Respondent Jeopardy and Hazards about People’s Privacy of Sensitive Problem in Randomized Response

• Main Authors: Yu-Chin Wu, 吳又親
• Other Authors: Ker-Chang Chen
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/44254391685444719496

碩士 === 銘傳大學 === 應用統計資訊學系碩士班 === 97 === “Sensitive question” has been an issue troubling researchers in the field of statistics. For questions relevant to privacy issue would lead to the consequence that people might answer the questions dishonestly or even refuse to answer, which makes it difficult to obtain the true information. Warner (1965) introduced randomized response model at first to generate enough reliable data to inference the true proportion of sensitive group, not only to reduce the probability of untrue answer, but also increase the sense of security of interviewees. Greenberg et al.(1969) introduced unrelated-question model. Christofides (2003) suggested a new model that interviewees didn’t need to answer directly “Yes” or ”No”, in other to reduce the probability of untrue answer. Recently, Wang and Tsai(2007) extended Christofides’s (2003) model and combined Warner’s(1965) and Greenberg et al.(1969) unrelated-question, then brought up a new random response model. Frederick et al.(1976) introduced respondent jeopardy in randomized response models, the purpose of the model was to judge that whether privacy of interviewees can be protected. Greenberg et al.(1977) introduced hazards function in the unrelated question randomized response model. It is important to protect people’s privacy in randomized response. The paper is to simulate not only the respondent jeopardy and hazards in different parameters situation, but also to discuss in which type of model people could get more sense of security indeed.