American Statistical Association
New York City
Metropolitan Area Chapter

Mailman School of Public Health
Columbia University
Department of Biostatistics Colloquium



EFFECTS OF RESIDUAL SMOOTHING
ON THE POSTERIOR OF FIXED EFFECTS
IN DISEASE-MAPPING MODELS
(RANDOM-EFFECT MODELS ARE STILL LINEAR MODELS)

by

James Hodges
University of Minnesota


Abstract

Disease-mapping models for areal data often have fixed effects to measure the effect of spatially varying covariates, and account for spatial clustering by including a random effect with a conditionally autoregressive (CAR) model. Adding the CAR random effect can cause large changes in the posterior mean and variance of the fixed effects. This talk explains why, and gives diagnostic tools to measure posterior variance inflation from collinearity between the fixed effects covariates and the CAR random effects, and to measure each region's influence, through the CAR random effects, on the fixed effect estimates. The larger point is that linear random effect models (including hierarchical models, among many others) are still linear models and thus subject to problems familiar from linear model theory, such as influence and collinearity.

This is joint work with Brian Reich of North Carolina State University.


Date: Thursday, May 10, 2007
Time: 4:00 - 5:00 P.M.
Location: Mailman School of Public Health
Department of Biostatistics
722 West 168th Street
Judith Jansen Conference Room (Room 425)
New York, New York

RESERVATIONS ARE NOT REQUIRED

Refreshments will be served at 3:30 P.M. in the
Biostatistics Conference Room (Room 627).


Home Page | Chapter News | Chapter Officers | Chapter Events
Other Metro Area Events | ASA National Home Page | Links To Other Websites

Copyright © 2007 by New York City Metropolitan Area Chapter of the ASA
Designed and maintained by Cynthia Scherer
Send questions or comments to nycasa@mindspring.com