|
Activity Number:
|
334
|
|
Type:
|
Other
|
|
Date/Time:
|
Wednesday, August 6, 2008 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
ASA
|
| Abstract - #302715 |
|
Title:
|
What Are the Crucial Error Rates To Consider in Sample-Size Analysis?
|
|
Author(s):
|
Ralph O'Brien*+
|
|
Companies:
|
Case Western Reserve University
|
|
Address:
|
Center for Clinical Investigation - BRB109, Cleaveland, OH, 44106-4961, USA
|
|
Keywords:
|
power analysis
|
|
Abstract:
|
We begin by reviewing the key concepts in classical power and sample-size analysis for regular (frequentist) hypothesis testing. More deeply, we ask: How should such planning help convince us that a significant p-value indeed reflects a true research hypothesis? By using judgments about the probability that the null hypothesis is false, we apply Bayes' Theorem to assess the probability that a significant p-value will be a Type I error or a nonsignificant p-value will be a Type II error. We dub these the "crucial" Type I and II error rates, and we show that they can differ greatly from their classical counterparts. All ideas are illustrated by examining a small preliminary study that tested a very speculative treatment for atherosclerosis and became enthusiastically reported due to its significant p-value supporting efficacy. Unfortunately, the crucial Type I rate may have been over 85%.
|
- 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 2008 program |
