|
Activity Number:
|
158
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Monday, August 3, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Health Policy Statistics
|
| Abstract - #303593 |
|
Title:
|
Nonparametric Inference Procedure for Percentiles of the Random Effects Distribution in Meta-Analysis
|
|
Author(s):
|
Rui Wang*+ and Lu Tian and Tianxi Cai and Lee-Jen Wei
|
|
Companies:
|
Harvard University and Stanford University and Harvard University and Harvard University
|
|
Address:
|
Department of Biostatistics, Boston, MA, 02115,
|
|
Keywords:
|
Bivariate beta ; Conditional permutation test ; Erythropiesis-stimulating agents ; Logit-normal ; Two-level hierarchical model
|
|
Abstract:
|
Suppose that the random effects distribution of the parameter of interest in meta analysis is completely unspecified. We propose a simple nonparametric interval estimation procedure for making inferences about the percentiles, for example, the median, of this distribution. In contrast to the existing methods which can only make inferences about the mean of the random effects distribution, the validity of the new proposal does not require the number of studies involved to be large. The new proposal is theoretically valid when the sample sizes of individual studies are large. Empirically we find that our procedure performs well even with moderate individual study sample sizes. The new procedure can be implemented with study-level summary statistics.
|
