Bias and efficiency of multiple imputation compared with complete‐case analysis for missing covariate values

By Ian R. White, J. Carlin. Statistics in Medicine, 2010

740 citations recorded by the source corpus.

Subfields: Management Science and Operations Research, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/36df255f7fc0490ed4195788c0cd322cb4a2f95c

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Management Science and Operations Research

1. K. Tilling
2. J. Heron
3. Rachael A. Hughes
4. P. Madley-Dowd
5. J. Sterne
6. Y. Liu
7. Ian R. White
8. A. Copas

In Statistics and Probability

1. Ian R. White
2. K. Tilling
3. J. Heron
4. Rachael A. Hughes
5. S. Seaman
6. P. Madley-Dowd
7. J. Sterne
8. A. Copas

Citations in the corpus

• Accounting for missing data in statistical analyses: multiple imputation is not always the answer (2019, methodology): “are many covariates each with small amounts of missing data, and individuals tend to have missing values on different variables, so that there is a large proportion of incomplete cases.(25) In this situation, MI can lead to substantial”
• Combining Multiple Imputation and Inverse-Probability Weighting (2012, background): “If weights W are all equal, and X = Z and the imputation and analysis models are the same, there is no benefit to IPW/MI over IPW/CC: it is more efficient to exclude individuals with missing Y (unless M = ∞, in which case exclusion and imputation are equivalent) (White and Carlin, 2010).”
• SAITS: Self-Attention-based Imputation for Time Series (2022, background): “Deletion introduces bias, while correctly specified imputation estimates are unbiased [14]; (2).”
• Using simulation studies to evaluate statistical methods (2017, background): “For empirical SE, relative % increase in precision, and relative error, the Monte Carlo SE formulas assume normally distributed ̂; for non-normal ̂, robust SEs exist – see (42).”
• The proportion of missing data should not be used to guide decisions on multiple imputation (2019, methodology): “Where data were simulated as MCAR, both MI and CCA are valid models [28].”
• Rebutting Existing Misconceptions About Multiple Imputation as a Method for Handling Missing Data (2019, background): “Only in very specific and rare cases of NMAR (Van Buuren, 2012, p. 48; see also Vach, 1994; White & Carlin, 2010) does listwise deletion give unbiased results of statistical analyses.”
• Multiple imputation of covariates by fully conditional specification: Accommodating the substantive model (2012, neutral): “The method of multiple imputation (MI) has become an extremely popular approach for accommodating missing data in statistical analyses, both generally(1) and in the specific context of partially observed covariates.(2) MI involves ‘filling in’ each missing value with draws from an appropriate distri”