What

Saw this tweet

About this paper

Difference between controlling for confounders and mediators

Notes

Can I use this as content for a video? Creating examples of “controlling” for the different items.

Graphics?

• The regression validity graphs
• Spreadsheet?

Potentially useful in criticizing other work

“controlling for an inappropriate variable can result in biased causal estimates”

• Not just that it is wrong, but it makes conclusions worse than not controlling for it.

“X is a cause of Y when an intervention on X (e.g., setting X to a particular value) produces a change in Y”

flowchart TD
	x[x]
	c[c]
	y[y]

	c --> x
	c --> y
	x --> y

c causes x, x causes y, and c also causes y.

“ it is also easy to add bias to an estimate by controlling for a variable that is either not a confounder or does not block a confounding path”