What is Simpson's Paradox?

An explanation of Simpson's Paradox, a paradox for drawing conclusions about relationships in statistical analysis.

Sponsors: Joshua Furman, Roman Leventov, NBA_Ruby, Antybodi, Federico Galvão, Mike Gloudemans, Eugene SY, Andrew Sullivan, Antoinemp1, Andreas Kurz, Ismail Fagundes, Joao Sa, Ploney, Tyler James, and Dennis Sexton. Thanks for your support!

Information for this video gathered from The Stanford Encyclopedia of Philosophy, The Internet Encyclopedia of Philosophy, The Cambridge Dictionary of Philosophy, The Oxford Dictionary of Philosophy, The Oxford Companion to Philosophy, The Routledge Encyclopedia of Philosophy, The Collier-MacMillan Encyclopedia of Philosophy, the Dictionary of Continental Philosophy, and more!

[Ianoxen]

Its good that you gave the medical situation as an example since it applies to all people and that majority often accept conclusions without looking at the specifics of the data. Great video!

[miglriccardi]

Before the disaggregation, within the first 1:30 of the video, you say Treatment 1 has 10% chance survivability and Treatment 2 has 40% chance survivability but treatment 1 “is clearly preferable” for Jane. I think you mean treatment 2 is preferable. Only after the disaggregation is treatment 1 preferable for Jane.

[muhammadbardosarhamaraur]

I would love to see videos on logical positivism explained by you.

[InventiveHarvest]

Good video. I would like to see more videos like this.

While you cannot be certain that you have included all of the relevant factors in your study (as well as not including all of the irrelevant variables), there is a theoretical side to economics to help. The interplay between data and theory is how we get closer to the truth.

Economists are well aware of limitations to their studies. Through peer review as well as presentation of a study to economists all over the world, the studies get more and more robust. So often at these presentations the conclusion of the study is: "we see what we expected to see - economic theory holds in this case."

[Pfhorrest]

Simpson, eh?

Once again, critical rationalism dissolves the problem. Yes, we are never completely certain -- and nobody ever possibly could have been, so that's not something to worry about. We might always be wrong, but that doesn't make all answers *equally* wrong. In the pairs of examples you give in this video, there is one clearly superior analysis of the data and one clearly inferior one. We can never know if any given analysis is *the best possible* one, but we can know when one is better than the other, and that is all we ever needed anyway.

[dexterharvey]

Is this linked to some degree with the inductive problem? I like Nassim Taleb’s observation that the unknown can have such an impact that it’s impossible to account for, this seems similar to what you talked about here; conclusions can even be reversed. Thanks for the video

[GeorgWilde]

I like the "doctors are statistics illiterate, yet they evaluate your health risks" paradox.

[deepashtray]

This a response to the anti-vaxxing movement? Sure fits.

[Arrakiz]

The paradox is how a show so bad manages to stay on air.