Configural and Metric Invariance Across Full-Form and Computerized Adaptive Versions of the Penn Computerized Neurocognitive Battery (CAT-CCNB)

Abstract

The Penn Computerized Neurocognitive Battery (CNB) is a collection of tests validated using neuroimaging, genetics, and other criteria. An updated version of the CNB was constructed in which all tests were converted to either computerized adaptive (CAT) or abbreviated forms. In a mixed community/clinical sample ( N = 307; mean age = 25.9 years; 53.7% female), we tested measurement invariance across the CNB and CAT-CCNB in terms of factor configuration and loadings. Multiple-Group factor analyses were conducted with appropriate constraints, and results supported the measurement invariance assumption, with non-significant differences in chi-square fit statistics for both correlated-traits (Δχ2 = 6.68; df = 12; p = .88) and bifactor (Δχ2 = 16.02; df = 27; p = .95) configurations. Results indicate that the latent constructs measured by the two versions of the CNB are roughly equivalent and comparable.