Sample surveys provide the fundamental basis for research in a wide range of social science disciplines. Nonresponse is a common problem in sample surveys, especially those involving surveys of human populations. Failure to account for nonresponse may seriously undermine the validity of survey results, particularly if the unobserved characteristics of the nonrespondents differ systematically from those of respondents. The last decade has seen the development of several new statistical techniques designed to handle the problem of nonresponse. However, further methodological and applied work is necessary before these techniques can be readily applied in practice. The aim of this project is to develop statistical methods for analyzing survey data with missing and/or incomplete responses, and apply the methodology to a large survey designed to study the costs associated with medical malpractice in New York State.
