Writing in the Washington Post, Leah Libresco, a statistician and former newswriter at FiveThiryEight, a site created by the famed statistician Nate Silver, admits that she reversed her stance on gun control after studying data on gun deaths in America.

Libresco begins by confessing that before she started researching gun deaths, she was incredibly frustrated by guns and blamed the National Rifle Association (NRA) for blocking the banning of assault weapons, restricting silencers and shrinking magazine sizes.

Then she analyzed data from the approximately 33,000 lives ended every year by guns in the United States. Evidence based reasoning was a revelation to her: “The best ideas left standing were narrowly tailored interventions to protect subtypes of potential victims, not broad attempts to limit the lethality of guns.”

Gun control advocates often bring up the experiences of Australia and the United Kingdom in tightening their gun laws. But data showed Libresco that this had little relevance for America: “Neither nation experienced drops in mass shootings or other gun-related crime that could be attributed to their buybacks and bans.”

In fact, “mass shootings were too rare in Australia for their absence after the buyback program to be clear evidence of progress.”

Libresco continues:

“When I looked at the other oft-praised policies, I found out that no gun owner walks into the store to buy an ‘assault weapon.’ It’s an invented classification that includes any semi-automatic that has two or more features, such as a bayonet mount, a rocket-propelled grenade-launcher mount, a folding stock or a pistol grip. But guns are modular, and any hobbyist can easily add these features at home, just as if they were snapping together Legos.”

Libresco says, “Silencers limit hearing damage for shooters but don’t make gunfire dangerously quiet. An AR-15 with a silencer is about as loud as a jackhammer.”

The majority of gun deaths in the United States would be difficult to prevent: “Two-thirds of gun deaths in the United States every year are suicides. Almost no proposed restriction would make it meaningfully harder for people with guns on hand to use them.”

The next largest set of gun deaths is young men aged 15 to 34 hilled in homicides, and Libresco concludes that “Few of the popularly floated policies were tailored to serve them.”

Libresco therefore “can’t endorse policies whose only selling point is that gun owners hate them. … I found the most hope in more narrowly tailored interventions.”

Libresco has some specific suggestions for combatting gun deaths:

“Older men, who make up the largest share of gun suicides, need better access to people who could care for them and get them help. Women endangered by specific men need to be prioritized by police, who can enforce restraining orders prohibiting these men from buying and owning guns. Younger men at risk of violence need to be identified before they take a life or lose theirs and to be connected to mentors who can help them de-escalate conflicts.”

Libresco concludes:

“A reduction in gun deaths is most likely to come from finding smaller chances for victories and expanding those solutions as much as possible. We save lives by focusing on a range of tactics to protect the different kinds of potential victims and reforming potential killers, not from sweeping bans focused on the guns themselves.”