|
Activity Number:
|
428
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Health Policy Statistics
|
| Abstract - #306960 |
|
Title:
|
Permutation Test and Its Application to the Youth Partners in Care (YPIC) Study
|
|
Author(s):
|
Lingqi Tang*+ and Naihua Duan and Joan Asarnow
|
|
Companies:
|
University of California, Los Angeles and University of California, Los Angeles and University of California, Los Angeles
|
|
Address:
|
10920 Wilshire Blvd., Suite 300, Los Angeles, CA, 90024-6505,
|
|
Keywords:
|
permutation ; ANCOVA ; nonresponse weights
|
|
Abstract:
|
Due to recent advances in computing power the permutation test has become a popular tool in the analysis of randomized controlled trials because it requires fewer assumptions than a parametric analysis. Although the permutation test is straightforward, there is no general agreement concerning an appropriate way of implementing the permutation test with analysis of covariance models (ANCOVA), especially when missing data complicate the problem. In this presentation, we demonstrate an application of permutation test to the Youth Partners in Care (YPIC) study. We conducted sensitivity analyses for intervention effects using a design-based permutation test to ascertain whether our findings from ANCOVA model are sensitive to model assumptions. Similar results from the permutation test and parametric model confirm that our findings are robust to model assumptions.
|
