|
Activity Number:
|
344
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 11, 2004 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #300682 |
|
Title:
|
Space-time Interaction Models for Mortality Data
|
|
Author(s):
|
James Leeper*+ and Jing Yu
|
|
Companies:
|
University of Alabama and University of Alabama
|
|
Address:
|
PO Box 870326, Tuscaloosa, AL, 35487-0326,
|
|
Keywords:
|
spatial-temporal ; space-time interaction ; general linear mixed model ; generalized additive mixed model ; mortality ; doubly repeated measures
|
|
Abstract:
|
When mortality rates are collected over time and space, modeling is a challenge because it is likely that observations from the same time unit are correlated, and it is also likely that observations from the same geographic area are correlated. Data with this structure are known as doubly repeated measures. Besides the fact that observations are correlated in two dimensions, space-time interaction issues are also of interest. Space-time interaction can be interpreted as different time trends in different geographic areas. This study is aimed at exploring spatial-temporal statistical methods for modeling mortality data with a concentration on the space-time interaction. This study extends a general linear mixed model (GLMM) for mortality data with a focus on space-time interaction. Data transformation is used to deal with the non-normality of mortality rates. The new model is compared to a generalized additive mixed model (GAMM), which does not assume normality, by simulation studies to compare power. The models are applied to county-specific infant mortality rates over a 30-year period in Alabama. These models are able to incorporate time-invariant and time varying covariates.
|
