LOGIN ENGLISH

제목 [학술세미나] [학과세미나] 12월 5일(화) 17시 학과세미나 안내
작성일 2017-11-29 15:42:05
내용 12.5(화) (17:00-18:00) 통계학과 세미나를 안내합니다.

(기존에 예정되었던 12.1(금) 세미나는 수시고사 건물통제로 취소되었습니다.)  

▪제목 : Bayesian Analysis of Sensitive Small-Area Proportions

▪연사 : Balgobin Nandram (Department of Mathematical Sciences, Worcester Polytechnic Institute)

▪일시 : 2017년 12월 5일(화) PM 17:00 – 18:00

▪장소 : 25동 405호

[초 록] 
In sample surveys with sensitive items, sampled units may not respond to these items, creating non-ignorable non-response, or they respond untruthfully. Usually a negative answer is given when it is actually positive, thereby leading to an estimate of the population proportion of positives (sensitive proportion) that is too small. We will review some of the survey designs that address this problem and will show why the Bayesian paradigm is attractive. In our application on college cheating, we have binary data from clusters within small areas, obtained from a version of the unrelated-question design with at least two different random mechanisms in each area. A hierar-chical Bayesian model is used to capture the variation in the observed binomial counts from the clusters within the small areas. We have shown that the joint posterior density of all the parameters, including latent variables, is proper and obtained an efficient blocked Gibbs sampler. In our example on college cheating and a simulation study, we have seen significant reductions in the posterior standard deviations of the sensitive proportions under the small-area model as compared to an analogous individual-area model. Surprisingly, the simulation study demonstrates that the estimates under the small-area model are closer to the truth than for the corresponding estimates under the individual-area model. Finally, we discuss many extensions to accommodate covariates, multiple sensitive items, optional response, finite population sampling and selection bias.
KeyWords: Blocked Gibbs sampler, Hierarchical Bayesian model, Latent variables, Non-identifiable parameters, Posterior propriety, Rao-Blackwellized estimates, Unrelated-question design.
 
파일 세미나 안내_171205_A4.hwp [16KB]