December 2011, Volume 45, Issue 3. | Stef van Buuren, Karin Groothuis-Oudshoorn
The article introduces the R package `mice 2.9`, which is designed for multiple imputation of incomplete multivariate data using chained equations. The package extends the functionality of the previous version, `mice 1.0`, by adding new features such as automatic predictor selection, data handling, post-processing imputed values, specialized pooling routines, model selection tools, and diagnostic graphs. The article provides a step-by-step guide on how to use the package to address missing data problems, including creating imputations, analyzing imputed data, and pooling results. It also discusses the modular framework of multiple imputation, the MICE algorithm, and various imputation models. The article emphasizes the importance of specifying the imputation model for each variable and provides guidance on selecting predictors, handling multilevel data, and dealing with perfect prediction. The package is demonstrated using a real dataset, and the article includes code examples to illustrate the implementation of the methods.The article introduces the R package `mice 2.9`, which is designed for multiple imputation of incomplete multivariate data using chained equations. The package extends the functionality of the previous version, `mice 1.0`, by adding new features such as automatic predictor selection, data handling, post-processing imputed values, specialized pooling routines, model selection tools, and diagnostic graphs. The article provides a step-by-step guide on how to use the package to address missing data problems, including creating imputations, analyzing imputed data, and pooling results. It also discusses the modular framework of multiple imputation, the MICE algorithm, and various imputation models. The article emphasizes the importance of specifying the imputation model for each variable and provides guidance on selecting predictors, handling multilevel data, and dealing with perfect prediction. The package is demonstrated using a real dataset, and the article includes code examples to illustrate the implementation of the methods.