International Society of Science and Applied Technologies |
|
Hierarchical Bayesian Modeling Estimation Method for Inferring the Population Proportion of Having a Sensitive Nature | ||||
Author | Hua Xin
|
|||
Co-Author(s) | Jianping Zhu; Tzong-Ru Tsai; Chieh-Yi Hung; Minzheng Jiang
|
|||
Abstract | In this study, a three-item randomized response design and a hierarchical Bayesian modeling (HBM) estimation procedure are proposed. A hybrid Gibbs sampling and Metropolis-Hastings algorithm is provided to obtain a reliable Bayes estimate of the population sensitive-nature proportion when its true value is small and closed to zero. Monte Carlo simulations were conducted to verify the performance of the proposed HBM estimation procedure. Comparing with existing counterparts, the proposed HBM estimation procedure is simple for use with less subjective assumptions. A data set regarding the homosexual proportion of college freshmen is used for illustration.
|
|||
Keywords | Bayesian estimation, Beta-Binominal model, Maximum likelihood estimation, Respondent protection, Randomized response. | |||
Article #: RQD26-59 |
Proceedings of 26th ISSAT International Conference on Reliability & Quality in Design |