On the Stationary Distribution of Iterative Imputations
Authors:
Preprint
Abstract
Iterative imputation, in which variables are imputed one at a time each given a model predicting from all the others, is a popular technique that can be convenient and flexible, as it replaces a potentially difficult multivariate modeling problem with relatively simple univariate regressions. In this paper, we begin to characterize the stationary distributions of iterative imputations and their statistical properties. More precisely, when the conditional models are compatible (defined in the text), we give a set of sufficient conditions under which the imputation distribution converges in total variation to the posterior distribution of a Bayesian model. When the conditional models are incompatible but are valid, we show that the combined imputation estimator is consistent.Keywords
All Related Versions
This publication has 12 references indexed in Scilit:
- mice: Multivariate Imputation by Chained Equations inRJournal of Statistical Software, 2011
- Multiple Imputation with Diagnostics (mi) inR: Opening Windows into the Black BoxJournal of Statistical Software, 2011
- Renewal theory and computable convergence rates for geometrically ergodic Markov chainsThe Annals of Applied Probability, 2005
- Multiple Imputation of Missing ValuesThe Stata Journal: Promoting communications on statistics and Stata, 2004
- Finite sample properties of multiple imputation estimatorsThe Annals of Statistics, 2004
- Miscellanea. Small-sample degrees of freedom with multiple imputationBiometrika, 1999
- Multiple-Imputation Inferences with Uncongenial Sources of InputStatistical Science, 1994
- On rates of convergence of stochastic relaxation for Gaussian and non-Gaussian distributionsJournal of Multivariate Analysis, 1991
- Comparing sweep strategies for stochastic relaxationJournal of Multivariate Analysis, 1991
- Asymptotic Results for Multiple ImputationThe Annals of Statistics, 1988