Response Rates and Response Content in Mail Versus Face-To-Face Surveys

Abstract

Two surveys were administered based on the same area probability sampling frame and with some of the same questions: one sample was used for hour-long face-to-face interviewing in the 1992 Detroit Area Study; the other sample received a much shorter questionnaire in the mail for sell-administration. The sample segments had previously been stratified in terms of the percentage that was black. For the predominantly white stratum, there was only a small difference in response rates due to mode of administration. For the predominantly black stratum, the mail survey obtained a considerably lower response rate then the face-to-face survey. Within the predominantly white stratum, there were no clear differences between results for the two modes of administration in demographic variables or in gross housing characteristics. However, the mail survey respondents expressed more negative attitudes toward racial integration and affirmative action than did the face-to-face respondents. Because the mail sample of the predominantly black stratum was small, it was not possible to carry out similar analyses of demographic or attitudinal differences, or to determine whether its lower response rate was due mainly to race, to correlates of race such as income or education, or even to problems with mail delivery in central cities.