층화 조건부 이산양적 확률화응답모형

Abstract

When the population is composed of several strata and the information we want to estimate is a sensitive discrete quantitative character we propose a stratified conditional discrete quantitative randomized response model using less than sensitive character. First, we apply stratified sampling to the model of Carr, Marascuilo.(1982) which used less than sensitive character and look the results of obtaining sensitive qualitative information. Then, we applly stratified sampling to the model of Lee, Hong, Son.(2009) which used less than sensitive character as a condition to obtain information for a sensitive discrete quantitative character and establish a theoretical system for obtaining quantitative sensitive information by stratum. We dealt with the proportional and optimal allocations are examined as a method of allocating samples to each stratum. We also propose a stratified Liu-Chow model by applying stratified sampling and compare the efficiency between the suggested two models. Finally, in case of numerical comparison between two models, we find that the values of ph and th are more increasing the stratified conditional discrete quantitative randomized response model is the more efficiency than the stratified Liu-Chow model.