Co-Calibration of Cognitive Domains Across Different Countries Without Common Linking Items
Abstract Modern psychometric methods allow for cocalibration of cognition across cross-national surveys, given the presence of common tests across studies. For narrow cognitive domains, there may not be common tests due to cultural and linguistic differences in testing. We developed a novel method to facilitate cocalibration that entails (1) identifying a common score across studies highly correlated with the focal domain; (2) deriving scores separately in each study for the domain of interest; and (3) applying stratified equipercentile equating to equate domain scores in (2) to the distribution of the common metric in (1). We tested this method by equating executive functioning in the Harmonized Cognitive Assessment Protocols (HCAP) in the US (N=3496), India (N=4096), and England (N=1273). The method preserves the rank order of executive functioning derived separately (r>0.99 in England; r>0.99 in India), while preserving between-study differences observed in general cognitive functioning. We discuss limitations and future directions.