Weighting for Period Perspective in Samples of the Homeless
Published in: American Behavioral Scientist, v. 45, no. 1, Sep. 2001, p. 80-104
Posted on RAND.org on December 31, 2000
For some sample-based studies of the homeless, it may be relevant to adopt a period perspective, in which analyses focus on the mix of persons who experience homelessness at least once during some period of time, not just the mix of persons at a point in time. Period perspective presumably can be enhanced by weights that account for the intermittent nature of homelessness, but such weighting invites implementation problems related to calculation, specifying the weighting model, measurement, and especially weight disparity. Some of these problems can be mitigated by a sample design that assures independence between selection and individual respondent behavior, but remaining difficulties, particularly weight variability, raise questions about how weighting should be used in this context.