Identification of Causal Effect with the Non-Compliance and Its <i>EM</i> Algorithm

• DOI: 10.12691/ajmsm-1-4-2
• Publication Date: 2013-07-04T08:08:57Z
• Authors: Li Xiaotong; Li Sichen
• Keywords: Identification; Causal model; Graphical model

Many practical studies in biology, medicine, behavior science and the social sciences seek to establish causal relationship between treatments outcomes, rather than mere associations. In this paper, we use a graphical model describe study its identification. For an unidentifiable model, introduce covariates which are always observed into so that it becomes identifiable. We then give identifiable condition of prove mathematically. Finally, <img src=image/abs1.png></img>algorithm for average effect outcomes accepted treatment example illustrate method algorithm.