Stats and guns

In my travels around the World Wide Web the other morning, I came across this:

Rob is very explicit that there is no positive correlation between violent crime and liberalized carry laws rather than a proven negative correlation. He feels that it will take more rigorous statistical analysis before this negative correlation could be said to be proven. [my emphases]

[I'm not sure I like the use of the word "proven." But that's a nit to pick another time.]

Everyone knows statistics lie, right? Wrong. Statistics are numbers. Numbers do not have brains and thus cannot tell a lie. Human beings, on the other hand, can. Humans can-- and do-- manipulate numbers, images, words, and so on in order to lie. What I like very much about Rob-- and I'm assuming Rob is a gun-friendly guy-- is that he's careful about what these numbers do and do not show.

US violent crime rate vs. % US population living in shall issue and Constitutional carry states

The quote and graph come from "Every Picture Tells a Story, Part Two" at No Lawyers, Only Guns and Money (h/t Instapundit).

What these numbers show is exactly what Rob says they show: no positive correlation between violent crime rates and liberalized, i.e., less restrictive, gun laws.

A positive correlation is a relationship between two sets of numbers such that as one set increases, in this case over time, so does the other. A negative correlation is the opposite, as one increases the other decreases. Note that in both sorts of correlations, neither set of numbers is necessarily "more important" than the other. In the abstract (without time), if there is a positive correlation between A and B we can say

As A increases B increases,

and we can also say

As B increases A increases.

It doesn't matter that A come before B in the alphabet!

Likewise, we can just as well say that these data do not show states' gun laws becoming less restrictive as violent crime rates have fallen. Either way, these data do not show a positive correlation between A and B. 

Assuming that more rigorous statistical analyses do reveal a (strong) negative correlation between violent crime rates and increased liberalism (less restriction) over states' gun laws, what does that mean? It means there's a (strong) negative correlation. Period.  No one-- including Rob it I'll bet-- would should then say that less restrictive gun laws cause violent crime rates to go down. Correlation is not causation. Duh. To put it another way, no one should now say violent crime rates went down because more people were able to get concealed carry licenses.

I say "should" because some will say that. And they'd be wrong, given these data. Does it make sense to say that armed potential crime victims are better able to ensure they are not actual victims. Sure! 

But think what data we'd need to back up the very much stronger claim that less restrictive gun laws, and therefore presumably more gun carriers, cause crime reduction. First, we'd have to talk about the presumption. Rob's project, the first part of which I've talked about before, deals with crime and law. He's right to proceed carefully and rigorously. 

~~

P.s. Sometimes correlation is just fine without causation.