Homeless spell exits and returns: Substantive and methodological elaborations on recent studies

This article builds upon a series of recent studies that have examined exits from and returns to homelessness. The data come from a three-wave panel study of homeless persons. The article's major substantive concern is the relationship between the type of housing situations to which homeless persons move upon exiting homeless spells and their likelihood of becoming homeless again. The issue of selection bias due to sample attrition, a serious methodological problem common to longitudinal research, is also addressed.