.png){kind=icon}
返回
Generalizing Trimming Bounds for Endogenously Missing Outcome Data Using Random Forests
.png){kind=icon}
OA
AI
DOI：10.1017/pan.2025.10001.png)
.png){kind=icon}
.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
We present a method for narrowing nonparametric bounds on treatment effects by adjusting for potentially large numbers of covariates, using generalized random forests. In many experimental or quasi-experimental studies, outcomes of interest are only observed for subjects who select (or are selected) to engage in the activity generating the outcome. Outcome data are thus endogenously missing for units who do not engage, and random or conditionally random treatment assignment before such choices is insufficient to identify treatment effects. Nonparametric partial identification bounds address endogenous missingness without having to make disputable parametric assumptions. Basic bounding approaches often yield bounds that are wide and minimally informative. Our approach can tighten such bounds while permitting agnosticism about the data-generating process and honest inference. A simulation study and replication exercise demonstrate the benefits.
Keyword:
causal inference
trimming bounds
partial identification
machine learning
random forest
experiments
AI总结
对已上传原文的论文进行重点信息的提取，主要内容包括：简要概述、研究摘要、背景介绍、关键亮点、图文解析、展望与总结。
期刊
IF:
5.4
论文数:
79
被引数:
6.7K
机构
引用论文
Training, Wages, and Sample Selection: Estimating Sharp Bounds on Treatment Effects培训，工资和样本选择: 估计治疗效果的界限OAAI
OAAI分享收藏分享收藏Priming Bias Versus Post-Treatment Bias in Experimental Designs启动偏差与处理后偏差在实验设计中的比较.png)
POLITICAL ANALYSIS
POLITICAL ANALYSISIF5.4
PREAI
分享收藏分享收藏
查看更多内容