Maximum likelihood analysis of generalized linear models with missing covariates

Citation
Nj. Horton et Nm. Laird, Maximum likelihood analysis of generalized linear models with missing covariates, STAT ME M R, 8(1), 1999, pp. 37-50
Citations number
41
Categorie Soggetti
Health Care Sciences & Services
Journal title
STATISTICAL METHODS IN MEDICAL RESEARCH
ISSN journal
09622802 → ACNP
Volume
8
Issue
1
Year of publication
1999
Pages
37 - 50
Database
ISI
SICI code
0962-2802(199903)8:1<37:MLAOGL>2.0.ZU;2-E
Abstract
Missing data is a common occurrence in most medical research data collectio n enterprises. There is an extensive literature concerning missing data, mu ch of which has focused on missing outcomes. Covariates in regression model s are often missing, particularly if information is being collected from mu ltiple sources. The method of weights is an implementation of the EM algori thm for general maximum-likelihood analysis of regression models, including generalized linear models (GLMs) with incomplete covariates. In this paper , we will describe the method of weights in detail, illustrate its applicat ion with several examples, discuss its advantages and limitations, and revi ew extensions and applications of the method.