Abstract
This study examines attrition from a large probability sample of older adults over five waves of data collection, each 6 months apart. Respondents to all five waves and groups differing in the wave or reason they dropped out were compared on initial psychological, health, social, demographic and life event measures. For men, in general, the better their initial conditions, the greater the number of interviews. For women, however, differences among groups participating for different lengths of time were nonlinear. For both men and women, these differences were accounted for by changing sources (types) of attrition across waves, plus strong differences among attrition types. In general, respondents to all five waves initially had better conditions than persons who died, became ill, or were not found at a later wave, but had equal or worse conditions than persons who lost interest in the study.

This publication has 1 reference indexed in Scilit: