Empirical Study And New Design Of Survey Methods For Sensitive Issues

The modern society after the rapid development of the Internet is full of complex and ever-changing information.Social surveys can be used to collect and process social information and to understand society more scientifically and efficiently.Most social problems do not need to be convenient or convenient for all members to survey.The sample survey can be used to extract the statistical value of the sample to estimate the population,so that the results can be obtained scientifically and conveniently.Sensitivity problems are often encountered in social surveys.Sensitivity issues involve privacy.The results obtained by direct investigation are less effective and the sample non-sampling errors are large,sensitive questions require professional questioning skills during the survey.Indirect survey methods are mainly divided into: randomized response techniques and non-randomized response techniques.Professional sensitivity survey methods can ensure the anonymity of respondents in surveys and improve the effectiveness of survey results.The randomized response technology has been developed very maturely.A large number of scholars have improved it.From the three aspects of the model itself,parameter estimation method and sampling method,the sensitivity of the survey is greatly reduced,and the authenticity and effectiveness of the survey results are improved.At the same time,randomized response technology has also been widely used in the investigation of various sensitive issues.Non-randomized response technology has been developed in the last decade.This technology does not require a randomized device.This feature makes it superior to randomized response technology in survey use,but the academic development of non-randomized response technology is still not enough.At present,it mainly focuses on the improvement of model parameter estimation method.Firstly,based on the two basic models of randomized response technology and the four models of non-randomized response technology,the six survey models commonly used in sensitivity problems are deduced in detail.Then two different sampling methods,stratified sampling and cluster sampling,are introduced into the four models of nonrandomized response technology respectively.Finally,the development of network survey and data quality are introduced,and the sensitivity problem is introduced.In the survey,an improved non-randomized response model was selected and combined with the network survey to implement the survey process on the computer using Monte Carlo simulation.Compared with the non-randomized response model under simple sampling,the non-randomized response technique under stratified sampling is more suitable for investigating the case where the overall situation is more complicated and the difference between units is larger,which can greatly improve the accuracy and efficiency of the investigation.The cluster sampling is suitable for many complicated investigation scenarios.Some scenarios can only select cluster sampling to conduct investigations,such as review after census.Therefore,the research of non-random response technology under cluster sampling is also necessary.In the empirical study,this paper uses the variant parallel model under stratified sampling to study the proportion of people who are relapsed among drug users,collects previous survey data to carry out Monte Carlo simulation,and successfully realizes the actual model under investigation.The process can show the situation under repeated investigations.The work of this paper provides a reliable and effective theoretical reference and guiding principle for the further promotion of sensitivity survey methods.