Generating randomized response by inverse mechanism

摘要

Following the pioneering work of Singh and Grewal we examine how their inverse approach of revising Warner's and Kuk's techniques of directly eliciting a randomized response to unbiasedly estimate a finite population proportion bearing a stigmatizing feature, fares as a viable competitor. We consider sampling by general schemes admitting positive inclusion probabilities for single and paired persons facilitating estimation. Our live-data based numerical presentations suggest Singh and Grewal's approach as quite promising.