Limiting attrition in longitudinal research on homeless adolescents: what works best?

The effectiveness of five tracking strategies (collateral contacts, Internet databases, driver's records, letters, and community visits) used in attempting to locate homeless and housed adolescents 4.5 years after they were first contacted was examined and compared. The study sample comprised 401 adolescents (252 homeless and 149 matched housed) from 8 counties in Southeast Michigan. Homelessness was defined as sleeping in a shelter, at a friend's home without parental permission, or on the streets for at least 1 night within 1 month before initial contact. We successfully located and interviewed 85% ( n =340) of our sample at the 4.5‐year time point. The most effective tracking procedure was collateral contact, which was important in locating 43% of participants in this study, followed by Internet databases (18%), community visits (11%), driver's records (6%), and letters (6%).