Estimation in Surveys with Nonresponse

There is increasing interest in the study of nonresponse in statistical surveys, as there is a need to reduce the level of nonresponse in order to increase the validity of the results, and to deal with the problem of nonresponse once it has arisen. It is possible to reduce the amount of nonresponse in a survey, but not possible to avoid it completely. Therefore there is a need for techniques that enable the survey statistician to estimate the level of nonresponse, and adjust for it accordingly in order to improve the results. This book presents a practical overview of techniques for estimating nonresponse in statistical surveys, and for adjusting for it once it has occurred. It also covers the related topics of frame imperfections and coverage errors in surveys, featuring many worked examples and implementation of the methods in a number of statistical software packages. The text is based upon a manual developed at Statistics Sweden, and used extensively by government statisticians working with data from real surveys. Features: Provides an overview of methods for the estimation of nonresponse in statistical surveys. Covers a range of methods for nonresponse adjustment, particularly reweighting and imputation. Includes guidelines for dealing with frame imperfections and coverage errors. Features worked examples throughout using real data. Includes implementation of the methods in a range of statistical software It will appeal to researchers and practitioners analyzing data from statistical surveys. Also suitable as supplementary reading for graduate students of survey statistics.