This interactive map shows the average number of gun deaths (including suicides, homicides, and accidents) per 100,000 residents, for the years 2004 through 2010, broken down by county:
Data is from the US Centers for Disease Control and Prevention. The total number of gun deaths in the US numbered 33,636 in 2013.
There are some interesting patterns here. Urban areas generally have fewer gun deaths than rural areas. Except that the border with Mexico, especially in Texas, California (and southern Florida) – prime areas for illegal immigrants and drug smuggling – ironically have very low rates of gun deaths.
Worst areas are the deep south, mountain states (Montana, Idaho, Wyoming, Nevada, southern Utah, Arizona, New Mexico, western Colorado) and the border between Oregon and Northern California (the “State of Jefferson“). Best areas are the rest of California, Washington State (except for the Olympic Penninsula), northern plains states, and the most of the northeast.
You can zoom in to see how your county does.
10 Comments
How ironic that the red areas vote red, the blue areas vote blue. I live in rural red that always votes Republican. I don’t worry because the rest of New York is blue.
Is it just me, or does anyone else have two different sound tracks come up with this site?
Hopefully, just you. I don’t have any soundtracks that I know of on this site.
I’m getting the same thing. For some reason the Daily Show clip in the Oct. 7th post is getting activated (which starts up with an ad).
Looking at the map, if we just wait long enough there won’t be that many Red states.
Gun loving states are colored the same color as blood. Go figure. The gun loving states have more guns. It’s seams pretty clear that more guns don’t make us safer. Dang!
I really like that someone has done this. It really gives you a good perspective of what is happening. I live in St. Louis County so I don’t think I need to check that one out. 🙂
Actually, this is more just an example of sampling bias. This scenario is a great example of work that Amos Tversky and Daniel Kahneman did. See their paper, “Belief in the Law of Small Numbers,” for example.
Here’s the main gist: Looking at county-level data is inherently misleading because they are fundamentally different samples. Notice that, in general, major cities avoid the extreme dark blue or dark red. If you couple this with your other post, you’ll quickly see that even very conservative cities with strong gun cultures (Colorado Springs, Arlington, Aurora, for example) have lower rates. It is simply the statistical fact that rural areas will exhibit more extreme behavior because they have a lot more variance, and the west and south have a lot more counties with sparse populations.
Furthermore, there’s also some blatant cherry-picking of data going on here. The premise seems to be that rural population + high gun ownership causes high gun-related death rates. However, the fact that the bluest regions include most of Vermont, Minnesota, and Wisconsin, all of which include a lot of rural space and high rates of gun ownership, casts doubt on this hypothesis. Simply put, this is all sampling bias. Sparse populations exhibit more extreme behavior on both sides of the coin.
Here’s an alternative hypothesis: The poverty rate of a region is a better predictor of gun-related death rates than political leanings, gun ownership rates, or rural vs. urban status. As an illustration, compare the map above with this county-by-county map of poverty rates.
In short, if we really want to address the problems of gun-related deaths, we need a lot more careful, nuanced analysis that doesn’t try to jump to easy conclusions by ignoring statistical biases.
As illustration, zoom in on Wayne (Detroit) and Monroe Counties in Michigan. Do you really think that Detroit proper is that much safer than Monroe? The county-by-county sampling gives that appearance, though, because Wayne’s population is 10 times that of Monroe’s, which makes Wayne’s rate look closer to the mean.
Michael, as I’ve often said, I am not against gun ownership (as long as it is responsible). I believe that the second amendment guarantees the right to own guns (but allows common-sense safety regulations).
In particular, I was not promoting the belief that high gun ownership leads to more deaths.
On the other hand, the colors were based on the death rate (gun deaths divided by population), so yes, one is less likely to die from a gun in Wayne county than Monroe county. Of course, gun deaths include suicides, so it is not just about suicides.
I do appreciate your comments. Thanks!
I disagree regarding Wayne vs. Monroe. The populations are dramatically different in size, so they have very different statistical properties, particularly variance and kurtosis. To truly compare the two counties, you must divide Wayne into multiple partitions that are equal in population size to Monroe. What you will most likely find, at that point, is that certain parts of Wayne have death rates just as high or higher than Monroe. Other parts would have lower rates than the cumulative Wayne rates. Comparing groups with different population sizes is inherently problematic.
Another example that comes to mind is Oak Park, IL. There’s a street that forms its northern border. Oak Park itself is one of the most affluent and safest places to live in the Chicago area. Cross that street and go two blocks farther to get to a neighborhood that was long overrun with gang violence and drugs. But both areas are part of Cook County, so they average out to being “okay” overall. The increased population masks the variability. It’s simply a statistical artifact.
I didn’t accuse you of attacking gun ownership. What I was attacking was this: the interpretation that rural communities are more likely to have higher death rates. This is demonstrably false, as many states in the upper midwest and northeast have very rural locations that are among the lowest death rates. That is, I find the hypothesis that rural vs. urban is the best predictor of death rates to be insufficiently supported by the data.