Your browser does not support JavaScript!
2016-12-23(五),主講人:黃名鉞 博士 (Department of Biostatistics, University of Washington)


統 計 學 研 究 所

專 題 演 講


講 題: Sufficient Dimension Reduction in Causal Inference

黃名鉞博士 (Department of Biostatistics, University of Washington)

時 間: 105年12月23日(星期五)10:40 - 12:00 noon (10:20 - 10:40 茶會於統計所821室舉行)
地 點: 綜合三館837室
摘 要: Investigating the causal effect of a treatment on an outcome is often the primary interest in medical and social studies. While the estimation of average treatment effects usually involves multivariate confounders, dimension reduction is often desirable and sometimes inevitable. In this talk, I first consider the Neyman-Rubin model and clarify the definition of a central subspace that is relevant for the efficient estimation of average treatment effects. This approach is also extended to the case with a continuous treatment. Second, when non-ignorable selection is present, instrumental variables are often used to nonparametrically identify treatment effects under various assumptions. To reduce the effect of curse of dimensionality, nonparametric identification of a variety of treatment effects under different sufficient dimension models are studied. Maximal dimension reduction for attaining efficiency is also studied for a binary instrumental variable, and is extended to multivariate, general instrumental variables. In practice, a cross-validation type estimation criterion is proposed to simultaneously estimate the structural dimensions, the basis matrices of the proposed central subspaces, and the optimal bandwidths for estimating the conditional treatment effects.


敬請公佈                                 歡迎參加

  • 友善列印
  • 分享
Voice Play