In the discussions about gun control following last week’s massacre at Sandy Hook School, I find myself strangely resistant to the “let’s just go ban the darned things” approach. Let me say that I’m all for rethinking how access to guns is regulated and such, particularly in terms of having arms available via collectivities such as militias, rather than an individual right of citizens. But I instinctively feel that that the just-ban-them response is the kind of simple-sounding public policies that may well intentioned but ultimately have far-reaching unintended consequences.
Barry Ritholtz’s website has an interesting chart that showing the relationship of gun ownership rates in industrialized countries to the firearm homicide rate. I don’t have the raw data to reproduce the chart here, but I do include a link:
A larger version of the graphic can be found here: http://www.ritholtz.com/blog/wp-content/uploads/2012/12/guns.jpg
The surrounding text is here: http://www.ritholtz.com/blog/2012/12/10-tuesday-am-reads-48/
The chart illustrates how higher gun ownership rates go hand in hand with higher firearm-related homicides. But there are two issues with this chart that are worth thinking about.
The first is a technical issue: the axes should be flipped. If we want to show that increased firearm possession leads to increased gun-homicide, gun possession should be on the horizontal axis, not the vertical (by convention). The way this reads, it looks like the chart is arguing that high homicide rates lead people to want to own more guns. This is actually a plausible interpretation of this data, as well, but probably not the way the author intended.
More importantly, the US is an outlier here, both in terms of gun deaths and in terms of gun ownership. The US’s presence in the chart actually skews the regression line in a noticeable way.
If we were to remove the US from the sample and re-run the regression, the line would be much steeper and probably explain more variance (i.e. a higher R^2). It would also suggest that gun ownership has a much smaller effect – though still a real one – on gun homicides than the chart currently suggests.
I don’t have access to the raw data, so I can’t run the regression myself, but I have tried to hand-draw the regression line to illustrate. This is imperfect, but good enough to make the point.
This chart estimates the regression line excluding the United States. Although I may have made small errors due to the fact that I am eyeballing the line rather than computing it from the raw data, the line is much steeper.
But assuming that high gun ownership causes high gun-homicide rates, what does this line predict about the United States? It suggests that the United States is substantially more violent than what would be predicted by its gun ownership rate. By my eyeball estimate, about three times more violent. Although guns are part of the story, they are not the whole story, and – in fact – they may not even be the most important part of the story.
Yes, let’s think about how to regulate guns. But the larger question is really “Why are we so violent?” I don’t know the full answer: common culprits are cinema, video games, media glorification, the drive of excessive competitiveness, etc..
Yes, let’s have a real discussion about sensible regulation, and I don’t see any reason why assault weapons need to be available to individuals for the asking, but I think a key underlying issue is also to figure out why we are so violent in the first place.