Randomized experiments can provide unbiased estimates of sample average treatment effects. However, estimates of population treatment effects can be biased when the experimental sample and the target population differ. In this case, the population average treatment effect can be identified by combining experimental and observational data. A good experiment design trumps all the analyses that come after. While most of the existing literature centers around improving analyses after RCTs, we instead focus on the design stage, fundamentally improving the efficiency of the combined causal estimator through the selection of experimental samples. We explore how the covariate distribution of RCT samples influences the estimation efficiency and derive the optimal covariate allocation that leads to the lowest variance. Our results show that the optimal allocation does not necessarily follow the exact distribution of the target cohort, but adjusted for the conditional variability of potential outcomes. We formulate a metric to compare and choose from candidate RCT sample compositions. We also develop variations of our main results to cater for practical scenarios with various cost constraints and precision requirements. The ultimate goal of this paper is to provide practitioners with a clear and actionable strategy to select RCT samples that will lead to efficient causal inference.
Using a Two-Parameter Sensitivity Analysis Framework to Efficiently Combine Randomized and Non-randomized Studies
University of Illinois Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校);University of Wisconsin-Madison(威斯康星大学麦迪逊分校);University of California Davis(加州大学戴维斯分校);University of California San Francisco(加州大学旧金山分校)
Estimating Individualized Treatment Effects in Acute Ischemic Stroke with Causal Transformation Models (TRAM-DAG): A Multi-Centre Observational Study with External RCT Validation
Thurgau Institute for Digital Transformation (TIDIT)(图尔豪数字转型研究所);HTWG Konstanz – University of Applied Sciences Konstanz(康斯坦茨应用科学大学);Department of Neurology, University Hospital Zurich (USZ)(苏黎世大学神经科部);University of Zurich (UZH)(苏黎世大学)
Observing the unobserved confounding through its effects: toward randomized trial-like estimates from real-world survival data
Operations Research Center, Massachusetts Institute of Technology(麻省理工学院运营研究中心);Sloan School of Management, Massachusetts Institute of Technology(麻省理工学院斯隆管理学院);Department of Surgery, Memorial Sloan Kettering Cancer Center(纪念斯隆凯特琳癌症中心外科部);Charité-Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin and Humboldt-Universität zu Berlin(柏林夏里特医学院,柏林自由大学和洪堡大学联合成员)
Finite-sample bias-variance tradeoff with variables related to trial participation inserted into causal forest models for ensuring generalizability
Division of Preventive Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School(预防医学系,内科,布莱根妇女医院和哈佛医学院);Department of Epidemiology, Graduate School of Medicine, Dentistry and Pharmaceutical Sciences, Okayama University(流行病学科,医药齿学研究院,冈山大学);Harris School of Public Policy, University of Chicago(哈里斯公共政策学院,芝加哥大学)
Budgeted Active Experimentation for Treatment Effect Estimation from Observational and Randomized Data
Didi Chuxing(滴滴出行);School of Statistics, East China Normal University(华东师范大学统计学院);School of Mathematics and Statistics, Beijing Jiaotong University(北京交通大学数学与统计学院);School of Computing and Information Technology, Great Bay University(大湾区大学计算与信息技术学院);School of Statistics and Data Science, Shanghai University of Finance and Economics(上海财经大学统计与数据科学学院);Institute of Data Science and Statistics, Shanghai University of Finance and Economics(上海财经大学数据科学与统计研究院)
Minimax Rates and Adaptivity in Combining Experimental and Observational Data
随机对照试验(RCT)是评价治疗因果效应的金标准;然而,它们通常具有有限的样本量,有时概括性较差。另一方面,来自大型管理数据库的非随机、观察性数据具有大的样本量和更好的概括性,但它们容易产生不可测量的混杂偏倚。因此,协调从随机对照试验和观察性研究中获得的效果估计值是相当有意义的,这些研究调查了相同的干预措施,可能会从两个领域中获得最佳结果。在本文中,我们从理论上描述了从极大极小的角度将观测数据整合到基于RCT的分析中的潜在效率增益。对于估计,我们推导出的最小最大收敛速度的均方误差,并提出了一个完全自适应锚定阈值估计,达到最佳的速度多对数因子。对于推断,我们的特点是极大极小率的长度置信区间和适应(未知的混杂偏差)一般是不可能的。一个奇怪的现象,从而出现:估计,从数据集成的效率增益可以实现没有先验知识的混淆偏差的大小;推理,同样的任务成为信息理论上不可能的一般。我们使用模拟和来自RCT DUPLICATE倡议的真实数据示例证实了我们的理论发现[Franklin et al.,2021 b]。
Covariate Adjustment in Randomized Experiments Motivated by Higher-Order Influence Functions
Shanghai Jiao Tong University(上海交通大学);SJTU-Yale Joint Center for Biostatistics and Data Science, Shanghai Jiao Tong University(上海交通大学-耶鲁大学生物统计与数据科学联合中心);Pfizer Inc(辉瑞公司)
Augmenting Limited and Biased RCTs through Pseudo-Sample Matching-Based Observational Data Fusion Method
Zhejiang University(浙江大学);Institute for Clarity in Documentation(清晰文档研究所);Inria Paris-Rocquencourt(Inria巴黎-罗克琴特研究所);Rajiv Gandhi University(拉贾·甘地大学);Tsinghua University(清华大学);Palmer Research Laboratories(帕勒尔研究实验室)
Class flipping for uplift modeling and Heterogeneous Treatment Effect estimation on imbalanced RCT data
Institute of Computer Science, Polish Academy of Sciences(波兰科学院计算机科学研究所);Faculty of Mathematics and Information Science, Warsaw University of Technology(华沙理工大学数学与信息科学学院)
Generalized Encouragement-Based Instrumental Variables for Counterfactual Regression
College of Computer Science and Technology, Zhejiang University(浙江大学计算机科学与技术学院);Emory University(埃默里大学);Gaoling School of Artificial Intelligence, Renmin University of China(中国人民大学高瓴人工智能学院);Carnegie Mellon University(卡内基梅隆大学)
Trustworthy AI/ML Regression and Unbiased Causal Inference for Real-World Data
University of Maryland(马里兰大学);Maryland Psychiatric Research Center(马里兰州精神病学研究中心);The University of Maryland Institute for Health Computing (UM-IHC)(马里兰大学健康计算研究所)
Experiment-selector cross-validated targeted maximum likelihood estimator for hybrid RCT-external data studies
Novo Nordisk(诺和诺德);Division of Endocrinology, Department of Medicine, University of North Carolina, Chapel Hill(北卡罗来纳大学教堂山分校医学院内分泌科);Department of Biostatistics, University of California, Berkeley(加州大学伯克利分校生物统计系);Biostatistics Research Branch, National Institute of Allergy and Infectious Diseases(国家过敏和传染病研究所生物统计研究分支)