Do We Really Know The Numbers On Gun Violence?

When our friends at the Gun Violence Archive (GVA) first started up, they were immediately attacked by Gun-nut Nation for all sorts of misdeeds, including the usual nonsense about undercounting all those instances when red-blooded Americans use a gun to stop a crime. But I notice increasingly that mainstream media sources now routinely reference the GVA and a pat on the back from Newsweek and The Washington Post, usually means you must be doing something right.

GVA             The problem that GVA has to deal with, of course, is that they generate all their data from what we refer to as ‘open sources,’ namely media and related coverage which appears online. The good news about such coverage is that it’s easy to do a search for online content, I have been using Google Alerts with keywords like ‘shootings’ and ‘gun violence’ for years. The bad news is that these sources can’t possibly cover all relevant events that would let us know the number of people who get shot every day

What we usually rely on for gun-violence numbers is the data produced by the CDC. After all, we assume that since medicine is a scientific exercise, at least since Louis Pasteur figured out that something called a microbe spreads disease, we also assume that medical science develops its practices using evidence-based facts. And what could be more of a fact than a dead body lying in the street?

Except there’s only one little problem.  When we take a look at the data on gun violence collected and published by the CDC, particularly when we go below the summary data which tells us how many people are shot and killed in the United States every year, all of a sudden we discover that the numbers not only aren’t so exact, but don’t even add up. Now you would think that something like gun violence, which allegedly costs us more than $200 billion a year in medical costs, lost wages and other various and sundry sums, would at least provoke some degree of concern about whether we actually are using valid numbers or not. Let me break it to you gently – we’re not.

In 2015 the CDC says that 35,476 people lost their lives because a gun went off and they didn’t duck; of this number, which is routinely reported by every gun violence prevention (GVP) group, homicides accounted for 12,979, suicides amounted to 22,018, another 484 were shot either by cops or armed citizens legal defending themselves, and 282 died but nobody’s sure how those deaths actually came about.

We know that the number of gun deaths that were ostensibly justified is probably undercounted by at least half. And let’s not forget the 489 unlucky folks who accidentally killed themselves or someone else with a gun, a number which is also probably well below the annual toll. But neither of those categories, even if doubled, would change the overall gun-death number by much. Let’s face it, gun violence in America is overwhelmingly a function of intentional injuries committed by the shooter against himself or someone else.

I have spent the last week comparing gun-death homicides furnished by the CDC to the numbers found in the National Violent Death Reporting System (NVDRS) which also happens to be an agency within the CDC, but draws its data from a wider pool of sources and is considered by scholars to be more reliable when it comes to counting bodies that wind up in the morgue. When we compare numbers, however, we discover that the numbers being used by the GVP community are perhaps 20% higher than the number published by the NVDRS.

The next time someone says that you can’t trust an online, open-source aggregator like the Gun Violence Archive, you might want to reply that the numbers we get from all those medical scientists might not be any more reliable, and in terms of accuracy, might even be worse. I’m as enamored of science as anyone else, but sometimes I wonder whether the science of gun-violence research actually exists.

Advertisements

2 thoughts on “Do We Really Know The Numbers On Gun Violence?

  1. I suppose the one thing we can do is just put bigger uncertainty estimates on the number. Something like a number plus or minus 20%. It doesn’t really change the arguments much.

Leave a Reply