Multiple Imputation in the Presence of High-Dimensional Data

• Activity Number: 378
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 11:35 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Domonique Watson Hodge* and Qi Long
• Companies: Emory University and Emory University
• Keywords: sparse principal component analysis ; missing data ; multiple imputation ; screening ; high dimensional data ; dimension reduction regression

Abstract

Missing data present challenges in the statistical analysis phase of research. Common naive analyses such as complete-case and available-case analysis may introduce bias, loss of efficiency, and produce unreliable results. Multiple imputation is one of the most widely used method for handling missing data which can be attributed to its ease of use. However, more research needs to be conducted to determine the best strategy to conduct multiple imputation (MI) in the presence of high-dimensional data. To address this concern, we evaluate several approaches for MI based on sure independent screening (SIS), sparse principal component analysis (sPCA), and dimension reduction regression (DRR). The performance of these methods is assessed through numerical studies.