|
Activity Number:
|
231
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 5, 2008 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #301269 |
|
Title:
|
An Approach To The Problem of Zeros In Mortality Rates
|
|
Author(s):
|
Anastasia Voulgaraki*+ and Rong Wei
|
|
Companies:
|
National Center for Health Statistics and National Center for Health Statistics
|
|
Address:
|
Room 3118, Hyattsville, MD, 20782,
|
|
Keywords:
|
State-level mortality ; mixed distribution ; two-part models ; hurdle models ; zero-inflated models ; bootstrap
|
|
Abstract:
|
In states with small subpopulations, the observed annual mortality rates are often zero, particularly among young ages. In estimating age-specific mortality, death rates are presented on logarithmic scales, and therefore zero rates are problematic. We use data from the NCHS public-use mortality files (1970-2002) to fit appropriate probability models and estimate expected values for replacing zero mortality. The probability models considered are based on mixed distribution and Poisson regression methods and are applied to mortality data from eight states with the problem of zero observed rates. The criterions used to compare the performance of the models are the root mean square error and the mean absolute error. The 95% confidence intervals are constructed using parametric bootstrap and asymptotic methods, where applicable. This paper describes the findings from our preliminary research.
|
