What Is Meant by "Missing at Random"?

Abstract

The concept of missing at random is central in the literature on statistical analysis with missing data. In general, inference using incomplete data should be based not only on observed data values but should also take account of the pattern of missing values. However, it is often said that if data are missing at random, valid inference using likelihood approaches (including Bayesian) can be obtained ignoring the missingness mechanism. Unfortunately, the term "missing at random" has been used inconsistently and not always clearly; there has also been a lack of clarity around the meaning of "valid inference using likelihood". These issues have created potential for confusion about the exact conditions under which the missingness mechanism can be ignored, and perhaps fed confusion around the meaning of "analysis ignoring the missingness mechanism". Here we provide standardised precise definitions of "missing at random" and "missing completely at random", in order to promote unification of the theory. Using these definitions we clarify the conditions that suffice for "valid inference" to be obtained under a variety of inferential paradigms.

Published in at http://dx.doi.org/10.1214/13-STS415 the Statistical Science (http://www.imstat.org/sts/) by the Institute of Mathematical Statistics (http://www.imstat.org)

References in corpus (1)

• Ignorability in Statistical and Probabilistic Inference

Cited by in corpus (26)

• On Inverse Probability Weighting for Nonmonotone Missing at Random Data
• On the consistency of supervised learning with missing values
• A matrix-based method of moments for fitting multivariate network meta-analysis models with multiple outcomes and random inconsistency effects
• Multiple imputation in Cox regression when there are time-varying effects of exposures
• Missing Data Imputation using Optimal Transport
• Study design in causal models
• BEST : A decision tree algorithm that handles missing values
• not-MIWAE: Deep Generative Modelling with Missing not at Random Data
• Diagnosing missing always at random in multivariate data
• Recoverability of Causal Effects under Presence of Missing Data: a Longitudinal Case Study
• Adaptive Optimization for Prediction with Missing Data
• Causal Inference: A Missing Data Perspective
• Gaussian Process Modelling for Improved Resolution in Faraday Depth Reconstruction
• Logistic Regression with Missing Covariates -- Parameter Estimation, Model Selection and Prediction within a Joint-Modeling Framework
• Sharing pattern submodels for prediction with missing values
• Assessing treatment effects in observational data with missing confounders: A comparative study of practical doubly-robust and traditional missing data methods
• On missing label patterns in semi-supervised learning
• Sequentially additive nonignorable missing data modeling using auxiliary marginal information
• Itemwise conditionally independent nonresponse modeling for incomplete multivariate data
• When is $Y_{obs}$ missing and $Y_{mis}$ observed?
• Clustering data with values missing at random using scale mixtures of multivariate skew-normal distributions
• A fresh look at ignorability for likelihood inference
• On the Fairness of Randomized Trials for Recommendation with Heterogeneous Demographics and Beyond
• Multiple imputation and selection of ordinal level 2 predictors in multilevel models. An analysis of the relationship between student ratings and teacher beliefs and practices
• Three issues impeding communication of statistical methodology for incomplete data
• Regression and Causality