Psychological Assessment, Vol 38(5), May 2026, 374-383; doi:10.1037/pas0001452
Establishing measurement invariance across racial groups is essential to ensure that psychological assessments function equivalently across diverse populations. This study examined the measurement invariance of the Minnesota Multiphasic Personality Inventory–3 Higher Order and Restructured Clinical scales across Black American and White American university students, with a focus on scalar invariance. Measurement invariance testing proceeds in sequential steps—configural, metric, and scalar—to determine whether cross-group comparisons are psychometrically valid. Establishing scalar invariance is necessary to assess whether observed mean score differences reflect latent construct differences rather than measurement bias. The sample included 766 students from two U.S. universities (331 Black American, 435 White American). We evaluated unidimensionality using confirmatory factor analyses and bifactor exploratory structural equation modeling, then applied multigroup confirmatory factor analyses to test measurement invariance. Results supported configural invariance for all scales and full or partial metric invariance for all scales, indicating structural comparability. However, full scalar invariance was rare; only Hypomanic Activation (RC9) achieved it. Partial scalar invariance was established for several scales—Behavioral/Externalizing Dysfunction (BXD), Low Positive Emotions (RC2), Antisocial Behavior (RC4), Ideas of Persecution (RC6), Dysfunctional Negative Emotions (RC7), and Aberrant Experiences (RC8)—with minimal latent mean changes between the full and partial models. Emotional/Internalizing Dysfunction (EID), Thought Dysfunction (THD), Restructured Clinical scales include Demoralization (RCd) and Somatic Complaints (RC1) did not demonstrate scalar invariance, limiting the validity of mean comparisons. Overall, the Minnesota Multiphasic Personality Inventory–3 Higher Order and Restructured Clinical scales showed mixed evidence of measurement invariance—consistent support for configural and (full or partial) metric invariance, but limited support for full scalar invariance. Hence, mean score comparisons should be interpreted cautiously, particularly for scales lacking scalar invariance. (PsycInfo Database Record (c) 2026 APA, all rights reserved)