Bi-Weekly Discussion - Understanding Gun Violence

UNDERSTANDING GUN VIOLENCE IN AMERICA

INTRODUCTION:

Since gun violence & gun control has once again become a major issue, I've decided to re-visit the topic with a slightly modified version of the discussion outline I used back in October of 2016. We will discuss risk perceptions of gun violence first, then move onto risk distributions and relative risks so we can get a sense of the overall problem, and then we'll address the distribution of murders & the distribution of guns in the U.S. and how much they overlap. I'm not going to directly address gun control policies in this meetup, nor do I want our members to get sucked into debating specific policies, since I think it's important to properly understand a problem before suggesting solutions to it.

Note that I'm excluding any specific focus on 2 highly publicized but statistically rare forms of gun deaths via mass shooting - terrorist mass shootings motivated by political or religious ideology & "lone wolf" active shooters who kill at random. (Recently, there's about 500 deaths/yr from all mass shootings, together accounting for about 4-5% of gun homicides, but only about 20% are random attacks - while about 50% are some type of domestic/family dispute and the other 30% are robberies.) There's enough material on terrorism & "active shooters" that they each deserve their own separate discussions. I'm excluding accidental gun deaths as well for the same reason - they're statistically rare (~500/yr) compared to gun homicides (~10-11,000/yr). I'm also excluding 2 common type of gun deaths from our discussion - suicide & police shootings - even though gun suicides outnumbers gun homicides by about 2-to-1 (so about 20-22,000 deaths/yr) and police shootings now result in about 1,000 deaths/year (so about 8-10% of all gun homicides). That's because both suicide & police shootings significantly different from standard criminal homicides in their causes, and both probably deserves their own separate discussions.

• NOTE: I'm hosting a Skeptics meetup immediately after this one which will focus exclusively on police shootings. To check out my discussion outline & RSVP, go here:
  https://www.meetup.com/Philly-Skeptics/events/249125196/

-----------------------------------------------------------------------------

DIRECTIONS ON HOW TO PREPARE FOR OUR DISCUSSION:

The preparation for this discussion will be a bit different than usual. Normally, I link videos for you to watch or articles to read. However, this time, I just couldn't find any good videos that addressed the aspects of gun violence from the angles that I wanted. I found 36 articles on this topic, and this is way too much to expect anyone to read. I took some of the important points from these articles and listed them below under the 4 main topic headings, and there's citations in the form of numbers in parenthesis at the end of sentences that correspond with the articles listed in a bibliography in our meetup's discussion section. Here's the link to the bibliography:
https://www.meetup.com/Philadelphia-Political-Agnostics/messages/boards/thread/50166073/0/#129652359

There were 2 tools I wasn't able to work into the discussion outline, but which are very pertinent. If you have time, please check them out:

(1) Slate's "Gun Deaths Map" which shows the distribution of gun homicides and fatal gun accidents (but not gun suicides) from June 13, 2015 to June 13, 2016:
http://www.slate.com/articles/news_and_politics/crime/2015/12/gun_deaths_map_how_many_people_have_died_or_been_injured_in_shootings_around.html

(2) CDC's WISQARS database, specifically the "National Violent Death Reporting System" (NVDRS), which allows you to roughly estimate a person's annual risk of homicide by type, location & demographics:
https://wisqars.cdc.gov:8443/nvdrs/nvdrsDisplay.jsp

In terms of the discussion format, my general idea is that we'll address the 4 topics in the order presented here. I figure we'll spend about 30 minutes on each section.

I. GUN HOMICIDE RISK PERCEPTIONS: MORAL FOUNDATIONS, LOCUS OF CONTROL, CULTURAL CONGITION, AND THE PROBLEM WITH LARGE NUMBERS

One of the most important things to understand is that there's a big divergence between gun crime rates & public perception of gun crime. Despite a dramatic decrease in gun crimes since the peak in 1993, 56% of Americans think it's worse and 84% think believe it's gone up or stayed the same. For some reason, women and the elderly were less likely to be victims of crime, but were more likely to believe gun crime had increased in recent years, while men, who were more likely to be victims, were more likely to know that the gun crime rate had dropped.(1)

So what psychological factors could be behind this divergence in risk perceptions as they relate to gun violence?

Jonathan Haidt's "Moral Foundations Theory" (MFT) helps elucidate the way that unconscious differences in our moral intuitions are often the cause of intractable political conflicts, and this certainly seems to be the case in the gun control debate. Liberals tend to rely heavily on the "Care/Harm" foundation and this makes them very concerned about gun violence, whereas conservatives & libertarians tend to rely more on the "Liberty/Oppression" foundation and this makes them more concerned about having their 2nd Amendment gun rights curtailed or eliminated. (2) A rhetorical study of pro- & anti-gun political speeches & newspaper op-eds at UC Berkeley seems to back this thesis up, and also suggests that conservatives arguing against gun control tend to invoke arguments related not only to "Liberty" but also to the "Loyalty" and "Authority" foundations as well. (6)

But why don't these moral foundations lead to the opposite conclusion? Hypothetically, liberals could see guns as a protective measure that helps innocent people (especially minorities) avoid harm from predatory criminals. Likewise, conservatives could see widespread gun ownership - especially by menacing or subversive members of their outgroup - as a trigger for their Loyalty foundation (which relates to ingroup/outgroup distinctions) and as a threat to their Liberty (in the sense of having their freedom of movement curtailed by fear of violence) and a subversion of their Authority (another moral foundation conservative tend to rely on). To a limited extent, this has happened historically. In the past, the NRA backed gun control legislation in the 1920s to curb Prohibition-era gangsters and in the 1960s to curb left-wing radical groups like the Black Panthers. And a few left-wing groups, like the Black Panthers and anarchist groups of the 1960s & '70s, and more recently some gay organization like "Pink Pistols," have advocated gun ownership for minorities to protect themselves.

Overall, however, the trend since the 1980s has been for conservatives to oppose gun control and liberals to support it. This may relate to the differences in the personality trait known as "locus of control", which relates to Neuroticism from the Five Factor Model. We know that liberals tend to score higher on Neuroticism than conservatives. In general, people with an internal locus of control tend to be low in Neuroticism and generally feel safer, but they can fall prey to the "just world hypothesis" and believe that bad things generally don't happen to good people and that they're mostly in control of their risk level. Conversely, people with an external locus of control tend to be higher in Neuroticism, and they can go to the opposite extreme and end up believing that good people are victimized all the time and that there's little one can personally do to increase one's safety, a view bordering on "learned helplessness".

https://en.wikipedia.org/wiki/Locus_of_control
https://en.wikipedia.org/wiki/Just-world_hypothesis
https://en.wikipedia.org/wiki/Learned_helplessness

It may be a historical accident that people with an external locus on control currently tend to identify as "liberals" while those with an internal locus of control tend to identify as "conservatives" or "libertarians". However, the split on gun control vies may also be due to the historical experiences of different groups in American society. The Yale psychologist Dan Kahan notes that the "unconscious emotional dynamics at work here" include the "predictably parochial nature of our empathic sensibilities, which lead us to attend selectively to forms of suffering and loss that are in fact universal in their nature and concentrated disproportionately on certain members of our society." Kahan has found that while African-Americans tend to score "liberal Democratic" in terms of their overall political values, when we account for ideology, "African Americans are more supportive of gun control... than whites of comparable political outlook." Kahan relates this to the life experiences many African-Americans have as members of urban communities with higher rates of gun violence. (7)

Kahan's research indicates that people often engage in something called "motivated cognition" or "identity protective cognition" and this effect individual's perception of gun risks. Women & minorities tend to fear all manner of risks more than white males, something researchers call the "white male effect". Kahan explains it partly in terms of differences in values: "Persons who hold egalitarian and communitarian worldviews worry more about crime and gun accidents, an anxiety that coheres with their negative association of guns with patriarchy, racism, and selfish indifference to the well-being of others. Persons of a hierarchical and individualistic worldviews, in contrast, tend to see guns as safe, and worry much more about the danger of being rendered defenseless against attack; this perception of risk coheres with their positive associations of guns with traditional social roles (father, protector, provider) and individualistic virtues (self-reliance, courage, physical prowess)." (2)

https://en.wikipedia.org/wiki/Motivated_reasoning
https://en.wikipedia.org/wiki/Cultural_cognition

Neither Haidt nor Kahan address this, but we also have to contend with the fact that people often have a hard time understanding risks when they're presented in terms of large numbers. Gun homicide statistics are often given in the media in terms of raw numbers of gun homicides in America; for example in 2013, 11,208 Americans were victims of gun homicide. National averages are another way to report this; for example, the US homicide rate in 2014 was 4.5 per 100,000. Raw numbers and rates can be hard for the average person to translate into personal risks. Even if you're good at math (which many Americans aren't), the human mind has a hard time fully comprehending large numbers and correctly calibrating their impact on an emotional level. When we think of over 11,000 Americans murdered every year by guns (and about twice that number in gun suicides) that sounds comparable to the aftermath of a horrible battle and most people find that very upsetting. (6) However, when we imagine a town of 100,000 people, which sounds like a lot of people when we try to visualize all of them packed into a football stadium, 4-5 murders a years among a large population like that doesn't seem nearly as scary to many people.

This is related to the "representativeness heuristic" that deals with how humans try to make judgments about the probability of uncertain events, and "scope neglect" is a cognitive bias that occurs when people can't seem to value something in a way that's proportional to its size. This mismatch between numbers of deaths & the public's emotional reaction recalls an infamous quote from Stalin: "The death of one man is a tragedy. The death of millions is a statistic."

https://en.wikipedia.org/wiki/Representativeness_heuristic
https://en.wikipedia.org/wiki/Scope_neglect

Lastly, we should consider how these psychological differences between liberals & conservatives may result in radically different policy preferences when it comes to gun control. Jonathan Haidt thinks that differences in moral foundations tend to make the gun control debate intractable: "I see no sign that there will be any left-right agreement on [gun control]," he said. Haidt explained the conflict between the Care/Harm & Liberty/Oppression foundations in layman's terms: "Conservatives tend to see the world more in terms of good-versus-evil and, for some of them, the nightmare is a disarmed citizenry that can be preyed upon by criminals. They know that having a gun in the house would increase the risk of an accident for a member of their family, but they’re willing to take that risk. Liberals are more prone to utopianism... Liberals are horrified by violence, and especially violence against children. So they demand a policy response. And while I want a policy response too, I think we have to make the response be based in the research on what will actually work." (5)

Dan Kahan thinks that the gun control debate is not merely about differing risk perceptions but also part of a broader "culture war" between liberals & conservatives: "The point [of the gun control debate] isn’t to save lives" but rather "to capture the expressive capital of the law." He argues that the role of guns in American history has vacillated between "overthrowing tyranny and perpetuating conditions of slavery and apartheid; taming the frontier and assassinating Presidents" which has "imbued guns with a rich surfeit of social meanings. Wholly apart, then, from the effect gun laws have (or don’t) on homicide, they convey messages that symbolically affirm or denigrate opposing cultural styles." Kahan goes on to argue that "we are a liberal democratic society, comprising a plurality of diverse moral communities [and] the individual liberty provisions of our Constitution forbid the State to 'enforce on the whole society' standards of private conduct reflecting any one community's 'conceptions or right and acceptable behavior'." Kahan argues that many people don't want the judiciary to be impartial and want to "use State power to resolve whose way of life is virtuous and honorable and whose vicious and depraved" and this leads us to fixate on laws that have ambiguous public-welfare consequences but express unambiguously partisan cultural meanings". (2)

II. GUN HOMICIDE RISK DISTRIBUTIONS: THE ECOLOGICAL FALLACY & ANECDOTAL EVIDENCE, THE "RISK SOCIETY" THESIS, PERFECT SOLUTIONS VS WORSE PROBLEMS, AND WHY POVERTY & RACISM ARE INSUFFICIENT EXPLANATIONS:

One of the common problems people have when they try to interpret crime statistics and estimate their own risk is that they just look at national averages or stats framed in terms of risk to the "average American" without realizing that risks can vary drastically from one locale or one demographic group to the next. This relates to something in statistics called the "ecological fallacy" where people make incorrect inferences about individuals based on statistics that apply to the group those individuals belong to.

https://en.wikipedia.org/wiki/Ecological_fallacy

Luckily, from data that is readily available online, we can find that the annual risk of homicide for the average American is 1 in 18,989, and the lifetime risk of death from homicide is 1 in 133. However, these overall stats for the "average American" hide large disparities. Over their lifetimes, the average black male have a 1-in-21 chance of being murdered; the average white male have a 1-in-131 chance; the average white female has a 1-in-369 chance; and the average black female has a 1-in-104 chance. Men tend to be killed during either the commission of a crime or as a result of a confrontation with another man, whereas women tend to be killed by an abusive male partner or ex-partner.

http://discovertheodds.com/what-are-the-odds-of-being-murdered/

http://www.nytimes.com/1985/05/06/us/around-the-nation-odds-of-being-murdered-are-1-in-133-study-says.html

Many people don't even look at data and just try to roughly estimate their risk by looking at what has happened to them & people they know. This is often called "anecdotal evidence" and social scientists usually caution against doing this because, especially when only one or a few anecdotes are presented, there is a larger chance that they may be unreliable due to "cherry-picked" or otherwise non-representative samples of typical cases.

https://en.wikipedia.org/wiki/Anecdotal_evidence

However, among social justice activist groups, anecdotal evidence is often valued as "lived experience" - often defined as "the first-hand accounts and impressions of living as a member of a minority or oppressed group." While relying on lived experience can leave you open to some of the problems of anecdotal evidence, it can also avoid the ignorance & oversimplification of social problems & inequalities that the majority does not see as a big problem. The majority view of society is often a naïve version of the ecological fallacy, where problems that don't show up in the official statistics are seen as inconsequential but actually have a big impact on the lives of minorities who often experience what might seem like a statistically unlikely series of risks & harms.

http://geekfeminism.wikia.com/wiki/Lived_experience

This concentration of risks & harms in poor & socially disadvantaged areas ties in with a concept the German sociologist Ulrich Beck called the "risk society" - a society's systematic way of dealing with hazards and insecurities induced and introduced by modernization. Beck explained that modern society is not only organized around the distribution of goods but also the "distribution of bads" - i.e. harms & risks of harm. Beck points out that governing elites have a tendency to focus mostly on the risks they cannot escape with their money & power - things like climate change & economic crises. In terms of violent crime, this would be things like terrorism or mass shootings in suburban schools or shopping centers. Risks that are concentrated among the poor & marginalized, like street crime or lead & asbestos in old buildings, are often ignored. For our purposes here, it's important to understand the implications of the highly unequal distribution of risks pertaining to gun violence.

https://en.wikipedia.org/wiki/Risk_society

To connect lived experience with gun homicide risks, it helps to look at what percentage of Americans know someone who's been shot - 22% of Americans know someone who was killed by another person with a gun. However, it's deceptive if you try to apply data gleaned from national averages to yourself or your friends, since you may not be statistically "average" and in fact crime research shows there's large disparities in gun homicide rates between people of different ages, sexes, races, and locations throughout the US. For example, only 3% of white Americans know a victim of gun homicide, while 47% of black Americans know someone killed with a gun, and 19% of black Americans have had a family member killed with a gun. (7)

In discussing gun homicide risk distributions, it's common to compare the US to other developed countries. UC Berkeley's Franklin Zimring and Gordon Hawkins have found that when the US is compared to 20 other developed countries, it's NOT an outlier in terms of violent crime rates aside from homicide, but that the US is an outlier when it comes to gun homicides which seems to drive the overall homicide rate in the US. (9)

However, if we extend our comparison of the US to other OECD nations besides the 20 wealthy (mostly) European ones used in Zimring & Hawkins' study or compare the US to nations that have a relatively high rank on the UN's Human Development Index (HDI), the US homicide rate doesn't look like an outlier anymore -- Mexico, Argentina, Bahamas, Panama, Uruguay, Costa Rica, Cuba, Russia, and the Baltic states all have higher rates. Considering that the US has a large number of recent immigrants from Latin America and shares a border with Mexico, there's disagreement among analysts as to whether the US should more resemble Western Europe or Latin America. (10, 12) As the political scientist Jason Brennan put it recently: "The US is a really messed up OECD country or alternatively a really high functioning post-racial colonial society. Compare us to Denmark and you think, 'What the hell?' Compare us to most the Americas or former colonized African countries and you think, 'Okay, doing better than expected'."

This disagreement about which type of countries should be used as a measure of relative success or failure in terms of homicide rates can be thought of as a conflict between two informal fallacies -- the "Nirvana Fallacy" (a.k.a. "Perfect Solution Fallacy") and its inverse, the "Fallacy of Relative Privation" (a.k.a. "Appeal to Worse Problems" or the "Not As Bad As" Argument). Essentially, one side sets a very high benchmark and declares that since things aren't perfect we need drastic reforms, while the other side sets a very low benchmark and argues that since things aren't horrible there's no need for reform.

https://en.wikipedia.org/wiki/Nirvana_fallacy
http://rationalwiki.org/wiki/Not_as_bad_as

Aside from the differences between national murder rates, individual US states also have large variations in murder rates. Many of the US states in New England, the Midwest and the Pacific Northwest have murder rates lower than much of Canada. At the state level, the US murder rate is being driven up by a few high-murder states such as Maryland, Louisiana, South Carolina, Delaware, and Tennessee. In particular, there is a clear trend toward higher violent crime rates in the Southern states. (10,11,14)

The US has large racial disparities in murder rates as well. The CDC's database shows that 5.2 out of every 100,000 Americans were homicide victims, on average, from 2010 to 2012. The annual rate of homicide deaths among non-Hispanic white Americans was 2.5 per 100,000 persons, meaning that about one in every 40,000 white Americans is a homicide victim each year. By comparison, the rate of homicide deaths among non-Hispanic black Americans is 19.4 per 100,000 persons, or about 1 in 5,000 people per year. Black Americans are almost eight times as likely as white ones to be homicide victims, in other words. So for white Americans, the homicide death rate is not so much of an outlier. It’s only modestly higher than in Finland, Belgium or Greece, for instance, and lower than in Chile or Latvia. But there’s no other highly industrialized country with a homicide death rate similar to the one black Americans experience. Their homicide death rate, 19.4 per 100,000 persons, is about 12 times higher than the average rate among all people. Instead, you’d have to look toward developing countries such as Mexico (22.0), Brazil (23.6), Nigeria (20.0), Rwanda (23.1) or Myanmar (15.2) to find a comparable rate. (12)

The disparity between black & white homicide rates is often explained by "concentrated poverty", along with other factors like the falling marriage rates & family breakdown that the Moynihan report noted in 1965, as well as the deindustrialization, urban decay, and proliferation of drug markets & gangs that began in the 1970s. These negative trends peaked during the "crack epidemic" of the 1980s & have since abated, but only partly. It's important to note that studies of poor black neighborhoods show that their residents have large disparities in homicide risk. A Yale study focused on poor black neighborhoods in Chicago with a litany of familar risk factors: 50% of all households were led by a single female; 43% of the 82,000 residents had less than a high-school education; a third of households were below the poverty line. And the homicide rate, over the 5 years of the study, was 55.2 per 100K/yr, about four times the citywide rate. But within this high-crime, high-risk neighborhood, there were very different levels of risk: Simply being arrested during this period increased the aggregate homicide rate by nearly 50%, but being in a network component with a homicide victim increased the homicide rate by a staggering 900% - from 55.2 to 554.1. (13)

It's difficult to attribute homicide rates directly to poverty because there's areas that are very poor but have low violent crime rates. For example, in Appalachia which is about 96% white and has an abysmal poverty rate, there's a great deal of drug use, welfare fraud, and the like, but the overall crime rate throughout Appalachia is about two thirds the national average, and the rate of violent crime is half the national average. One major social difference is that the majority of impoverished households in Appalachia are headed by married couples, not single mothers. They're also living in rural areas where the poverty is less concentrated. (15)

Ironically, if we look at white homicide rates in other areas of the American South and in the American Southwest, they are much higher than Appalachia despite being wealthier. In many of these state, white people have non-gun homicide rates of 1.5-2 per 100,000 which along would put them ahead of the total homicide rates (gun & non-gun) of whites in the Northern states. Steven Pinker ascribes this North-South divide in violent crime rates to the South's historic "culture of honor" which encourages men to take the law into their own hands and settle personal disputes with violence (14).

It's also difficult to attribute homicide rates directly to racial discrimination because there's areas that have large minority populations but have low violent crime rates. Ron Unz compared the average crime rates for the 5 most heavily Hispanic cities—Albuquerque, Dallas, Los Angeles, San Antonio, and El Paso—to the those of the 5 whitest—Oklahoma City, Columbus, Indianapolis, Seattle, and Portland. He found that the more Hispanic cities are the ones with the lower crime rates—10% below the white cities in homicide and 15% lower in violent crime. A particularly remarkable result is that gigantic Los Angeles—50% Hispanic and frequently perceived as a dangerous city riven by street gangs— now has violent crime rates close to those of Portland, Oregon, the whitest major city in the nation at 74%. (16)

Despite public fears of immigrants causing crime to rise, this is not borne out by crime data which shows the "immigrant paradox" - i.e. immigrants tend to be poorer & more socially disadvantages but commit less crime than the native population. This trend applies to immigrants from Europe, Asia, Africa & Latin America. U.S.-born adult men are incarcerated at a rate over two-and-a-half times greater than that of foreign-born men. From 1991 to 2008, when nearly 3.7 million foreign-born people, about a third of whom were illegal immigrants, moved to California, the state’s violent crime rate fell by 55%. This negative correlation between upward trends in immigration levels & downward trend in crime can also be found on a national level during the 1990s & 2000s, and some criminologists think the immigration of people with lower crime rates may have played a role in decreasing the overall US crime rate. (17)

III. RELATIVE RISKS, HOMICIDE TYPES, NEGLIGENCE VS MALICE, AND GUNS VS CARS, ALCOHOL & MEDICAL ERRORS:

One way to understand gun violence in the US is to compare the risk of gun homicide to other common & uncommon risks. At the outset, we should be aware that subtle differences in the way these risks are posed by those reporting them can create "framing effects." By comparing the risks of gun homicide deaths to other common risks that offer either much worse or much better odds, this can feed into the 2 fallacies mentioned above: the "Nirvana Fallacy" & "The Fallacy of Relative Privation" -

https://en.wikipedia.org/wiki/Framing_effect_(psychology)

When we compare gun homicides to non-gun homicides in the US, we see that in recent years homicides committed with guns tend to be a little over twice as common as those committed by other means. In 2013, there were 11,675 Americans murdered with guns (502 in mass shootings) and 4,962 murdered without guns.

http://www.oregonlive.com/data/2015/10/gun_deaths_other_causes.html

The vast majority of gun homicides are committed with handguns. While there's been a lot of focus recently on "assault weapons" like the AR-15 because of their use in mass shootings, rifles in general & shotguns kill less Americans than knives. In 2013, knives were used to kill 1,490 American, rifles were used to kill 285 Americans, and shotguns killed 262. Personal weapons (hands & feet) killed 656 Americans, blunt objects killed 472, and "other weapons not stated" account for 903 . For comparison, there were 7,105 murders by handgun in America in 2016. (As the article linked below notes, there's a lot of uncategorized murders, but if we apportion them based on the types that are categorized the overall picture stays mostly the same.)

http://dailycaller.com/2018/02/19/knives-gun-control-fbi-statistics

In terms of overall number of deaths per year, gun homicides - or even all homicides deaths - do not break into the top 10 causes of death for the overall population, which is dominate by diseases & accidents, along with total suicides. (However, homicide is in the top 10 causes of death for age groups under 40.) However, everyone has to die from something and most of the top 10 causes primarily effect the elderly, and so there's a common argument that common causes of death for younger people have a disproportionate impact on the wellbeing of the population as a whole, and this includes things like homicides & car accidents.

Many people raise the argument that homicides are different than the other causes of death here because they are intentional, and they affect innocent third parties. This is a strong argument against risks people voluntarily assume for themselves that can result in their own death (e.g. drug or alcohol abuse, dangerous recreational activities), but this claim does not work as well for risks which affects innocent third-parties. The fact that deaths from things like drunk driving accidents or medical errors occur through negligence rather than malice is small comfort to the dead and their families. (6, 23) Truly reckless acts fall into a grey area between malice & negligence, referred to as "gross negligence" in legal terms. Opinions vary about how we should think of this grey area in moral terms - Hanlon's Razor suggests that we "never attribute to malice what could adequately be explained by stupidity," but Clark's corollary points out that "any sufficiently advanced incompetence is indistinguishable from malice". http://rationalwiki.org/wiki/Hanlon's_razor

When we compare the US deaths from car accidents to gun deaths, we find that within the last decade both kill about 30-35,000 people per year. We also find that DUI deaths are roughly comparable to gun homicides, and that gun ownership & car ownership are roughly equal. In the United States in 2007 there were 12,632 gun homicides. Accurate statistics of gun ownership is not available, but most estimates are around 200-250 million legally-owned firearms in the U.S. Note that because gun ownership is more concentrated than car ownership -- only about 55 million Americans own guns. (36). During 2007, there were 247 million registered automobiles in the U.S. and 13,041 deaths from drunk driving. In 2006 there were 22,073 alcohol-related deaths, which does not includes deaths from DUI, and this number is roughly equal to the number of gun suicides. This means that when we add DUI deaths & other alcohol-related fatalities, alcohol is directly involved in about as many deaths as guns. This comparison is interesting in large part because gun control is politicized in a way that car control & alcohol control are not (20, 23)

One of the major ways that car deaths & gun deaths are not comparable is the way in which they are distributed geographically & demographically. Car deaths are more common in rural & suburban areas, whereas gun homicides are more common in urban areas. (Suicide is also more common in rural areas.) Rural areas are also further from hospitals, which means injured people are more likely to die from their wounds. Overall, the risk of death from injuries was 1.22 times more common in rural counties than in urban ones. (21)

The annual deaths from medical errors appear to dwarf gun homicides, and even total gun deaths, possibly by an order of magnitude. An older report from the Institute of Medicine in 1999 estimated deaths from preventable medical error at 98,000/year. However, a recent report in the BMJ declared that the estimated deaths from medical errors in the US has now risen to more than 250,000/year, making preventable medical errors in hospitals the 3rd-largest cause of death in the country in 2013. However, this estimate was based off stats gleaned from patients many of whom were very sick, and critics of the study argued they shouldn’t be compared with the general public, but needed to be compared with other sick, hospitalized patients. Considering there are about 2.5 million deaths each year in the United States, about 700,000 of which are hospitalized patients, if medical errors in hospitals caused 250K death/yr this would have to account for up to 10% of all deaths, or up to more than a third of hospitalized patients, which is hard for many researchers to fathom. The jury appears to be still out on this question... (22) It's interesting to ponder why medical deaths aren't more publicized or politicized - perhaps because of the higher status of doctors vis-à-vis gun owners and the fact that medical errors are unintentional while gun deaths are mostly intentional.

IV. DISTRIBUTION OF MURDERERS & GUN OWNERS, AND THE "HOBBESIAN" & "ROUSSEAUAN" VIEWS OF CRIME:

One of the key points of disagreement in debates about gun violence is what it says about the average American: Is the average person in every state & locale in America a ticking time bomb waiting to go off if they have access to a gun? Or is gun violence mostly limited to the inhabitants of poor urban neighborhoods? Or is violence actually hyper-concentrated among a relatively small number of recidivist criminals that can be identified & targeted?

Ultimately, the way this issue is addressed can lead to very different views of humanity, the "state of nature" in which people exist without police oversight, and the political philosophies one favors to remedy the defects in the state of nature. The classic example of these conflicting views is found in the Enlightenment era philosophies of Thomas Hobbes - who saw humans as naturally selfish & violent and in need of authority - and Jean-Jacques Rousseau - who imagined humans as naturally good & sociable but corrupted by civilization.

https://en.wikipedia.org/wiki/State_of_nature

In a politically polarized climate like present day America, arguments about human nature & crime tend to be split based on identity politics and ideological allegiances. There's a common assumption that the members of our in-group are basically good, every success story is testament to our inherently good nature, and if one of us occasionally commit a crime it's either a mistake or a result of unique circumstances or they are just a "bad apple," a rare exception to the rule. However, there's also a common assumption that members of our out-group are basically bad, every crime they commit shows this, and if occasionally one of them is a successful & productive member of society they must be the exception to the rule. In debates over America's high level of gun violence, you can see this double standard with liberals blaming it all on "gun-toting rednecks" and "racist cops" and conservatives blaming it all on "inner-city thugs" and "illegal immigrants", but both sides hold themselves & their allies blameless. This ties in with a cognitive bias we've discussed before -- the "ultimate attribution error".

https://en.wikipedia.org/wiki/Ultimate_attribution_error

According to David Frum, "most gun casualties occur in the course of quarrels and accidents between people who would be described as “law-abiding, responsible gun owners” up until the moment when they lost their temper or left a weapon where a 4-year-old could find it." (25) The economist John Lott takes issues with this, citing statistics that show the majority of murders are not committed by people who had up until then been law-abiding citizens. About 90% of adult murderers have a violent criminal record, and about 89% of juvenile murderers have a criminal record for serious crimes. (26)

A routine finding in the criminological literature is that about half of the crime is committed by a very small fraction of the population, around 5-8% depending on the sample and methodology used. This finding has been replicated in many different studies around the world. (24)

We've got recidivism stats for murder in Chicago, which was America's "murder capital" in 2012. More than 200 people were arrested for homicides that happened in Chicago in 2012 and about 15% were minors 14-to-17 years old, while the other 85% were adults. Out of the adults, at least 90% had an arrest history in Chicago and more than a third of them were previously charged with weapons-related offenses, according to police data. (Unfortunately, since over half of the 509 murders in Chicago in 2012 went unsolved, it's difficult to extrapolate these results and determine what percentage of all homicides in the city were committed by recidivist criminals.) (28)

According to the criminal justice professor David Kennedy, research on violence shows it's concentrated in "hot spots", rather than "dangerous neighborhoods" -- blocks, corners, and buildings representing just 5-6% of an entire city will drive about 50% of its serious crime. The same is true about people. Homicide and gun violence are overwhelmingly concentrated among serious offenders operating in groups: gangs, drug crews, and the like representing under 0.5% of a city's population who commit 50-75% of all murders. We also know some reliable predictors of risk: individuals who have a history of violence or a close connection with prior victims are far more likely to be involved in violence themselves. Hot groups and people are so hot that when their offending is statistically abstracted, their neighborhoods cease to be dangerous. Their communities aren't dangerous; these criminals are. Kennedy suggests that by targeting just the few dangerous people in high-crime hot spots, police can stop tactics like routine stops that harass the general public in poor minority neighborhoods that they've previously viewed as "high crime" and forge better relations with community leaders who can help rebuild the social fabric. (29, 30)

To understand gun violence, we also have to look at the distribution of gun owners. Pew Research did a recent study on American gun owners and found, perhaps not surprisingly, that gun ownership is less common in the Northeastern states and is more common in the South, Midwest & Western states. Republicans are over twice as likely to own a gun as Democrats (49% vs 22% ownership rates; Independents are at 37%). People in rural & suburban areas (51% & 36% ownership rates) are more likely to own guns than those in urban areas (25% ownership rate). There's also significant racial disparities in gun ownership -- 41% of non-Hispanic whites in the US own guns, versus only 19% of blacks. Pew Research notes that this racial disparity presents a paradox - black Americans are significantly more likely to be victims of homicide (19.4 per 100K versus 2.5 per 100K for whites) but overall are less than half as likely to own guns. (33) It's worth noting that gun ownership is tightly concentrated; while about 30% of American households -- which translates to 22% of American adults (~55 million) -- legally own at least one gun, about 3% of those owners are gun dealers & collectors who own an average of 17 guns each and collectively own about half of all the firearms in the U.S. (36).

Another thing to consider is the correlation between gun ownership rates and homicide rates. This is a hotly contested issues among both researchers and pundits. When we examine statistics on a state-by-state basis, there does appear to be a correlation between a state's gun ownership rates & "gun death" rates (i.e. gun homicides + gun suicides + fatal gun accidents). However, that appears to be mostly a factor of gun suicides -- when we exclude gun suicides (and the small number of fatal gun accidents), there does not appear to be a correlation between a state's gun ownership rate & its total homicide rate (14, 32). However, when you control for robbery rates (which is taken as an indication of a state's underlying level of violent crime), there does appear to be a correlation between a state's gun homicide ratesand household gun ownership. (31) This result is often taken to mean that "more guns equals more murders" but that's not quite certain. Notice that the while higher gun ownership rates appear to drive the gun homicide rates up, it's not clear that it increases the overall homicide rate (gun homicides + non-gun homicides). This suggests that when people want to commit murder and cannot obtain a gun, they may simply use another weapon or method, and that motive and opportunity rate than means has a much larger effect on the violent crime.

• NOTE: Many people are surprised by the idea that the availability of guns might not be the limiting factor for homicides, since it seems to many like it would be much harder to kill someone without a gun. However, it pays to consider that historians have estimated that medieval London had a homicide rate somewhere between 18-36 per 100K/yr. The reason for the higher homicide rates in pre-modern societies seems to be social norms that encouraged violence (e.g. dueling, feuding) and less reliable law enforcement.

http://bedejournal.blogspot.com/2011/11/steven-pinkers-medieval-murder-rates.html

When we look for a direct overlap between legal gun owners and violent criminals, overall crime statistics suggest that the vast majority of America's 55 million gun owners never commit a gun crime. It's hard to find accurate statistics on this, but it looks like if we take data on "concealed carry killers" from the Violence Policy Center (a pro-gun control group), only 0.0012% of legal gun owners with a concealed carry permit ever use their gun to commit murder. (37) Studies suggest that concealed carry permit holder are highly unlikely to commit a crime -- significantly less likely than even a police officer: “With about 685,464 full-time police officers in the U.S. from 2005 to 2007, we find that there were about 103 crimes per 100K officers. For the U.S. population as a whole, the crime rate was 37 times higher—3,813 per 100K people. We find that permit holders are convicted of misdemeanors and felonies at less than a sixth the rate for police officers. Among police, firearms violations occur at a rate of 16.5 per 100K officers. Among permit holders in Florida and Texas, the rate is only 2.4 per 100K. That is just one-seventh of the rate for police officers.” (35).

However, the majority of gun owners in America do not have concealed carry weapons (CCW) permit. So how do we calculate the percentage of legal gun owners in American (both those with & without CCW permits) who at some point commit a violent crime with their gun? We could start by looking at the fact that about 5% of Americans will be incarcerated at some point in their lives, and about 40% of prisoners have a current or prior violent offense, so that means about 2% of Americans have been convicted of a violent crime. (38)

So how can we use this information to estimate the percentage of Americans that will commit a violent gun crime? Well, that's difficult, because less than 10% of all violent crimes are committed with a firearm, although guns are used over 50% of murders & robberies. However, less than half of all violent crimes result in an arrest, much less a conviction & imprisonment. We can make some assumptions - let's say that 75% of those in jail for a violent crime used a gun. We'll also assume that the police catch & prosecutors convict 75% of people who've committed a gun crime at some point, since recidivism is probably higher among gun-wielding criminals, police probably try harder to catch them, and prosecutors are more likely to push for jail time. This would means that 1.5% of Americans go to prison for a violent gun crime during their life, and another 0.5% commit a gun crime but never serve time.

However, while 2% of Americans might commit a violent gun crime during their life, we have to account for the fact that only about 20% of gun crimes are committed by legal gun owners. (26) . Therefore, a rough estimate of the percentage of Americans that are legal gun owners that commit a violent gun crime at some point in their life comes to about 0.4%. Of course, in raw numbers this still sounds like a lot of people since 0.4% of 325 million Americans is 1.3 million people that obtained a gun legally but used it to commit a crime at some point in their life. When we account for the fact that about 55 million Americans are legal gun owners, this means about 2% of legal gun owners may commit a gun crime at some point. (NOTE: Since this is my personal guesstimate as I couldn't find any studies that showed what percentage of legal gun owners commit a gun crime, I suggest taking it with a grain of salt.)

As for my assumption above that only 20% of gun crimes are committed by legal gun owners, that comes from a recent University of Pittsburgh study. They found that in approximately 80% of gun crimes, the perpetrator was not a lawful gun owner but rather in illegal possession of a weapon that belonged to someone else. More than 30% of the guns that ended up at crime scenes had been stolen; however, more than 40% of those stolen guns weren't reported by the owners as stolen until after police contacted them when the gun was used in a crime. This suggests that there may be a problem with legally purchased gun ending up on the black market, either through theft or straw purchases.(34) Despite the dramatic drop in violent crime over the last 20 years, gun thefts have increases significantly, nearly doubling from 230,000/yr in 1994 to 400,000/yr in 2015. This sounds like a huge number, but keep in mind that represents about 1-2% of all the guns in the U.S. (36)