Using three different studies, the paper demonstrates that by merging a large scale national data base with survey respondent and non-respondent name and address files, characteristics which distinguish these two groups can be identified. The respondent sample can then be weighted to adjust for some of the bias of non-response.
Symposium: 1999: Florence, Session 4 - The Response Rate Challenge
Authors: Appel, Valentine, Cable, Virginia, Jennings, Dan
Organisations: Consultant, The Wall Street Journal
Topics: Analysis Issues, Response Rates