Attrition in a longitudinal study of an aged population

This report focuses upon attrition in longitudinal studies of older persons using a Florida survey as an illustration. In 1959 interviews were completed with a probability sample (N = 2544) of older noninstutionalized residents (age 65+) of a retirement community. Data were collected on health status, health practices, social factors, and activities. Nine years later a follow-up study of the same sample was undertaken. In this report 1959 data for 2497 of these persons are analyzed by results of the 1968 effort (reinterviewed, lost to follow-up, or deceased). About one third of this highly migratory sample could not be located in 1968; death certificates were located for another one third; and the remaining one third was reinterviewed. Social characteristics, housing arrangements, activities, and chronic conditions of the three subpopulations are compared. Persons reinterviewed were found to differ from dropouts in age, sex, type of dwelling unit, activity patterns, and prevalence of chronic conditions. No significant differences were found in educational level, household composition, usual occupation, and retirement status (when controlled for sex).