Child Behaviour Checklist in the ABCD Study.

Abstract

The Child Behaviour Checklist (CBCL) is a widely used measure of psychopathology in developmental research and serves as the primary measure in the landmark Adolescent Brain Cognitive Development study (N = 11,861; 8-11 years). However, its psychometric properties in this sample have not been assessed in detail. Analyses using a multi-method framework (CFA, B-CFA, FMM, IRT, ESEM) indicate that a commonly used bifactor structure of the CBCL does not fit the data and instead the scale can be adequately described as unidimensional. Moreover, seven of eight subscales demonstrate insufficient construct validity and are not valid indicators of any higher-order models. All exhibit poor measurement precision below the mean of the latent traits, and item-level exploratory methods produce unstable factors with weak discriminant validity. These findings suggest caution in using the CBCL for dimensional measurement in this dataset and highlight the need for measures explicitly developed and validated for dimensional psychopathology in population-scale research. SUMMARY: The CBCL is a widely used measure of child and adolescent psychopathology, but its psychometric properties in the ABCD study cohort remain unclear. Using a multi-method approach we show that the CBCL is unidimensional and is not well represented by the commonly used bifactor model. CBCL subscales demonstrate insufficient construct validity and demonstrate poor measurement precision across the latent trait continuum. In combination, our findings caution against the use of the CBCL for dimensional measurement of developmental psychopathology in the ABCD study cohort.