|
Activity Number:
|
165
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 9, 2004 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #300566 |
|
Title:
|
Anchoring Questions in the Respondent-generated Intervals Protocol
|
|
Author(s):
|
Judith M. Tanur*+ and S. James Press and LiPing Chu
|
|
Companies:
|
SUNY, Stony Brook and University of California, Riverside and University of California, Riverside
|
|
Address:
|
PO Box 280, Montauk, NY, 11954,
|
|
Keywords:
|
anchoring ; Bayes surveys ; bounds ; RGI ; hierarchical model
|
|
Abstract:
|
The Respondent-Generated Intervals protocol (RGI) has been used to have respondents (Rs) answer a numerical factual question by giving both a point estimate and also bounds within which they feel almost certain that the true value of the quantity being reported on falls. Some new anchoring questions elaborate the RGI protocol, aiming to improve the accuracy of estimators derived from it. Because point estimates of Rs who give short intervals are weighted more heavily in the Bayesian RGI estimator than point estimates of Rs who give longer ones, it is advantageous to cue and encourage more accurate Rs to give shorter intervals and less accurate Rs to give longer ones. We describe preliminary results of an experiment embedded in a survey to test this new thinking. We introduce mechanisms to direct confident (thus presumably accurate) Rs to give shorter intervals and less confident (thus presumably less accurate) Rs to give longer ones. The experimental design varies the instructions about how Rs should construct their intervals. We also present a new Bayesian estimation procedure that depends upon both the means and the standard deviations of the bounds to further improve accuracy.
|
