Philip J. Cook
Terry Sandford Institute of Public Policy, Duke University
“Evidence-Based Gun Policy”
March 12, 2003

Well, certainly medicine has a very important craft aspect to it, and it’s not entirely scientific, it’s definitely much more scientific than policy-making in almost any arena that we can think about, and there isn’t in fact a tradition of a systematic evidence-based approach to policy making whether it’s in the gun area or in most any other area that we can think about. I teach in a public policy program at Duke University and, of course, there that is what we are trying to impart to the student is that it is possible even in very contentious policy areas such as gun policy to look at the evidence and to be guided by the evidence, at least up to a point, so that’s the message that I have for today is that even in this area where the debate roars on year after year and is dominated by strong assertions and over-the-top rhetoric, that even in that area, dispassionate analysis is possible and, in fact, interesting, at least to some of us, so I hope I can convince you of those.

Jens Ludwig and I have just put a book last week called Evaluating Gun Policy published by the Brookings Institution Press and much of what I’m going to do today, or talk about today, is based on an essay that we included in that text. Let’s talk about the burden of gun violence before I go on to talk about evidence related to what we might be able to do about it. In the year 2000, the most recent year we have the vital statistics on this subject, as you can see we have almost 29,000 fatalities. It might be a surprise to some people who have not looked at this area that the bulk of them are suicides and that homicides are these days running at about 39%. The top segment there is for unintentional injuries resulting in death from guns, so this is what the picture looks like as of the year 2000.

Another way to look at it is in terms of years of life lost before the age of 65. Homicide and accidents fall greatly disproportionately on teenagers and people in their 20s and so that each of those deaths accounts--in the homicide area--accounts for disproportionate number of years lost. Suicide is more shaded towards the older people, but as you can see, in terms of this particular measure, homicide and suicide are both among the top causes of years of life lost before age 65 and it turns out that guns play a very important role in both homicide and suicide. A majority of both are committed with guns; about 60% of suicides and about two-thirds of homicides, and so the gun death rate then would put it as perhaps the fifth leading cause if we classified things that way.

But 2000 turns out to be quite an unusual year relative to recent history and so if we look at the pattern of homicide since 1950 as this chart, the top line shows you the overall homicide rate and what you see that starting in the early 1960s there was a huge increase that peaked in 1980, peaked again in 1991 and then has plunged since then, and so we during the 1990s actually went through a dramatic decline in the homicide rate and the gun homicide rate increased in the early years faster than the overall homicide rate became more important relatively speaking and has since then pretty much tracked it all the way along and what you see here at the bottom is the non-gun homicide rate, so what you can see from this is that the year 2000 is a relatively benign year in terms of homicide compared with the last 20 years or 25 years.

Suicide--there’s not quite so clear a picture. Suicide rates have been much flatter over this entire 50-year period, but guns have become increasingly more important especially during the early period of that. As you can see from the fact that that back in the 1950s, it was about a 50/50 between gun and non-gun and then it increased so that gun suicide became predominant, so that’s by way of sort of orienting you to the public health statistics in this area, what’s going on.

There’s a lot of questions that come up about what we can do about this severe burden on our society, questions that are personal and questions that are policy. Let me just run through a few of them. A personal question about whether your own household should have a gun in it, perhaps for self-protection or for recreation, but from the self-protection point of view, does that make it more or less safe? What about if your neighbors get guns? That’s a question that people often answer differently than they do about their own household and so they say, yeah, I feel more safe with a gun in my house, but I’d just as soon that the people living around me didn’t have guns because they’re just a little-- Drink too much or they’re a little too scary from time to time, and of course if you’re talking about policy, you’re probably going to be included in whatever the policy decision is and it’s hard to make that distinction.

The current law requires that you be 18 years old in order to buy a long gun from a dealer, that you be 21 years old in order to buy a hand gun. An interesting question about whether those and other prohibitions about gun possession are appropriate or should they be expanded or contrac[tion]. A very hot issue over the last 15 or 20 years has been the question about people’s legal ability to carry a concealed gun and most states now have gone with a law that mandates that local officials issue permits to carry concealed guns almost to anyone who applies and meets very minimal restrictions. Virginia is one of the states that has joined that bandwagon and in fact has one of the least restrictive arrangements on carrying concealed guns. I understand recently there was an extensive debate about whether people who had the permit to carry a concealed gun should be allowed to carry it into bars and restaurants and that kind of thing, so it’s an interesting contentious issue that has occupied the time of a lot of politicians and some scholars.

A lot of momentum from the public health community about the issue about whether gun design should be regulated like other consumer products, like we regulate motor vehicles or like the Consumer Product Safety Commission regulates most consumer products. When the Consumer Product Safety Commission was formed, it explicitly excluded the gun industry and so in fact guns are exempt from that kind of regulation. There is no prior review and so it’s only up to the courts occasionally to review the design of the gun and perhaps find liability if it’s very unsafe, but for the most part, guns are unregulated and then, of course, the question about illegal sales and trafficking of guns--what can be done about that and about the interesting question about whether the current law gives the law enforcement agencies enough information to allow them to connect a gun that they might find at the scene of a crime or know to be connected with a crime to the person who was the shooter or the illegal user of that gun. Should we have ballistics fingerprinting so-called, so there’s a whole series of issues around that which have come up, so this is just to kind of orient you to some of the policy issues that might come along and the approach, I like to think of it as a pragmatic one that is turned to evidence instead of rhetoric and we borrowed a line from William James, the philosopher from the turn of the last century, about pragmatism where he says the pragmatist turns away from bad a priori reason from fixed principles and from dogma, artificiality and the pretense of finality and truth and turns towards facts.

The facts, once you start looking at them carefully don’t always line up neatly to favor one side or another in the debate and that's the only problem with this approach, that if you have a prior commitment you may be disappointed. It’s much easier to argue your way rhetorically to a conclusion than it is empirically, but our ambition for the book and in my own work over the last 25 years is to let the evidence be the guide.

Let me start with what the evidence shows us in kind of a very general sense, just three or four conclusions. This isn’t about particular policies but it’s background for any kind of policy discussion. The first thing is that the prevalence of gun ownership seems to have very little effect or relationship to the crime rate and so we often hear it said that if there are more guns in a community, then that's going to result in more crime or some people say it will result in less crime, but the evidence the way I see it pretty much supports the idea that the gun prevalence in a community has very little relationship to the amount of robberies or assaults or other kinds of violent crime in that area, or burglary, for that matter. So that whatever the problem is with guns, it’s not really that guns are increasing the crime rate but that does not seem to be correct. On the other hand, guns are playing a role in crime and the role that they play is that when there is an assault, if there is a robbery, if that crime involves the gun, then the outcome is likely to be different than if it involves a knife or a club or bare fists, and so that the concern--it seems to me, the proper concern about guns--is that they intensify violence, that they increase the case fatality rate and that to the extent that guns are being used in those robberies and assaults on the street, the homicide rate is going to increase accordingly.

Of course, you’re all familiar with the bumper strip that says Guns Don’t Kill People, People Kill People. My favorite variation on that is Guns Don’t Kill People, They Just Make It Really Easy and that does seem to be what’s behind the patterns that we’re seeing and the evidence.

In terms of gun prevalence and gun ownership, it may be surprising to some of you that only--only--about 35% of households in this country own guns. It turns out that it’s mostly a male thing; 40% of the men in the country own a gun and 10% of the women, so it’s a huge imbalance between the two of them, and that rate differs widely across regions. It differs widely between rural and urban areas and in other respects, and that gives us a basis for finding another evidence-based conclusion and that is that that when you study the connection between the prevalence of gun ownership in a community and the weapon choice by criminals or by potential suicides, what you find is a close connection and so the communities that have high prevalence of gun ownership, you find that the robberies there are much more likely to involve guns and the assaults are much more likely to involve guns and certainly the suicides, and so if there’s a lot of guns in a community, there’s a lot of gun use relative to the overall crime rate with that kind of intensification that I’m talking about, so more guns, more deaths.

Now I’d like to turn briefly to looking at three of the most prominent policies of the 1990s in this area and talk about what evidence can tell us about the effectiveness of those policies. Those three are Project Exile which was in Richmond, the Brady Law adopted by Congress in 1993, and then Operation Ceasefire, a policy that was started in Boston back in the early 1990s, so let’s start with Project Exile in Richmond, something that I’m sure many of you have heard about. The program was started in response to a very high and very rapidly increasing homicide rate in Richmond and the decision was made to try to discourage convicted felons from carrying guns by instead of the proverbial slap on the wrist when they were arrested, instead to divert them to federal court where there’s a mandatory five-year prison sentence for being a felon in possession, so if you’re convicted of it. The idea was to greatly increase the severity of sentencing for felons who were caught illegally carrying a gun and federal law prohibits felons from owning a gun or possessing a gun. The rhetoric around this was that Project Exile has been a great success. It reduced gun homicides supposedly in Richmond by 40% between 1997 and 1998. In fact, homicide did drop that much in that period of time; that it as a result was embraced by the full political spectrum as we know it, and it became one of the models for the predominant gun policy that we have now from the Justice Department and that’s called Project Safe Neighborhoods, so that has been touted by all sides and it has become a centerpiece of the current administration. Further, Governor Erhlich in Maryland, the newly-elected governor there, is promoting that as an alternative to the gun control measures that have been popular there under the Democrats.

All right. So the problem is that nobody ever bothered to do a real systematic evaluation of it, but what we’ve had is a number or two thrown around that wasn’t really tied to the event itself, no real good study until Jens Ludwig, my co-author, and Steve Rafel, took a look at it a year ago, so here’s the background you need--that it was implemented in February 1997. The first thing that happened was that actually the homicide rate increased by 40%, so the rate in 1997 was 40% higher than it was in 1996 and it was only in the next year after it had already been in place that we saw the drop. That’s a little disturbing, but much more persuasive than that, I mean, you might say, well, what was going on was it just took a while for the word to get onto the street before the felons started to leave their guns at home and all the rest of it, and for that reason what the evaluators did was they looked at the changing gun homicide rate from the year before to the year after and came up with this result for Richmond. Sure enough, 31% drop in gun homicide. It’s not 40%, but it is 31%. That’s very impressive, until you realize that something very similar was happening in many other cities at the same time that did not adopt anything like Project Exile during that period, so the lesson here don’t be hasty until you’ve looked at the control group. If we adopt as a control group other cities on the eastern seaboard, then we find that Richmond is in a dead heat with them or if we look at all U.S. cities of similar size, in fact, Richmond didn’t do quite as well during that period, so it seems like Ockham’s razor would suggest that it was not Project Exile that accounts for the big reduction in gun homicide but rather whatever it was that was affecting the U.S. as a whole and a number of other cities like Richmond that would be responsible, so at least this evaluation introduces a control group which we didn’t have in the previous rhetoric and leads us to the conclusion that Project Exile may have had no affect whatsoever. At least, it’s not clear.

Another thing that is disturbing in terms of claiming effectiveness for Project Exile is that the non-gun homicide fell by almost as much as the gun homicide and of course Project Exile could logically have had no affect on the non-gun homicide rate, so--

All right. Here’s the second example--Operation Ceasefire in Boston. This is the other principal foundation piece for the Project Safe Neighborhoods that is our current policy. Again, a surge in homicide, this time almost entirely involving youth gangs in Boston and almost entirely with guns, something like 85% or 90%. What happened was that led by a group at Harvard, there was a multi-agency task force put together to try to solve this problem, to analyze it and say what can we do to intervene, and what they discovered was that the people who were being killed were gang members. The people who were doing the shooting were gang members, that the police knew who they were, they knew the gangs and they knew the individuals in those gangs and so what they devised was a very innovative policy which was what they started to call Retail Deterrence and what it was is that they would bring these gangs, one gang at a time, into the stationhouse, confront them with the U.S. Attorney with a representative from the Bureau of Alcohol, Tobacco and Firearms, the local police, the probation office, and all the rest of them, and say we knew who you are, we know what you’re up to, and we just want to tell you that there’s a new regime in place and that is that next time we hear about you or anybody in your gang using a gun, then we’re going to come down on all of you because all of these kids are basically in trouble with the law and it’s easy to find ways in which they’re in violation in terms of their car registration or in terms of probation conditions that they’re violating or lots of other things, so that that was the threat. And it was delivered in person by this impressive array of law enforcement agencies and then they backed it up by having a couple of cases that got a lot of attention where stiff sentences were handed out for minor violations and the result, again, was that the programs were-- This program was deemed a great success and that the multi-agency problem-solving approach has been picked up by a number of other cities and has been incorporated into the Operation Safe Neighborhoods.

Let me show you some of the evidence in support of this. If you look at what was happening in Boston from 1976 to 2001, now this is just youth victims because that’s what that was all about, what you see is, okay, it’s bumping along during this period at about 20. It suddenly spikes up over 70. That’s what got them busy, got them concerned about it. It’s still going along in the high 40s and then where you see the dotted line is the date at which they introduced this Retail Deterrence program and there was a sudden dramatic drop following that that didn’t seem to have any other explanation for what might have been going on, and so this is-- They also brought in various kinds of control groups and the evidence appears very solid in this area, not 100% conclusive but certainly plausible that it was this particular intervention at this time in this place that accounted for reducing gun use and gun homicide during that period, so here’s the case where a get tough policy seems to have worked and it actually persuaded these gang kids to leave their guns at home and stop using them when they were-- They were still dealing drugs, they were still getting into hassles of all kinds, but they were doing it now without guns because they had been told that that would get them into a lot of trouble.

One more example and that’s the Brady Act, the most important federal legislation of the 1990s, supported by Sarah Brady of Handgun Control, Inc., the wife of Jim Brady who was shot at the same time Ronald Reagan was because Jim Brady was his press secretary and was traveling with him at the time, so the Brady family has made it a personal crusade to get this law adopted and after seven years of effort, they were finally successful in 1993 with the Democrats coming into power at that time. What it did was to expand the requirements of the Gun Control Act of 1968 by requiring that licensed gun dealers do a background check on hand gun buyers. This is something that was never required by federal law previously. Previously, lie and buy was the routine lots of time for people who were disqualified from owning a gun. All you had to do if you, say, had a felony conviction or were under indictment, go to the gun dealer, fill out the form, check all the boxes correctly that says, no, I’m not a felon, promise, and sign your name at the bottom of it and the gun would be handed to you immediately.

Under the Brady Act, in fact, gun dealers all over the country had to do a background check. They had to contact the chief law enforcement officer in the area who then used the information that was available on the individual to see whether that person was in the criminal history files that they had. If so, then go ahead and stop the sale. Furthermore, there was a mandatory five-day waiting period introduced as a result of this.

Now, an interesting possibility for those of us who like the idea of doing evaluation is that this really set itself up like a natural experiment. The reason it did was because this national law only applied to 32 states. The reason was that 18 states and the District of Columbia already had restrictions in place that were equivalent or more stringent than what was required by the Brady Act. For those 18 tight control states, there was no change in the regulatory system. For the 32 states, they introduced for the first time this minimum requirement of doing a background check, beginning in 1994.

And what we heard about from the Clinton administration, from Janet Reno as Attorney General and all the rest of it, was that this law was working. The evidence that it was working was kind of remarkable actually and that is that more than 60,000 people were being turned back each year on the average who were attempting to go there, who lied about their own criminal history and then were caught as a result of this background check and thus stopped from making the purchase at that point. They would add, quite plausibly, that thousands of others may have been deterred from even going to the dealer in the first place. After all, they were committing a felony by going there and lying on that form. So in addition to the 60,000 there might have been tens of thousands of others that were not getting a gun as a result of it.

Of course, on the other side, was the sad fact about the gun markets and that is there’s a huge secondary market that doesn’t involve dealers and was completely unregulated by the Gun Control Act and the Brady Act and everything else and something like 40% of all handgun transactions are taking place in that secondary market, as I say, unaffected.

So it remains to be seen whether it was effective. What we did was to take advantage of this natural experiment and compare what happened to homicide rates in the states that changed with the states that didn’t change and there’s the answer. If you look at those two groups of states and look at the average gun homicide rate, what you’ll see is that it’s going along very much the same prior to the intervention. The intervention happens at the beginning of 1994. Suddenly the treatment states which start out on the top there a little bit, are now in a different circumstance than they were before whereas the control states remained the same. Did it make a difference? Well, if you look at this, it’s hard to see that it made any difference at all for the first few years and in the years since then, if anything, the control states have done better with respect to the gun homicide rate than the treatment states so, again, based on the homicide rate which is a much more appropriate measure, it seems like the ultimate outcome rather than the number of people turned down in the effort to buy a gun, it does appear that the Brady Act was largely ineffective in reducing it, so those are the three most prominent policy initiatives of the 1990s. Each of them lent itself to doing this kind of evaluation and I think we reached some conclusions and, of course, my real interest here is not only to talk about those policies but to point out the general lesson here and that is that even in an area as contentious as gun policy, it is possible to get some data and to look at it carefully to find a control group of some kind and to see whether the changes we’re seeing in homicide and suicide are really the result of the intervention and if so, what kind of conclusion can we reach.

We are not in the position of the drug manufacturers of doing randomized placebo controlled double blind experiments in the field. You’re not going to do that in the area of gun policy, but you can do at least something in this area to begin to get an idea of what policies work and what don’t. And what we needed is that if you’re going to pay attention to outcomes, then you need to have a good measure of outcomes and a control group and that the evidence of what works or what might work is not sufficient to reach an ultimate conclusion about these things since clearly there are very important values at stake in terms of controlling guns and controlling gun policy which also need to play a role in good.