1
$\begingroup$

In this video, I can see that confounding variable is a variable that is correlated with two other variables:

enter image description here

But this image tells that confounding variables is causally related to other two variables:

enter image description here

So is confounding variable defined as something that is correlated or causally related?

$\endgroup$
1
$\begingroup$

A variable whose value would be correlated with, but not causing any response in a target value is one that is caused by your target. So, correlation seems like a poor criterion for a confounder.

Every causal confounder will be correlated with the target - more or less by definition - so chances are the video creator either phrased things poorly or didn't think it through.

$\endgroup$
2
  • $\begingroup$ Thanks! So the correct definition is the one on the second picture, right? $\endgroup$ Dec 18 '19 at 10:53
  • 1
    $\begingroup$ I would say so; I don't like the fact the arrows only go from the confounder there though. They can go in both directions as well (bidirectional causation), just as long as they don't only go from the target to the confounder. $\endgroup$
    – jkm
    Dec 18 '19 at 11:01

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.