A Modified EM Algorithm for Regression Analysis of Data with Nonignorable Nonresponse

• Activity Number: 192
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Yang Zhang* and Gong Tang
• Companies: Amgen and University of Pittsburgh
• Keywords: Missing-data analysis ; Non-ignorable nonresponse ; EM algorithm

Abstract

The often unknown mechanism for the missing data process may be associated with the underlying values. Standard statistical methods, including likelihood-based methods and weighted estimating equations, require a model for the missing-data mechanism and incorporate it in the estimation and inference. Misspecification of the missing-data model often causes biased estimates and wrongful conclusions. EM algorithm is an iterative algorithm that is often used to find MLE for the likelihood-based methods. In E-steps, given a current estimate and a model for the missing-data mechanism, the conditional expectations of the sufficient statistics are calculated. Under the premise that the current estimate is consistent, we found that those conditional expectations could be approximated from the empirical data without the need for modeling the missing-data mechanism. Subsequently we proposed a modified EM algorithm regardless of the potential missing-data mechanism. Our simulation studies showed that the parameter estimates had negligible bias and were more efficient than the initial values obtained from external data. The increment of log-likelihood function and Q function were monitored.