CROSS-NATIONAL HARMONIZATION OF COGNITIVE FUNCTION: THE HRS AND INTERNATIONAL SISTER STUDIES

摘要

To make appropriate cross-national comparisons of cognitive function between the the HRS and international sister studies, we evaluated conceptual and statistical equivalence leveraging item response theory (IRT) methods. We obtained granular, item-level data on cognitive function from interviews conducted in 10 international studies (N=120,099). We assessed configural invariance and dimensionality, and co-calibrated cognitive performance across studies and time. Unidimensional IRT models fit well to cognitive data in each study (all RMSEA<0.07; all CFI>0.91). To co-calibrate across study and time, we selected as a reference sample the HRS CODA sample (N=2,320) and scaled cognitive functioning for HRS waves and other studies to that cohort. We describe distributions of cognitive scores across country. Without careful consideration of the conceptual and statistical equivalence of different assessments, data derived from these studies are incomparable. Effective public health planning and informative research requires harmonized cognitive assessment data from cross-national studies.