Respondent Cooperation and Requests for Contacts in Longitudinal Research

Colter Mitchell, University of Michigan
Vaughn R.A. Call, Brigham Young University

Researchers contemplating panel research designs regularly face the problem of panel attrition. A vital tool in reducing panel attrition is first wave information, which can be used to locate respondents who move. We analyze data from the 1988 National Survey of Families and Households (N = 13,007) to examine respondent cooperation to requests made during the initial interview for names and addresses of friends and relatives who might help researchers locate respondents who move. About 90 percent of respondents provided at least one contact. Compared to respondents who provide three contacts, respondents who provide one or no contacts have 26% or 60%, respectively, lower odds of being found the next year. Utilizing censored Poisson and sequential logistic regression we find that respondents with greater interest in the survey, and larger family and social networks, supply more contacts. Respondents from racial minority groups and respondents in the East provide fewer contacts.