|
Activity Number:
|
423
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistical Computing
|
| Abstract - #309944 |
|
Title:
|
On the Asymptotic Properties of the Signed Composite Likelihood Ratio Statistic
|
|
Author(s):
|
Zi Jin*+ and Nancy Reid
|
|
Companies:
|
University of Toronto and University of Toronto
|
|
Address:
|
100 St George St, Toronto, ON, M5S 3G3, Canada
|
|
Keywords:
|
Composite likelihood ; Pairwise likelihood ; Asymptotic efficiency ; Higher order asymptotics ; Signed composite likelihood ratio statistic
|
|
Abstract:
|
Composite likelihood has received increased attention in handling large dimensional data sets when the computation of the full likelihood is highly complex. In particular, we concentrate on pairwise likelihood, which arises by considering two-dimensional marginal densities. The pairwise likelihood estimators are generally consistent and asymptotically normally distributed. We summarize different test statistics, as well as their asymptotic distributions. Higher order asymptotic properties of the signed composite likelihood ratio statistic are outlined. Some examples are analyzed in detail, and simulation studies are presented to further explore the asymptotic performance of the composite likelihood ratio statistic.
|
