Inference for imputation estimators

By James M. Robins, Naisyin Wang. 2000

302 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Statistics and Probability

View Paper: https://www.semanticscholar.org/paper/3214538c562c010bdbb055582e04fc3a4d7b6ce9

How researchers see this paper

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

In Artificial Intelligence

In Statistics and Probability

Citations in the corpus

Calibrated Bayes, an alternative inferential paradigm for official statistics (2012, background): “Multiple imputation does not yield consistent estimates of variance under particular forms of model misspecification (Meng 1994; Rao 1996; Fay 1996; Robins and Wang 2000; Kim et al., 2006).”
Maximum Likelihood Multiple Imputation: Faster Imputations and Consistent Standard Errors Without Posterior Draws (2012, unknown): “We know of only one other approach that can produce consistent variance estimates under misspecified and incompatible imputation and analysis models [22] — but the calculations are relatively complicated and require statistics, including but not limited to the score function, that users often lack a”
A new algorithm for haplotype‐based association analysis: the Stochastic‐EM algorithm (2004, methodology): “…non-parametric or Bayesian-based, for handling missing data has received much attention in the last decades (Deltour et al. 1999; Meng, 1994; Robins & Wang, 2000; Rubin, 1996; Wang & Robins, 1998, Wei & Tanner, 1990) and investigating how such methods, other than the SEM al-gorithm, can be…”
Stable Weights that Balance Covariates for Estimation With Incomplete Outcome Data (2015, background): “This work was supported by a grant from the Alfred P. Sloan Foundation. ity in the estimates (Robins and Wang 2000; Kang and Schafer 2007).”
Multiply Robust Estimation in Regression Analysis With Missing Data (2014, methodology): “When π̂ (X, S) is close to zero for some subjects whose outcome is observed, the weight R/π̂ (X, S) used by the IPW and the AIPW estimators becomes extremely large, leading to skewed and highly variable sampling distribution of those estimators (Robins, Rotnitzky, and Zhao 1995; Scharfstein, Rotnitz”
Comment: Performance of Double-Robust Estimators When “Inverse Probability” Weights Are Highly Variable (2007, background): “Indeed, Robins and Wang (2000) in their paper “Inference for Imputation Estimators” stated the following: If nonresponse is ignorable, a locally semiparametric efficient estimator is doubly protected; i.e., it is consistent if either a model for nonresponse or a parametric model for the complete…”
Using simulation studies to evaluate statistical methods (2017, methodology): “New methods are sometimes published but not implemented in any software (for example, Robins and Wang 33 and Reiter 34 ), implemented poorly, or implemented in unfamiliar software.”
Covariate balancing propensity score by tailored loss functions (2016, background): “th data point. The IPW put weight w ion this data point, where (15) w i= ( p^ 1 i if T i= 1; (1 p^ i) 1 if T i= 0: This weighting is useful to estimate the ATE ˝ 1; 1. Several previous articles (e.g. Robins and Wang, 2000; Kang and Schafer, 2007; Robins et al., 2007) have pointed out the hazards of ”
Removing GPS collar bias in habitat selection studies (2004, methodology): “We calculated a RSF for each of the 30 data sets and plotted the mean coefficient against the number of simulations conducted to discern how many iterations were needed to achieve stable estimates (Rubin 1996; Robins & Wang 2000).”
Combining Multiple Imputation and Inverse-Probability Weighting (2012, methodology): “Bias also occurs when IPW is used at stage 1 instead of MI (IPW/MI*), but is smaller than that of MI*/MI*, and indeed of MI*/MI. Theorems 1 and 2 of Robins and Wang (2000) enable the asymptotic (N → ∞) percentage bias in V̂ M to be calculated when M = ∞ (see Web Appendix).”
Multiple Imputation with Diagnostics (mi) in R: Opening Windows into the Black Box (2011, methodology): “Examining the implications of imputations is particularly important because of the inherent tension of multiple imputation: that the model used for the imputations is not in general the same as the model used for the analysis (Meng 1994; Fay 1996; Robins and Wang 2000).”
On the bias of the multiple‐imputation variance estimator in survey sampling (2006, background): “Robins and Wang (2000) extended the result to nonparametric models and proposed an alternative variance estimator.”
Calibrated Bayes (2006, background): “In a recent discussion of these ideas, Robins and Wasserman (2000) argued for reexamination of the likelihood principle in the context of an infinite-dimensional example described by Robins and Ritov (1997).”
Multiple imputation using chained equations: Issues and guidance for practice (2011, background): “Alternatives to proper imputation [6] are not considered here.”
Proper and Improper Multiple Imputation (2003, background): “Robins & Wang (2000) give a more general argument.”