Estimation Parameter for Two-stages Randomized Response Technique in Logistic Regression Model

• Main Authors: CHANG,PEI-CHIEH, 張珮婕
• Other Authors: LEE,SHEN-MING
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/ave633

碩士 === 逢甲大學 === 統計學系統計與精算碩士班 === 106 === 　　When the survey study is related to sensitive issues, such as political orientation, sexual orientation or income, the interviewees may not be willing to respond truthfully which leads to bias results. In order to protect the interviewees privacies and improve the willingness to provide the correct answer, Warner (1965) proposed the randomized response (RR) technique in which the respondents randomly selects questions by means of devices in order to ensure that their are protected. Huang (2004) took into account that some interviewees dishonestly the questions, and proposed a two-stage RR. Not only this method can estimate the proportion of sensitive ethnic groups, but also estimate the honest answer rate. Maddla (1983) proposed a logistic regression model to investigate the effect of covariates on the proportion of sensitive features. Later, Scheers and Dayton (1988) discussed the RR technique in Warner (1965). Greenberg at al. (1969) designed under the RR technique without correlation, and estimated those proportions by using the maximum likelihood estimation method. This article uses Huang’s (2004) two-stage RR technique for a logistic regression model to investigate and estimate the effect of covariates on the proportion of sensitive features and the honest response rate.