Activity Number:
|
475
- Understanding Threats to People, Data, and Privacy
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, July 31, 2019 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Social Statistics Section
|
Abstract #305302
|
|
Title:
|
A Curious Variation on the Warner Device for Use in Randomized Response
|
Author(s):
|
Stephen Sedory* and Zakry Zapata and Sarjinder Singh
|
Companies:
|
Texas A & M University-Kingsville and Texas A&M University-Kingsville and Texas A&M University-Kingsville
|
Keywords:
|
randomized response technique;
estimation of proportion;
response bias
|
Abstract:
|
The use of randomized response began with Warner's 1965 paper in which he introduced a method for eliciting answers to questions on sensitive issues that preserved the privacy of the respondent. The method utilized a device that randomized the question to which the interviewee was to respond. The interviewer would not know which question was being asked, which, it was hoped, would remove any motivation for evasive answer bias. Both the interviewer and interviewee would know the fixed probability of being confronted with the sensitive question. This paper introduces a variation on such a device for which the associated estimator of the population proportion of a sensitive attribute is free from such a fixed probability. It is expected that the privacy of the respondent would be further protected and bias in the response further reduced. The paper determines properties of the estimator and compares it with that associated with the original Warner device.
|
Authors who are presenting talks have a * after their name.