Methods Of Sensitive Problems

Sensitive problems are private individual or organization problemswhich are not easy to tell the outside world, such as whether evading tax,whether being addicted to drug, whether gambling, etc. For suchproblems, if the survey methods were used directly to get the answers ofthe questions, in order to protect their own privacy, respondents oftendeliberately make wrong answers or refuse to answer.To get reliable sample data of the sensitive issues, it is necessary touse Randomized Response Techniques, which was developed after1965,when Warner model was proposed by Warner. In allusion to randomizedresponse models of the sensitive problems of property characteristics, inthis paper, the author introduced RRT models(including Warner model,Simmons model and sum of random variables response model) to solvethe binary classification sensitive problems and RRT models(includingmultiple sample model, a single sample model and stratified randomsampling model) to solve the muti-class sensitive problems in details; Inallusion to quantitative sensitive questions, the author introduced theaddition model, multiplication model, independence model, transitionmodel, and two-stage sampling on addition model in details. A unifiedlogistic regression model is presented and analyzed in allusion to solvethe sensitive problems of property characteristics.