Conjectures and refutations in cognitive ability structural validity research: Insights from Bayesian structural equation modeling

Area of Science:

• Psychometrics
• Cognitive Assessment
• Structural Equation Modeling

Background:

• Traditional factor analytic approaches have limitations in fully capturing the theoretical structure of intelligence tests like the WISC-V.
• Prior research has favored a four-bifactor model for the WISC-V structure.
• The need for advanced methods to explore complex cognitive interrelationships is crucial for IQ theory.

Purpose of the Study:

• To investigate the theoretical structure of the WISC-V using Bayesian structural equation modeling (BSEM).
• To compare the fit of a five-factor higher-order model against the previously preferred four-bifactor model.
• To explore BSEM as a tool for advancing IQ theory and addressing replication issues in cognitive assessment.

Main Methods:

• Application of Bayesian structural equation modeling (BSEM) to the WISC-V dataset.
• Specification of all cross-loadings and correlated residual terms within the BSEM framework.
• Comparison of model fit indices between the proposed five-factor model and the existing four-bifactor model.

Main Results:

• A five-factor higher-order model, including a correlated residual between Visual-Spatial and Fluid Reasoning, demonstrated superior fit compared to the four-bifactor model.
• No other statistically significant cross-loadings or correlated residual terms were identified in the measurement model.
• The WISC-V ten subtest primary battery achieved simple structure, supporting the interpretation of its index scores.

Conclusions:

• BSEM provides enhanced insights into the WISC-V structure, surpassing traditional factor analytic methods.
• The findings support the interpretation of WISC-V index scores as per the manual.
• BSEM offers a novel tool for intelligence researchers to explore complex cognitive relationships and improve structural validity research in cognitive assessment.