How many mass shootings did the United States experience between 2013 and 2020? It’s a simple question. And it seems like it would be easy enough to answer. But it isn’t that simple. There is not federal definition of “mass shooting.” The closest the federal government comes is with incidents the F.B.I. defines as “mass murders” in their Supplementary Homicide Report. Those are incidents that involve at least four fatalities, happen in a single location, do not involve a “cooling off” period, and more. Sometimes, scholars borrow that definition to define “mass shootings”, selecting only those that relied on the use of firearms (though even here, many do not count incidents that could be classified as gang violence, drug violence, family or intimate partner violence, those that include multiple shooters, or those that occur across multiple settings – so-called “spree killings”).
Different databases exist that rely on slightly different definitions, and they result in a different count. Because mass shootings a collection of mass shootings have been so horrific in the past few weeks, there will be lots of stories and claims about “how common” this problem actually is. And the truth is, regardless on your political goals or agenda, there’s a database for you.
This problem of definition has serious consequences and severely limits what we know about mass shootings. The best way of answering the “how many?” question is to say that any estimate we have is a conservative estimate; we just don’t know by how much.
The past few years, Tara Leigh Tober and I have worked to try to address this issue. As a part of this, we have a manuscript under review that examines just how much/little overlap there is between incidents in different databases. We hope to have this out soon. Spoiler, there’s much less overlap than you might think.
As a part of that work, it allowed us to provide an estimate of how many mass shootings occurred during a time period where for which we were able to collect data from all five databases – 2013-2020. Our findings documented 3,155 separate incidents in at least one of the five most commonly relied upon databases. One important caveat is that the definitions used to populate each of these databases are different. Below, I made a time-lapse map charting incidents over this period by date. 3,155 incidents displayed in about one minute. It’s horrifying. But it helps present the scale and scope of the problem in ways any of these individual databases on their own is not completely capable.
Collecting this larger body of data also allows us to appreciate new patterns in this uniquely American problem. For instance, beyond the “how many?” questions, we can start to look for patterns beyond the fact that almost all of this violence is committed by men (regardless of database or definition) (see here).
Just by way of example, below I charted frequencies of mass shootings by month for each year of data (see figure below). I shared it on Twitter, and thought I’d share it here as well. With a larger sample, appreciating patterns in this form of gun violence is more possible. Like other forms of violent crime, mass shootings are more and less common at different times of the year. These last few horrific weeks including the NYC subway shooting, a terrifying mass shooting in a mall in South Carolina, the white supremacist mass shooting at a grocery store in Buffalo, New York, and the school shooting at Robb Elementary in Uvalde, Texas. And since, 2013, mass shootings have been more common from May-August. The pattern is extremely consistent. But fully appreciating this requires a less restrictive definition.
Gun control is obviously a crucial component in addressing mass shootings. The Canadian Prime Minister just enacted new legislation that will put a cap on the number of handguns in Canada as well as a ban on assault rifles (HERE). This is an important move. And there are diverse gun cultures in Canada that will react to this news in different ways. Tara and I also wrote a report as a part of Canada’s Mass Casualty Commission as well (HERE). And there, we recommend the Canadian government collect data on mass shootings systematically and make those data public and easily accessible. We can do better.
Because we live in a data-saturated era, when we don’t have data on something about which we absolutely should, we should pause. Sometimes this results from a mismatch between incentives and resources – those with the incentive to collect the data lack the resources while those with the resources have no incentive. Sometimes the burdens of data collection make producing a data set on a topic unfeasible. Some kinds of data resist easy measurement—data that resist being categorized and classified. Finally, some data don’t exist because specific groups do not want them to exist. All of these issues play a role in data on mass shootings in one way or another. But they shouldn’t.