Econometrics is not Science

Alex Tabarrok has posted a bit of troll bait on firearms and suicide. It’s troll bait because these studies are created to get attention. They do that by confirming some aspect of the narrative and decorating it with the veneer of science. A PhD, claiming a correlation between something lefty sees as a great evil and something that is bad, will get the the usual suspects buzzing about the author. Often, the authors of the paper will put together a cover letter for the media, so they don’t have to read the actual study.

In this case, guns and suicide is the issue. If it were a paper claiming to prove that religion makes you a bigot or lead in the water makes you hate socialized medicine, it would have a better chance of landing on TV, but guns still work in certain quarters of the Left. The paper itself is not worth reading, but the on-line debate is worth a read. The content of the debate is not all that interesting. It’s the nature of the debate. It reveals a lot about the sorts of people who are prone to accepting these sorts of studies at face value.

Everything from the soft sciences starts with statistical correlation between one thing and some other thing. The idea is to create the belief that a causal relationship exists, where only (maybe) a correlation exists. The most famous of which is the claim that marijuana use leads to hard drug use. The fact that people prone to drug addiction would use the most common drugs first is hardly surprising. Claims that weed causes heroine use are easily disproved by the tens of millions of weed smokers who never use heroine.

The statistical methodology in this paper is sound and it is hardly shocking that more guns would mean more successful suicides. Guns are a great tool for killing yourself. It is exceedingly rare for someone to survive a gun shot to the head. People often panic and call for help when taking pills or slashing their wrists. They have ample time to contemplate their act before and during its commission. That and they can always call for help after committing the act. That’s not happening after the hammer goes click.

That said, it does not show that guns cause suicide. That would mean forming metal into a particular shape has some influence over the psychological well being of humans. Only a lunatic would think such a thing. Instead, they modify the claim to, “well the availability of the gun can lead some people to rashly elect suicide when they otherwise would have time to be talked out of it.” That assumes legal gun make gun more available. If you are planning to shoot yourself, you still have to buy the gun and that takes time.

In real science, cause and effect are tested. In fake science like this, cause and effect are inferred or implied, depending on the claim. In this case, the implication is guns cause suicide as the author goes out of his way to make the claim that other forms of suicide declined in his study. Even if the observation is accurate, the way it is framed is intended to lead people to a causal relationship. It’s not an outright lie, but it is dishonest, because the intent is to convince people of soemthing that is false.

That’s the issue with the soft sciences. They are too prone to these sorts of shenanigans by the Left. Much of it never replicates and much of it is politically motivated. The result is nutty ideas are lacquered with the respectability of science, despite being ridiculously wrong or misleading. That’s not to say there is no value in discovering statistical correlations. It’s just that it is a starting point to begin thinking about causal relationships and more complex statistical relationships.