在醫學與社會領域中，許多研究不只考慮處理 (Treatment) 對結果 (Outcome) 的因果效應，也開始著重於中介變數對結果的因果效應。主分層是一個處理中介變數因果效應的機制，其概念是藉由定義聯合潛在中介變數，分成子群計算群內的因果效應。

然而，隨著科技技術的快速進展，使得我們面臨高維度的困境，因此進行變數選取是必要的手段。本論文聚焦於高維度主分層因果分析的問題。對於高維度的資料，我們首先使用三階段模型選擇的方法 CGA+HDIC+Trim 分別對三個模型進行模型選擇，選擇出相關變數，再利用準確率(Accuracy)、幾何平均數(G-mean)、曲線下面積(Area under curve) 三個指標進行投票，票數較高表示選模結果較好，最後使用此模型當最終模型，將其帶入三重穩健估計式中，得到估計主分層因果效應。為了驗證本論文方法的有效性，我們提供多個模擬以及一筆實際資料作印證。

In medicine and sociology, much research not only concerns the average treatment effect on the outcome but is also interested in the underlying mechanism through an intermediate variable. Principal stratification is such a mechanism to deal with an intermediate variable by targeting subgroup causal effects within the principal strata defined by the joint potential values of an intermediate variable. In addition, due to the rapid development of science and technology, we are inevitably facing high-dimensional dilemmas. This thesis aims to provide a high-dimensional principal stratification causal inference to tackle this challenge. We first use a three-stage model selection method (CGA+HDIC+Trim) to choose relevant variables. Then, we use three indices, accuracy, G-mean, and the area under the curve, to evaluate whether the model selection result is satisfactory. Moreover, we use a multiply robust estimator to estimate the principal stratification causal effect. Finally, in support of the usefulness of the proposed method, we provide simulations and an application for the analysis of longitudinal survey data.

Chen, Y.-L., Dai, C.-S., and Ing, C.-K. (2019). High-dimensional model selection
via chebyshev greedy algorithms. Working paper.
Ding, P. and Lu, J. (2017). Principal stratification analysis using principal scores.
Journal of the Royal Statistical Society: Series B (Statistical Methodology),
79(3):757–777.
Frangakis, C. E. and Rubin, D. B. (2002). Principal stratification in causal inference.
Biometrics, 58(1):21–29.
Gilbert, P. B. and Hudgens, M. G. (2008). Evaluating candidate principal surrogate
endpoints. Biometrics, 64(4):1146–1154.
Ing, C.-K. and Lai, T. L. (2011). A stepwise regression method and consistent
model selection for high-dimensional sparse linear models. Statistica Sinica,
pages 1473–1513.
Jiang, Z., Ding, P., and Geng, Z. (2016). Principal causal effect identification and
surrogate end point evaluation by multiple trials. Journal of the Royal Statistical
Society: Series B (Statistical Methodology), 78(4):829–848.
Jiang, Z., Yang, S., and Ding, P. (2020). Multiply robust estimation of causal
effects under principal ignorability. arXiv preprint arXiv:2012.01615.
Ning, Y., Sida, P., and Imai, K. (2020). Robust estimation of causal effects via a
high-dimensional covariate balancing propensity score. Biometrika, 107(3):533–
554.
Rubin, D. B. (2006). Causal inference through potential outcomes and principal
stratification: application to studies with” censoring” due to death. Statistical
Science, pages 299–309.