2010년 2회 통계세미나 개최 안내 통계연구소에서는 다음과 같이 통계 세미나를 개최하오니 많은 참여 바랍니다. 일시: 2010년 3월 24일(수) 오후 5시 장소: 고려대학교 교양관 111호 연사: Hanna Yoo (Institute of Statistics , Korea university) “Joint model of Clustered Failure Time Data With Informative Cluster Size” In many toxicity experiments, offsprings from the same mother share common characteristics and are analyzed in the context of clustered data. f their mother mice suffered from the toxicant as a treatment, two outcomes related with offspring must be taken into account. The first is low fetal resorption resulting a low birth rate and the second is the transition of toxin to the litters which affects their survivals. Therefore, the smaller size of offsprings tend to have shorter survival times. This phenomenon occurs when cluster size is related with response variable and is termed as an informative cluster size. Another example is found at dental study where response variable of interest is teeth survival times from initiation of therapy (Spiekerman and Lin (1998), Cong et al. (2007)). In their study it is shown that survival distributions were different according to the number of the teeth. That is, groups with more teeth tend to have longer teeth survival times. The ignorance of this phenomenon can cause biased results in statistical inference. Our interest is to consider a joint model of the clustered failure time data and cluster size. We utilize a random effect to connect these two processes. Simulation studies are performed to assess the finite sample properties of the estimators and a lymphatic filariasis study is analyzed as an illustration. 고려대학교 통계연구소 |