Multiple imputation with diagnostics (mi) in R

Opening windows into the black box

Yu Sung Su, Andrew Gelman, Jennifer Hill, Masanao Yajima

Research output: Contribution to journalArticle

Abstract

Our mi package in R has several features that allow the user to get inside the imputation process and evaluate the reasonableness of the resulting models and imputations. These features include: choice of predictors, models, and transformations for chained imputation models; standard and binned residual plots for checking the fit of the conditional distributions used for imputation; and plots for comparing the distributions of observed and imputed data. In addition, we use Bayesian models and weakly informative prior distributions to construct more stable estimates of imputation models. Our goal is to have a demonstration package that (a) avoids many of the practical problems that arise with existing multivariate imputation programs, and (b) demonstrates state-of-the-art diagnostics that can be applied more generally and can be incorporated into the software of others.

Original languageEnglish (US)
Pages (from-to)1-31
Number of pages31
JournalJournal of Statistical Software
Volume45
Issue number2
StatePublished - Dec 2011

Fingerprint

Multiple Imputation
Imputation
Black Box
Diagnostics
Residual Plots
Bayesian Model
Conditional Distribution
Prior distribution
Demonstrations
Model
Multiple imputation
Black box
Predictors
Software
Evaluate
Estimate
Demonstrate

Keywords

  • Chained equations
  • Mi
  • Model diagnostics
  • Multiple imputation
  • R
  • Weakly informative prior

ASJC Scopus subject areas

  • Software
  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

Multiple imputation with diagnostics (mi) in R : Opening windows into the black box. / Su, Yu Sung; Gelman, Andrew; Hill, Jennifer; Yajima, Masanao.

In: Journal of Statistical Software, Vol. 45, No. 2, 12.2011, p. 1-31.

Research output: Contribution to journalArticle

Su, Yu Sung ; Gelman, Andrew ; Hill, Jennifer ; Yajima, Masanao. / Multiple imputation with diagnostics (mi) in R : Opening windows into the black box. In: Journal of Statistical Software. 2011 ; Vol. 45, No. 2. pp. 1-31.
@article{ed00c16917c54051811abfa2f1934658,
title = "Multiple imputation with diagnostics (mi) in R: Opening windows into the black box",
abstract = "Our mi package in R has several features that allow the user to get inside the imputation process and evaluate the reasonableness of the resulting models and imputations. These features include: choice of predictors, models, and transformations for chained imputation models; standard and binned residual plots for checking the fit of the conditional distributions used for imputation; and plots for comparing the distributions of observed and imputed data. In addition, we use Bayesian models and weakly informative prior distributions to construct more stable estimates of imputation models. Our goal is to have a demonstration package that (a) avoids many of the practical problems that arise with existing multivariate imputation programs, and (b) demonstrates state-of-the-art diagnostics that can be applied more generally and can be incorporated into the software of others.",
keywords = "Chained equations, Mi, Model diagnostics, Multiple imputation, R, Weakly informative prior",
author = "Su, {Yu Sung} and Andrew Gelman and Jennifer Hill and Masanao Yajima",
year = "2011",
month = "12",
language = "English (US)",
volume = "45",
pages = "1--31",
journal = "Journal of Statistical Software",
issn = "1548-7660",
publisher = "University of California at Los Angeles",
number = "2",

}

TY - JOUR

T1 - Multiple imputation with diagnostics (mi) in R

T2 - Opening windows into the black box

AU - Su, Yu Sung

AU - Gelman, Andrew

AU - Hill, Jennifer

AU - Yajima, Masanao

PY - 2011/12

Y1 - 2011/12

N2 - Our mi package in R has several features that allow the user to get inside the imputation process and evaluate the reasonableness of the resulting models and imputations. These features include: choice of predictors, models, and transformations for chained imputation models; standard and binned residual plots for checking the fit of the conditional distributions used for imputation; and plots for comparing the distributions of observed and imputed data. In addition, we use Bayesian models and weakly informative prior distributions to construct more stable estimates of imputation models. Our goal is to have a demonstration package that (a) avoids many of the practical problems that arise with existing multivariate imputation programs, and (b) demonstrates state-of-the-art diagnostics that can be applied more generally and can be incorporated into the software of others.

AB - Our mi package in R has several features that allow the user to get inside the imputation process and evaluate the reasonableness of the resulting models and imputations. These features include: choice of predictors, models, and transformations for chained imputation models; standard and binned residual plots for checking the fit of the conditional distributions used for imputation; and plots for comparing the distributions of observed and imputed data. In addition, we use Bayesian models and weakly informative prior distributions to construct more stable estimates of imputation models. Our goal is to have a demonstration package that (a) avoids many of the practical problems that arise with existing multivariate imputation programs, and (b) demonstrates state-of-the-art diagnostics that can be applied more generally and can be incorporated into the software of others.

KW - Chained equations

KW - Mi

KW - Model diagnostics

KW - Multiple imputation

KW - R

KW - Weakly informative prior

UR - http://www.scopus.com/inward/record.url?scp=84863414529&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84863414529&partnerID=8YFLogxK

M3 - Article

VL - 45

SP - 1

EP - 31

JO - Journal of Statistical Software

JF - Journal of Statistical Software

SN - 1548-7660

IS - 2

ER -