|
Activity Number:
|
228
|
|
Type:
|
Invited
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
SSC
|
| Abstract - #307972 |
|
Title:
|
Default Priors for Frequentist and Bayesian Inference
|
|
Author(s):
|
Donald A.S. Fraser*+
|
|
Companies:
|
University of Toronto
|
|
Address:
|
100 St. George Street
Dept Statistics, Toronto, ON, M5S 3G3, Canada
|
|
Keywords:
|
Bayesian ; frequentist ; default prior ; Marginalization paradoxes
|
|
Abstract:
|
A location model has a natural default prior, the constant or flat prior in the location parameter. This gives a posterior interval which has standard coverage properties for a scalar parameter linear in the location coordinates but typically has biased coverage when it is nonlinear in the location components; this is the marginalization paradox of Dawid, Stone and Zidek (1973). For a general model with moderate continuity we develop second order flat priors for full and for component parameters; these provide second order coverage properties, respectively. This focuses the source of the marginalization paradoxes which is parameter curvature; and it shows how coverage bias can be avoided using priors targeted on the interest parameter.
|
- 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 |
