Multiple imputation of missing values in the wave 2007 of the IAB Establishment Panel

The basic concept of multiple imputation is straightforward and easy to understand, but the application to real data imposes many implementation problems. To define useful imputation models for a dataset that consists of categorical and of continuous variables with distributions that are anything but normal, contains skip patterns and all sorts of logical constraints is a challenging task. In this paper, we review different approaches to handle these problems and illustrate their successful implementation for a complex imputation project at the German Institute for Employment Research (IAB): The imputation of missing values in one wave of the IAB Establishment Panel.