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

Area of Science:

• Developmental psychology
• Psychometrics
• Neuroscience

Background:

• The Child Behaviour Checklist (CBCL) is a primary measure in developmental research.
• It is the main psychopathology measure in the Adolescent Brain Cognitive Development (ABCD) study (N=11,861).
• Detailed psychometric assessment of the CBCL within the ABCD cohort is lacking.

Purpose of the Study:

• To rigorously assess the psychometric properties of the CBCL within the ABCD study.
• To evaluate the fit of commonly used structural models for the CBCL in this large cohort.
• To determine the validity and precision of CBCL subscales for dimensional measurement.

Main Methods:

• Multi-method psychometric framework including Confirmatory Factor Analysis (CFA), Bifactor CFA (B-CFA), Factor Mixture Modeling (FMM), Item Response Theory (IRT), and Exploratory Structural Equation Modeling (ESEM).
• Analysis of data from the Adolescent Brain Cognitive Development (ABCD) study (N=11,861; ages 8-11).
• Evaluation of unidimensionality, bifactor structure fit, construct validity, measurement precision, and factor stability.

Main Results:

• The commonly used bifactor structure of the CBCL did not fit the ABCD data.
• A unidimensional model adequately described the CBCL scale.
• Seven of eight subscales exhibited insufficient construct validity.
• Poor measurement precision was observed below the mean of latent traits.
• Item-level analyses yielded unstable factors with weak discriminant validity.

Conclusions:

• The CBCL's psychometric properties in the ABCD study are suboptimal for dimensional measurement.
• The scale functions as unidimensional rather than fitting a bifactor structure.
• Findings caution against using the CBCL for dimensional assessment of developmental psychopathology in large-scale research.
• Need for measures specifically validated for dimensional psychopathology in population studies.