Abstract
OBJECTIVES: The Harmonized Cognitive Assessment Protocol (HCAP) is a neuropsychological assessment for dementia that is used to derive national dementia prevalence estimates through a substudy of the Health and Retirement Study. We aimed to evaluate the degree of measurement invariance of the HCAP across Hispanic/Latino-a-e-x (H/L) and non-H/L white (NHW) older adults in the United Study. METHOD: We combined HCAP data from two large cohort studies. We used multiple-group confirmatory factor analysis to evaluate measurement invariance across H/L adults assessed in English (n = 259), H/L adults assessed in Spanish (n = 236), and NHW adults assessed in English (n = 2,462). RESULTS: We observed measurement differences in the HCAP across H/L and NHW older adults for the memory, language, executive functioning, and orientation, but not visuospatial, domains. Effect sizes for measurement differences in HCAP items ranged from small to large. Not accounting for measurement differences led to meaningful underestimation of cognitive function ranging from 5% (language domain) to 82% (executive functioning domain) of H/L individuals assessed in English and 11% (memory domain) to 96% (executive functioning domain) of H/L individuals assessed in Spanish. CONCLUSIONS: The interpretation of ethnic differences in cognitive function with the HCAP may be confounded with measurement differences. Measurement equivalence in cognitive assessment instruments across H/L and NHW populations cannot be assumed. (PsycInfo Database Record (c) 2026 APA, all rights reserved).