How well does AIC perform in partially observed data?
*Ashok Chaurasia, Department of Statistics, University of Connecticut 
Ofer Harel, Department of Statistics, University of Connecticut 

Keywords: Model selection, incomplete data, multiple imputation, AIC.

      Many model selection criterions (e.g. AIC, BIC) proposed over the years have become common procedures in applied research. However, these procedures where designed for complete data. Complete data is rare in applied statistics, in particular in medical, public health and health policy settings. Incomplete data, another common problem is applied statistics, introduces its own set of complications in light of which the task of model selection can get quite complicated.
      Recently, few have suggested model selection procedures for incomplete data with varying degrees of success. In this poster we explore model selection in the multivariate regression setting in the presence of ignorable missing data via multiple imputation. We explore this issue via simulation to illustrate the performance of various statistics under different data configurations.