Learning from our mistakes in aging research

Abstract

Gerontologists were: (1)cautioned against transforming data when interactions are primary hypotheses; (2) advised to entertain several approaches to measuring change; (3) reminded that incorrect conclusions can result from cross-sequential analyses when the cohort variable and the time-of-measurement variable cover disparate time domains; and (4) appraised of potential sources of bias when independent samples of the same birth cohort are drawn at different times, and a way to avoid such biases was suggested.