Activity Number:
|
320
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, August 14, 2002 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Biopharmaceutical Section*
|
Abstract - #301553 |
Title:
|
Comparison of Several Exact Tests for Testing the Hypothesis of Efficacy with a Non-Zero Lower Bound for Comparative Studies
|
Author(s):
|
Pralay Mukhopadhyay*+ and Roger Berger
|
Affiliation(s):
|
North Carolina State University and North Carolina State University
|
Address:
|
Box-8203, Raleigh, North Carolina, 27695, USA
|
Keywords:
|
Exact test ; Binomial distribution ; Confidence interval ; Two-by-Two table
|
Abstract:
|
Exact tests of efficacy (1-p1/p2) with a non-zero lower bound based on two independent binomial proportions are considered. These tests are desirable for studies with small to moderate sample sizes. They are used to demonstrate the efficacy of a new treatment over an existing treatment. We have proposed a test, which is a modification of the standard exact unconditional test and is seen to perform much more well than the standard test in many situations. Our proposed test is compared with the standard test and an approximate test, with respect to size and power. We will invert these tests to get confidence intervals. Comparison of these intervals will be considered on the basis of individual sample points and coverage probability plots.
|