Friday, May 08, 2009

How's it go again? "If you outlaw guns, only ...?"

I'm kind of meh on the whole gun control thing these days, but if I were still a staunch advocate of gun control, I'd be waving these numbers in your face:

Hawaii, with its strong gun laws and low rate of gun ownership, has the lowest gun death rate in the nation, the Washington D.C.-based gun control group Violence Policy Center said Wednesday.

The islands had a per capita gun death rate of 2.58 per 100,000 people in 2006, according to the center's analysis, based on the most recently available national data from the U.S. Centers for Disease Control and Prevention's National Center for Injury Prevention and Control. The national average was 10.3, it said.

Only 9.7 percent of Hawaii households own guns, compared with 45.6 percent in Louisiana, which topped the nation in per capita gun deaths at 19.5, the center said.


Louisiana was followed by Alabama (57.2 percent household gun ownership, 16.9 deaths), Alaska (60.6 percent, 16.3), Mississippi (54.3, 16.3) and Nevada (31.5 percent, 16.2).

Each of the five states had a per capita gun death rate far exceeding the national rate of 10.3 per 100,000, and each state has weak gun laws and higher gun ownership rates.

In contrast, Hawaii and other states with strong gun laws and low rates of gun ownership had far lower rates of firearm-related death.

Massachusetts followed Hawaii with household gun ownership of 12.8 percent and a gun death rate of 3.2, ahead of Rhode Island (13.3 percent, 4.4), Connecticut (16.2 percent, 4.9) and New York (18.1 percent, 5.2).

Something to think about.

Stand by for a ream of contradicting statistics from Adam L in 5..., 4..., 3...

(h/t: Attaturk)


Adam L said...

I actually don't disagree with the statistical analysis.

I was looking for factors that had a good R^2 value when compared state by state with murder rates since I wrote a script to automatically generate 2-dimensional continuous information to draw information about a state.

I looked at gun control laws... R^2 was about 1.3%; poverty, which I thought would be a massive hit, was only 17%, which was surprisingly low to me: the title of this post is a joke, by the way. And then, in a post that I had to couch as being a fake racist post that was stupid and I won't even link to directly I just didn't want to talk about it bluntly, I found something surprising.

Now I have recently taken further steps to remove any crumbs that might lead google and other tools to ferret out the connection between my real name and my blog so that I can express non-PC opinions without worrying about pissing off my bosses in acaemdia, but still, posting this on my own blog and being honest about it was something I was not prepared to do. The statistic I looked at to find a correlation with murder rates? The percentage of the population that is white. R^2 for all 50 states is surprisingly larger than for poverty, at 22%, but what was really astounding was how much an outlier Hawai'i was against a very clear general trend, and when I looked into this, it became clear that the reason that Hawai'i did not fit this pattern is because it's got a very low percentage of white people (in the context of all 50 states) but instead of other states where whites are a smaller percentage of the population instead of having large Hispanic or African-American populations, it has a totally sui generis demographic situation for a U.S. state where Asians and Pacific Islanders make up a huge percentage of the population.

So I ran the regression with Hawai'i tossed out as an outlier, for the remaining 49 states R^2 = 37%. The percentage of state's population in the language of a statistics textbook accounts for more than twice the variability in the states murder rate accountable to poverty and something like 20 times the variability accountable to the rankings of gun control laws by the Brady Campaign Against Gun Violence.

So while I can't find a flaw with what is written there, I would say that the underlying statistical association that I'm finding is one that is very politically correct, indeed. Not only does a state's murder rate have almost nothing to do with how strict the gun control is, but furthermore a politically correct variable to explain what does correlate with murder rate--poverty-- explains the variability a lot less well than how many non-whites you have in a state, with the outlier of Hawai'i pointing to the even more offensive conclusion that it's not just non-whites (who you might argue are apprehended less frequently or discriminated in favor of because they run society) but also that Asians and Pacific Islanders also appear to not be responsible for violent crime to the same degree as other demographic groups.

Standard disclaimers about statistics: this is correlation, not causation, and even an R^2 of .37 is not exactly dead to rights correlation or even a strong correlation (although it's at a level that certainly doesn't appear due to random chance.)

I would say with regard to Hawai'i that unlike the other non-contiguous state of Alaska the population there has a lot of residents who weren't basically Republicans who couldn't find enough space in Montana. It doesn't have the same gun culture as the mainland in the same way that Japan doesn't.

If you cited similar statistics about Japanese gun control you could make a convincing case that the Japanese experience with gun control proves that banning guns works very well... in Japan. But Japan is not the U.S. and I think to a pretty large extent Hawai'i is not the mainland U.S.

Adam L said...

It might be helpful to have posted the link to my percentage white/murder rate map.

Also... I meant to write that black and Hispanic populations being a great explanatory factor for murder would be a politically incorrect explanation for how gun crime varies geographically.