The seminal work here is a 1999 book by Berkeley's Franklin Zimring and Gordon Hawkins, called Crime Is Not the Problem. Zimring and Hawkins set out to examine what was, at the time, the conventional wisdom: that America had a uniquely terrible crime problem, one without any parallel in other developed democracies.
More from Vox:
A huge international study of gun control finds strong evidence that it actually works
Iran's government has some pretty harsh words for Trump
Australia confiscated 650,000 guns. Murders and suicides plummeted.
They found, pretty definitively, that the conventional wisdom was wrong. "Rates of common property crimes in the United States are comparable to those reported in many other Western industrial nations, but rates of lethal violence in the United States are much higher," they write. "Violence is not a crime problem."
Zimring and Hawkins determined this by looking at 20 developed countries' overall crime rate and rates of violent death. They found virtually no connection between the two, indicating that a country's level of violent death wasn't determined by its overall crime levels:
The lowest death rate country (England) has a crime rate just over average. The next lowest violence nation is Japan, which has the lowest crime rate also. The third lowest death rate country is the Netherlands, in the highest crime rate group.
"This data set provides a multinational example of the central point that lethal violence is the crucial problem in the United States," Zimring and Hawkins write. "It shows the United States clustered with other industrial countries in crime rate, but head and shoulders above the rest in violent death."
Why does this happen? It's not because, as you might think, American violent criminals are just more likely to kill people. "Only a minority of Los Angeles homicides grow out of criminal encounters like robbery and rape," they find (there's no reason to believe the pattern would differ in other cities). So even if it could be shown that American robbery and rape rates are across-the-board higher than those in similar countries (which doesn't appear true today), that still wouldn't explain why America has so many more homicides than other countries.
Again, Zimring and Hawkins's LA data was revealing. "A far greater proportion of Los Angeles homicides grow out of arguments and other social encounters between acquaintances [than robbery or rape]," they find.
This is where guns enter the story. The mere presence of firearms, according to Zimring and Hawkins, makes a merely tense situation more likely to turn deadly. When a gang member argues with another gang member, or a robber sticks up a liquor store, there's always a risk that the situation can escalate to some kind of violence. But when people have a handheld tool that is specially engineered for killing efficiently, escalation to murder becomes much, much more likely.
And indeed, that's what Zimring and Hawkins's data found.
"A series of specific comparisons of the death rates from property crime and assault in New York City and London show how enormous differences in death risk can be explained even while general patterns are similar," they explain. "A preference for crimes of personal force and the willingness and ability to use guns in robbery make similar levels of property crime fifty-four times as deadly in New York City as in London."
Guns, not criminality per se, are the problem.