A Monte Carlo Comparison of Four Estimators of the Dispersion Matrix of a Bivariate Normal Population, Using Incomplete Data
COLORADO STATE UNIV FORT COLLINS DEPT OF STATISTICS
Pagination or Media Count:
Consider a random vector X1, X2 distributed as a bivariate normal with mean vector zero, and dispersion matrix sigma sigma sub ii. Suppose the authors are given samples of sizes n1 and n2, respectively, from the marginals of X1, X2, and a sample of size n3 from the bivariate population of X1, X2. Suppose the problem is to obtain a good estimator of sigma based on the above incomplete sample. In the paper, four estimators of sigma are compared using Monte Carlo methods, and it is found that a certain relatively simple estimator of sigma is the best or close to the best in almost all situations.
- Statistics and Probability