A couple of weeks ago, I got pulled into a Twitter argument about the following graph:
This graph, which conveniently does not include the United States (which would pull the trend line closer to level) and which does not take into consideration different cultures, different population sizes, or any countries you would call war torn, posits that more guns equals fewer murders. (Not to mention it assumes something as complex as the interaction of gun ownership on human activity must be a linear relationship.) I don’t buy it, but then, there’s nothing really to buy. Yes, with this sample, there is a trend. I could cherry pick my own countries and generate a line in the other direction. This, however, doesn’t advance the debate.
I really think the biggest factor that this graph ignores is culture. Japan is a monolithic culture, England and France are awash in immigrants, and Estonia was part of the USSR for fifty years there. (BTW, where is Russia on this graph?)
So what about the US? We are (red/blue arguments aside) a pretty reasonably homogenous group. Or at least, homogenously heterogeneous, if you get my meaning. So I wanted to evaluate these statistics using the various states of our fair Republic. If you graph gun ownership percentage against murder rate you get this:
Wow. It doesn’t get any flatter than that. This indicates there is essentially ZERO correlation between gun ownership and murder. Good to know.
But wait. Something doesn’t feel right about this analysis either. There is something substantively different between Alaska and Connecticut. That’s population density. I wonder if there’s any correlation between the density of gun owners and murder rate? In other words, does simple proximity between gun owners increase the likelihood of murder?
Awesome! Point proven. More guns in a smaller area increases the murder rate.
Wait. No. That doesn’t prove anything. Just living cheek by jowl with our fellow humans might, guns aside, have an impact on the murder rate, right? If we look at generic population density against murder rate, it’s probably just as much of a trend, probably more. Let’s find out…
What do you know. Slight trend downward. So, close proximity to humans is slightly less conducive to murder… unless there are guns involved.
(Full disclosure. The murder rates I got for 2008 from the Justice Department. Gun ownership data I got from http://www.uscarry.com, based on 2007 data. Size and population data for the states courtesy Wikipedia. If anyone has any better resources, I’d love to crunch the numbers. Also, any gun ownership data on DC would be instructive, I think, since it’s obviously both highly dense, and highly murdery.)
brenthawthorne
David Gaw
reviewsbyrussell
Rip Rowan
Cuz We Said So 
It’s a slippery slope to implicate the mere presence of an inanimate object to the murder rate, even if the object can be used for murder. It completely removes the human element and thus the culpability of the humans.
I think it would be interesting to see how the murder rates compares to other human-oriented statistics similar to what you did with the density comparison. Perhaps: per capita income, education levels, size and type of minority populations, gang activity, rates of other violent and non-violent crime, etc.
And, of course, there are loads of other intangible characteristics that are a lot harder to quantify and chart (such as respect for fellow man, etc) but I think would also have a distinct impact on an area’s murder rate.
After all, it’s human holding the gun that I worry about – not the gun itself.