A multivariate technique for multiply imputing missing values using a sequence of regression models

By Trivellore E. Raghunathan, James M. Lepkowski, John Van Hoewyk, P. Solenberger. 2001

2,370 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/13b30e35b9a54dad07094cfe4f50d40ff15d8370

How researchers see this paper

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

Citations in the corpus

Using Calibration to Improve Rounding in Imputation (2008, background): “Alternatively, sequential imputation (Raghunathan et al. 2001; Van Buuren and Oudshoorn 2000) can accommodate data of miscellaneous types by imputing under a collection of univariate conditional models, at the risk of inconsistencies due to the absence of a single joint model.”
A Note on Bayesian Inference After Multiple Imputation (2010, background): “Here, Y (l)mis is a draw from the posterior predictive distribution of (Ymis|Yobs), or from an approximation of that distribution such as the approach of Raghunathan et al. (2001).”
Handling Missing Data by Maximum Likelihood (2012, methodology): “An alternative algorithm, recently introduced into PROC MI in Release 9.3, is variously known as the fully conditional specification (FCS), sequential generalized regression (Raghunathan et al. 2001), or multiple imputation by chained equations (MICE) (Brand 1999, Van Buuren and Oudshoorn 2000).”
Multiple Imputation of Multilevel Missing Data—Rigor Versus Simplicity (2015, background): “These findings are discussed in more detail in Reiter et al. (2006) and Andridge (2011), and we refer the interested reader to these articles.”
Using CART to generate partially synthetic public use microdata (2005, unknown): “Such conditional approaches are related to the sequential imputation algorithms of van Buuren and Oudshoorn (1999) and Raghunathan et al. (2001) for parametric modeling.”
Missing data techniques for structural equation modeling. (2003, background): “A rather different approach to multiple imputation with nonnormal data is sequential generalized regression( Raghunathan, Lepkowski, Van Hoewyk, & Solenberger, 2001 ).”
World income inequality databases: an assessment of WIID and SWIID (2015, unknown): “…based on matching such as the hot deck, for example), but also the way in which the SWIID’s combination of one-step, two-step, and loess regressions compares with methods commonly used to impute multiple missing variables sequentially such as ‘chained equations’ (Raghunathan et al. 2001).”
Multiple imputation using chained equations: Issues and guidance for practice (2011, methodology): “Results from NORM agree asymptotically with those from MICE when all imputation models are linear [11], but may perform poorly with non-Normal or categorical data [60].”
Multiple imputation: review of theory, implementation and software (2007, methodology): “• IVEWARE—[49] Imputation and Variance Estimation software performs single or MIs of missing values using the Sequential Regression Imputation Method described in [50].”
Frailty in Older Adults: A Nationally Representative Profile in the United States. (2015, methodology): “To address these, we employed multiple imputation (10 replicates) using chained equations (30) incorporating all variables defined above.”
MICE: Multivariate Imputation by Chained Equations in R (2011, methodology): “Two general approaches for imputing multivariate data have emerged: joint modeling (JM) and fully conditional specification (FCS), also known as multivariate imputation by chained equations (MICE).”
Small-sample degrees of freedom for multi-component significance tests with multiple imputation for missing data (2007, background): “The basic idea is to fill in the missing data, Ymis, with draws from the conditional distribution of Ymis given the observed data, or from approximations of that distribution such as those of Raghunathan et al. (2001).”