USING RANDOMIZED RESPONSE TO ESTIMATE THE TRUTHFUL REPLIES

Abstract

Variety of work has been done in estimation of the unknown parameters of the innocuous variables but the estimation and inferences of the sensitive variables is relatively new field. Lots of gratitude is due to Warner (1965) who pioneered the technique of randomized response. This thesis contributes to eliminate the evasive answer bias while dealing with the qualitative sensitive study variables. We have worked on the Kim and Warde (2004) model to improve its methodology and proposing a new mixed randomized response model for estimation of population proportion of the qualitative sensitive variables. We have introduced a variant in the Kim and Warde (2004) model and increased the efficiency of estimation of population proportion of the qualitative sensitive variables. The proposed model is also extended in the stratified random sampling scheme, and is proved to be efficient in this atmosphere as well.