A Cultural Neuropsychological Approach to Harmonization of Cognitive Data Across Culturally and Linguistically Diverse Older Adult Populations

Objective: To describe a cultural neuropsychological approach to prestatistical harmonization of cognitive data across the United States (U.S.) and Mexico with the Harmonized Cognitive Assessment Protocol (HCAP). Method: We performed a comprehensive review of the administration, scoring, and coding procedures for each cognitive test item administered across the English and Spanish versions of the HCAP in the Health and Retirement Study (HRS) in the U.S. and the Ancillary Study on Cognitive Aging in Mexico (Mex-Cog). For items that were potentially equivalent across studies, we compared each cognitive test item for linguistic and cultural equivalence and classified items as confident or tentative linking items, based on the degree of confidence in their comparability across cohorts and language groups. We evaluated these classifications using differential item functioning techniques. Results: We evaluated 132 test items among 21 cognitive instruments in the HCAP across the HRS and Mex-Cog. We identified 72 confident linking items, 46 tentative linking items, and 14 items that were not comparable across cohorts. Measurement invariance analysis revealed that 64% of the confident linking items and 83% of the tentative linking items showed statistical evidence of measurement differences across cohorts. Conclusions: Prestatistical harmonization of cognitive data, performed by a multidisciplinary and multilingual team including cultural neuropsychologists, can identify differences in cognitive construct measurement across languages and cultures that may not be identified by statistical procedures alone.

Automated summary

This study presents a cultural neuropsychological approach to prestatistical harmonization of cognitive data across the United States and Mexico. By comparing linguistic and cultural equivalence, the researchers identified confident and tentative linking items. Results showed that a significant proportion of the linking items exhibited measurement differences across cohorts. The study highlights the importance of multidisciplinary and multilingual teams in identifying differences in cognitive construct measurement that may not be detected by statistical procedures alone.