Factor analyzing the Norwegian MATRICS Consensus Cognitive Battery

• Published in: Psychiatry and clinical neurosciences 2017-05, Vol.71 (5), p.336-345
• Main Authors: Mohn, Christine, Lystad, June U., Ueland, Torill, Falkum, Erik, Rund, Bjørn R.
• Format: Article
• Language: English
• Subjects: Adolescent; Adult; Aged; Case-Control Studies; Child; Cognition; Cognitive ability; Cognitive Dysfunction - complications; Cognitive Dysfunction - diagnosis; Discriminant analysis; Factor analysis; Factor Analysis, Statistical; Female; Humans; Male; MATRICS Consensus Cognitive Battery; Mental disorders; Middle Aged; Models, Psychological; Neural coding; Neuropsychological Tests; neuropsychology; Norway; Psychometrics; Schizophrenia; Schizophrenia - complications; Schizophrenic Psychology; Short term memory; Visual discrimination learning; Young Adult
• ISSN: 1323-1316; 1440-1819
• DOI: 10.1111/pcn.12513

Aim The MATRICS Consensus Cognitive Battery (MCCB) assesses seven cognitive domains with 10 subtests. This domain structure has not been demonstrated. Three factors have been produced in US samples. We examined the dimensional structure of the Norwegian MCCB. In addition, we studied the contribution of each subtest to the battery sum score. Methods The participants were 131 patients with schizophrenia spectrum disorders and 300 healthy controls. Their Norwegian MCCB test scores were subject to exploratory and confirmatory factor analysis and regression analysis. Results The theoretical MCCB factor structure was not shown. In the patient group, three‐factor and two‐factor models had acceptable fit. In both groups, the Symbol Coding, Spatial Span, Letter–Number Span, and Visual Learning subtests contributed most to the sum score. Conclusion The theoretical domain structure of the MCCB could not be demonstrated in these Norwegian participants. Consonant with US studies, models with three and two factors had mediocre fit, and in the schizophrenia spectrum disorder group only. In both groups, the subtests Symbol Coding, Working Memory, and Learning were the most sensitive in tapping general neurocognitive performance, supporting US results. We conclude that in both Norway and the USA, the MCCB generates the same cognitive domains through factor analysis, but that these domains are not the ones suggested by the MATRICS project.