Imputation methods for handling item‐nonresponse in practice: methodological issues and recent debates

Abstract

Nonresponse is a major problem often faced by social scientists when analysing survey data. A range exists to impute the missing responses but the choice between these methods may be difficult. This paper reviews advantages and disadvantages of a range of imputation methods and provides guidance on how to use such methods in practice. The paper introduces the reader new to the imputation literature to key ideas and methods. For those already familiar with imputation, the paper highlights some new developments and recent debates. The paper discusses an example from the social sciences, applying several imputation methods to a missing earnings variable. The objective is to illustrate in a real data example basic considerations when choosing between methods and to advise practitioners in the use of such methods.

Keywords:

• item‐nonresponse
• fractional imputation
• multiple imputation
• hot deck imputation