|
Activity Number:
|
204
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 7, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistical Computing
|
| Abstract - #305507 |
|
Title:
|
A Note on Testing of Hypothesis of Kroneckar Product Covariance Structure in Doubly Multivariate Data
|
|
Author(s):
|
Anuradha Roy*+
|
|
Companies:
|
The University of Texas at San Antonio
|
|
Address:
|
Department of Management Science and Statistics, San Antonio, TX, 78249,
|
|
Keywords:
|
covariance structure ; Kroneckar product ; maximum likelihood estimates ; mixed effects model ; PROC mixed
|
|
Abstract:
|
The decision of whether to use Kronecker product variance covariance matrix over an unstructured variance covariance matrix is an important issue in doubly multivariate data in small sample set up. In this paper, we talk about statistical tests pertaining to Kronecker product variance covariance matrices. The tests can be tested by the MIXED procedure of SAS, in two steps. As PROC MIXED has the potential problem of convergence of the maximum likelihood estimates, algorithms are discussed to calculate the test statistics using the IML procedure of SAS. We demonstrate the tests with four real datasets and compare the results with that of PROC MIXED. Simulation studies are conducted to examine the needed sample sizes. A comparison study is presented when one of the matrices in the Kronecker product is either an unstructured or a compound symmetric or an autoregressive of order one.
|
