Activity Number:
|
311
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, August 14, 2002 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Statistics in Epidemiology*
|
Abstract - #301591 |
Title:
|
Variance Estimation for Rate Estimates Using Data From Two Independent Surveys
|
Author(s):
|
Lei Li*+
|
Affiliation(s):
|
RTI International
|
Address:
|
3040 Cornwallis Road, P.O.Box 12194, Research Triangle Park , North Carolina, 27709-2194,
|
Keywords:
|
rate ; relative risk ; variance estimation ; Poisson Model ; Survey data
|
Abstract:
|
Rate and relative risk are commonly estimated to compare the exposure effect in epidemiology. The numerator and the denominator for calculating rate are sometimes based on data from two independent surveys. For example, to study the effect of smoking on death rate, the death rate for smokers and nonsmokers are estimated using the number of death from the National Mortality Followup Survey and the number of smokers in a population from the National Health Interview Survey. The rate estimated in this way can be treated as a ratio estimate and its variance estimate can be derived. When there are confounding variables to be controlled for, Poisson models are often used, and rate and relative risk estimates can be easily derived from the model parameters. However, the estimation of the standard errors of the model parameter estimates needs to take into consideration of the design features of two different surveys. The methods for the variance estimation in this context are discussed and a method based on Binder's sandwich estimator (1983) is developed. Numerical results are given as examples.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2002 program |