Counting the Lives Lost to School Shootings Since Sandy Hook

With the memory of the massacre in Uvalde, Texas, still fresh, Senate negotiators on Sunday night proposed a cautious bipartisan compromise on gun control legislation. The framework includes measures both to bolster background checks for those under age 21 and strengthen security at schools.

While the latest catastrophic flare-up of gun violence may have mobilized lawmakers, the tragedy at Robb Elementary School on May 24, which claimed the lives of 19 children and 2 adults, was far from the first fatal school shooting in recent years. But precisely how many lives have been lost to those shootings is a figure for which there is little consensus. There is no federal definition for such incidents, leaving it up to researchers, advocacy groups, and publications to craft their own.

To examine why it is so difficult to agree on this grim tally, TIME reviewed the roster of nearly 1,000 incidents of gunfire at schools and colleges published by the gun-control advocacy non-profit Everytown for Gun Safety, spanning the nearly 10 years since the December 2012 massacre at Sandy Hook Elementary School in Newtown, Conn. Everytown’s figures are commonly cited—alongside a handful of organizations that comb and aggregate media accounts of gunfire on school grounds—and use a broad definition of “any time a firearm is discharged inside a school building or on a school campus or grounds.”

Read more: Mass Shootings Are Only a Small Part of America’s Deadly Problem With Kids and Guns

Other organizations commonly filter such lists based on a handful of variables such as the number of casualties and the time of day, often focusing on “Newtown-like” events. Advocacy groups like Everytown, meanwhile, argue that such a lens discounts many traumatic events and lost lives that occur in contexts like school sporting events.

To capture this tricky calculus, TIME took the 101 incidents on Everytown’s list of on-campus gun violence at preschools and K-12 schools that involve an assault and at least one fatality, and coded each by the variables that are commonly used to arrive at a measure of the impact of school shootings. You can adjust this filter to see how the tally changes based on any given distinction in what constitutes a school shooting.

The above incidents are only a subset of the Everytown universe, which also includes nearly 250 events in which no one was killed but at least one person was wounded, sometimes severely. The list also includes a number of episodes in which a person–usually a student–took their own life on campus, often during the school day, as well as a number of accidents, typically when a student brought a gun to school and discharged it in the course of showing it to another. The above visualization is not meant to discount the impact of such incidents, but is limited by the often spare information available as to when and where they took place and who was involved.

Notably, a fair portion of gunfire on school grounds involves confrontations between adults, often at night and on weekends and almost always outside. But any effort to draw a clear distinction between these episodes and ones that endangered a student turn up borderline cases. Of the 101 incidents on this list, 50 occurred outside school hours (“school hours” here includes school-sponsored sporting events), and of the 53 people killed in those incidents, 11 were under age 18.

On a Monday afternoon last October, for example, a 21-year-old opened fire in front of a pizza shop directly across the street from Lincoln High School in Philadelphia, critically injuring a 16-year-old student and killing a 65-year-old man who had just dropped off a dress for his granddaughter. Such an incident will fall under some definitions of a school shooting and not others, though activists and gun control proponents argue that such incidents are less different than a “typical” school shooting than they may seem–even when a shooting occurs long after students have departed for the day.

“I would contend that children know what happened on the school grounds,” says Sarah Burd-Sharps, senior director of research at Everytown. “They’re each different, and children receive them differently.”

Read more: School Shootings Are Raising Anxiety and Panic in U.S. Children

What’s most apparent from this list is the urgent need to protect schools, even from violence in which students were never the intended targets.

“Most of the shooters who shoot in schools are students or former students at that school. It’s where they’ve spent time, it’s where they have their relationships, it’s sometimes where they’ve had failures,” says Burd-Sharps. Schools, further, are “a harder space to close off and protect” than many other venues.

On a statistical scale, school shootings are still sufficiently rare that, even by the broadest definition, they tend to defy the efforts of traditional data analysis. Some months and years are much worse than others. And any effort to trim the list to a smaller number of qualifying events, while not to the liking of the activists who compile them, involves a certain practical, and tragic, recognition on the part of those doing the counting: Any attempt to tally American school shootings without excluding any is likely to fail. At some point, there are just too many to count.