Effects of Demographic and Health Variables on Rasch Scaled Cognitive Scores

Objectives: To determine whether demographic and health variables interact to predict cognitive scores in Asset and Health Dynamics of the Oldest-Old (AHEAD), a representative survey of older Americans, as a test of the developmental discontinuity hypothesis. Methods: Rasch modeling procedures were used to rescale cognitive measures into interval scores, equating scales across measures, making it possible to compare predictor effects directly. Rasch scaling also reduces the likelihood of obtaining spurious interactions. Tasks included combined immediate and delayed recall, the Telephone Interview for Cognitive Status (TICS), Series 7, and an overall cognitive score. Results: Demographic variables most strongly predicted performance on all scores, with health variables having smaller effects. Age interacted with both demographic and health variables, but patterns of effects varied. Discussion:Demographic variables have strong effects on cognition. The developmental discontinuity hypothesis that health variables have stronger effects than demographic ones on cognition in older adults was not supported.