Date of Award
2007
Degree Type
Thesis
Degree Name
Master of Science
Program
Epidemiology and Biostatistics
Supervisor
Dr. John Koval
Second Advisor
Dr. Guangyong Zou
Abstract
The main focus of this thesis is on finding a practical way to impute binary missing values in a data set with arbitrary missing pattern using standard statis tical software. We considered cases where two and three variables with missing values. The method we propose is to use multiple imputation (MI) by Markov Chain Monte Carlo (MCMC) to transfer an arbitrary missing pattern to mono tone missingness, and then apply proper MI for binary data. The simulation study shows that our methods produce more accurate parameter estimates than the rounding method and complete-case analysis. We illustrate our methods with a real data set involving high-school students smoke behavior
Recommended Citation
Huang, Wenyi S., "EVALUATION OF IMPUTATION METHODS FOR NON-MONOTONE MISSINGNESS IN LOGISTIC REGRESSION" (2007). Digitized Theses. 5103.
https://ir.lib.uwo.ca/digitizedtheses/5103