Bias in a Directory Sample for a Mail Survey of Rural Households

Abstract

This study examines bias in a mail survey of rural households that used telephone directories as the sampling frame and compares it with a parallel survey that used an independent sample selected from field listings. A higher response rate was obtained with the directory sample, which was primarily attributed to poorer address information for the listed sample. Although coverage of the study population was higher for the listed frame than for the directory frame, no significant differences were found for eight respondent and seven household characteristics. Also, no significant differences were found for seven key content variables regarding health behaviors and attitudes. The findings support the use of telephone directories as the sampling frame for general population mail surveys of rural areas where such coverage of the population is considered adequate. Frederick J. Kviz is an Associate Professor in the Epidemiology-Biometry Program, School of Public Health, and Research Associate Professor in the Survey Research Laboratory at the University of Illinois at Chicago. This research was supported by Grant RO 1 HS 02778 from the National Center for Health Services Research. Sampling, data collection, and data processing were conducted by the Survey Research Laboratory at the University of Illinois. The author is grateful to Johnny Blair, Linda B. Lannom, and Jutta P. Sebestik for assistance in tabulating some of the results. Public Opinion Quarterly Vol. 48:801-806 ? by the Trustees of Columbia University Published by Elsevier Science Publishing Co., Inc. 0033-362X/84/0048-801/$2.50 This content downloaded from 157.55.39.243 on Thu, 06 Oct 2016 04:53:25 UTC All use subject to http://about.jstor.org/terms 802 FREDERICK J. KVIZ However, an evaluation of bias in mail surveys using directory samples seems never to have been reported. It would be useful to examine the size and nature of bias in such surveys and to consider whether that bias might be substantially reduced. To that end, this paper examines bias in a general population mail survey of rural households using telephone directories as the sampling frame in comparison with a mail survey using an independent sample selected from field listings of the same population. The listed sampling frame was expected to provide more complete coverage of the study population because it would avoid exclusion bias of directories due to nonsubscribers to the telephone service and unpublished telephone numbers.