The interpretation of shared age-related variance among factors in cross-sectional cognitive aging studies

Abstract

Many cross-sectional correlational studies in cognitive aging have focused on explaining age-related variance. It has been assumed that variables sharing variance with both cognition and age may be the key explanatory variables underlying the cognitive decline in normal aging. Statistical biases intrinsic to this approach have been described by Hofer and Sliwinski and a narrow age cohort design proposed. The present paper aims at explaining how Hofer and Sliwinski's criticisms apply to a specific type of research design in cognitive aging where the goal is to identify underlying aging processes, but does not apply to more general gerontological research. Methods to estimate bias in cross-sectional studies are required as is greater awareness of this potential bias.