Estimating a finite population proportion bearing a sensitive attribute from a single probability sample by item count technique.

*(English)*Zbl 1365.62051
Chaudhuri, Arijit (ed.) et al., Data gathering, analysis and protection of privacy through randomized response techniques: qualitative and quantitative human traits. Amsterdam: Elsevier/North Holland (ISBN 978-0-444-63570-9/hbk; 978-0-444-63571-6/ebook). Handbook of Statistics 34, 387-403 (2016).

Summary: Randomized Response Techniques (RRTs) initiated by S. L. Warner [J. Am. Stat. Assoc. 60, No. 309, 63–69 (1965; Zbl 1298.62024)] have some disadvantages. Some participants either do not understand the RRT procedure or suspect revelation of privacy. Sometimes they end up concluding the randomization procedure as a foul trick. To overcome these problems as well as to shape the whole procedure in the form of canvassing a survey questionnaire, D. Raghavarao and W. F. Federer [J. R. Stat. Soc., Ser. B. 41, 40–45 (1979; Zbl 0453.62011)], J. D. Miller [“A new survey technique for studying deviant behavior”, Ph.D. thesis (1984), http://www.worldcat.org/title/new-survey-technique-for-studying-deviant-behavior/oclc/12768382], and J. D. Miller, I. H. Cisin and A. V. Harrel [“A new survey technique for studying deviant behavior: item count estimates of marijuana, cocain and heroin”, paper presented at the annual meeting of the American Association for public opinion research, St. Petersburg, Florida (1986)] introduced the item count technique also known as the List experiment or the Block total response or the Unmatched count technique which is user friendly. This method was revised by A. Chaudhuri and T. C. Christofides [J. Stat. Plann. Inference 137, No. 2, 589–593 (2007; Zbl 1102.62130)] which was an improvement over the original item count technique in terms of protection of privacy of the respondents drawn from the population using a general sampling design, whereas the original method was restricted to simple random sampling. But a serious disadvantage is that it requires the selection of two independent samples costing more time and money. Also it needs the knowledge of the population proportion of an innocuous characteristic unrelated to the sensitive characteristic whose proportion is to be estimated. This chapter avoids both the problems.

For the entire collection see [Zbl 1349.62001].

For the entire collection see [Zbl 1349.62001].