|
Activity Number:
|
213
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 13, 2002 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section*
|
| Abstract - #300454 |
|
Title:
|
Resolving a Controversy Cver Weighted Versus Unweighted Means
|
|
Author(s):
|
Jitendra Ganju*+
|
|
Affiliation(s):
|
Chiron Corporation
|
|
Address:
|
4560 Horton Street, M/S U-140, Emeryville, , California, 94608, USA
|
|
Keywords:
|
interaction ; stratum sizes
|
|
Abstract:
|
Consider a trial conducted in osteoporotic female and male patients for the purposes of estimating and testing the difference in mean response to two drugs. Suppose the main effect is calculated by weighting cell means by proportionate frequencies. What is the variance of this main effect? In calculating its variance, the cell sizes (the number of females and males per drug) are treated as constants and the cell means are treated as random variables. But given how experiments are often designed, the cell sizes also are random variables. The variance of the main effect should reflect not only the variability of cell means, but also the variability of cell sizes. Answers to the following types of questions are needed: Was the number of females and males fixed by design? If not, was sex a pre-stratified or a post-stratified variable? This talk argues that the choice of a main effect should be governed by the experimental design. How the variance calculation is influenced by different designs is also explored.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2002 program |
