|
Activity Number:
|
290
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 5, 2003 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #301958 |
|
Title:
|
A Comparison of Two ADAM Weighting Schemes
|
|
Author(s):
|
Michael Yang*+ and Dean Gerstein
|
|
Companies:
|
NORC at the University of Chicago and NORC at the University of Chicago
|
|
Address:
|
1350 Connecticut Ave., NW, Washington, DC, 20036-1722,
|
|
Keywords:
|
poststratification ; bias ; variance ; design effects ; cost efficiency
|
|
Abstract:
|
The Arrestee Drug Abuse Monitoring program (ADAM) collects substance abuse and urinalysis data from the arrestee population in 35 sites across the U.S. ADAM sampling and data collection take place on a daily basis due to the dynamic nature of the arrestee population. The existing ADAM weighting scheme adjusts for unequal selection and response rates through post-sampling stratification, which requires the collection and processing of extensive population data of booked arrestees at each site. Our analyses indicate that ADAM poststratification weighting has been expensive, inefficient, and causing administrative difficulties. We therefore proposed and tested an alternative two-stage weighting scheme where unequal selection probabilities and unequal response rates are adjusted separately. Under this new scheme, the estimation of the base weight more accurately reflects the sampling process, and the estimation of nonresponse adjustment is based on factors that are known empirically to affect response rates as well as substance abuse behavior. This paper compares the two weighting schemes and shows that the new scheme achieves superior cost efficiency and statistical efficiency.
|
