Multiply Robust Imputation Procedures for the Treatment of Item Nonresponse in Surveys

• Activity Number: 284
• Type: Invited
• Date/Time: Tuesday, August 11, 2015 : 8:30 AM to 10:20 AM
• Sponsor: IMS
• Author(s): David Haziza* and Sixia Chen
• Companies: University of Montreal and Westat
• Keywords: Nonresponse ; Imputation ; Survey data ; Variance estimation ; Multiply robust procedures

Abstract

Item nonresponse in surveys is often treated through some form of imputation. We introduce the concept of multiply robust imputation procedures in the context of finite population sampling, which is closely related to the concept of multiple robustness proposed by Han and Wang (2013). Multiple robustness can be viewed as an extension of the concept of double robustness. In practice, multiple nonresponse models and multiple imputation models may be fitted, each involving different subsets of covariates and possibly different link functions. An imputation procedure is said to be multiply robust if the resulting estimator is consistent if all but one model are misspecified. A jackknife variance estimator is proposed and shown to be consistent provided that the sampling fraction is negligible. Extension to random and fractional imputations are discussed. Finally, we present the results of a simulation study assessing the performance of the proposed point and variance estimators.