A Method for Estimating Sensitive Attribute Proportion

This paper considers the problem of unbiased estimation of proportion of sensitive attribute in human society. In order to protect the respondent’s privacy, various techniques of randomized response (RR) rather than direct response are available. It is well-known that RR techniques are designed for sensitive characteristics to improve the respondent cooperation. In this paper we briefly describe Warner’s model and compare it to the survey with face to face interview. Then we examine two methods of randomized response of unrelated question and crosswise model. In the end, with the use of Warner's technique, we investigate the proportion of the usage of alcoholic drink in one of the universities.