借助优势比信息识别不可忽略缺失数据的模型参数

Identification of graphical models for nonignorable nonresponse in terms of information on odds ratios

由不可忽略缺失机制引起的缺失数据,常使得模型变得不可识别。对于那些不可识别的模型,可以通过添加协变量和借助其他来源的外部数据来达到识别的目的。本文探讨不可忽略缺失机制下,利用外部获得的优势比估计,来达到识别联合概率的方法。

In the case with missing data from a mechanism of nonignorable nonresponse, a graphical model usually is not identifiable. In this paper we present a method for identiying graphical models in terms of external information on odds ratios.