Selective attrition effects in bereavement research: A three-year longitudinal analysis

The study examined the effects of selective attrition on the picture of bereavement adjustment painted by longitudinal studies. Originally, 185 conjugally and nonconjugally bereaved adults (M age = 50, range 20-82) were administered a battery of self-report measures indexing psychological functioning and adjustment to loss. Complete data gathered six months later were available from 166 persons (133 completers, 33 dropouts), while for a follow-up conducted three years later, complete data were available from ninety-six completers and seventy dropouts. Based on the women in this sample, discriminant function analysis conducted at the six-month follow-up suggested that nonconjugally bereaved completers had been older, had been more recently bereaved, had experienced more loneliness, had reported their health relative in age peers to be less adequate, and had been more active copers. At three years, nonconjugally bereaved completers had been older, more affluent, and had been more lonely. No differences between refusers and completers were found for the conjugally bereaved at either the six-month or three-month occasions. Regarding attrition effects, these data suggest that distinctions between short-term and long-term time frames and between conjugal and nonconjugal bereavement are necessary. These data also indicate that longitudinal studies of nonconjugal bereavement may be presenting an inaccurate picture of post-death adjustment, while attrition bias in studies or conjugal bereavement may be less distorting in its impact.