Bayesian Techniques in Randomized Response

Abstract

The issue of truthfulness of the answers given to socially or personally sensitive questions appeared at the very beginning of the construction of the surveys. The respondents often try to avoid answering a sensitive question or they do not tell the truth. Various research and analysis techniques have been developed to adjust the surveys results in order to assess the real opinion of the respondents. The work introduces a bayesian analysis of the Randomized Response technique. It presents arguments for choosing various a priori distributions, both non-informative and compensating for the preliminary information described by parametrized distributions.