In 2018, a mass shooting in Toronto killed two young women and injured 13 others, shocking the city and prompting policy-makers to draft urgent solutions to Canada’s growing gun violence problem. But in the wake of the massacre, all levels of government revealed they didn’t track where crime guns originate from. The culmination of a year-long reporting effort to source Canada’s crime guns, our story revealed police forces’ poor data practices, bringing questions of gun policy to the fore in last year’s federal election campaign and challenging leaders to devise a way forward informed by numbers rather than cheap politicking.
We deduced that firearms sourcing data existed, but was being held by dozens of municipal forces across the country, so we went about trying to do the work government had shied away from, asking for these databases through freedom-of-information requests. Ultimately, we filed more than 40 requests to governments and the largest police forces in the country. The plan was to collate the gun databases into a cohesive, national picture. But what we got back from these requests was a mess: no two police forces in the country seem to collect their data the same way.
That plan failed. But this failure revealed, in part, why the gun policy debate in Canada has remained at an impasse: there’s not enough data to support arguments for or against the tightening of gun regulations. Because the data is so poor, both sides tend to reach a stalemate.
Our work did yield some unexpected findings. A ballistics database acquired from the Royal Canadian Mounted Police showed that assault-style rifles are rarely involved in violent crime in the country, making up just 0.66 per cent of the guns that have been analyzed on murder and attempted murder cases. The day after our story ran, Prime Minister Justin Trudeau took to the campaign trail to advocate a full ban on assault-style rifles as a way of combatting gun crime. Anyone who’d read our story knew it was a political ruse. We also obtained a cache of Toronto Police documents that showed legally-imported handguns make up a sizable portion of firearms used in the city’s crime.
In the lead-up to this story, our reporting also prompted Ontario’s regional police force to change its inspection policies for gun shops, and revealed a government-run survey on gun control had been gamed by an anti-gun control activist.
The story – the culmination of a year of source-building and publishing stories on the state of gun crime and gun regulation in Canada – relied heavily on freedom of information requests. In total, we filed more than 40, and extracted gun databases from many major metropolitan police forces and the federal government.
To analyze the data, we relied on tools like R and the tidyverse set of packages, along with a data journalism R template developed called startr, which we developed in-house and which was open-sourced last year.
The story, and the methodology piece breaking down how we failed to arrive at the national gun sourcing figures we hoped to compile, underscore the messiness of the data journalism process: sometimes, no matter how much you wrangle sources, the language of freedom of information requests and the data itself, the answer you hope to find doesn’t exist. But that non-answer in itself is a finding, and one we couldn’t have arrived at without tools like R, startr, and the dozens of sources who spoke to us.
What was the hardest part of this project?
In Canada, devising gun policy boils down to one giant problem: data, or rather, a lack of it. In the wake of the Toronto’s Danforth Avenue massacre, all levels of government revealed that they didn’t track where guns used in shootings originate from. That has huge policy implications. If the shooters are using guns legally imported for sale in Canada, then gun control can be a helpful measure. But if the weapons are arriving via U.S. smuggling routes, then placing new restrictions, or even bans, on Canadian gun sales would do little to stem the flow of firearms into criminal hands. Without those sourcing figures, Canada can never move forward on evidence-based gun policy.
We also spoke with countless law enforcement sources and balanced their views against dozens of gun enthusiasts, many of whom attended a Vegas-style gun show in the greater Toronto area featured in the article.
But while gathering the data might seem like the hardest part of putting this story together, that wasn’t it. It was speaking with those affected by gun violence.
Throughout the story, clarity comes in the figure of Patrick McLeod, an ex-cop whose daughter was caught up in the Danforth shooting. A former gun enthusiast, he says the shooting convinced him that all handguns need to be banned, data be damned. He bases his argument on 18-year-old Reese Fallon, his daughter’s childhood best friend, who died in the shooting. The gun that killed her was Canadian. He says he doesn’t need any other data to form his opinion. Why should Canada’s dearth of data be an excuse for even one more death?
What can others learn from this project?
The story was only possible due to the many pieces and reporting we did in the year leading up to the federal election. Without context from law enforcement, government and gun industry sources, and the stories that built on each other and staked The Globe and Mail as a publication regularly reporting on issues of gun policy, we couldn’t have crafted the definitive picture of the gun regulation debate during the federal election. Investing the time to meet with sources, and listening to them on what freedom of information requests we should file and what stories we should look into were crucial.
We also became adept at filing freedom of information requests that extracted databases from police forces, no easy feat in Canada, where freedom of information laws are much stricter than in the United States. As our open requests piled up, we figured out the language we needed to use and the things we needed to ask for that would get us the data we wanted. Anyone looking to harvest data across multiple government departments via freedom of information requests should be patient, file as many requests as they can up front, and learn from their successes (and failures) while other files remain open.