Block-Conditional Missing at Random Models for Missing Data

TL;DR

This paper introduces block-conditional MAR models that interleave subsets of variables and missing data indicators, providing new methods for estimation, including EM algorithms, especially for categorical and exponential family data.

Contribution

It proposes a novel class of block-sequential models called BCMAR, extending missing data analysis beyond traditional MAR assumptions with new estimation strategies.

Findings

Block-conditional MAR models generalize MAR assumptions.

Block-monotone reduced likelihood often yields consistent estimates.

EM algorithm effectively estimates parameters in these models.

Abstract

Two major ideas in the analysis of missing data are (a) the EM algorithm [Dempster, Laird and Rubin, J. Roy. Statist. Soc. Ser. B 39 (1977) 1--38] for maximum likelihood (ML) estimation, and (b) the formulation of models for the joint distribution of the data ${Z}$ and missing data indicators ${M}$, and associated "missing at random"; (MAR) condition under which a model for ${M}$ is unnecessary [Rubin, Biometrika 63 (1976) 581--592]. Most previous work has treated ${Z}$ and ${M}$ as single blocks, yielding selection or pattern-mixture models depending on how their joint distribution is factorized. This paper explores "block-sequential"; models that interleave subsets of the variables and their missing data indicators, and then make parameter restrictions based on assumptions in each block. These include models that are not MAR. We examine a subclass of block-sequential models we call block-conditional MAR (BCMAR) models, and an associated block-monotone reduced likelihood strategy that typically yields consistent estimates by selectively discarding some data. Alternatively, full ML estimation can often be achieved via the EM algorithm. We examine in some detail BCMAR models for the case of two multinomially distributed categorical variables, and a two block structure where the first block is categorical and the second block arises from a (possibly multivariate) exponential family distribution.