|
Activity Number:
|
543
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 10, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #305890 |
|
Title:
|
A New Approach for a Linear Combination of K Multinormal Mean Vectors
|
|
Author(s):
|
Shu-Hui Lin*+ and Jack C. Lee
|
|
Companies:
|
National Taichung Institute of Technology and National Chiao Tung University
|
|
Address:
|
7F5 390 Wen Shin South 2 Road, Taichung, 408, Taiwan
|
|
Keywords:
|
coverage probability ; generalized confidence region ; generalized pivotal quantity ; heteroscedasticity
|
|
Abstract:
|
We consider the problem of constructing a confidence region for a linear combination of K multivariate normal populations when covariance matrices are completely unknown and unequal. A new generalized pivotal quantity for constructing a generalized confidence region is derived. If only two populations are considered and b1=1, b2=-1, then our model is reduced to the multivariate Behrens-Fisher problem. The generalized confidence region is illustrated with a numerical example and the merits of the proposed method are numerically compared with those of the existing methods with respect to their expected hyper-volumes, coverage probabilities under different scenarios in simulation studies.
|
