基于混合模型的敏感性问题研究概率论与数理统计专业论文.docxVIP

I I 基于混合模型的敏感性问题研究 中文摘要 基于混合模型的敏感性问题研究 中文摘要 在抽样调查中，敏感性问题的研究日益受到重视。自从沃纳、西蒙斯应用随 机化技术分别提出沃纳模型及西蒙斯模型以来，敏感性问题的研究得到了快速的 发展。近年来，计算机及网络的普及与发展为敏感性问题的研究提供了新的手段。 对于某一敏感性问题，具有敏感性特征的被调查者作出不真实回答时往往会 有些迟疑，这会导致此类人群与无敏感性特征的人群在回答该问题的时间上具有 差异性。通过设计好的计算机程序记录每位被调查者回答敏感性问题所用的时间， 本文建立了一种基于回答敏感性问题所用时间的混合模型，且针对该模型，给出 了估计比例参数的 Gibbs 抽样算法。模拟算例表明，利用该模型可以较好地识别未 如实回答的敏感性特征人群的比例。 关键词：敏感性问题；混合模型；Gibbs 抽样 作 者：李 昶 指导老师：汪四水 II II Abstract Study on Sensitive questions Based on the Mixture model Study on Sensitive questions Based on the Mixture model Abstract The research on sensitive questions is getting more and more attention in the sample survey. Since Warner and Simmons developed Warner model and Simmons model applying randomized response technique, the research on sensitive questions has gained rapid development. In recent years, the development and popularization of computer and network provide a new method for researching sensitive questions. For a sensitive question, respondents with sensitive characteristics will always delay giving the untruthful response when were investigated. This leads to the difference in the time of answering these questions between them and those who are not with the sensitive characteristics. Through computation records of time by each respondent answering relevant questions, this thesis establishes a mixture model based on reaction time of respondents, and for the model, presents Gibbs sampler to estimate the proportion parameters. The simulation examples show that the model can be used to identify the proportion of respondents who answer the question untruthfully. Keywords: Sensitive questions; Mixture model; Gibbs sampler Written by Li Chang Supervised by Associate Prof. Wang Sishui 目 录 HYPERLINK \l _bookmark0 第一章 引言 1 HYPERLINK \l _bookmark1 第二章 敏感性问题的传统处理方法 3 HYPERLINK \l _bookmark2 §2.1 沃纳随机化回答模型 3 HYPERLINK \l _bookmark3 §2.2 西蒙斯随机化回答模型 4 HYPERLINK \l _bookmark4 第三章 敏感性问题混合模型的建立 6 HYPERLINK \l _bookmark5 §3.1 模型的假设 6 HYPERLINK \l _bookmark6 §3