While I generally agree with your broader point, I doubt this relationship is causal.

Agreed. It's entirely possible that high murder rates drive support for the death penalty, for example.

magashna: Punishment is not a deterrent. rbut: Punishment is a deterrent and has been proven ... magashna:murder rates are consistently higher in states with the death penalty. joshuamorton: I doubt this relationship is causal.

Magashna is clearly not arguing causality, but rather that the death penalty has failed to be a deterrent to murder.

This doesn't show that though. You have to examine the counterfactual: would the crime rate be even higher without the death penalty? If yes, it is a deterrent.

How can one examine a couterfactual in cases like this? Barring access to Marty McFly’s time machine unless you have two identical copies of a state to experiment on how would you go about this?

Edit: legitimately curious, not trying to be nitpicky

You probably can't, which is my point.

Generally the best ways to analyze the impact of similar social changes is to compare two similar locations, one that makes a change and one that doesn't, over the same time period, and hope that controls for broader social change.

It's not perfect by any means though.

A confounding factor in this statistic is that even if you were a rational actor, weighing the deterrence value of the death penalty, the actual application of the death penalty is far from certain, so you'd appropriately assign it low probability. The distance from the present in time, and absolute probability are likely key contributors to this not being a good stat. A good look at deterrence would be Singapore, where probability of swift punishment seems to materially reduce crime.