← 返回起始笔记位置

知识卡片

普通读书笔记卡 · 6592.d · 生成模型:GPT-6

安慰剂测试须说明零效应

理解等级:中级

打乱或重建处理变量,破坏其原有因果联系后,期望估计接近预设零效应。须核对置换单位、依赖结构与具体零假设;测试失败提示流程问题,但与其他估计器比较也不能唯一诊断是哪一项统计假设出错。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      11.5.3 Replacing treatment with a dummy We can also experiment with replacing the treatment variable with a dummy variable. This is analogous to giving our causal effect inference workflow a “placebo,” and seeing how much causality it ascribes to this fake treatment. Since this dummy variable will have no effect on the treatment, we expect the ATE to be 0. Let’s try this with our

    参考来源

    - 位置:《Causal AI》第11章《构建因果推断工作流》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Causal AI/_epub-src/OEBPS/Text/chapter-11.xhtml) - 结论依据:摘录中关于安慰剂测试须说明零效应的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:11.5.3 Replacing treatment with a dummy We can also experiment with replacing the treatment variable with a dummy variable. This is analogous to giving our causal effect inference workflow a “placebo,” and seeing how much causality it ascribes to this fake treatment. Since this dummy variable will have no effect on the treatment, we expect the ATE to be 0. Let’s try this with our