Degree Name
Master of Science
Graduate Program
Mathematics
Advisor
Dr. David J. Olive
Abstract
A robust multivariate linear regression estimator can be obtained by replacing the least squares estimator with the robust hbreg estimator. Then the robust multivariate linear regression estimator is asymptotically equivalent to the classical multivariate linear regression estimator since the probability that the robust estimator is equal to the classical estimator goes to one in probability as the sample size n tends to infinity for a large class of iid zero mean error distributions. This paper discusses the robust estimator and some tests using the robust estimator that are asymptotically equivalent to those using the classical estimator.