|
Activity Number:
|
546
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 10, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #306276 |
|
Title:
|
Control of the Family-Wise Error Rate for Multiple Correlated Test Outcomes: the Effect of Stress on Cytokine Production Prior to Spaceflight
|
|
Author(s):
|
Alan Feiveson*+ and Satish Mehta and Duane L. Pierson
|
|
Companies:
|
National Aeronautics and Space Administration and Enterprise Advisory Services, Inc. and NASA Johnson Space Center
|
|
Address:
|
Johnson Space Center, Houston, TX, 77058,
|
|
Keywords:
|
multiple testing ; family-wise error rate ; cytokines ; spaceflight ; correlated p-values
|
|
Abstract:
|
This study compared levels of cytokine production measured in astronauts' blood samples at time periods shortly before and after NASA space missions with corresponding levels under normal conditions. The problem was how to choose a reasonable cutoff p-value for deciding which cytokines showed a significant change, taking into account the highly-correlated cytokine measurements and extremely unbalanced design of the study. The approach was to construct a pseudo-sample of p-values representative of the case where all 21 null hypotheses (corresponding to zero logistic regression coefficients) were true, while maintaining the dependence of p-values induced by the original measurements. From this pseudo-sample obtained by bootstrapping, it was possible to estimate the distribution of the minimum p-value. The 5-th percentile of this distribution was then used as the p-value cutoff.
|
