Activity Number:
|
384
|
Type:
|
Contributed
|
Date/Time:
|
Thursday, August 15, 2002 : 8:30 AM to 10:20 AM
|
Sponsor:
|
General Methodology
|
Abstract - #300792 |
Title:
|
Local Sensitivity to Nonignorability in Frequentist Inference
|
Author(s):
|
Guoguang (Julie) Ma*+ and Daniel Heitjan
|
Affiliation(s):
|
Merck & Company, Inc. and Columbia University
|
Address:
|
BL3-2, Blue Bell, Pennsylvania, 19422,
|
Keywords:
|
Frequentist Inference ; Missing data ; Sensitivity analysis
|
Abstract:
|
The problem of missing data is very common in biomedical research. When observations are nonignorably missing, standard inferences can give misleading results. In this paper, we develop a method for computing the sensitivity of frequentist data summaries to small departures from an ignorable model. Specifically, we use Taylor-series expansions to evaluate the local sensitivity of quantities such as the mean and variance of the distribution, the power of a test, and the coverage probability of a confidence interval. We illustrate the analysis by simulation studies and a real-world example.
|
- 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 2002 program |