The Qualitative Sensitive Question Investigation Technique Research

sensitive question investigation has been continued since middle ofthe last century when the Warner model and Simmons model werebrought up. Many advanced methods around the two classical modelshave been brought up by the later scholars. Theoretical researches andApplications in this field have made good progress. However there is stillroom for improvement in Sensitive question investigation. The traditionalRandom Response Technique (RRT) is still faced with thenot-high-enough technical accuracy problem. In nature RRT is tosacrifice a certain amount of the sample to get respondents' trust andcooperation in return. Sensitive question investigation essentially meansthat the problem must be solved in certain estimation accuracy to reducethe exposure risk of respondents answering sensitive questions and toraise the level of respondents' cooperation.To resolve the above problem, we should open thinking fromdifferent ways. First, from the implementation of research methods, wecan choose some useful research methods which can be propitious to sensitive question investigations. Such as mailed questionnaireinvestigation, this can avoid the embarrassment of face-to-face, networkinvestigation, not only of confidentiality, but also there are otheradvantages, such as lively, fast, convenient and so on. The projectiontechnique originated from psychological measurement methods are mostappropriate applied to the investigation of the respondents' feel,viewpoints, and motivation and so on, which would be consistent with thenature of sensitive question investigations. However, each investigationmethod has its shortcomings, such as projection technique, it is still in itsinfancy. Points in the implementation and standardization work are notsatisfactory, a lack of scoring objectivity and representation norm. Theseare depending further research.In addition to investigating methods, RRT is still a mainstreamtechnique. In view of the shortcomings of RRT that technical accuracy isnot high, based on a wide range of reading and study many scholars'literatures in this area, I propose a new category improved method—thestack improvements respectively Warner model and Simons model. Theprinciple of multi-stack Warner improved is to prepare more than one setof device such as several packs of cards with the positive or negativequestions on each card's back, and the opposite questions are settled as acertain proportion. Each respondent takes out one card from each pack,and then adds the outcome of each card in the mind, only tells investigators the total outcome. Simmons model operating principle issimilar. Due to the outcome's probability of each card is independent, soin essence, the stack improved method is to reuse of waste samples toimprove accuracy of the estimates. Theoretically it is proved that whenthe multi-stack tends to infinity, the estimation variance tends to the valuewhich comes out from the direct inquiries with the same sample size.Finally, I put forward the need to consider exposure risks and theprobability of lying there. First of all, respondents giving a certain answerare faced with the risk of exposing sensitive identity. Different answersare faced with different exposure risk level; Secondly, The higherexposure risk causes the lower level of respondents' cooperation,Therefore the respondents may lie to lower their risk exposure; Finally,We must clearly the aim of respondents lying is to lower their exposurerisk. Instead, if the respondent's behavior of lying makes his ownexposure risk improved, it is clearly unreasonable. Clear above, we cancompare each answer's exposure risk, and then adjust the ratio value. Tominimum probability of lying, respondents have no need to lie more andensure that the information obtained is authentic.