The Washington PostDemocracy Dies in Darkness

Bill Clinton’s claim that the assault weapons ban led to ‘a big drop in mass shooting deaths’

There's some data that supports former president Bill Clinton but he overstates the case. (Video: Sarah Cahlan/The Washington Post)

“How many more people have to die before we reinstate the assault weapons ban & the limit on high-capacity magazines & pass universal background checks? After they passed in 1994, there was a big drop in mass shooting deaths. When the ban expired, they rose again. We must act now.”

— Former president Bill Clinton, in a tweet, Aug. 5, 2019

In the aftermath of the mass shootings in El Paso and Dayton, Ohio, Clinton renewed a call to reinstate the assault weapons ban that he signed into law in 1994. It was in place for 10 years until President George W. Bush let it lapse.

Even supporters of the law acknowledged that it was riddled with loopholes, such as allowing copycat weapons to be sold, that limited its effectiveness. A 2004 study for the Justice Department found that the ban’s impact on gun violence was mixed, at best, because of exemptions written in the law; if the ban were renewed, the “effects on gun violence are likely to be small at best and perhaps too small for reliable measurement.” The report said that assault weapons were “rarely used” in gun crimes but suggested that if the law remained in place, it might have a bigger impact.

In 2013, we gave Clinton Three Pinocchios for incorrectly claiming “half of all mass killings in the United States have occurred since the assault weapons ban expired in 2005, half of all of them in the history of the country.” But here, his claim is a bit more nuanced — that there was “big drop” in mass-shooting deaths during the ban, and then deaths rose after it expired.

Is this claim more credible? Let’s take a look.

The Facts

Angel Urena, Clinton’s spokesman, was quick to supply the source of his claim — a paper published in January in the Journal of Trauma and Acute Care Surgery.

(We should note there is no standard definition of a mass shooting, though the study relied on four or more deaths, not counting the shooter. Contrary to popular perception, there is no FBI definition of a mass shooting, though the FBI defines a mass murderer as someone who kills four or more people.)

The authors, led by Charles DiMaggio, a professor of surgery at New York University’s Langone Medical Center, studied mass-shooting data from 1981 to 2017 that were common to three open-source sets of data. They concluded that assault rifles accounted for 430, or 85.8 percent, of the 501 mass-shooting fatalities reported in 44 incidents but that there were nine fewer mass shooting-related deaths per 10,000 homicides during the federal ban period. Assuming the calculations were correct, the authors said an assault weapons ban would have prevented 314 of the 448, or 70 percent, of the mass-shooting deaths during the years when the ban was not in effect.

That’s an interesting finding, but the study has been disputed. In a letter published by the journal, Louis Klarevas, a research professor at Teachers College of Columbia University, said the “authors misidentified the involvement of assault weapons in roughly half of the incidents.” He wrote that “when the erroneous cases are recalibrated, the number of incidents involving assault weapons drops 62 percent from 34 to 13, and the number of fatalities resulting from such shootings drops 46 percent from 430 to 232.” He also said one incident did not qualify for inclusion because it involved three shooting deaths, not four.

In a response that will be published, the authors said: “We make no claim to have retroactively determined whether these guns would have been illegal under the original statutory language. We question Dr. Klarevas’s claims to be able to do so.” Klarevas told The Fact Checker that he stood by his conclusion that “they conflate many semiautomatic firearms with assault weapons.” As he put it, “assault weapons are by nature semiautomatic, but semiautomatic firearms are not necessarily assault weapons.”

Klarevas, who defines a gun massacre to be six or more deaths, says his own research for the book “Rampage Nation,” however, backs up Clinton’s statement. Compared with the 10-year period before the ban, the number of gun massacres during the ban period fell by 37 percent, and the number of people dying from mass shootings fell by 43 percent. But after the ban lapsed in 2004, the numbers in the next 10-year period shot up again — a 183 percent increase in mass shootings and a 239 percent increase in deaths.

“I assessed only high-fatality mass shootings when I examined the impact of the federal assault weapons ban. I did not have comprehensive data on mass shootings that resulted in less than 6 killed, which is why my assessment is limited to high-fatality mass shootings,” he said in an email. “My data-set of high-fatality mass shootings shows that the federal AWB seems to have had an impact in reducing the frequency and lethality of gun massacres.”

Another expert who has assembled a database on mass shootings is Grant Duwe, director of research and evaluation for the Minnesota Department of Corrections. Duwe also has concerns about the DiMaggio study, in addition to the points raised by Klarevas. He said they did not statistically control for any other rival causal factors that may have influenced mass-shooting deaths, he questioned the statistical technique they selected, and he said it was “odd” to restrict the cases to only those appearing in three data sets. His own data set shows there were 95 more cases (139 vs. 44) from 1981 to 2017 and more than twice as many fatalities (1,014 vs. 501).

DiMaggio acknowledged that the researchers discovered one data set, from the Los Angeles Times, “was limiting the cases, but having set this restriction a priori we felt obligated to adhere to our analysis plan.”

Duwe provided The Fact Checker with a spreadsheet of mass shootings from 1976 to 2018. Duwe defines a mass public shooting as an incident in which four or more victims are killed publicly with guns within 24 hours — in the workplace, schools, restaurants and other public places — excluding shootings in connection with crimes such as robbery, drugs or gangs. (Note that this would exclude a number of “mass murders” that sometimes get lumped into the data, such as the Beltway sniper who killed 10 people over a three-week period in 2002.)

But beside just looking at raw numbers, Duwe examines trends on a per capita basis, both the incidence and severity (total number of victims) of mass public shootings. Finally, because there are some big fluctuations in the data from year to year, he includes a “moving average” to help clarify the direction of trends over time. (His spreadsheet included both a three-year and five-year moving average.) For instance, in 1999, the Columbine High School massacre caused a spike in the annual numbers, for a mass public shooting fatality rate of 19.07 per 100 million. A five-year average smooths that out to 7.49.

“When we look at the attached data, especially the charts, there’s not strong support for the notion that the per capita incidence was much lower during the late 1990s and early 2000s,” Duwe said. “There’s more support, however, for the idea that the per capita severity (the rates at which victims were killed or shot in mass public shootings) was lower during this period of time. But what’s even clearer from the data is that there has been an increase in both the incidence and severity of mass public shootings (on a per capita basis) since the latter part of the 2000s.”

Here’s his chart. Note the slight dip in the fatality rate between 1995 and 2004, compared to the years before and after the ban. Note also the sharp increase after 2014, 10 years after the ban was lifted.

Still, Duwe says it remains an open question whether the somewhat lower severity in public mass shootings during the 1995-2004 period was the result of the ban.

“No single observational epidemiological study, ours included, can establish causality,” DiMaggio agreed. “But what we can say, and what our study shows, is that fewer people in the U.S. were killed as a result of mass shootings during the ban period.”

Another developing area of research is the effect of lifting the ban, which included limits on high-capacity magazines. Christopher S. Koper, an associate professor of criminology at George Mason University, was the author of a 2004 DOJ study, referenced above, that concluded the impact of the law was mixed.

“One thing that has changed from my prior writings is that we have accumulating evidence on the effects of lifting the old federal ban,” Koper said. He, along with colleagues, has documented a recent rise in the use of high capacity semiautomatic weaponry across a number of local and national data sources on crime guns. His recent work “suggests that the spread of these types of weapons may have greater impacts on shootings than suggested by my earlier research,” he said.

Koper said it was a “more subtle and nuanced policy argument,” but “collectively, I think these studies underscore the old federal ban’s preventive value in capping and gradually reducing the supply of assault weapons and large-capacity magazines, despite its limitations.”

The Pinocchio Test

The assault weapons ban existed for only 10 years, and there are relatively few mass shootings per year, making it difficult to fully assess its impact. Definitions also matter.

The DiMaggio study cited by Clinton’s spokesman appears to be missing many cases and mislabeled some weapons. Clinton’s case appears to become stronger when high-fatality (six or more) shooting incidents are studied, such as by Klarevas, compared with the more common four deaths or more. Duwe’s research, which adjusts for population and also looks at five-year moving averages, suggests only a slight dip during the ban.

We frequently remind readers that correlation does not equal causation. Clinton cited a “big drop in mass shooting deaths” during the ban, but that’s going too far. Given the way the results shift depending on what mass shootings are catalogued and studied, Clinton should not be so definitive.

We wavered between One and Two Pinocchios, but ultimately leaned toward Two. There’s increasing evidence the ban may have had some impact — but at this point it should not be overstated.

Two Pinocchios

(About our rating scale)

Send us facts to check by filling out this form

Sign up for The Fact Checker weekly newsletter

The Fact Checker is a verified signatory to the International Fact-Checking Network code of principles