'Let me show you a picture,’ says law professor John Donohue. He jumps to his feet, comes round his desk and plops down a graph. He’s explaining a statistical technique he uses to analyze the relationship between violent crime and laws that allow more people to carry guns in public. His work shows that when more people carry guns, violent crime increases.
Donohue’s goal is to influence how laws are made, using empirical studies of how things work in the real world. “I think of myself as somebody who is just trying to figure out what the impact of laws and legal change is, so we are in a better position to know if we want to change the law or policy,” he says.
Right now, he’s focused on the statistics. He is explaining a technique called “synthetic controls,” a way of comparing crime rates in states with different gun laws. “It’s actually a pretty cool technique,” he says. He’s pointing to a line on the graph, comparing Pennsylvania with a weighted average of Delaware, Hawaii and Maryland. “And then you see, well, what happens when Pennsylvania gets the right-to-carry law there? Their violent crime rate goes up.”
Donohue, the C. Wendell and Edith M. Carlsmith Professor of Law, is on a roll. He’s an expert witness for the state of California in four different federal court cases intended to weaken California’s strict gun regulations. He’s pulling together a lifetime of knowledge and skill, and sharing it broadly with the public. And he’s working on the strongest version yet of a research project that meticulously probes the connection between homicide rates and gun laws.
A printout of the draft manuscript sits on his desk in front of him, and during a long interview, he glances down at it frequently and comments on it. Though he couldn’t be more cordial, it’s clear that all he wants to do is get back to it.
He looks down again, explaining, “This is the paper on right-to-carry laws. I’ve been working on the issue, in one form or another, for 20 years, maybe longer. This is the culmination of that work. But, you know, last night I was thinking up new ways of exploring the issue, so my team is working on that right now.”
Donohue is part of a decades-long movement to transform law, economics and policy from a set of fields largely powered by doctrine, theory and precedent to one guided by modern empirical research. That means less rhetorical flourish and more data and better statistics. For Donohue, it’s meant assembling the necessary data and refining statistical techniques to answer carefully crafted questions — often about contentious issues such as crime, guns and abortion.
His field, law and economics, has long been dominated by theory. In science, though, hypotheses and models are just a first step toward understanding how things work. Theories aren’t accepted until they’ve been tested against real-world data. It’s like the difference between a thought experiment and a real one. There’s a veritable chasm between the two. John J. Donohue III was born to bridge that chasm.
An Empiricist is Born
Looking back on his childhood in the Bronx, Donohue recalls his two grandfathers with affection and admiration. “My one grandfather was maybe the toughest human being alive, and my other grandfather was a great champion of social justice. I’m sort of a combination of those two people,” says Donohue. His maternal grandfather was a World War I hero and an eminently practical guy, able to build or fix anything with his hands. His paternal grandfather was a pacifist, an intellectual and a supporter of women’s rights.
Donohue combines the intellectual strength of one grandfather with the genius for getting things done of the other. His longtime friend and colleague Pamela Karlan, the Kenneth and Harle Montgomery Professor of Public Law, compares Donohue’s skills in both law and empirical analysis to those of an Olympic biathlete. “He’s right at the intersection. He can both shoot and ski.”
‘Without John Donohue, Connecticut might still have the death penalty.’
As an undergraduate at Hamilton College, Donohue immersed himself in the abstract reasoning of a major in mathematics, yet was already thinking about the nitty-gritty of public policy. He went straight from his undergraduate math studies to Harvard Law School. And, after a few years of legal work, he enrolled at Yale for a doctorate in economics, wrapping up his education with a data-intensive dissertation on the relative job mobility of men and women.
His old friend Peter Siegelman, a law professor at the University of Connecticut, says Donohue played an important role in raising the level of discourse in empirical studies of the law, which first emerged in the 1960s.
“He showed that we can evaluate the way laws work by going out into the real world and carefully collecting data, then analyzing it with statistical empirical methods. That’s a pretty big thing. It still hasn’t completely swept through law schools, but it’s now considered to be a very, very acceptable way of doing law.”
“I don’t want to sound too harsh, but a lot of legal academics are just good at making clever arguments,” Siegelman continues. “And if those clever arguments turn out not to actually be based on facts, well, nobody really cares about the facts that much anyway. John’s always been interested in the truth and finding out what’s actually going on.”
Donohue has never shied away from controversial topics. In 2011, he published an empirical study demonstrating that Connecticut’s death penalty was being applied in a way that was biased in favor of white people and fundamentally arbitrary. The report triggered vigorous debate in the Connecticut legislature and led to the abolition, a year later, of the state’s death penalty. “Without John Donohue, Connecticut might still have the death penalty,” says Karlan.
In a 2001 paper, he and University of Chicago economist Steven Levitt (author of the bestselling book Freakonomics) advanced the then-shocking suggestion that the legalization of abortion had lowered national crime rates. Starting in the late 1950s, the U.S. crime rate had risen from about 160 violent crimes per 100,000 people to 758 per 100,000 in 1991. Then the rise in crime reversed itself, and for nearly a quarter century, the rate of violent crime dropped. But why?
Donohue and Levitt argued that one contributor was Roe v. Wade, the 1973 U.S. Supreme Court decision that rendered abortion legal nationwide. In their paper, “The Impact of Legalized Abortion on Crime,” they argued that greater numbers of abortions in the 1970s could explain as much as half of the decline in crime in the 1990s.
The two bolstered their statistical arguments by looking at individual states. First, for example, they showed that states such as New York and California, which had already legalized abortion before 1973, had had higher abortion rates than states where it was still illegal. Donohue and Levitt then showed that states individually repeated the same overall pattern: More abortions preceded a drop in crime two decades later.
Did the increased availability of abortion let people make choices that allowed them to better parent the children they did have? It was a provocative idea, and the blowback was intense. Some accused the two professors of promoting abortion and even eugenics. In a rapid back-and-forth series of academic papers, Donohue and Levitt argued with other researchers over confounding variables such as the crack epidemic.
In competing theories, some have argued that the drop in crime resulted from more police, zero-tolerance policing, three-strikes laws, gun control and more prisons. Others have argued persuasively that lead in paint and gasoline — phased out in the 1970s — exposed children to toxic levels of lead that predisposed them to criminal behavior when they grew up. Still another theory, in the same spirit as Donohue and Levitt’s hypothesis, was that the availability of birth control pills contributed to the drop in violent crime.
The decline could have multiple causes, of course. “I still think abortion legalization is part of the story,” says Donohue. “I don’t think it’s the whole story.” Time has brought more and better data and better statistical techniques. “Steve Levitt and I have a new paper coming out on 20 years of additional data on abortion and crime,” Donohue says. “And the results are quite strong. I’m not quite finished with that paper. I would have been if there hadn’t been all these mass shootings going on.”
Since the mass shooting at Marjory Stoneman Douglas High School, in Parkland, Fla., in February, Donohue has been besieged by calls from journalists from every corner of the world. He’s made talking to reporters a priority.
Throughout his career, Donohue has engaged the public and fellow academics through lectures, debates and op-eds. The day after the Parkland shooting, Donohue was interviewed by Reuters about the shooting and gun law; three days later, the Law School posted a Q&A with him on one of its blogs; and four days after that, he was a guest on the legal podcast Lawyer 2 Lawyer. The following day, Scientific American published a column in which he explained why he thought the president’s suggestion to arm teachers was a bad idea.
“The FBI analyzed 160 cases of active shooters over the period from 2000–2013,” Donohue wrote, “and not one was stopped by a concealed carry permit holder who was not active duty military, a security guard, or a police officer. 21 were stopped by unarmed civilians. There are certainly beneficial uses of guns by permit holders that thwart or even deter crime, but these positive influences are outweighed by all of the ways — often not well understood — in which gun carrying elevates violent crime.”
In July, he was still giving public talks and still writing about guns.
Donohue is dogged intellectually; once he’s got hold of something, he doesn’t let go. But personally, he is upbeat and mild. So antagonizing pro-life groups and gun-rights activists seems like it could be anxiety inducing. “I take some precautions,” says Donohue. “I try not to be a fearful person. But you know, it’s a joke here at the Law School. They say, ‘I’m putting a sign on my door saying John Donohue’s office is very far from here.’ ”
Guns in Public
Donohue has used empiricism to test the commonly held belief that the widespread carrying of guns — whether openly or tucked away out of sight — reduces crime in states with permissive right-to-carry laws. Such laws make it easy for individuals to acquire a permit to carry a gun in public; in some states, no permit is required at all. Donohue says his empirical work demonstrates that liberal right-to-carry laws increase violent crime.
The idea that carrying guns prevents violence gained ground with the 1998 publication of economist John Lott’s influential book More Guns Less Crime, now in its third edition. Even as violent crime was in the midst of its long decline, Lott argued that laws allowing virtually anyone to carry a gun reduce violent crime. He contended that crime fell in the 1990s because more people were carrying guns. The argument was an instant success with individual gun-rights advocates and with the gun industry.
In 1986, some 41 states either did not issue right-to-carry permits at all or did so on a limited, case-by-case basis. But by 2017, 42 states had permissive right-to-carry laws. And, in 2008, the U.S. Supreme Court held for the first time that the Second Amendment was not a collective, state-based right to operate a militia, but an actual, individual (albeit not unlimited) right to keep and bear arms. That case was about guns in the home; the question of the right to carry guns in public has not yet been resolved at the federal level.
California, one of the last states left with restrictive gun laws, has been the site of an intense legal struggle over the right to carry, and Donohue is frequently called upon to testify as an expert. In mid-May, for example, a federal trial judge rejected a challenge to California’s restrictions on carrying guns in public, citing Donohue in his decision:
“California relies on the expert report and testimony of Professor John J. Donohue III of the Stanford Law School. . . . Based on the evidence California has submitted, it has shown that the State reasonably could have inferred that there was a relationship between prohibiting individuals from carrying firearms openly in public and promoting and achieving the important governmental objective of public safety. That these objectives would be advanced could be inferred from Donohue’s findings that the enactment of right-to-carry laws lead to increased violent crime rates. . . .”
Two federal circuit courts have ruled in favor of laws that restrict who may carry a gun, and two other circuit courts have ruled against such laws. As such cases are decided, says Donohue, the possibility that one of them will end up in front of the U.S. Supreme Court increases. “Every one of those issues will end in the Supreme Court,” he says. “Since they’re being litigated in federal courts all around the country, you never know exactly which case will end up in the Supreme Court. They could take a bunch of them and decide them all at the same time, or they could say, well, this year we’ll decide whether you have the right to carry.”
More Guns, More Crime?
So, if everyone’s armed, does that cause less crime or more crime? Or does arming people have no effect on crime? The answer lies inside the arcane worlds of statistics and causality. As far back as 2004, a report by the National Research Council concluded that Lott’s more-guns-less-crime conclusions were unsupported by his analysis. A fierce debate over the statistics of crime and gun laws has burned ever since.
Today, researchers have crime data all the way up through 2014, far more than they had in 1998, when More Guns Less Crime was first published. And statistics and its cousin, econometrics, have advanced. Now Donohue has more years of data, more states with right-to-carry laws to examine and better stats. Equipped with three full-time research assistants who do the bulk of the number crunching, Donohue has concluded that in states that pass right-to-carry laws, violent crime increases by 13 to 15 percent.
Donohue emphasizes that trying to use observational data to tell a causal story isn’t easy. “When you’re trying to look at these complex social phenomena and control for all of these things that are changing at exactly that moment and anticipate all of the influences and see the ways in which they interact, it’s really, really tough,” he says.
One strength of Donohue’s approach is to crunch the data in several different ways to see if he can get a different outcome. For example, in the first edition of More Guns Less Crime, Lott used what’s called “panel data analysis,” a kind of longitudinal study of multiple variables. Modern panel data analysis evaluates statistical significance more accurately, says Donohue, and, on reanalysis, most of Lott’s original results were not statistically significant. So, for the new data set, Donohue’s team uses the upgraded version of panel data analysis, and two newer methods.
Donohue has concluded that in states that pass right-to-carry laws, violent crime increases by 13 to 15 percent.
One of those two methods involves “synthetic controls,” a statistical approach in which researchers construct an artificial control group that’s analogous to a real control group.
To understand why this is necessary, imagine an eighth grader who is designing a gold-standard interventional experiment on gun control laws for her school science fair. To study the effect of right-to-carry laws on crime, she randomly gives 25 U.S. states liberal right-to-carry laws and, for comparison, gives the other 25 states laws that severely limit the right to carry a gun.
She waits three decades, then measures the violent crime rate in each group of states. For good measure, she swaps which states are assigned which laws and waits another 30 years. By the time she finishes her science fair project, she’s long past retirement age. For that and a number of other reasons, that particular experiment is never going to happen.
Instead, Donohue and his team average together the violent crime rates of several states that don’t have right-to-carry laws to construct a “synthetic control.” At the beginning of the “experiment,” the synthetic control’s crime rate is similar to that of a test state that passed a right-to-carry law. It’s easy to compare what happens to the test state’s crime rate compared with that of the synthetic control.
Donohue’s never satisfied, though. He keeps coming up with new ways to test the idea that right-to-carry laws increase crime. Is there a way to punch a hole in the idea? Is there a way someone else could show it’s wrong? So far, the synthetic controls approach supports the hypothesis that right-to-carry laws increase violent crime, as does the Lasso technique, which was refined by Stanford statisticians.
The work totally engages Donohue. But it can be frustrating, too, he says. “You can be in this mode where, let’s say, you’ve tried something nine ways, and all nine point in a certain direction, and you’re saying, ‘Wow, this is great!’ And then you say, ‘Oh, there’s another way to try this,’ and that way, the results are not the same.
“And sometimes you can then figure it out, and that’s a lot of fun. And then sometimes you never quite are able to see why doing it that way leads to this muddier set of results.
“And then, of course, it’s really interesting when your hypotheses are either confirmed or rejected. There are many times when I was thinking the world was one way, and then did the analysis and found the world was quite different from what I expected.”
Donohue clearly loves his work. At 65, he has no interest in retiring. “I feel like I’m just sort of hitting my stride, in a sense. It’s a funny thing,” he says tentatively.
Ever the empiricist, he explains that when economists at the University of Chicago profiled the productivity of different types of professions, they found essentially two types of creative intellectuals. On one hand were people who have blinding insights, which tend to come early in life. On the other hand were artisans who spent a lifetime perfecting their skills and whose best work came late in life. Theoretical economists tended to do their best work in their 30s, while empirical economists, like Donohue, did their best work in their 60s.
“I’m starting to kick it right now,” he says, smiling.