Guns and Gun Violence: Through a Public Health Lens

Full video, transcript and powerpoint presentation; “Guns and Gun Violence: Through a Public Health Lens”; May 29, 2015. 

Click the icon at the bottom of this page to download "The Epidemiology of Firearm Violence in the Twenty-First Century United States " by Garen Wintemute.

VIDEO

POWERPOINT

SESSION TRANSCRIPT

Bruce Shapiro:  We are incredibly lucky to have people of the caliber we have here. Garen Wintemute is a professor of Emergency Medicine, Director of the UC Violence Prevention Research program, Susan P. Baker/Steven P. Teret Chair in Violence Prevention at UC Davis. I have known his work for years. His work is on the epidemiology of gun violence. I will leave it to you to take us where we need to go.

Garen Wintemute: Good morning. I go by Garen. You can call me Dr. Wintemunte if you want, in which case, I will have to figure out what your last name is.

A couple of introductory comments. I am going to be talking about some of the science of firearm violence. I'm a practicing emergency physician at a Level I trauma center. That's what made me a scientist. I was interested in preventing some of the things that brought patients into the emergency room, looking at the screen, and all the events that brought people into the emergency department, and I have been in more ways than one swimming upstream on this issue for more than 30 years now. 

Two other things that you should know -- because I'm a clinician, I'm of a practical bent. I'm interested not just in doing science for the sake of the science, but to make the world a better place. I work with journalists all the time. All the time. There're really smart ones, and the ones who are just here for the day, and you can tell the difference. But I really enjoy working with journalists, and I enjoy contacts from any of you. 

The other thing that I should mention right up front is as Bruce kind of alluded to, there was a little bit of controversy about the funding of this conference. It receives some of its funding from an organization that does advocacy, and I do not take funding directly or indirectly from advocacy groups. You should know that I'm here on my own nickel. 

The other thing I should mention right now is that I'm going to go through the material that I've brought at an indecent rate of speed because I am interested in the interactive Q&A part of this that will come at the end. It will come before you expect it. It's going to come during the talk. I think part of the fun of being a scientist, part of the fun of being a journalist -- we're all investigators, right -- is questioning established wisdom. It gets me up in the morning. And we're going to go through criticism of a very widely-cited study that in my judgment was reported on uncritically. We're going to go through and exercise, together. And this is going to be interactive. If somebody is tweeting, I'm going to walk over and say, "You get to address the following question." So if you want to tweet now, you can tweet now. When we get to the interactive part, and you're tweeting, you are up for grabs. 

So having said all of that -- we all are used to thinking, or getting used to thinking of firearm violence as a public health problem. That's an understanding that's actually of quite recent origin. And just over 20 years ago, when it was a hard case to make -- people had not been thinking about it in that way -- the then the head of the Centers for Disease Control, he had just become head of the Centers for Disease Control, was interviewed in the New York Times and was asked about the nature of his job, and interviewed about violence and the other issues that he talked about. He tried to make the case that violence was a public health problem. He gave one of the best reasons, I think, that has been offered. He said simply, "If it isn't a public health problem, then why are all these people dying from it?" Simple enough. 

So I'm going to be approaching violence as a public health problem. We're going to look at some epidemiologic data. I believe that you have, or at least have access to a paper (link to PDF of paper) that we published just a couple of months ago that has in it many of the slides that you're about to see. 

Beyond the epidemiology, we are going to talk a little bit about policy. I was asked to talk about the science, so there's not as much policy here today as I would ordinarily put in there. I love policy. So as policy questions. I do a lot of policy work. I had mentioned the critique, and I am going to close with this slide. Please understand that it was not a prescriptive slide. It's some suggestions. The last slide is going to be a list of what I think are some of the big stories that are coming. I guess I'll put it this way: if you see something on that is already of interest to you, I am telling you to go for it, because there's going to be a lot of attention paid to it. 

With that, before we move further, what I'd like to do is just talk about all those people that the CDC mentioned. 

The basic numbers: in the most recent ten years for which we have data, we have lost more than 300,000 civilians in the United States to firearms, suicide, and homicide. For the sake of this talk, I am going to be using the term "firearm violence" to refer to suicide and homicide. I'm not going to be talking about unintentional shootings that are relatively uncommon. We can certainly talk about them during the Q&A. We're not going to be talking about the deaths of undetermined origin and so on. 

About 30,000 deaths a year, nearly 70 percent of all homicides in the United States are committed by firearms. More than half, and an increasing majority of suicides in the United States are committed with firearms. And the costs were estimated in 2010 to be what I think most of us would consider real money, the kind of money that you can express as a percentage of GDP. If it's okay with you, I'm going to hold questions …

BS: One question. This is updated data? I just wanted to make sure. 

GW: Yes. Mother Jones and ProPublica -- they have actually the same economist, Ted Miller who generated this estimate, collaborated with Mother Jones on the revision, which was published earlier this month. And this number is over the span of the ensuing five years. This number is now 223, and if I were to stay on script, a popular rifle caliber. 

But by now, some of you may be wondering why does the guy have Arlington in the background? We've just marked Memorial Day. As an epidemiologist, I am always looking for comparisons, some sort of way to ground really abstract numbers like this in a context that will speak to a more general readership. We all in the room have faced the challenge how to get what we've just learned but nobody else knows how in a way the people will not only read it, but they will remember it and be energized by it. 

I was looking for comparisons and came up with the following. In the last ten years, we have lost more civilians to firearm violence than the United States had combat fatalities, all mechanisms of injury put together, in World War II. We have lost in the last ten years, more civilians to firearm violence than the United States has experienced in all other conflicts, not including World War II, in its history, from the Revolution to the current conflict in Iraq and Afghanistan combined. 

Arlington has more than 400,000 graves. We have been burying people in Arlington for 151 years. We could fill a parallel Arlington with civilians somewhere between 12 and 14 years. 

I'm going to digress for a moment. We can come back to this in Q&A. A couple of comments on mass shootings -- this is Aurora -- and mental illness. Mass shootings were the lead news story in 2012. They outpaced the election. We were talking before the meeting -- things are different since Sandy Hook, and indeed, things are. Just like things like this are happening and they are well-attended. But there are times as a researcher, someone who is guided, you need to personally come to realizations and publicize realizations that run contrary to the public's understanding in ways that are sometimes painful for the person making the disclosure, painful for the people period. And there are a couple of those here.

Mass shootings are not the problem. Mass shootings are the most easily described portion of the problem, and that theme has been mentioned a couple of times already. But look at it this way. Columbine and Virginia Tech and Aurora and Tucson, where Gabby Giffords was shot just down the road, and Sandy Hook. All together, 90 people died, other than the shooters. And I am not minimizing those 90 deaths or the tragedies that continue to unfold among their survivors. But we lose on average 89 people every day, on average, day in and day out. A very small percentage of the problem that we are here to talk about are represented by these sorts of events. And no two of these events are alike. 

The policy implication is we cannot base policy on events like these. If we do, we won't prevent the next such event because they're all different. And we won't be doing much about the far larger part of the share of the problem as several of you mentioned. 

Point number two: mental illness. There is a very strong, and increasingly strong perception among the public that crazy people are behind firearm violence, and that is just not true. When it comes to interpersonal violence, one person against another -- less than four percent of cases are attributable primarily to mental illness. I'm going to come back to this theme in a minute. Mental illness plays a huge role in firearm violence, just not in the way we think it does. And I will come back to that point. 

Okay. You have these. And if you don’t, let me know, and I will send them to you. We're going to look at some of the data. I'm going to offer in, and there will be a lot of comparisons here. That's the bookmark. As you all know, I used to give this as a quiz. Over time, everybody knew the answers. Firearm violence is predominantly a suicide problem? It was predominantly a suicide problem even back in the period of time between the mid-1980s and about 2000, when rates of firearm homicide were far higher than before or after. Note what's happening here, over the now what's close to ten years. Note this divergence. Notes of firearm homicide have been on balance, I think, slowly declined. We would fit a downward sloping line to that, where rates of firearm suicide have been going steadily up. I'm going to come back to that several times before we're done. 

I should mention one of the first papers I ever wrote 35 years ago now, when I was a punk, looked at the same sort of comparison for the 20th century. And this was true going back all the way to the 1930s, with very rare exceptions. Fatal firearm violence is not predominantly a crime problem. It is primarily a self-directed violence problem. It was the fact that the crime part was dealt with by criminologists and the suicide part was dealt with by mental health experts, and the unintentional shootings were dealt with by safety experts and hunters -- it was the fact that there were these three seemingly separate problems dealt with by separate groups of people that in fact had a common denominator, the technology, that led to the first efforts to portray firearm injury and death as a public health problem. 

Let's look at a little bit more detail. We've shifted to 2012. I just haven’t had time to update all the slides as new data have come out. We are looking here at death rates from firearm homicide only from males in the year 2012, by age, by race/ethnicity, and we are looking at deaths per 100,000 people. This is a risk-based graph. The image of firearm violence that most people have. You are familiar with this graph, even if you’ve never seen it. This is a graph that says firearm violence is predominantly a problem of young African-American men, and if we are talking about homicide, that's true. In fact, you have to look hard. I have not found one, but not for the lack of effort, you have to look harder than I have looked so far, in order to find a major medical or public health problem for which risk is so concentrated as is firearm homicide on young African-American men.

Now let's go directly to a policy implication. Let's take the phrase "Black Lives Matter." That phrase has the peculiar resonance that it does in this culture because all of us know, that for many of us, that's not true. We have to admit that, whether it's true among ourselves or not, we have to admit it's true about our society, that not all lives matter equally. And I bring that uncomfortable point up because it is a daily part of trying to do something with these data at the policy level that one recognizes that I am people who consider themselves the policy makers, people who consider themselves not to have a stake in the issue, because it affects people who are not like me. 

Now don’t take my word for that. I'm going to read you a quote. This quote comes from a man named Jeff Cooper. Jeff is dead now. He lived to a very old age. Jeff was a Marine. He was one of the architects of modern combat handgun tactics, and an official of the National Rifle Association on his retirement from the military and other things. He ran a ranch that taught people how to use firearms under combat conditions in Nevada. The ranch still exists, and I'm told it's a very fascinating place to visit. 

Jeff Cooper had this to say about this phenomenon. He said that here when the crack epidemic was in full swing. "The consensus is that no more than five to ten people in 100 who die by gunfire in Los Angeles are any loss to society. These people fight small wars amongst themselves. It would seem a valid social service to keep them well supplied with ammunition." That quote appeared in his column in Guns and Ammo. He was a thought leader in a particular aspect of the gun culture in the United States. 

Gun culture in this country is a misnomer. There are all kinds of different sets. I don’t use the term "gun culture" just like I don’t use the term "gun control." They are both misleading, overly simplistic, overly general. 

So, homicide. Let's look now at suicide. Same general display. Suicide, males 2012, death rates by age, the picture is completely different. Homicide. Suicide. In the case of suicide, risk for white males is highest beginning at the first age at which risk is measured. There are suicides among five to nine year olds in the United States. The differential increases, and then it keeps on increasing. The age-related increase in risk in firearm suicide among white men is greatest as they get older. 

Now I'm going to aggregate homicide and suicide. Now we're talking about firearm violence. But let me go back a couple slides. Watch the resemblances. So this red line looks very much like the red line two slides hence. Most firearm deaths among black males are homicides; most firearm deaths among white males are suicides. I'm not going to get all the way into that, but my point here is this. As we aggregate, what you see is essentially a homicide curve here, with a relatively smaller number of suicides here, and what you see here, is essentially a suicide curve with a relatively speaking smaller number of homicides. So if I can alternate again, this line right here pretty much is this line right here. It's just like been squished down toward the horizontal by the far larger risk that exists among African-American males.

Everybody's with me to this point, right? Now I'm going to mix it up some. Up until this slide, I have been taking a risk-based approach to the problem. And there's much to be said for a risk-based approach to problems like this. The group at highest risk potentially has the most to benefit from things that you might do about the problem. So it is traditional in public health to start with identifying the high-risk groups, and then potentially to focus our interventions on them. 

But there is a complementary approach to problems like this. It's not contradictory. It is complementary. They work really well together. This approach is called the population health approach, as opposed to the public health approach. The population health approach stems from this observation that the greatest number of adverse health events, I don’t care what we're talking about -- actually heart attacks are the prime example. The greatest number of adverse health events can occur among members of the population who are low risk, not high risk, if the low risk group is big enough. 

Most heart attacks occur among people who don’t have a boatload of risk factors for heart disease, because there are so many such people. Everybody has that, right? This is risk. The approach that I just talked about, in place of risk, substitutes what is called burden. The burden of illness. Where are the most cases? So the next slide is going to show you these exact same events. The only thing that's going to change is I'm going to get rid of the denominator. And instead of talking about race, we're just going to talk about number of cases. So risk. Burden. There are a lot of white males in the United States, such that they are relatively low risk for firearm violence, predominantly suicide, leads to the picture you see here. So burden. Risk. Same events. Risk. Burden. I can do this all day. The red line carries over. It's a small group, but risk is so high, it still shows up when you look at burden as an obvious peak. 

But the visual message here is obvious. By the time you get to age 35, how many people have not yet hit age 35? So incidentally, I meant to say that what I really like about this group is that there are some old people in the room, and I can say that safely because I'm the oldest person in the room, and there are some people just getting started, and that's perfect. I'm sorry I'm not going to be here tomorrow, because that changes a lot of things. In any case, by the time you hit age 35, most deaths from firearms violence among males in the United States occur among white males. By the time you hit age 60, an age I last saw quite some time ago, more than 90 percent of deaths from firearm violence in the United States among males occur among white males. They're suicides; they're not homicides. 

Now two more points to make about that. One is, this problem -- the slide in the paper, I didn't put it in for the sake of time -- homicide rates among young African-American males, this peak right here, this 30 year age span, have increased in the last roughly ten years by about five percent. We'd like to see a decrease, but the increase is about five percent. The rate of suicide among white males ages 35 to 64 in the last ten years has gone up by 30 percent. This is the edge of the wave, right here. 

The other observation to make is that this is where mental illness shows up. I mentioned that about four percent of interpersonal violence is primarily attributable to mental illness. For suicide, the estimates range from between 45 percent and 75 percent, directly attributable to mental illness. But it's not psychosis. It's not "the voices are telling me to kill myself" -- it's depression, which is much more common than we think it is. And there is a whole body of literature to buttress the statements I just made. 

One other point. The question comes up -- What are we going to do about this, if anything? We could adopt the Jeff Cooper approach. It's really simple, actually. All you have to do is just change a preposition. And we could say, "These people fight small wars within themselves." For this group over here, if we felt like it, "Besides, they're going to get old. They are old. They're going to get sick soon, and their being sick is going to cost us a lot of money, so why are we intervening?" And we could conclude that it was a social service to keep them well-supplied with ammunition. I'm going to argue that that's not the approach we're going to take here, and I would argue that that's not the approach we should be taking over here either, although to some extent, I think we still do take that approach. 

Okay, end of discussion about science. We are going to move into the science of policy because I can't resist. 

What I want to talk about is a couple of the main risk factors for firearm violence. The ones I'm going to talk about other than age and sex, which we've talked about already, and race/ethnicity, are alcohol abuse -- I'm not trying to talk about controlled substances. We're going to talk about alcohol, and we're going to talk about a prior history of criminal activity, in particular a prior history of violent criminal activity because there are some good data.

Before we do that, I want to talk to you about a policy issue. It's been mentioned already and I will mention it again that comes into play here. A couple of mentions were made of polarization. And there is polarization on this issue. It is not between the NRA and other people. And where that has most clearly been established is on the question of comprehensive background checks. If you ask people, "Do you support gun control?" people tend to say no. I say to that, because I don’t know what it means. But if you ask people about specific policies, level of support goes up, quite dramatically. 

So for example, for comprehensive background checks, support is somewhere between 85 and 92 percent-ish percent of the general population. It's above 80 percent for firearm owners. It's above 70 percent for members of the National Rifle Association. The divide is not so much between the people who own guns and people who don’t, it's between the leadership of organizations that have one sort or another of a vested interest in maintaining an appearance of polarization when in fact no such reality exists. 

Sorry, that was a digression.

The reason I brought up background checks at this point, apart from the fact that they're an advancing story is there is widespread agreement, that people at really high risk for committing firearm violence should not be allowed to have access to violence. There are legitimate differences of opinion over what constitutes sufficiently high risk, but the devil is in the details. 

Background checks make enforcing such policies possible. It's just that simple. You can't enforce a policy that says a person who is at really high risk of imminent violence can't buy a gun. You can't enforce that policy if at the moment of purchase, you don’t know who the members of those high risks groups are. And that frankly is all the background checks do: they provide a mechanism for ensuring that as I'm about to sell this gentleman a gun, I can one way or another, know if he's prohibited or not, the very next time notwithstanding. He tells me he is. 

These risk factors, by the way -- prior violence, alcohol abuse, controlled substance abuse -- apply to the mentally ill as they do to the rest of us. And that's another polarization that I think does not serve us well. But let's move on. Let's talk about prior violence first. This is work that we did in California, looking at the question, "Is a history of prior criminal activity a risk factor for future criminal activity among people who legally purchase handguns in California, where we have the data that allows us to do this. The comparison will be here to people who bought guns, but had no prior criminal record. In California, it's possible to administratively enroll, if you will, large populations of people who have purchased handguns -- we know the exact date on which that purchase occurred -- look at criminal records before and after, and come up with the answers to these questions. 

So let me lay out this slide a little bit. In this study, I should mention about 4000 handgun purchasers in the study. We followed them for 15 years after they purchased their firearms. In characterizing their prior histories, we looked just at convictions. By the way, there are no felons in this study because these people bought their handguns legally, and if you're a felon, you can't buy a handgun legally. So we looked at prior misdemeanor convictions, and these categories here were assigned hierarchically. So if you were eligible for this one, that's the one that you were in. So people who had prior convictions for some kind of misdemeanor that involved neither firearms nor violence. One of them, or two of them. Prior convictions involving firearms, but not violence, illegal carrying, one or two. Prior convictions involving violence whether firearms were involved or not. 

And across, looking at outcomes over 15 years, we looked at something called relative risk. I'm going to assume that not everybody here is comfortable with what a relative risk is. When I say "relative risk," you think times as likely as. It's not percent. It's a factor of. I'll finish off definitions in just a second. 

We looked at relative risk of arrest for various kinds of crimes. Violent crimes index, murder/rape/robbery/aggravated assault. So when I say that compared to people with no prior criminal record, those legal handgun purchasers who had only one prior conviction and that for an offense involving neither firearms nor violence had a relative risk of 5.9 for arrest for any offense. You think 5.9 times as likely to be arrested for, etc. 

So what you can see here, if you use convictions as our benchmark. There is no level of prior activity below which risk for future criminal activity is not significantly increased by a factor of basically five among people who purchase handguns legally. Is everybody with me? 

I'm going to point out just a couple of other things on this slide. Notice what we call the response relationship. As we go from one conviction to two or more convictions, the risk ratio grows. That's a very common finding in research like this, and it lends plausibility to the results because it's consistent with whatever research has shown. The badder you were, the badder you are likely to be. 

BS: So Garen, just to be clear. Convictions without a gun and without violence, you're still five times more likely…

GW: Five times as likely.

BS: Five times as likely …

GW: To be arrested. 

BS: To be arrested as someone with no priors whatsoever.

GW: Correct.

BS: And it moves up from there.

GW: That is correct. 

Okay. So I'm just going to go down to the bottom here. People with prior violent misdemeanor convictions, two or more of them, among legal purchasers of handguns -- anywhere from 10-ish to roughly 15 times as likely as people with no prior criminal record to be arrested in the future. That is a big difference. 

All of these people bought those handguns legally. Through a licensed retailer. That's how we know about them, and that's how we have records. In most of the country, all of these people still can buy handguns or other kinds of firearms legally, because the laws haven’t changed. California changed its law, and not as a result of this research; the change in the law was made before this was done. But we have the natural experiment to evaluate and we did. What California did was say, not having these data, but knowing as the way people do, "This is unacceptable." California made it illegal to purchase handguns, or possess handguns if you had a prior conviction for a violent misdemeanor. 

And here's what happened as a result. You're used to this slide. In this case, are reference group is "People who had their purchases denied," I'll back up in just a second and do the backstory. And we are looking again at relative risk of arrest among violent misdemeanors. 

Here's the setup. The state changes its policy. We take two groups of people. Everybody in both groups has a violent misdemeanor conviction, at least one. Everybody in both groups tries to buy a handgun. Group one, old policy, purchase approved. Group two, new policy, purchase denied. The groups were, frankly, eerily similar in demographics and everything else. The only difference between these two groups of people was Gun yes, Gun no. And here's what we found.

You know what the relative risk is. We looked at relative risk for crimes involving firearms or violence. And again, we are comparing the people who were handgun approved, or if you will, exposed to a handgun purchase, and, if you will, our unexposed group are the people whose purchases were denied. So that exposure was associated with a 30 percent increase of risk of arrest for a crime involving firearms or violence. At age 21 to 24, when the absolute rates of arrest were highest, you don’t have that here, but take my word for it, and it is concordant with your experience, risk was increased, risk associated with a handgun increased by 40 percent, blah blah blah.

Here's a group with which we saw no effect. People with three or more misdemeanor offenses involving firearms or violence. Denial of purchase was not associated with benefit in that group. I'm not actually surprised by that. I didn’t predict it, but I'm not surprised. These, I suspect, are people -- three or more prior convictions, lord knows how many offenses -- with an established pattern of behavior that involves the sorts of activities that leads to things like this. This is speculation, but I think it's plausible. 

These are the people who don’t change that pattern of behavior. These are the people who, if they are denied a handgun from a licensed retailer, will find some other way to get a gun because they've got a pattern of behavior to maintain. But this is a very small group, and the overall effect is here. I could turn this upside down and say that risk among those who were denied was about 70 percent of risk among those whose purchases were approved. And frankly, if I had it to do over again, I would have done the analysis that way. But you stick with what you started with. 

But a small group, overall effect. When I'm talking to policymakers about this stuff, I say, "Okay, so that didn’t work for that high-risk group of continuing offenders." That's what we have cops for, and what we have other approaches for violence prevention for. There will be no single approach to violence prevention that will prevent all violence. I hear people say, "Let's just get rid of all the guns," and everything will be fine. And my response, which I only verbalize when they're not around in a setting like this, is "What have you been smoking?" It is just not going to happen.

Okay. Alcohol. I should just mention at this point, by the way, we are releasing a study basically today -- it's a pretty lengthy review article. Everything you are about to see here is in that article on the relationship between alcohol misuse and risk for firearm violence. There are press releases. Kate has copies. 

GW: So we'll just add these to the resource table and if you're interested, you can grab a press release. 

So let's talk about alcohol. First off, boy, do we drink in the United States. A 30 day history of alcohol use; information obtained by government surveyors. So I'm willing to say to a representative of the federal government I am willing to provide you information about my alcohol behaviors. In a given month, more than half the population drinks some; 17 percent binge drinks. That's five or more drinks in a single setting for men; four or more drinks in a single setting for women. That's not sexism -- that's body weight. Six percent drink heavily on a chronic basis two or more days for men, one or more day a week for women. So lots of exposure to alcohol.

Second bullet point. Lots of exposure to firearms. There are maybe 50 million, 60 million -- I use numbers like this, and Phil Cook, who you're going to hear from next, is really good at criticizing them. So I don’t know if you're going to criticize this one or not. But the General Social Survey, which gathers information on a regular basis, asks the same questions over and over again, "Is there a gun in your home and if there is, do you own it?" … this is an estimate from their data. Don't worry about the number. The point is this. There are a lot of firearm owners in the United States, and there are a lot of people who drink in the United States. And you will remember Venn diagrams; if two exposures are really common, the overlap between those exposures is likely to be common too. And indeed it is. 

As a result of the study, the best estimate is this: if we assume that firearm owners drink alcohol and do these other sorts of things no more commonly than the general population does, it's just a matter of back of the envelope math to estimate that in a given month, something like nine million firearm owners binge drink at least once. Something like 2.5 million firearm owners drink heavily on a chronic basis in any given month. 

But there's a range here. The reason for the range is this. I will point back to the data from a CDC survey project the Behavioral Risk Factor Surveillance System. Not from 2010, we had to use data from the mid-90s because that's the last time the CDC had the guts to ask these questions. After adjusting for age, race, sex, and state of residence, firearm owners are not just more likely to drink, they are substantially more likely to binge drink and to drink heavily on a chronic basis than are people who don’t have firearms at home. And the high range of the estimates come from applying those data from BRFSS. So if you believe the high range, close on 12 million gun owners binge drinking in a given month; 3.5 million drinking heavily on a chronic basis.

Okay. So let's assume for the sake of argument that there are a lot of firearm owners who drink out here. Does it matter? It matters first off in this way, just to look at size of the problem. This again is in the new paper that's out today. We used again some data from CDC from its Alcohol-Related Disease Impact Service, to compare alcohol-attributable deaths from motor vehicle crashes and various kinds of violence over these four years. It's an interactive website. You can do this too. 

And what we see -- I'll just focus on the results for males -- that based on the best data available for estimating … I actually tried to derive the data a couple of other ways and came up with essentially exactly the same answer. There are among men at least as many alcohol associated deaths from firearm violence as there are from motor vehicle crashes. So where’s the MADD for firearm violence?

Okay. So it's common. What's the risk? We'll go back to the public health side. So some of you in the room will be comfortable with the difference between prevalence and risk. Prevalence simply refers to how common something is. So up here, we've got some information is acute alcohol intoxication among various sorts of people who either do bad things or have bad things happen to them? So 37 percent of violent felony perpetrators, by their own report -- and by the way, these surveys are from the Bureau of Justice Statistics. They are done on inmates who are already in prison. There is no secondary gain in saying, "I was drunk, reduce my sentence." They are serving their sentences already. 

For people who die from firearm violence, we have autopsies, and we're not asking questions. And so acute intoxication is very high. That's prevalence. We need to think again about risk. Here's risk. 

Maybe it's the case that 25 to 30 percent of the general population is drunk at any one time. Or at night. Let's say that these things might be more likely to happen, and there is no risk. It's just in the general population. How many people are drunk in the room at the moment? But here's risk, and you've already read to the end of the slide. And somehow, that slide got reformatted a little bit. Oh well. 

In any case, we're back to relative risks. So the relative risk for perpetrating interpersonal violence goes up by a factor of three to six, 300 to 600 percent, associated with prior history of alcohol misuse, victimization the same, kind of the same for suicide for pre-existing. Look at this, for the relative risk of committing suicide with a firearm associated with acute intoxication. And the way this is done is look at blood alcohol levels in a bunch of people who have just shot and killed themselves, and look at blood alcohols in otherwise similar people who live nearby at the same is the thinking. Risk is up by a factor 75 to 85. That's a lot. That's just obscene. 

So with junk science -- I don't know if you can see it that's a picture of my copy of America's first freedom. And if I recall correctly, both Phil and I are mentioned in this article. 

I was going to use this to launch a critique of a study and I talked too long. So I'm going to stop. Questions? Comments? 

Audience Member: Just off the top -- your focus on your presentation has been on males. Is that simply because gun violence isn’t a significant problem among females?

GW: I'm just trying to figure out a quick way to get to the slides. The longer version that doesn’t have the policy part presents the slides for women as well. The rates of fatal firearm violence are far higher among men than they are among women, after adjusting for everything else; testosterone poisoning is a real phenomenon. But the patterns tend to be the same. The lines aren’t smooth because there are fewer deaths. But for homicide, African-American women are at greatest risk. 

In fact, there are some interesting interactions. African-American women are at such high risk for firearm homicide that their risk is greater than the risk of white men. For suicide, white women dominate from start to finish. But there isn't the continuing up with age curve we saw for men. 

Audience member: So in the California data, we were looking at a lot of comparisons between people who had guns and were denied guns, things like that. Are there any data for people who actually even -- I know, given the data set you were working with, but any data for people who choose not to have guns around at all?

GW: Got it. Everybody's hearing the questions? I don’t need to repeat them, I'm assuming? Okay.

So we didn’t do that. And the reason is this. It derives partly from the data we have, and partly from the data we don’t have. We have good data on people who purchase guns from a licensed retailer. We know that those people have guns. We have talked for years about doing a study that might involve, let's say using our sampling frame, licensed drivers. Or in California, people with an ID, because they are all in the same database. We could do that, and we could even get fancy and we could delete from a list of people taken from that data set, the people who are in the gun dataset. But we still have no assurance that the people left over don’t have guns. 

Because the data for gun ownership in California, until the beginning of last year, were only for handguns, so we don’t know about people who have rifles and shotguns. Those purchases go through dealers. They generate background checks; there's just no archive. The archive started in 2014. So years from now, maybe we can do that. 

But then there are the really illegal as in criminal intent, the nominally illegal but no criminal intent purchases that go not through a licensed retailer, and we don’t have that information. 

BS: So if I'm a local/regional reporter, and I'm trying to evaluate whether some policy proposal either by an agency or the legislature or an existing law is capturing the right risk pool -- through your lens, the question I should be asking is "Is this capturing people who have committed multiple misdemeanor acts of violence with our without guns?" That would be the first question, one of the first questions I might ask?

GW: Definitely. There's no for any of this, you all know that.

That would be one of the questions. If the intent of the policy is to prevent violent crime. Actually, I'll just mention about the study that we're not going to look at in detail. They kind of did this in a silly oversimplified way. They took a list of 28 different laws that states had passed, and they ranked states by how many of these laws they have had, beginning and end of assessment, basically. And they looked at rates of death from firearm violence. With encouragement, they looked separately at suicide and homicide. They weren't going to do that. But when they did that, what they found was that having a bunch of these laws -- the laws were intended to prevent crime. All of the effect was on suicide. And that wouldn't have been known except the reviewers wouldn’t let the article go through without that addition. 

And the other piece of it is this: something else they didn't do -- all they did was say, "How many laws do you have, and what kind of death rates do you have?" The other thing they didn’t do originally was include in the statistical modeling, "How common is gun ownership in your state?" When they did that, when they were forced to do that, all of the effects that they saw went away. And they refused to talk about that in the published paper. 

But in any case, you have to think about what's the intended effect of the law. You have to look for those effects. And then you have to look for the unintended effects. 

Audience member: I have several questions. 

GW: You're going to get to ask one question, because other hands went up. You can ask one question, and it cannot be a compound question. I will take the first part of a compound question. 

Audience member: I can't ask about this data? I will ask it offline. 

GW: Yes. I will give you my card. We will start a beautiful relationship, but other hands went up. 

Audience Member: My question is then the research. I mean, how do you as a reporter -- we are into data. And clearly because even research becomes the focus of criticism, that you didn’t have the right approach, or that you excluded certain data sets or certain people. How is the reporter, there has been a real movement by the NRA to keep crime data from guns that I was once able to access, now we can't. Congress won't let me. So what recommendations for data sets would you as a reporter look for in order to find a holistic approach to the problem with guns or not having a problem of guns?

GW: Good. So here are really good resources. We have good mortality data in the United States. And those mortality data are available for your use interactively in real time. The CDC's WISQARS system. Web-based Interactive Statistical Query and Response System or something, but they have a cute little mountain lion that wiggles her whiskers as you look at the website. 

The National Violent Death Reporting System has very, very detailed data on violent deaths in the states that participate. Many states do not participate. 

Audience Member: What was the name of that?

GW: The National Violent Death Reporting System. It's available at the same website that brings you WISQARS. The same webpage. And those are the two that we most talked about. On the crime side, uniform crime reporting data by the FBI, we have pretty good crime statistics. Ask Phil about this. They're the best that we have. They're about the best in the world, but they leave a lot to be desired, particularly when it comes to studying firearm crime. We have the National Crime Victimization Survey. Another interactive where you can do your own analyses. 

On the private side, I'm going to pick one and then stop. The General Social Survey. I mentioned that before. It's maintained by the University of Chicago. I mentioned it in part because the data are really good and carefully maintained, and in part because they also have an interactive website, and you can do your own analysis. Their website is not user friendly, and if you work for an organization that has data geeks employed to do that kind of stuff, this is one time that you might need their help. The others you can do on your own. 

Audience member: This may be a little bit more of a philosophical question. But I think that a lot of news organizations, mine included, don’t do a lot about suicide unless it's a public figure or a public place or someone has committed a murder-suicide. We don’t typically report on suicide. So it's surprising to see how big of an issue gun suicides really are. And I'm just wondering how you think reporters should be more involved. 

GW: Perfect. So think about it. The whole thing is right there. “My organization doesn't report about it," but I heard the audience reaction when I put this up. Nobody's organization reports about it. Nobody knows about it. And potentially, if we had known about it sometime back, we wouldn't today have the problem that we have. So I would argue not just to you with the question, but to all of you as investigators, as bringers of the light and truth to people who don’t have the ready access that you do. 

Suicide is a story to investigate. How do you go after it? I think, a couple of possibilities. One is if you can take an aggregate approach, and there are a lot of these deaths. Great hooks are easy to find. You can lead with local data that you can get from your county medical examiner or coroner. You can interview those families if your organization allows you to do that kind of stuff. You can talk about the very large bigger picture of which your anecdotes are a part, and you can talk about isn't it interesting how we've ignored this problem? And as we've ignored it, it's been increasing year by year. What's wrong with all of us? So the inattention becomes part of the story. 

BS: Frances, the Dart Center actually did a workshop like this one on covering suicide a couple years ago. And on our website there are a lot of resources generated by that workshop, including experts and tip sheets and all kinds of stuff. 

Audience member: Just in general, how much or little is known or is there a way to gauge a number of deaths by illegally versus legally obtained firearms?

GW: No. Here's the problem. We don't have a registry of firearms in this country. We don’t even have them in California. We used to have access to gun tracing data. Guns used in a crime, local law enforcement sends information to ATF, and aggregated across hundreds of thousands of records, you could develop a picture of the operation of a criminal gun market because you could see where guns started and where they end. 

Cheryl, say something.

Cheryl Thompson: I'm going to talk about it tomorrow. 

GW: Beautiful. Are you going to talk about how we used to have the data but now we can't. 

CT: Yes.

GW: So y'all have to come tomorrow.

BS: And I'm putting up here a parking lot for issues that we don’t quite have time to talk about now, but maybe at some point in the day we will.

GW: Okay. Sir?

Audience member: Are you familiar with John Lott's research? And what do you think about it?

GW: The question was if I'm familiar with John Lott's research and what do I think about it? I'm familiar with his research. I know John Lott. We have met a number of times. 

BS: Maybe quickly summarize for people who don't know … 

Audience member: What would you say to journalists who are looking to his research in gun law …

GW: If I could say one thing to journalists who are looking to use his work? John Lott's major supporter is Ted Nugent. That's the one thing I would say. A great deal of time and energy has been well-spent pointing out, sort of like I would have done here, the obvious easy fatal mistakes in John Lott's research. And the third thing I will say is John Lott is one of a large number of people investigating violence in general, and firearm violence in particular, and his conclusions are unique. Nobody agrees with him. Think about that.

Okay, next. Sir?

Audience member: I will try to ask my question as well as I can. You talk a little a bit about the criteria that we know are associated with interpersonal gun violence. For example, you have multiple convictions, you are probably more likely to use a gun than you would if you didn't have violence. It seems to me, and correct me if I'm wrong, but we know less about the actionable risk factors for self-directed violence are a little less clear. I was wondering what are, there have been some movements in policy to address that. I know in some states, if you've been committed to a mental institution either voluntarily or involuntarily, you can't legally own a firearm. I think Pennsylvania is one of those states. 

I was wondering what's the current thinking on that?

GW: A couple of things. One is, on the crime side, we have been willing to say that if you are a member of a high-risk group, we are going to curtail things that you might otherwise be allowed to do. We are less willing to adopt that group approach when it comes to mental illness. I'm going to leave that aside.

It turns out that most of the main risk factors for interpersonal violence are also risk factors for self-directed violence. Sex. Age. Except it works differently. Old guys who are self-directed. Young guys who are interpersonal. But we can't do anything about that. We do it for young people. But we're not going to say if you're over 65, you can't buy a gun, I suspect. 

But alcohol. I didn’t talk about the policy implications, because I wanted to do that critique. But there are efforts to say, "If you have a bunch of prior misdemeanor convictions for offenses involving alcohol, you shouldn’t be allowed to buy a gun." The article that we have coming out today is about that potential issue. And alcohol applies equally to interpersonal and self-directed violence. 

So does controlled substance use. We have law about that, but it's really crummy and unenforceable. 

And then with regard to critical points in the history of mental illness, on the one hand, the statement I made about interpersonal violence is true. A small portion of the problem. It is also true that most people with serious mental illness never engage in violence. Those who do are likely to do so again. There are critical points in the history of mental illness that are associated with a very, very high short-term risk of violence in the future, one of which is a hospitalization. So in California, for example, federal law does not do this. We do. If you are hospitalized involuntarily for dangerousness to self or others on the basis of mental illness, you are prohibited from purchasing or possessing firearms. Period. 

That prohibition sunsets in five years. The policy recognizes that the conditions that led to the need for the prohibition may be temporary. That approach is becoming more common. And the last thing I'll say about this is, and I know this more as a clinician than as a researcher -- there are all those people who are clearly, something clearly bad is going to happen with them, but there is no crime involved, and they are not holdable. We can't involuntarily admit them. I'll give you an example; it happens all the time. 

A guy comes in, he's really drunk. He's lost things in his life lately. He's suicidal. And at two in the morning, we put a hold on him. At eight in the morning, he's sober. He's got a headache, but he's not suicidal. And he's really not suicidal. And he is no longer holdable and we lift the hold. And that can happen several times in a month. And the hold gets lifted every time, and then he kills himself.

There is a new mechanism in California that has been enacted. It is the law. It won't take effect until January. We are going to evaluate it. It's called a gun violence restraining order. It is modeled very closely, consciously, on the domestic violence restraining order policies in California. For a guy like him, this new mechanism will allow either immediate family members or law enforcement to go to a judge. There are rules of evidence, information that the judge must consider, information that the judge may consider, but it's fairly specified in advance. And they can petition for a gun violence restraining order. 

The risk is high. We can't quantify how high, but we have a problem. And the judge is allowed to consider the evidence and basically say, "Look, I don’t have all the answers, but I am persuaded that risk is high here. Let's get the guns out of the equation while we figure it out." The order is only good for two weeks. It's renewable, like DV orders are. It can be extended for up to a year, etc. We don’t know how this is going to operate. It hasn't started yet. 

We don’t know how else it's going to be used, but we'll find out. 

BS: We'll do one more question and then we have to break.

GW: Okay. In the back.

Audience member: I will just preface this by saying that methamphetamines are just ripping up my community. 

GW: My peeps. 

Audience member: And the California voters recently decided to make methamphetamine possession a misdemeanor instead of a felony, which allows these people, and my law enforcement people tell me that people are previously convicted are saying, "Now I can get a gun."

GW: That's right.

Audience member: I'm interested in, I guess my question is are you studying this and what the ramifications for this would be?

GW: Not yet, but we're going to. 

So you're in Fresno. I'm in Sacramento. We have the same story. We are in the methamphetamine capitals of the world. Crank is a really bad drug. I'd much rather people be on cocaine than crank if I had to pick. But the larger problem is that we converted a bunch of prohibiting offenses into non-prohibiting offenses, not just for drugs. And some of those people are probably going to try and have the prohibition expunged and purchase guns and we'll see.

There is, however, a very quick and widely-supported effort underway not to say that those prior felonies have to be felonies again, but to extend the prohibiting misdemeanors so that I can get my felony knocked down to a misdemeanor, but I'm still going to be prohibited. Because everybody sees the risk in a bunch of cranksters having guns. 

I'll stop there.

BS: Thank you. 

We have five minutes. Before we go, I just wanted to add one quick though on the question on researchers. At the Dart Center, we face all the time this question on what researchers to listen to on a number of fields. Gun violence is one of them, but obviously, neurology, traumatic psychology, public health, suicide, veterans -- all of the stuff that we do workshops on, we're trying to assess all the time. Who do we talk to? Who can best inform our tip sheets for coverage?

And one of the things that I want to go back to is the very last thing that you pointed out because as reporters we face this all the time. Do we listen to X person or not? It's the question of replicability of their results. Not that people who are outliers are automatically always wrong, there are prophets in the wilderness, but in general, the strength of peer review, the replicability of results, the ability of other researchers to run their primary sources to the ground is one very good gauge of whether any given bit of academic research should be central to coverage.

We just actually saw this on an unrelated field, a very interesting example of this on marriage equality. On gay marriage, there was a research team at Yale University that put out some very interesting data about public opinion on gay marriage, and it turned out that one of the grad students was cooking the books. We found out about that because peers at another institution attempted to replicate the results, couldn't, went back to the original data, and the primary researcher on it went to his grad student and said, "Well, what the fuck?" 

That's a direct quote. 

BS: He failed the "what the fuck" test. I think the notions of peer review and replicability of results are always a good guide when you're dealing with evidence-based research on very complex public policy and science-related issues. 

Five minutes. 

GW: Can I say one thing in response to that? It was a subtext, sometimes explicit -- I want all of you guys to be more critical of your sources. A big problem with that study was that it got published all the hell over the place without anybody really looking under the hood. That was wrong. Journalists did a disservice to the country, I think, by not being critical in their coverage of that particular piece of research. 

So, something for you to look up, a resource. Google the phrase: Bradford Hill criteria. An epidemiologist by the name of Austin Bradford Hill, toward the end of his career, after he had seen any possible mistake he made many, many times, came up with a list of about eight very simple tests to use in evaluating observational research. Are the results plausible? Are they replicable? And others. Google Bradford Hill criteria. His original paper is about three pages long, this teeny little thing. Those criteria are my best friends. If I'm reading a piece of work, and it doesn't start satisfying these criteria, I stop reading, because it's not worth my time. If I'm going to do a piece of work, I am very careful to try and make sure that I meet as many of those criteria as I can, particularly if I'm doing work that's never been done before and I know I'm going to fail replicability. 

Thank you again. 

BS: Thanks all.