Abstract:
|
Susanne Raessler was a wonderful friend, colleague, and collaborator. Our earliest collaborations with Susie occurred when she invited us to present workshops on missing data and multiple Imputation at conferences she organized in Nuremberg in 2003 and 2004. Since then, we have collaborated with her in many ways, such as participating in additional conferences she organized, working with her on applied and methodological research, co-authoring papers with her, and serving on doctoral committees for her students. In this presentation, we will discuss research, inspired by our collaborations with Susie, on three different topics. One is data fusion -- the subject of a book authored by Susie -- which is particularly relevant in this era of big data. Another is Sequential Hierarchical Regression Imputation, or "SHRIMP," which creates multiple imputations for clustered data by using chained hierarchical prediction models. The final topic is split-questionnaire designs, which allow surveys to cover a large number of items efficiently while limiting the burden on respondents.
|