Improvement of the Gupta and Thornton Scrambling Model through Double Use of Randomization Device

To eliminate or reduce the extent of effect of falsified reporting, randomized response models are now being used as a survey tool in almost every field of science. To contribute in the literature on randomized response models, we propose an improved randomized response model utilizing two responses from each respondent. The proposed model provides an unbiased estimator and may be thought of as more shielding in term of privacy protection. The proposed estimator is a weighted estimator having the minimum possible sampling variance. The responses are obtained through the Gupta and Thornton (2002) model. The suggested weighted estimator is unconditionally more efficient than all of the scrambled response estimators suggested until now.