Multiple Imputation Using the Fully Conditional ... - SAS

Paper 2081-2015. Multiple Imputation Using the Fully Conditional Specification Method: A Comparison of SAS , Stata, IVEware, and R. Patricia A. Berglund, University of Michigan-Institute for Social Research ABSTRACT. This presentation emphasizes use of SAS to perform Multiple Imputation of missing data Using the PROC MI Fully Conditional Specification (FCS) method with subsequent analysis Using PROC. SURVEYLOGISTIC and PROC MIANALYZE. The data set used is based on a complex sample design. Therefore, the examples correctly incorporate the complex sample features and weights. The demonstration is then repeated in Stata, IVEware, and R for a comparison of major software applications that are capable of Multiple Imputation Using FCS or equivalent methods and subsequent analysis of imputed data sets based on a complex sample design. INTRODUCTION. Paper 2081-2015 presents a detailed example of Multiple Imputation of missing data from a complex sample design with the Fully Conditional Specification (FCS) method of PROC MI with subsequent analysis Using PROC SURVEYLOGISTIC and PROC MIANALYZE.

The FCS method is also labeled the sequential regression algorithm (Raghunathan, et al. , 2001) in IVEware or the “chained equations” approach (van Buuren et al., 1999; Royston, 2005; Carlin, et al., 2008) in Stata and R. Broadly described, each of these algorithms is based on an iterative algorithm. Each iteration (t=1,…,T)

Loading..

Tags:

  Methods, Iteration, Imputation

Information

Related search queries