RESUMO
The authors implemented strategies to maximize cohort retention to avert loss of statistical power and minimize bias in a longitudinal evaluation of a middle school HIV/STD prevention intervention. A retention rate of 80% of the baseline sample (n = 2,975) at six months and 73% at 18 months was achieved despite high reported rates of student mobility and a major system reorganization in one urban district. The strategies increased retention but did not eliminate differences in demographic characteristics and behaviors between the groups of retained and lost students. Results confirm the need to implement retention strategies early and to maintain them throughout data collection. Information from a tracking data base can be used to prioritize students for follow-up to reduce bias from sample loss.
