Saturday, December 06, 2008

Lying With "Statistics" on Gun Laws

You may have read LeftyRowen's blog announcing this study; here's a brief refutation.

...The study compares the average crime rates for states with and without certain laws and claims that for certain states the crime rates are lower in states with certain regulations. Many of these laws I have looked at myself using panel level data (e.g., requiring background checks on the private transfer of guns), and I couldn't find an effect.

Let me give you a simple example of the problems with using purely cross-sectional data as used in this study. Suppose that high crime rate cities are the ones that adopt the strictest gun controls (which is generally true) and that crime rates fell (just for the sake of argument assume that is the case), but the drop is not enough--so that the crime rate in these cities is still higher than the low crime rate cities that didn't adopt the laws. Looking across the cities it would appear that the cities with the gun control laws had the higher crime rates, but does that mean that gun control caused higher crime? No. Obviously in the example that we gave the opposite really happened, but looking at simple averages gave a very misleading result. The only way that you can figure out what happened is to follow those cities over time and see how their crime rates changed relative to what they were in the past and to see how they changed over time relative to the places that didn't change their laws...

In other words, it's like quality control. You have to examine each element of a given process over time to determine the actual effects of changing a given element. (It's also like sighting in a rifle; one uses "groups" of shots to adjust sights, not just one shot, to establish a valid sample, before adjusting the sights.)

Let's face it. That study was commissioned by the Bloomberg Group of Mayors (Tom Barrett among them.) The results were what those Mayors wanted to see--and if that means that the methodology was flawed, who cares?

Not the MSM.


Scott said...

What if crime rates dropped in areas where concealed carry was instituted? Would that indicate that such policies had a positive influence on crime? Not so fast. Perhaps crime was dropping in general--for all areas. One would have to factor in larger trends like that before making a claim.

Me, I've never seen any evidence at all that marginal gun policies do anything for or against crime or violence rates.

Dad29 said...

That 'perhaps crime was dropping in all areas' supposition would certainly invalidate claims that CCW reduces crimes.

On the other hand, there is Chicago--and DC. They enjoy a high crime-rate AND a ban on guns (albeit DC's has been overturned.)