利用生态度量数据探索多站点试验中跨站点影响方差的来源
应用统计
2021-12-03 v1 统计计算
摘要
提出了一种新方法，用于探索社会干预多站点试验中的跨站点影响方差来源。该方法利用处理组中参与者关于处理经历的汇总报告来定义潜在中介变量，并利用对照组中参与者关于关注结果的汇总报告来消除由未测量的局部情境因素引起的混杂。特别关注纠正中介变量中由测量误差导致的偏差。给出了该方法有效性的渐近理论。给出了两个示例应用，一个为模拟的，一个为实证的。
引用
@article{arxiv.2112.01338,
title = {Using Ecometric Data to Explore Sources of Cross-Site Impact Variance in Multi-Site Trials},
author = {David R. Judkins and Gabriel Durham},
journal= {arXiv preprint arXiv:2112.01338},
year = {2021}
}
备注
Note: After reviewer comments, we will be making several changes to the narrative structure of the paper, as well as including the results of a Monte-Carlo expansion of the simulation analysis discussed in section 4.1. Updated paper to be uploaded in the coming months