Privacy Concerns, Too Busy, or Just Not Interested: Using Doorstep Concerns to Predict Survey Nonresponse Nancy Bates 1 , James Dahlhamer 2 , and Eleanor Singer 3 Using newly available paradata, this article explores the use of “doorstep concerns” to predict interim and final refusals in the National Health Interview Survey (NHIS). Using ten weeks of automated contact history records, we analyze privacy and burden concerns but also examine other verbal and nonverbal interactions recorded by interviewers during contact with households. We conduct a multi-model multinomial logit analysis starting with a social environmental model (e.g., region, urbanicity), followed by the addition of process variables (e.g., number of noncontacts, mode of contact), and finally include the new household-level doorstep concerns (e.g., privacy concerns, too busy). The study found that the doorstep concerns greatly improved models predicting nonresponse relative to models including only environmental variables and basic contact history measures. Privacy concerns were significant in predicting interim refusals, but not final refusals. The effects of burden differed depending upon the particular doorstep concern used as an indicator of burden. Key words: Respondent burden; paradata; CAPI surveys; nonresponse; doorstep concerns; privacy. 1. Introduction In their study of nonresponse in household interview surveys, Groves and Couper develop what they call “a conceptual framework for survey cooperation” (Groves and Couper 1998:30). The framework consists of four blocks of variables, two within and two beyond the researcher’s control. Those within the researcher’s control include survey design variables, such as topic and mode of administration, and interviewer characteristics, including experience. The two beyond the researcher’s control include the social environment in which the survey takes place, including economic conditions, the survey- taking climate, and neighborhood characteristics; and characteristics of household(ers) q Statistics Sweden 1 U.S. Census Bureau, Demographic Surveys Division, Washington, DC 20233-9100, USA. Email: nancy.a.bates@census.gov 2 National Center for Health Statistics, 3311 Toledo Road, Hyattsville, MD 20782, U.S.A. Email: fzd2@cdc.gov 3 Survey Research Center, University of Michigan, P O Box 1248, Ann Arbor, MI 48106-1248, USA. Email: esinger@isr.umich.edu Acknowledgments: This research was made possible through the tireless efforts of approximately 400 Census Bureau Field Representatives who collected CHI data for the 2005 NHIS. The authors also thank Eric Lars Buckland, Charles Lin, and Norm Johnson for their assistance with the multinomial modeling. We also acknowledge Marcie Cynamon, Jane Gentleman, Cheryl Landman, Jennifer Madans, Jeffrey Moore, and Yves Thibaudeau for reviewing and improving earlier versions of the article. Disclaimer: This report is released to inform interested parties of ongoing research and to encourage discussion of work in progress. The views expressed are those of the authors and not necessarily those of the U.S. Census Bureau. Journal of Official Statistics, Vol. 24, No. 4, 2008, pp. 591–612