時 間：114年11月20日（星期四下午15:10~16:00）
地 點：管理學院新大樓M103
主講人：黎長日 教授
(Applied Analysis Research Group, Faculty of Mathematics and Statistics, Ton Duc Thang University, Ho Chi Minh City, Vietnam)
講 題：Logistic regression for data acquired via two-stage generalized randomized response technique
Abstract
We propose a two-stage generalized randomized response (GRR) design, incorporating the first-stage design of Huang’s two-stage RR design as the first-stage design and Christofides’ GRR design as the second-stage design. Our design allows estimation of the proportion of the population with a sensitive attribute and the population proportion of sensitive attribute respondents providing a truthful response to a directly asked sensitive question in the first stage. We derive the large-sample properties of the maximum likelihood (ML) estimators for logistic regression parameters and assess the finite-sample performance of the ML estimators via Monte Carlo experiments. The practical application of our methodology is illustrated by using the 2016 survey data on the sexuality of local freshmen of Feng Chia University in Taiwan.