|
Activity Number:
|
390
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #305917 |
|
Title:
|
Design Effects in Randomized Experiments Based on Sample Surveys
|
|
Author(s):
|
K. P. Srinath*+
|
|
Companies:
|
Abt Associates Inc.
|
|
Address:
|
4550 Montgomery Ave., Bethesda, MD, 20814,
|
|
Keywords:
|
complex sampling design ; sampling weights ; treatment and control groups
|
|
Abstract:
|
Evaluation of new social or educational programs is frequently done using randomized experiments based on samples from surveys that use complex designs. Such evaluations may involve specific geographic areas like state, Census regions or could be at the national level. Dividing a sample of last stage sampling units to treatment and control groups at random in a multi-stage design is common. A question of interest is what is an appropriate design in terms of the number of units at each stage, the overall probability of selection and the sampling weight that minimizes the design effect. The desire is to minimize the design effect as much as possible to maximize the power of statistical tests for testing differences. We examine alternative designs including the self weighting design both from an operational point of view and design effect.
|
- 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 2006 program |
