#0568 · 方法
相关不等于因果,先排三关
看到两个指标一起变,先依次排查三关:
1. 有没有共同的原因(第三个变量);
2. 因果方向会不会反过来;
3. 是不是样本筛选造成的假象。
三关都过了,才允许写「导致」,否则只能写「相关」。
When two metrics move together, check three things in order:
1. Is there a common cause (a third variable)?
2. Could the causation run the other way?
3. Is it an artifact of how the sample was selected?
Only if all three are ruled out may you write "causes"; otherwise write "is associated with".
小杜
已贴 1,040 张上墙共收到 0 个章