Estimation of Sensitive Characteristic using Two-Phase Sampling for Regression Type Estimator

Abstract. The problem of underreporting and non response on the sensitive issues are very common in most of the surveys due to our social setup. The randomized response (RR) models reduce rates of non-response and biased response that would ensure respondents ’ privacy if they respond truthfully concerning personal questions. Here RR device is proposed for regression type estimator in which two independent samples are drawn from the population. The estimator of population mean of sensitive variable has been developed. Its bias and variance and optimum variance have also been derived.