|
Activity Number:
|
44
|
|
Type:
|
Invited
|
|
Date/Time:
|
Sunday, July 29, 2007 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #307731 |
|
Title:
|
Does Mis-specification of the Random Effects Distribution Affect Predictions of Random Effects?
|
|
Author(s):
|
Charles E. McCulloch*+ and John Neuhaus
|
|
Companies:
|
University of California, San Francisco and University of California, San Francisco
|
|
Address:
|
185 Berry Street, Suite 5700, San Francisco, CA, 94107,
|
|
Keywords:
|
mixed models ; misspecification ; generalized linear models
|
|
Abstract:
|
Statistical models that include random effects are commonly used to analyze longitudinal and clustered data. Predicted values of the random effects are often used, e.g., in profiling of physician or hospital performance. In typical applications, the data analyst specifies a parametric distribution for the random effects (often Gaussian) although there is little information available to guide this choice. Whether inferences about regression parameters are sensitive to this specification is of considerable debate in the literature. However, there has been little work on whether the misspecification affects prediction of random effects. Through theory, simulation and an example, we show that misspecification can have a moderate impact on predictions of random effects and develop simple ways to diagnose such sensitivity.
|
- 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 2007 program |
