On Facebook, several people I know have shared this photo from Glen Beck, taking a picture of some book:
It’s from Glen’s book Control. And as you can see, he thinks that this take is a good argument to argue that having more guns doesn’t make gun murders increase. Unfortunately, his table shows he doesn’t understand much of anything about statistics.
First off, he’s saying that the gun-related murder victim per capita is greater than 1 in any given year? That would mean, for example, 5 people in the US were killed in 1996 for every person in the US. Unless immigration is happening at the most fantastic of rates, this is what we in the business call absolutely impossible. More likely, this is the murder rate with guns per 100,000. That is how the statistics are usually provided. So already, his ability to gather data is suspect.
But let’s also note the next column, which is number of guns owned in the US. This is an absolute number, but the first column is a relative number. One is normalized for the US population, the other is not. That is important because the number of guns owned could be going up just because of population increase. So if anything, you would want to compare gun murders per 100,000 and guns per capita. In other words, Glen isn’t even comparing appropriate measures of things.
But even such a comparison doesn’t seem to be all that sensible. What if an increased number of guns per capita was happening because gun owners were owning more guns, while people who don’t have guns aren’t becoming gun owners? Let’s consider two scenarios and see which would theoretically make more sense when it comes to making us safe. Take a group of ten people and ten guns. Is this population more likely to be able to protect itself if one person owns all ten guns or if each person has a gun. Guns per capita will measure this the same way, but the latter would seem to be the better in making the population safer. Also, there is a depreciating value of of each gun if one person owns many. After all, you can only hold and fire so many guns at once. So, if anything we should be comparing the number of gun murders per 100,000 and the percentage of residences that own guns.
So, let’s assume Glen collected the murders per 100,000 correctly, and now let’s look up what the percentage of residences who say they own a gun. That data I found here and here. Now, in some years the question was asked multiple times, and there is some notable variances in the percentages. In 1996, there is a variance of 6 points. Of course, that’s an issue with sample selection and self-reported data. But let’s take it and compare it to the gun murder rate and see what’s the correlation.
You can put this data into Excel and do a linear fit to see what results I got. Using the data from 1993 to 2011, I found a positive correlation; that is, more residences with guns, the more gun murders. In fact, the variance of the data that could be explained by this factor was over 40% (R^2>0.4)! And the result was statistically significant (p<<0.05). So, using the data that is an appropriate measure and with actual statistics (rather than just lookey here), it would indicate that more guns very much explains more gun-related murders. However, that would be rather incomplete analysis. One thing I noted was how the year 1993 was such an outlier. If you do not include that data point, then the correlation is barely positive and the variance is very small. So, in the best case scenario for gun-right advocates, more guns makes us a little less safe.
However, this is no where near a complete analysis. For one thing, there are many other factors to consider. In this same time period, there were changes in law enforcement which could have simply put more bad guys in jail before they could do more murders. The population has become older, and most of the violent crime is done by young men; few young men per 100,000 means fewer likely criminals. There have been community efforts to decrease gang influence and numbers. Etc., etc., etc. A proper analysis would need to include all these factors and more, measure how strong an effect they have, and then you could say how much gun ownership played a role. But those results are hard to get, both from the assumptions in the sociological models to the data gathered. I already noted how the data about gun ownership is a bit problematic.
So I don’t make any conclusions from what I have done, other than Glen Beck is pretty much spreading propaganda, and even a basic understanding of statistics shows how wrong and incompetent he is. Still, lots of people are going to be using this argument, even though the data, when more correctly considered, gives exactly the opposite result. It’s crappy arguments like this that make people believe all statistics are invalid (except for the ones they like), and in this area of politics it means lives saved or lost. So to me, this is serious.
What’s the best solution? I don’t know. I’d need to do real research. You know, that thing Glen doesn’t do. If you want to figure things out, at least look at Google Scholar and see what’s in the peer-reviewed literature. They at least know what ‘per capita’ means.