|
Activity Number:
|
144
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 7, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #306386 |
|
Title:
|
Smoothing U.S. State Life Tables for Years 1999--2001
|
|
Author(s):
|
Rong Wei*+ and Lester R. Curtin and Robert Anderson and Elizabeth Arias
|
|
Companies:
|
National Center for Health Statistics and Centers for Disease Control and Prevention and National Center for Health Statistics and National Center for Health Statistics
|
|
Address:
|
3311 Toledo Road, Room 3114, Hyattsville, MD, 20782,
|
|
Keywords:
|
US mortalities ; life tables ; reliability criteria ; smoothing methods ; variances of LT functions
|
|
Abstract:
|
The NCHS publishes race-gender specific life tables for individual States decennially. Many of those tables were not published due to concern over the statistical reliability of the table functions. In constructing State life tables for 1999-2001, two different statistical models were considered: a Bayesian method that borrows strength from the national data and a two-dimensional local weighted regression to use 30 years of State data. Both methods eliminate the arbitrary nature in the past smoothing technique and allow a greater number of life tables to be published. To evaluate the new life table functions, minimum total deaths, number of adjusted death and changes in life expectancy at birth being used for criteria of reliability are calculated. Restrictions on relative standard errors for probabilities of dying and life expectancy at birth are also justified for the new methods.
|
