Nonresponse: What it is and what we do about it

Article type: Research Article

Authors: Bailar, Barbara A.¹

Note: [1] Barbara A. Bailar is the Associate Director for Statistical Standards and Methodology at the U.S. Bureau of the Census. Almost all of her professional work has centered on the problems of nonsampling errors in sample surveys in censuses, including work on effects of interviewers, recall periods, nonresponse, and undercount in a census. She holds a Ph. D. from the American University in Washington, D.C., in mathematical statistics.

Abstract: Nonresponse is a perennial problem in sample surveys. Differentiation between unit nonresponse and item nonresponse is important because each calls for different strategies. Longitudinal surveys may bring forth the need for new strategies to cope with both unit and item nonresponse. Nonresponse strategies also vary depending on whether the survey is an establishment surveyor a household survey, and whether estimated distributions are to be released publicly. Though statistical solutions to coping with nonresponse are important and help in making data more useful, attention must also be given to motivation and training to improve response rates.

DOI: 10.3233/SJU-1984-2405

Journal: Statistical Journal of the United Nations Economic Commission for Europe, vol. 2, no. 4, pp. 381-392, 1984