|
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 - #301058 |
|
Title:
|
Statistical Methods in Estimating Mortalities for Ages with Zero Observed Deaths
|
|
Author(s):
|
Rong Wei*+ and Anastasia Voulgaraki
|
|
Companies:
|
National Center for Health Statistics and National Center for Health Statistics
|
|
Address:
|
Room 3222, Hyattsville, MD, 20782,
|
|
Keywords:
|
Mortality estimation ; Life Tables ; small area ; Mixed Distribution ; Two-part Model ; Bayesian method
|
|
Abstract:
|
Annual age-specific mortality is a key estimate in constructing the US State-Specific Life Tables. The estimates are based on the observed number of deaths in each of the race-gender subpopulations. However the observed deaths are often zero for some young ages in small States during the data collection period. To estimate the mortality rates with "zero" deaths, several strategies have been applied: 1. taking an average of data from three consecutive years; 2. "borrowing" strengths from other States in the same data year using Bayesian methods; 3. LOESS smoothing with a two dimensional window of age and time; 4. applying probability models based on data from 30+ years; and 5. smoothing age mortality curves using Heligman-Pollard equation. In this paper we compare results from above methods applied on eight US States' data. A problem of "zero inflation" in estimation is also addressed.
|
