Managing missing data: concepts, theories, and methods

Selecting an effective missing data method to provide acceptable results is essential in applied research. The following chapter provides readers with an introduction to some of the more traditional methods that are still used in the discipline (including case-wise deletion, averaging available items and single imputations such as mean imputation and regression imputation), as well as more modern methods including multiple imputation and maximum likelihood. Evaluations of missing data methods suggest that although the modern methods are not a panacea for handling missing data, they fare better under more circumstances compared to their traditional counterparts. Additionally, advances in both theory and computing power allows the development and implementation of new efficiency designs which enables missing data to be actively planned into research designs. Two prominent planned missing data designs are discussed: the three-form design and the two-method design.