The 6th workshop on the analysis of incomplete data (AID) held at the Leibniz Institute for Educational Trajectories (LIfBi) on July 19, 2017, illustrated the different strategies that a public data provider can adopt—namely, to either provide the tools for imputing within the analysis, or to impute first and then analyze.
In the first session, Ph.D. students from the LIfBi methods group and the Chair of Statistics and Econometrics at the University of Bamberg as well as researchers from the GfK, Nuremberg, presented their research. The session covered talks addressing current issues in multiple imputation—for example, the handling of missing values in compositional data and design of split-questionnaire surveys.
In the second session, recent work in the field of incomplete data analysis using computer-intensive simulation techniques were presented. The talks showed results from research of the DFG-funded project "Analyzing relations between latent competencies and context information in the NEPS" as part of the DFG Priority Programme 1646 "Education as a Lifelong Process" addressing the possibilities of handling missing values in confirmatory item factor analysis with up to ten latent dimensions and latent heterogeneity.
The last section enjoyed a lively discussion between the approximately 20 participants with Professor Trivellore Raghunathan, Ph.D. (Survey Research Center, University of Michigan, USA) about current challenges in the statistical analysis of data with missing values and the resulting requirements that data providers of large-scale survey data have to cope with.
The event was organized by Professor Dr. Susanne Rässler, Head of the Chair of Statistics and Econometrics at the University of Bamberg, and Dr. Christian Aßmann, Head of the LIfBi Department “Research Data Center, Method Development”.