Date: December 2005
Creator: Rhodes, Anthony Ryan
Description: Longitudinal methods have become an improved and essential means of measuring intra-individual change over time. Yet one of the greatest and most hazardous drawbacks studying participants over multiple sessions can be the loss of participants over time. This study attempts to illuminate the problem of attrition in longitudinal research by estimating the mean effect sizes for participant loss across 57 studies published in 13 prestigious journals which regularly use older participants. Results estimate overall attrition to be around 34% of the original sample. The subsequent break down of attrition into its subtypes yield mean effect sizes for attrition due to Refusal (8%), Loss of contact (10%), Illness (6%), and Death (14%) in studies sampling from adults 50 years or older. Analyses were then conducted via meta-analytic one-way ANOVA and weighted regression to identify possible moderators of overall attrition and their four subtypes.
Contributing Partner: UNT Libraries