|
Activity Number:
|
236
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #306703 |
|
Title:
|
The Effect of Collinearity on Parameter Estimation in Bayesian Spatially Varying Coefficient Models
|
|
Author(s):
|
David Wheeler*+
|
|
Companies:
|
The Ohio State University
|
|
Address:
|
100 W. 2nd Ave., Columbus, OH, 43201,
|
|
Keywords:
|
MCMC ; spatial statistics ; regression ; simulation study
|
|
Abstract:
|
The recognition in the sciences that relationships between explanatory variables and a response variable in a regression model are not always constant across a study area has led statisticians to develop Bayesian regression models that allow for spatially varying coefficients. The interest in the application of these models is the interpretation of the coefficients in addition to overall model fit. There has been little focus, however, on the accuracy of the estimated coefficients, particularly in the presence of explanatory variable collinearity. This presentation will investigate the effect of collinearity on the regression coefficients through use of simulation and MCMC estimation. The results show the Bayesian regression model is overall fairly robust to moderate levels of collinearity but degrades substantially with strong collinearity.
|
