A Proxy Variable View of Shared Confounding
A Proxy Variable View of Shared Confounding
复制标题
共享混杂的代理变量视图
DOI：
--
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
D. Blei
中科院分区：
文献类型：
--
作者：
Yixin Wang;D. Blei
来自观测数据的因果推断可能会受到未观察到的混杂因素的影响。混杂因素--既影响治疗又影响结果的变量--在两者之间产生虚假的非因果关联。在没有附加条件的情况下，未观察到的混杂因素通常使因果量难以识别。在本文中，我们着眼于有多个处理共享混杂的情况，并研究在什么条件下因果识别是可能的。关键的观察结果是，我们可以将治疗的子集视为未观察到的混杂因素的代理，并确定其余治疗的干预分布。此外，虽然现有的代理变量识别公式涉及求解积分方程组，但我们表明，通过直接对数据进行建模，可以绕过对此类解的需要。最后，我们将这些结果推广到具有其他混杂因素和选择变量的因果图的扩展类。
Causal inference from observational data can be biased by unobserved confounders. Confounders—the variables that affect both the treatments and the outcome—induce spurious non-causal correlations between the two. Without additional conditions, unobserved confounders generally make causal quantities hard to identify. In this paper, we focus on the setting where there are many treatments with shared confounding, and we study under what conditions is causal identification possible. The key observation is that we can view subsets of treatments as proxies of the unobserved confounder and identify the intervention distributions of the rest. Moreover, while existing identification formulas for proxy variables involve solving integral equations, we show that one can circumvent the need for such solutions by directly modeling the data. Finally, we extend these results to an expanded class of causal graphs, those with other confounders and selection variables.
具有功能混杂因素的因果估计。
DOI：
--
发表时间：
2020
期刊：
Advances in neural information processing systems
影响因子：
--
作者：
Puli,Aahlad;Perotte,AdlerJ;Ranganath,Rajesh
通讯作者：
Ranganath,Rajesh
用未衡量的混杂因素的代理变量识别因果效应。
影响因子：
2.7
作者：
Miao W;Geng Z;Tchetgen Tchetgen E
通讯作者：
Tchetgen Tchetgen E