So What If The CDC Gun-Violence Numbers Are Wrong?

Our friends at The Trace dropped a real bombshell last week by exposing the fact that the gun-injury data published by the CDC may or may not be correct. And if the data which we use to understand gun violence isn’t any good, then how can we figure out what to do about the problem that kills and injures more than 125,000 Americans every year? Or maybe the real number is as high as 160,000, or maybe it’s around 90,000; according to the intrepid Trace reporting team, the CDC estimate could be off either way by as much as 30 percent.

cdc             To arrive at this shocking state of affairs, the reporters from The Trace enlisted the help of another organization whose expertise, when it comes to figuring out data, is above reproach.  I am talking about the help they received from a team that works for Nate Silver’s Five Thirty Eight website – remember how accurate they were in predicting the outcome of the Presidential election in 2016?

With all due respect to this earthshaking news about the possible lack of precision in CDC gun-violence numbers, I published not one, but two columns on exactly this problem back in June. I also was unable to get a list of the hospitals that provide the data for the injury estimates, but I did find a national map which indicates the location of each medical center whose injury numbers are gathered by the Consumer Product Safety Commission and then forwarded to the CDC.  The Trace quotes the editor of a major medical journal as stating that if the hospital sample includes sites which treat lots of gun injuries, this will influence the overall estimate in significant ways. This is nothing more than someone saying something because he’s been asked to say something; if you look at the map of participating hospital sites, you discover that, au contraire, some of them are located in places where gun violence rarely occurs.

Of course no article on anything having to do with public health gun research can be published by The Trace unless they get some quotes from some usual suspects such as major public health researchers like David Hemenway and Philip Cook. The former says he doesn’t trust the numbers, the latter was actually trying to figure out how many gun injuries end up resulting in deaths (the ‘case-fatality’ rate) which is a somewhat different issue than just trying to validate the non-fatal injury rate itself.

This report is grist for the mill of groups and organizations who are trying to re-start gun-research funding from the CDC. And I would certainly never (read: never) state or insinuate that there should be any kind of prohibition on such funding in any way. But this article raises two concerns that The Trace team does not appear to acknowledge or understand, nor are these issues found on the radar screen of their friends at 538.

With the exception of live births and deaths, both of which must be reported as accurately as possible to the U.S. Census so that we know how many people are actually living in the United States, virtually all of the data produced by the CDC on illness and health issues are estimates, and while the CDC does a much more comprehensive tracking of heart attacks than gun-shot wounds, take a look at the data on heart disease and you’ll notice that the reported numbers always are preceded by the word ‘about.’

More important the fact is that while everyone keeps barking up the tree about how we need more funding, I see little attention paid to another branch, namely, research not on the victims of gun violence, but on the perpetrators of this dread disease. If 75,000+ individuals didn’t pick up a gun every year and use the weapon to try and kill someone else, we wouldn’t experience a gun-violence rate that is many times higher than the rest of the OECD. Any chance The Trace might try to figure that one out?



One thought on “So What If The CDC Gun-Violence Numbers Are Wrong?

  1. I think you asked the question once. Given the large number of people who engage in interpersonal violence, and the ease of getting a gun, why do so few perps actually shoot someone? As we say in cone head land, why not study the source term?

Leave a Reply