Murders per gun, which countries have the most and least

This is a lengthy email from an MR reader who wishes to remain anonymous.  These are his words, not mine, everything which follows:

Back in December I asked you knew of any naive measures of  “gun murders / # of civilian guns” per country, and seeing where the US falls in this distribution.

Some time after I found this WaPo data set compiled in 2012 from the UN, Small Arms Survey, and others. (There is a “data caveat” I’ll point out after the plots below.)

Here’s what I did: dropped all data into a spreadsheet and calculated the number of homicides per 100,000 guns — simply (homicide by gun) / (total guns) * 100000. Call this “H” for simplicity.

This produces numbers in a ~0-20 range for “western countries. So “H = 2.5” –> 2.5 gun-caused homicides per 100,000 guns for the year in question (2005 I believe), by country.

Here are three very quick, ugly kernel density plots from R.

The raw data plots is pasted in the end of the email (apologies for messyness!)

The vertical lines are:  mean=green,  median=blue,  USA=red.

Total world, as a limiting case:


US is below mean and median:  US = 3.7,  mean = 91.0,   median = 6.0

EU countries (density excludes US): 

image (1)

US is just above mean, above median:  US = 3.7,  mean = 3.1,   median = 1.9

“Post-WWII westernized” countries:

image (2)

US is just above mean, above median:    US = 3.7,  mean = 3.1,   median = 1.5

The “westernized” countries were somewhat arbitrarily those with a long “westernizing” history post-WWII. Chosen quite ad-hoc and off-the-cuff; largely it means Eastern European countries were replaced by Canada, Australia, Japan, etc.

Immediate data caveat: My earlier spot-checking against the cited sources turned up a number of discrepancies, which I couldn’t quickly figure out — mostly small, some large. Eg. I recall that some EU countries saw order-of-magnitude differences when I put in “direct from source” numbers, which is worrisome. Unfortunately I never had time to examine these further (hence the delay in reply), but perhaps some enterprising undergraduate student would be interested!

The broad strokes are still interesting. Here are some quick ‘surprises’ for me:

– The US is no longer a massive outlier, although still above average for “westernized” plots.

– Japan’s H-score is much higher I expected, ~10x the number in Norway (and higher than England, Northern Ireland, Czech Republic – the later has much less strict control and lower H).

– The Netherlands came out ~6-7x higher than median (of EU/”westernized”); ~3x the US

– Taiwan seems quite high:  ~11x median of “westernized,” ~5x US.   Was not expecting this (but not sure why).

– Ireland is ~4.5x higher than Northern Ireland

– Belgium is closest to the US in the EU states, Belgium=3.9 vs US=3.7

– All surprises, perhaps all data size related:  Denmark, Netherlands, Japan, Taiwan, Ireland, Italy (higher than US, was not expecting that), Belgium, Luxembourg

There are many possible data concerns. Sample size is very important (the few data I spot-checked varied significantly over time). Measurement is almost certainly an issue, and I dread looking into differences in the definition of “homicide” for these countries. I suspect, however, that clever methods and data collection could still provide useful information about ranges of these values (an enterprising undergrad could probably make quite the impact with careful data examination/collection and some Bayesian “Locomotive Problem“-style work).

Because of data issues, I don’t think of this as “the final word” but rather an interesting first pass.


Can someone clarify the axes here? If I understand it, the X axes are "homicides per hundred guns". What are the Y axes?

What about a plot of "guns per person" (or per household) vs. "homicides per person per year". Surely that would be easier to read -- or there some caveat in such a plot that I am missing?

A density plot is like a histogram, so the y axis is the number of countries.

Wow - talk about confusing graphs, appreciate this was a quick and dirty but the point of a graph is to present the data in an intuitive form. You shouldn't have to spend anytime figuring it out.

Well, if the raw data were added to the blog, someone could plot “guns per person” (or per household) vs. “homicides per person per year” on a scatter plot

So what does this measure indicate? The propensity in a population to use a gun to kill someone? A gun in Japan is more likely to be used to kill someone, so I guess a gun is more likely to be possessed in Japan for violent purposes? In the US, the curve is shifted toward people who own guns for non-aggressive purposes?

No, it's a measure of how efficient the gun killing economy is by ciuntry.

The US is extremely inefficient, needed to spend many times more on guns than any other nation to kill people. But not to worry, the US spend so much more that it can kill at higher rates.

Other nations have a limit number of fly swatters, but the US kills more flies by firing off millions of rounds of ammo.

But economies are zero sum. All the spending on guns and bullets puts lots of money in workers pockets making them, which then gets spent buying guns and ammo. Think of it as a great jobs program, more satisfying than paying one group of people to dig holes, and a second set to fill them in.

Increasing productivity of gum killing would merely create unemployment, unless the number of murders was greatly increased.

Postscript. I note that this data explodes the NRA good guy with a gun killing a bad guy with a gun argument. I note that violent Islamic extremists are far more efficient in killing while good guys need to out number them ten to one and fire ten times as many bullets to kill the Muslims.

Muslims are just more productive with guns than American Christians.

Allah is stronger than god.

But that merely states what Rumsfeld said - the Muslims have the power of superman chewing through steel cables and being able to kill with their minds and eyes so they can't be allowed in the US. Gitmo has commie unproductivity juju to counter their super power.

Are you a charter member of the Coalition to Criminalize Gun Owners?

They are not Uber mensch. Muslims pick places (gay bars, CA gov. offices, etc.) where congregate sheep that not only are unarmed, but are mentally incapable of doing anything except call 911 and wait for help.

Obviously, there was no "NRA good guy with a gun" there. Imagine one Obama-supporter, democrat with one gun shooting at 200 people. He had to stop and reload, say, every 30 shots. No one did a thing while he pulled out a mag and fumbled around to put in the next. That is the definition of a liberal, Obama-worshipping idiot.

There are 200 million guns in US citizens' possession and 12 trillion rounds of ammunition. You can find the numbers of gun murders (do not count suicides, police related, accidents, etc.) and use 200 million as a denominator. See what you get.

Plus, (racism alert!) let's make America comparable (homogeneous populations) to Japan and Western Europe, and only use the numbers murders committed by guns wielding whites, not NRA good guys either. I bet you will be surprised.

I'm sure your magic good guy accuracy is impeccable when you're dancing drunkedly in a crowded, loud, dark room. Do you sit at the bar and face the entrance all night in anticipation of a shooter? I have a feeling you're kind of a drag to party with.

Related: if the US had an average number of gun deaths per gun, how many homicides would we have? We have far more guns.

Fewer. The graphs clearly show that the US is above average in deaths per gun.

Why do all the commenters here seem to suggest that the graphs show the opposite? Confirmation bias?

The graphs clearly show that the US is above average in deaths per gun.

The graphs show that "above average" entirely depends on what your sample space is. The US is "below average" compared to all countries. The US is "above average" compared to selected European European countries, or to selected "Westernized" countries, the latter using a perhaps dubious definition in order to exclude high income oil states as well as middle to high income Latin American states like Chile, Argentina, and Panama that have income (PPP) per person greater than that of some European Union members like Croatia.

You dudes ain't so good with numbers. None of you saw coming the 2007 economic financial crash, either.

The charts are 100% bullshit. The researcher commences with his conclusion and sells it as fact. It's propaganda. If Americans' 200 million guns and 12 trillion rounds of ammunition were the problem, you wouldn't need bullshit charts and massaged statistics. The bodies would be piled three deep.

All gun deaths is a bullshit starting point. Leave out suicides, accidents, war/jihad, black/Latino gang banger murders and likely it would be somewhere close to comparable with (homogeneous populations) Japan and Western Europe.

Close not to Europe in general, but to an area with higher homicide rates, like Naples. The homicide rate in the U.S. is very elevated (4x) compared with the norm of (non-Russian) Europe. Other sorts of crime, not so much (or quite the opposite in the case of auto theft).

Re hispanics in the United States, I think the elevated offense rates apply to Puerto Ricans and Dominicans much more than they do to Chicanos, but I could have that wrong.

In the non-Latin Caribbean, homicide rates vary from about 2.7 per 100,000 (Martinique) to 40+ per 100,000 (US Virgin Islands, sadly). Biology doesn't get you vary far trying to make sense of this or plan a programme of action.

It would surely be less, but this analysis is totally meaningless to answer that question.
The fact that the US is below average on homicides/100,000 guns is mainly because they have most than twice the number of guns than any the next highest 'westernized' country (Switzerland), not because they're better gun owners, or less violent.
If I owned 20 cars and had 5 accidents a year, my accident to car ratio (C) is pretty low (0.25), but that doesn't mean I'm a better driver than a person with 1 cars that has 1 accident a year (C=1).
If I only owned 1 car, I'm unlikely to only have 1 accident every 4 years as my C would imply. I might actually still have 5 accidents, because I'm just a really bad driver...

By "more than twice the guns", I mean, more than twice the guns per capita. The total number of guns obviously much greater as India and China (the 2 countries with higher populations), have about 1/20th the guns per capita that the US does.

Why are we always compared to Europe when it comes to gun deaths? Our demographics are nothing like Europe. Let Europe keep it's borders open for another decade or so and we'll see how it's homicide rate is doing then.

Dude, I love the MR website, but this post looks desperate in twisting the stats to support a deeply entrenched prior viewpoint.

If you look at homicides per person per year for culturally/economically similar countries the stats are pretty clear:

US = 3.8 per 100,000
Canada = 1.4 per 100,000
UK = 1.0 per 100,000
Australia = 1.0 per 100,000

The US has almost 4x the murder rate of comparable countries, the obvious difference being gun control laws.

For what possible useful purpose does an ordinary citizen need an assault rifle? Madness.

"In response to my FOIA request, U.S. Army experts at Picatinny Arsenal were kind enough to categorically state that semi-automatic rifles are not 'assault rifles'."

So Shut the F*** up. Also, you probably think the AR in AR-15 stands for "Assault Rifle." (hint: google it). Moron.

"When the debate is lost, slander becomes the tool of the loser." - Socrates

Slander? That may the word for blaming me and millions of law-abiding gun owners for a massacre with which we had no part. Another word comes to mind: "collective guilt." Hitler used that for 6,000,000 Jews.

My assault rifle hasn't suddenly jumped up and killed anyone. Although, he (his name is Ho) nudges me when I read fascist comments like so many here. And, my high magazine capacities (JoAnn, Sue, Bertha, and Clara) keep whispering sweet nothings.

I'm a card carrying NRA member, and I always wince when I hear some gun-lover thinking he's won an argument because he catches someone misusing the word "Assault Rifle", "clip", "magazine", "shell", etc.

You haven't won the argument just because someone misuses a term, and it doesn't help the 2nd Amendment to interact this way.

Dear Dimitri - Maybe you could take your panties out of a knot long enough to meditate on what the etymological roots of the name AR-15 (originally ArmaLite Rifle) have to do with cogency Mr. Jones argument. Or what reasons there might be for the differences in the usual civilian definition of an assault rifle from the military definition.

HINT: The AR-15, is a (semi-automatic) civilian version of the fully automatic M-16 military assault rifle. It's usually difficult for civilians to legally buy fully automatic weapons, but you don't have to be a wizard gunsmith to modify the AR-15 to make it fully automatic.

You might want to google how to make a coherent argument.

So what? It's easy to make a bomb out of a pringles can. Does that mean I bought a bomb from the convenient store?

A Pringles can is too small. Fertilizer and diesel fuel makes a sweet IED.

The gun control chimera serves to divert attention from the jihad and Obama's/Hillary's disastrous policies which made America a far more dangerous place than it was December 2008, when Congress and the states enacted the Second Amendment.

"you don’t have to be a wizard gunsmith to modify the AR-15 to make it fully automatic"

And how many deaths each year are due to people armed with AR-15s converted to automatic? I cannot remember one, and since all rifles put together are responsible for 3% or less of deaths by firearm (with AR-15s a fraction of that), your comment seems fairly irrelevant.

Just a note that in California the "bullet button" would slow fast magazine changes in this situation.

Incremental change to law is possible.

Now the sad thing is that "enthusiasts and 2nd amendment activists" try for work arounds to the button. And would-be mass shooters can easily find their websites.

For some, safety is not desired.

The AR 15 is considered as a semi automatic but can easily be shot in automatic mode. There are legal add ons to fire the gun in basically full auto mode. Google slide fire ar 15 and bump fire AR 15. In the meantime have a look at these videos

Funny how so much more important nomenclature is to gun control than it is to other public policy issues like foreign policy, social services or taxes.
We all use technically incorrect words and phrases when talking about these topics but we're generally well-understood.
Except when it comes to weaponry.

The fecundity of Automatic Rifles by the cowering masses, which is all this study represents, gives no measure of the repeatable lethality of a subset of weapons in the wrong hands in a given area, which is the pertinent information needed. The study and it's graphs are junk. As far as morons go, you claim more expertise. I can call an AR-15 a fairy wand and it would be no less lethal, no matter Army lingo. It is an insanely dangerous weapon. I can get off 4 aimed rounds a second in semi-automatic mode. Didn't you hear that cadence in Orlando? In full automatic I can take no aimed shots, just spray wildly and inaccurately. If I took the AR-15. and you took the assault rifle in a showdown, the moron would not be around for long. And it wouldn't be Jack Jones. It might be me, if you are one really lucky guy.

You don't think armed revolution is a right of the people?

It is an interesting question. Perhaps overwhelming majority revolution is justified. Minority insurrection is just costly and painful.

Think about the III%ers and their self recognized minority status.

Well, the Bolsheviks and Mao certainly thought so.

I am quite sure that there has never been a state in which armed revolution has been other than a grave crime. Monopolization of violence is one of the chief points of a state. You couldn't have a state in which armed revolution was a "right".

New Hampshire Constitution: "Government being instituted for the common benefit, protection, and security, of the whole community, and not for the private interest or emolument of any one man, family, or class of men; therefore, whenever the ends of government are perverted, and public liberty manifestly endangered, and all other means of redress are ineffectual, the people may, and of right ought to reform the old, or establish a new government. The doctrine of nonresistance against arbitrary power, and oppression, is absurd, slavish, and destructive of the good and happiness of mankind."

The most obvious difference between the U.S. and those other countries is demographics. The U.S. has a much higher proportion of blacks, who shoot other blacks at a tragically high rate.

It is important, but it can't be separated from the fact that those shootings are facilitated by easy access to guns (both legal and illegal, which usually start as legal purchases).

Follow up question: do blacks in other countries shoot each other at similar rates in the US?

Sure looks like it.

In the UK, black individuals are 367% more likely to be killed in a homicide than white individuals
Likewise, black suspects are even more overrepresented among homicide suspects

On the flip side, the American "white" homicide rate is still something around 150-200% of the Western European average "white" homicide rate.

I have not been able to find statistics, but I have been told that the murder rate for Japanese of Korean descent is higher than that of ethnic Japanese in Japan.

Frankly, the guns and murder angle is pretty shoddy when you look at state by state data. The 5 states with the lowest murder rate are:
1. New Hampshire
2. Vermont
3. Iowa
4. Idaho
5. North Dakota

None of these have restrictive gun laws and most have much higher than average per capita firearms ownership. They all, however, are rural and very lacking in diversity.

The 5 worst states are:
1. DC (a clear outlier in terms of both racial composition and murder)
2. Louisiana
3. Maryland
4. Missouri
5. South Carolina

Pretty much all of these have a large urban area (DC, New Orleans, Baltimore, St. Louis, Columbia) that has a very high minority percentage and is at least a local nexus of the drug trade.

Another telling sign is the fact that California, with all its gun laws and bans has exactly .1 fewer murders per 100K than Texas with all its gun culture ... and Californians use guns in their homicides slightly more often than Texans (3.4 vs 3.2).

Some data shows that the most dangerous places are highly mixed neighborhoods without a dominant racial group.

Now none of this data says why this is the case. Could be institutional racism. Could be mult-generational poverty. Could be all manner of flaky things, but the US shows so much internal variability that gun count seems like a really poor predictor of violence. Demographics, while still not great, are a bit better.

"Another telling sign is the fact that California, with all its gun laws and bans has exactly .1 fewer murders per 100K than Texas with all its gun culture … and Californians use guns in their homicides slightly more often than Texans (3.4 vs 3.2)."

By the way, it's not widely known, but it makes a lot of sense when you stop and think about it: Hollywood liberals such as Steven Spielberg tend to be huge gun nuts:

It doesn't have much measurable effect on the homicide rate, but wealthy areas in West L.A. appear to have high rates of gun ownership for an area with very little hunting culture. Spielberg, for example, was introduced to clay pigeon shooting by John Milius, and he drops about $100k to have a hand crafted shotgun made in Italy to commemorate each movie he completes. Spielberg doesn't talk to the press about his love of guns, but his protege Shia LeBeouf has been frank about the subject.

The tradeoff is largely between homicides and property crimes. The English were very law-abiding for most of the 20th Century but in the late 20th Century turned into "A Clockwork Orange," with high rates of home invasion and car theft.

Interestingly, the British government responded the crime wave forecasted by Burgess and Kubrick much like in that amazing 1962 novel and 1971 movie: with technocratic innovations, such as video cameras and ASBO legal orders banning troublemakers from sources of temptation.

I'd be very interested in understanding more about how these innovations have worked out.

On the flip side, the American “white” homicide rate is still something around 150-200% of the Western European average “white” homicide rate.

And there's an enormous difference between the American white homicide rate among white Southerners versus among white New Englanders (and Mormons in Utah).

Blacks tend to be more "into the moment" than other races. It's a big part of why they are so mediagenic on average compared to other people. A downside of being into the moment is the moment can last the rest of your life if there are guns around.

Blacks and guns aren't a good combination.

Blacks also have "more rhythm" and are better at "landing white women" than other races.

Professsional running is also dominated by blacks which surely implies institutional barriers to white children who would like to run competitively.

Blacks also have “more rhythm” and are better at “landing white women” than other races

No. Intermarriage rates across that color bar are lower than across the other color bars. As for popular music, Broadway, adult contemporary, country, folk-acoustic, orchestral pops, and new age are not the issue of the domestic black population or the Caribbean. Rap, rock, jazz, R & B, Gospel, calypso, and reggae are. Not bad for 13% of the population.

The most obvious difference is easy availability of guns.

The comparison of white homicide rates in the US with rates in other countries is flawed. Blacks in the US tend to be quite poor as a group. Per capita income for blacks is around 2/3 or less that of whites.

So before doing the comparison we should remove from other countries' numbers homicides committed by an equivalent 13% of the population, allowing for other demographic aspects as well. Otherwise, drawing racial conclusions is nonsense.

The comparison of white homicide rates in the US with rates in other countries is flawed. Blacks in the US tend to be quite poor as a group. Per capita income for blacks is around 2/3 or less that of whites.

No. Personal income per capita for blacks is about 1/3 below the national mean, or about 38% below the Anglo-Caucasian mean. They're not poor on any international or historical scale. Personal income levels are quite similar to national means in France and Britain and quite similar to those of white Anglos ca. 1985. Life expectancy among blacks is also about what it was among white Anglos ca. 1985. A larger share of blacks hold drab wage jobs. However, the really large difference o'er the color bar is in the exposure to crime (which has a corollary in exposure to school disorder). The blacks' probability of dying by homicide exceeds that of the remainder of the population by 5 or 6 fold. One part of the country it does not is NYC, where only two police precincts have homicide rates which exceed the mean of American metropolitan centers as it was in 1980 and where no police precinct has a homicide rate more than about 3.5x the citywide average. (The worst precinct in the city has a homicide rate better than the citywide averages in Rochester and Buffalo).

You know, it's as if ferocious policing like the NYPD under Bloomberg really is better for law-abiding blacks.

Bloomberg made gun control work: the cops stop and frisk sketchy looking guys and send them to prison if they have an unlicensed gun on them. At the end of his 12 years in power, a liberal judge ruled that was illegal due to disparate impact racial discrimination, but then Bloomberg got a higher panel to put a kibosh on her. So it was left kind of in legal limbo when he retired.

Blacks are overwhelmingly supportive of gun control measures for this reason. The white insistence on gun nuttery might make white communities marginally safer while making minority communities much more dangerous.

The US has almost 4x the murder rate of comparable countries, the obvious difference being gun control laws.

Uh, no. The difference is an African-American population that commits 60% of the homicides. It's difficult to tease out the Mestizo/Amerindian component, because they are so often counted as "white" when they're criminal suspects or defendants. Anglo-European Americans may be slightly more truculent than their overseas cousins but not by much.

This is the sort of back-flipping that goes on constantly in this debate.

My guess is that white people in the U.S. kill each other more than in Europe, although the absolute numbers aren't all that high. Guns are an overly efficient way to kill.

The high rate of gun ownership, however, protects against the kind of crimes that are more common in Britain, such as drunken assaults and home invasions.

One interesting way to study this would be to compare Koreans in Los Angeles and in Seoul. My guess is from watching video of Korean shopkeepers in 1992 that Koreans own a _lot_ more guns in L.A. than in Korea.

How much more do Koreans shoot each other here than there?

"The high rate of gun ownership, however, protects against the kind of crimes that are more common in Britain, such as drunken assaults and home invasions."

Not true.

What's not true? British law and social practice promote home invasions and they're wretchedly common, as are rural burglaries. The distinction in crime rates between the U.S. and England and Wales is intense regarding the homicide rate, less intense regarding robbery and rape, and, if anything, to our advantage re burglary. (And car theft is vastly more common in England and Wales than in the US).

The difference is an African-American population that commits 60% of the homicides

No, 53%. And it's largely intramural.

Unintentional hilarity.

If you look at homicides per person per year for culturally/economically similar countries the stats are pretty clear...
The US has almost 4x the murder rate of comparable countries, the obvious difference being gun control laws.

On the very page you linked, you can click "show" to view the breakdown by state for the US.

If you look at the homicide rate for culturally/economically similar Northern US states, it's very close to Canada, despite a large number of guns. If you look at the homicide rate for Southern US states, it's much higher. That goes back hundreds of years. (And yes,it's equally true among Southern whites.)

So the obvious difference is not gun control laws.

Typical response from both pro-gun and anti-gun types. It's either "guns are bad so why do even even need to discuss it" or "Second Amendment so why do we even need to discuss it."

Or, "it's black people so why even discuss it." As if it that makes it not a problem or something we can't/shouldn't address.

Well if they're saying "it's black people so let's set aside this notion that the problem is the gun"...that would make some sense.

Ok, asthma is the the same dynamic. Black kids end up in the ER because they can't breathe at a much higher rate than others. So...should we say, "let's set aside this notion that asthma is the problem"?

That's racist, Jan. Giving black kids more medical scrutiny because you believe their lungs are inferior to white lungs. It's 2016.

We don't live i. A world where me can discuss "the negro problem" with an I toward actually making life better for anyone of *any* race.

You do realize that an argument that the US has a normal or low amount of violence compared to its number of guns is at first pass an argument for reducing the number of guns? Indeed, an argument that the US violence is very high compared to guns would be an argument against gun control.

I think that a weakness of these graphs are arbitrarily excluding the middle income countries with high violence as inherently not westernized. It's a self fulfilled prophecy.

Canada is not demographically similar to the USA as far as comparisons of homicide rates go. If there where as many "First Nations" people in Canada as there are Native Americans plus African Americans in the USA, Canada would have a higher homicide rate. Those 3 groups vary so greatly in regard to homicide rates that it messes up the comparison. So you should compare people descended from Europeans in the 2 countries. When you do the USA is still higher but the difference is much smaller. Maybe gun control is a factor but am not very confident on that.

Canada has a lot more guns per capita than Europe.

Here's a way to look into the effect of sheer number of guns: count up the number of murder-suicides among white Americans versus highly gun controlled European countries. Back in 2010 I read through Jill Leovy's L.A. Times' homicide blog of all the murders in giant Los Angeles county for 3 years. A lot of the homicides with white victims involved immediate suicides by the shooter -- e.g., a middle aged guy who lives with his mother shoots his mom and then shoots himself. Some of these were no doubt planned, but I imagine some of them wouldn't have happened without the abundance of guns in Los Angeles. If you compared the number of domestic murder-suicides among whites in Los Angeles County to the number in, say, Scotland or Israel, you might get a reasonable effect size for effect of a large number of guns.

The phrase "assault rifle" is a prime example of how authoritarian extremists seize control of the vocabulary to advance their agenda. In reality a very small number of murders in the US are committed using long rifles of any description. The campaign against them mindlessly distracts from meaningful initiatives that would have positive outcomes. Nevertheless, they are effective as a purely symbolic rallying point for the goosesteppers.

So the 'obvious'. Difference is gun control laws, huh? Not the large inner city underclass, gang wars, the war on drugs, racial divides, or any of dozens of major cultural, geographical, or sociological factors. Nope, it has to be gun control, because the narrative needs it to be.

How about comparing the gun homicide rate between U.S. States and their Canadian counterparts of similar population density and racial diversity? How does Alberta stack up against Montana? British Columbia vs Washington State?

If you want to get even better data, try controlling for economic status, population density, gang activity, etc.

And if you want to get really sciency about it, you'd check your 'gun control the only difference' belief by looking at the rate of other violent crimes. For example, the robbery rate in the U.S is 42% higher than it is in Canada. It appears that gun control is not the only difference between us.

Oh, and by the way, Canada has a LOT of guns - about 30% as many per capita - and 'assault rifles' can be purchased over the counter at Cabelas without even having to register the gun, so long as you have a firearms aquisition license, which you can get after an afternoon of training at a gun range and a background check.

The other problem with your thesis is that the U.S has been setting new records for gun sales almost every year for the past 10 years - while the gun homicide rate has been falling.

"The other problem with your thesis is that the U.S has been setting new records for gun sales almost every year for the past 10 years – while the gun homicide rate has been falling."

I think you have the wrong denominator. It's not the number of guns but the number of gun owners which should be correlated with the homicide rate.

I don't understand why there's such a focus on the number of guns per murder. I doubt the likelihood of someone committing murder increases as they add to their collection.

Guns per murderer? You can only murder with so many guns at once.

Any why not suicides? And attempts? Saving gun victims from gun deaths has gotten much better as of late.

Ordinary people in Japan do not have guns. Just about the only guns used to kill people are guns obtained illegally by yakuza to kill other yakuza. The situations in Japan and the US are in this respect so vastly different as to render this kind of exercise meaningless.

Most non black people in the US don't shoot each other. Just about the only people who kill each other at significant rates with guns are American blacks. So the situation in the US and Japan are not that different.

So is the point that guns in the US, on average, are less likely or about as likely to be involved in a homicide? That's great news! We're not a nation of psychopaths!

Unfortunately, we're up to our armpits in metric shit-tons of guns, so more of us are going to die from guns. Oh well.

Just shows that trying to regulate guns does not accomplish much. As in suicide, rates are about the same, and if a gun isn't present the user just goes to the next means.

Not true for homicides, obviously.

I honestly don't undderstand this. Is the argument here that if we make 10 million more guns tomorrow we'd see in increase in the number of gun deaths equal to the above reported rates?

Alternatively, is the claim that we'd see a significant reduction in the gun-deaths/100000 guns if we confiscated half the existing guns?

My guess is that in reality we'll simply see the rate per number of guns change much more than the actual gun death related numbers. Wonder if there's a new study on that (not looking for pointers to Lott on this question right now)

It would be interesting to compare homicides rates for Asians and Asian-Americans.

I'm confident that Tyler's point in plotting murders per gun instead of gun murders per person is that guns kill people. Clearly our guns are slightly more violent than Japan's because they have a different demographic.

According to the Obama Administration's Bureau of Justice Statistics, from 1980 to 2008 African Americans committed 56.9 percent of all gun homicides.

So the black gun homicide rate per capita is about 8.8 times the rest of the population's.

It sounds like America basically has a black gun problem.

In this century, New York City drove down its homicide rate to exceptionally low levels by having its huge police force stop and frisk vast numbers of youngish male blacks and Latinos. But NYC can get away for longer with tactics that would be instantly condemned as racist due to disparate impact if practiced by police departments in less famously liberal cities with less important people living there.

Here's the Obama Administration's 2011 report "Homicide Trends in the United States, 1980-2008" by Alexia Cooper and Erica L. Smith, BJS Statisticians

Under the Bush Administration, the predecessor of this report was available in HTML but now you have to open a PDF to read the eye-opening numbers.

But the Homicide rate of decedents of Europeans is unusually high in the USA.

Yes, in a relative sense. On the other hand, how big is the absolute difference between U.S. whites and, say, Scots or Irish?

We could calculate the difference in years of life expectancy. The numbers for homicide are available. I don't know what they are. I'd guess about 0.3 years of life expectancy. Anybody know?

I'm trying to think of anybody I've known who was killed with a gun ... Actually, now that I think about it, my ne'er do well uncle John Smith was murdered about 50 years ago during a dice game in Harlem.

Looks like Ireland is at about 1.1 and last I saw USA whites where at about 2. That is a huge difference, almost 2x.

I knew 2 people who murdered. One was a part of gang warfare. Our police do not do a very good job, more and better police could help with that sort of murder. The other was a kid that I worked with bought some crack from a guy who gave him something other than crack and took his money, he took out a gun and shot him right there. Ending the war on drugs might help with that sort of thing. Some might see that latter case as justifiable, are people in the USA more polite due to guns?
I also once saw a young thug that I knew get stabbed, I think the news said 18 times, but he survived.

You must have lead a very sheltered life. I went to "bad" schools. If I remember correctly one thug kid at high school died in police custody. I one saw a cop hit a young thug with his car sending him sprawling but the kid got up. I got some stories.

You must have lead a very sheltered life.

I grew up in an incredibly average (non-Southern) area with a population of about 1.1 million. About 90% of the population lives in Zone A, which has a homicide rate which averages to 2.4 per 100,000. The other 10% lives in Zone B, which has a homicide rate of 35 per 100,000. There's some geographic variation in Zone A (between tract developments and rural areas and small towns, for example). You're not 'sheltered' if you live in Zone A. Zone A is what's normal.

I think if you unpack it you'd discover it's unusually high in the South. For instance, in Upstate New York's non-metropolitan counties, the homicide rate is about 1.14 per 100,000. That's higher than you'd expect in comparable areas in England, but not a great deal higher. I have family in Franklin County, Va., a largely exurban area with no obtrusive black population. The homicide rate in Franklin County Va is 3.5 per 100,000 with little year-to-year variation. (I'd wager virtually every homicide is a domestic dispute or a bar fight).

Yes, interestingly North Dakota before the oil boom had a Europe level of murder despite being awash in guns.

Also: It is interesting to me that the white homicide rate is much higher in the south than the northeast but that the black homicide rate is much lower in the south with 2 exceptions Louisiana (high black homicide rate) and NY (low black homicide rate). NY could be because many of the blacks are from recent immigrant families.

Which part of New York? The non metro counties have small black populations which nestle there unobtrusively. They are not distinct as regards to their origins.

Overall, about 9% of the black population consists of immigrants. I doubt that has more than a mild effect on homicide rates.

Well-behaved black immigrants have an effect on black homicide rates in NYC, especially in Queens.

North Dakota until the energy boom had a relatively old population, and older people don't shoot each other much. Similarly, poor, dysfunctional West Virginians have lots of guns but their homicide rates don't stand out unless you do a lof adjustments. They're white and old.

West Virginia's population is 90% exurban, small town, and rural. For a place with little in the way of a metropolitan-urban population, their homicide rates stand out a great deal.

If the NRA were really serious about making Americans safe, they would be focusing on getting blacks to join the NRA and then start buying lots of guns and bullets.

Clearly blacks globally need five times as many guns as they have now, minimum!

And Muslims need to join the NRA and buy lots and lots of guns.

But the group that really needs to join the NRA and really armor up is the LGBT community because they suffer the most per capita.

Might as well, bring in the Mexicans as well.

If I could, I'd work real hard to get Obama to restart the Black Panthers for Self Defense, lead the movement to get blacks to flood the NRA with members on a sustained basis to take control of the NRA. When Obama and Al Sharpton and Tavis are the spokesmen for the NRA, and Michelle is pitching guns, then Congress will act on gun legislation.

And your point is what?

That we should subject some people to arbitrary search an seizure, even though many studies suggest stop and frisk and a relatively modest role?

That if gun violence in our country affects a minority group with more pigment in their skin, the rest of us should not care?

Or are you the type of guy who likes to walk up to people and say "you look really ugly" and feel good because you think your being honest?

This is how I've understood America's Gun Problem - equivalent to the rest of the western world, possibly lower, with pockets of extraordinarily high violence which raises the national average. I also think including Suicides in gun violence is extremely misleading. In other words, that it's not really a problem for the NATION itself. Some cities/neighborhoods have big problems, but not states or the nation.

I'm a strong 2nd A supporter, and grew up in the countryside where bringing a hunting rifle to school wasn't a big deal (kept locked in your truck, of course), because they'd go hunting/skeet shooting after school. Or actually during school, we had skeet shooting during PE sometimes. There was a CULTURE of responsible gun ownership there, and that is being lost. They are tools, to be respected and treated carefully. Always assume it's loaded. Never point at another person, even if empty. Look behind your target. Don't threaten. Frankly I view the open carry folks as... not wrong, just insecure asses. I don't know how to reverse cultural trends, but even among gun owners that respect seems to be, slowly, fading, as the mere act of owning a firearm becomes a bold political statement--which it should not be.

We don't have a gun problem, we have a people problem. With the high murder rate, it's either due to poverty/gangs and evil/mental illness. The turd that shot those people in Orlando was evil. Not mentally ill, but straight up evil. I don't know how you stop people from being evil. I don't know how you create opportunity for the blacks in Chicago such that gangs are no longer an attractive prospect. Banning guns is mindlessly easy, and doesn't address any of the actual problems.

When I was at UCLA in 1981, practically all the letters to the editor in "The Daily Bruin," were about gun control, drugs, or abortion. In fact, many were about all three. It was common for readers to write in about how it was impractical for the government to elimiate one or two of those three, but, for some reason, it was extremely rare for anybody to say it would be difficult to effectively ban all three.

This would seem especially true for guns, which are after all a consumer durable of which there are currently 200 million or more in the country. The current inventory of illegal drugs in American probably turns over once or twice per year, since the point of drugs is to consume them. But guns don't get consumed. They last a long time.

It might be possible, as Daniel Patrick Moynihan, suggested to crack down more on ammunition. But then Weekend-in-Chicago levels of accuracy probably don't suggest that criminals are currently spending a lot of time practicing their marksmanship at the range anyway.

Mayor Bloomberg of New York, however, showed that it _was_ possible for an extremely well-funded police force to get more guns off the street: you just tell your cops to stop and frisk suspicious-looking individuals. Granted, they almost always turned out to be black or Latino males. But for most of Bloomberg's 3 terms in office he didn't have much problem with civil rights lawsuits because, well, important people live in New York City and they liked the results, disparate impact be damned.

Steve, I'm more open to your writing than a lot of folks around here, but come on dude. I spent four years at UCLA and would bet that not a single letter to the Daily Bruin discussed the triad of guns, drugs, and abortion. Either a lot changed between your years and mine, or you're just completely making shit up.

I never make up stuff.

That's what the Daily Bruin letters' section was like in 1981. Go look it up.

Gun control was a particularly big deal in 1981 because of the shooting of John Lennon on December 8, 1980. For example, I went to see economist John Kenneth Galbraith give a speech (he was a rather inept speaker) and he felt compelled to begin with a plea for gun control that didn't have anything to do with the rest of his address.

Maybe those three topics are what the kid who was Letters Editor in 1981 was into, I don't know. But, obviously, I do have decent memory and pattern recognition skills for opinion journalism. It's not a very august claim to be able to remember the obsessions of college newspaper letters to the editor columns after 35 years, but that's what I can do.

Actually even when you adjust for socioeconomic status America comes out worse in the gun murder area, driving our murder rates higher than other countries.

It sounds like, until a member of your family commits suicide with a gun sold as a self protection device by the NRA hucksters, it is not much of a problem for you.

The second amendment is not intended to promote stupudity.

What's the difference if a family member commits suicide by gun or by hanging himself? Seems you're more concerned about the gun than the family member.

Except it's been proven time and again, using various methodologies, that more guns = more suicides. It's sad that the CDC isn't even allowed to study the issue, but good research nonetheless has been produced.

An important distinction involving guns and suicides is between depression-driven suicides and terminal illness-driven suicides. I actually don't know what the numbers are.

Just hope it isn't a murder suicide, or that you have to clean up the mess.

Also more anal sex with strangers means more suicides. Oh, wait, you don't actually care? Nevermind.

But if the first amendment isn't intended to promote stupidity should we get rid of it?

We used to have this widespread concept of a gun as a tool, that's no big deal to own and use provided you were trained and treated it with respect. Not that different from a table saw. But there's been a real split from the middle - on the one hand, people who look at a gun like it's radioactive, or is literally going to bite them of its own accord. On the other hand this group of "enthusiasts" for whom it's an object of obsession.

It's a way for white people to be angry at each other over the bad behavior of blacks:

The endless gun-control brouhaha, which on the surface appears to be a bitter battle between liberal and conservative whites, also features a cryptic racial angle. What blue-region white liberals actually want is for the government to disarm the dangerous urban minorities that threaten their children’s safety. Red-region white conservatives, insulated by distance from the Crips and the Bloods, don’t care that white liberals’ kids are in peril. Besides, in sparsely populated Republican areas, where police response times are slow and the chances of drilling an innocent bystander are slim, guns make more sense for self-defense than in the cities and suburbs.

White liberals, angered by white conservatives’ lack of racial solidarity with them, yet bereft of any vocabulary for expressing such a verboten concept, pretend that they need gun control to protect them from gun-crazy rural rednecks, such as the ones Michael Moore demonized in “Bowling for Columbine,” thus further enraging red-region Republicans.

Steve, white people with an interest in crime control live in mixed neighborhoods in the inner city, or in neighborhoods adjacent to the slums, or are commuters through troubled areas. Suburban whites are insulated and often quite contemptuous (which I discover in fora like this). The cultural problem in my home town is composed of four phenomena which re-inforce each other: suburban irritation and contempt, white liberal stupidity and virtue-signalling, the stupid emotional neuralgia of the black bourgeoisie and parts of the black working class, and a mix of demobilization and embarrassment which prevents the rest of the black population from asserting its interests. White liberals babble about gun control because they cannot and will not acknowledge the utility of police and prison guards, who are men who have sensibilities and interests quite different from white liberals as a rule.

Vox has done their own charts, generally coming to the opposite conclusion to those stated within this post.

I can't embed the video directly here, the comment box is plain text, but I provide the link to YouTube here

The state of gun violence in the US, explained in 18 charts

I think this is the Vox article you are mentioning:

I bookmarked it awhile back because it was so good with statistics on the subject, unlike this website.

That article generally comes to the same conclusion as this post.

???? John, please dont mislead the audience. Your statement is false and unsupported.

Here is the first paragraph of the Vox piece:

"Among developed nations, the US is far and away the most violent — in large part due to the easy access many Americans have to firearms. These charts and maps show what that violence looks like compared with the rest of the world, why it happens, and why it's such a tough problem to fix."

So, you think Tyler's argument is the same conclusion as the Vox article.

You can quote any part of the Vox piece to show me that the Vox piece "comes to the same conclusion as this post."

What part of this post do you think constitutes "argument" ?

Did you switch to Vox once Jon Stewart stopped feeding you confirmation?

Notice that John Thacker could not offer evidence even though he was given the opportunity to do so.

I guess you didn't notice the game they were playing with terminology. They keep switching between gun related homicides and the more general gun deaths, depending on which fits their agenda better. At least the charts presented here stick to the same one all the way through. Adjusted for that, I suspect their numbers would indeed agree with the ones here.

Also, I don't understand why in the US context, "guns" go together with "freedom".
Being shot by a bullet is not voluntary. It is an act of force. The state has a legitimate role in maintaining both internal and external peace.

From the UK the American obsession with guns is just bizarre.

A gun is just a tool. Maybe we just don't share your bizarre emotional responses to inanimate objects.

These numbers, along with previous posts about suicide rates show America basically has the same rates (murder and suicide) as the rest of the world. The obsession with method makes one think that the person who dies is not really what you care about.

No. The 1934 law established limits and differences between civilian and military/police weapons.

The great game you play is that the existing division does not exist, can not be modified.

How is this at all surprising?

Criminal groups almost everywhere will have access to guns. The rest of the gun owning population will a) mainly act as a dampener and b) its share of the total population will basically be a function of regulations and GDP per capita.

The US is very rich and has low restrictions and a vibrant gun culture so despite a relatively heavily criminalized population (if one concentrated amongst certain demographic groups) gun murders are drowned out by the sheer prevalence of guns.

If you torture the data long enough, it will confess to anything ( Darrell Huff )

Is this math the NRA does to feel better?

I would be curious if Tyler would explain why he felt this was interesting. I confess, I read just the title and a enough sentences to see that I could not care how many murders per gun we have.

I find it very curious that so many seem to believe this chart was an insidious attempt to promote a pro-gun agenda. It seems very neutral to me, just a data dump basically. In fact I don't think it's very enlightening either way.

Ok. Since it doesn't appear to promote any specific agenda or help in forming policy, why do the analysis? And why recommend that people spend time studying it?

A common refrain I protest against in the business world, is "in order to make this analysis sound, we will generate some statistics." The statistics are useless without a logical line of reasoning to the business or policy conclusion.

I agree. But you need to consider the statistics before you can even begin to reach a conclusion. You seem to assume it works in the opposite direction.

In any event, even assuming that this post is an attempt to push one side or another, it seems to me that the logical conclusion is that the USA's main problem is not the rate at which guns are used, but the sheer number of guns. Which implies a solution more in line with your priors.

Anyway I've never seen anything from Prof. Cowen that suggests he's the pro-gun type. He's a middle aged urban professor who wears powder blue oxford shirts and hangs out with Ezra Klein. Not exactly Larry Roy sitting down at his hunting camp with a collection of 15 rifles of various caliber and a XXL can of dip.

A few points which hope to add (but I’m sure will distract),
1) It is not often that we think about measures only in the unit efficiency. I had no preconceived notion of what the figure should be coming into this. While it is reasonably low, the figure also discounts the (relatively) long unit life.

2) While acknowledging that gun unit efficiency is only one measure, it seems that it makes most sense to first understand the whole problem before jumping in with individual measures which seek to show only a part of the issue

3) The ‘whole’ issue seems to be most simplified by:

A: Gun unit efficiency: Deaths / Gun
B: Gun penetration: % of households who own a gun
C: Gun prevalence: # of guns/ Gun owning household
D: Population measure: Households in a country

A*B*C*D= E
E= Gun deaths per year

4) I suspect, but don’t know, that the USA is most extraordinary compared to our peers in the B & C measures.

5) It seems to me that death is something that we all should seek to minimize, thus the use of this measure rather than, for example, homicides. Accidental death and suicide is also important to minimize.

6) Secondly, while it may be that certain groups have higher prevalence of gun use or death, it seems that any race identification distracts from the problem rather than seeks to improve the issue. Like point 5, I don’t care who is dying only that it happens less often.

Ok, a pretty consistent story, though I think it makes more sense as plotted here ( The number of homicides per gun makes sense, and gives a peak in density, in that having more guns is correlated with more gun murders. There are deviations, due both to the number of guns (where you might expect diminishing returns) and exogenous factors like demographics, economics, culture etc. (americans shoot a lot of people even more than the number of guns would predict).

Policy-wise, if H was a constant across countries (an infinitely sharp density peak) the case for gun control would be clear, if there was no relation (a broad distribution) it would be clearly pointless.

Let's analyze cirrhosis deaths per 100k and and give it the same treatment, thus......divide by alcohol consumption per capita :)

Data from the WHO (2010). Russia is the top 3 of alcohol consumption per capita. If you divide the male cirrhosis death rate per 100K (48.7) by liters of pure alcohol per capita per year (15.1) the result is 3.2. On the other side of the world, Argentina and Brazil have similar ratios: 3.2 and 3.31. But these ratios come from 29.8/10.3 and 28.8/8.7. On one hand: being drunk is an utility. The reason the cirrhosis deaths per 100K inhabitants are higher in Russia is because the average consumption is higher, approx happier. On the other hand: cirrhosis deaths per 100K in Russia is almost twice compared to Brazil.

Something similar happens with guns. There's utility in buying, owning, collecting, shooting and fooling around with a gun. But, more guns yield a higher homicide with gun per 100K stat.

However, there's a difference between alcohol and guns. It's been acknowledged for a long time that alcohol abuse is not good for the individual or society. When friends tell you you're drinking too much you can answer in two ways: a) thanks buddy for helping in the fight, avoid vomiting an officer, and taking me home, I'll drink less/slower next time or b) I'm an adult you can't tell me what to do. I'm well over drinking age, it's my constitutional right....etc. I think the situation is closer to b. Don't be the annoying self-destructive alcoholic.

There isn't really any evidence that gun control would bring down homicide in the US. In a cross-state regression, gun ownership enters non-significantly or negatively. The important correlates of homicide are % in poverty and % black:

So surely the correct interpretation of this data is the the death rate per gun is fairly consistent (ie. To say 'guns kill people' is probably more true than 'guns don't kill people, people kill people'). Thus an increase in the number of guns in an absolute sense will strongly correlate with an increase in the number of gun homicides in an absolute sense. Therefore less guns is better than more guns.

So a country with 100 guns may have about 300 gun homicides, but a country with 1,000 guns would have about 3,000 gun homicides.

Further, this appears to completely refute the argument that guns are effective protection (if only the criminals have guns...). Were that the case we should expect to see the US coming in well below the gun control countries in relation to gun homicides per gun, as in gun control countries only criminals have guns, whereas in the US both criminals and law abiding citizens have guns.

I'm sorry, but, why is normalizing the number of murders to the _number of guns_ even slightly reasonable?

Because the gun-control argument is that guns are the problem. But we have many more guns than a lot of countries and a lot of countries have far higher murder rates. And if you disaggregate the stats by race, white America is comparable to those way more homogenous white nations. So it would appear to be a which-people-have-guns problem.

Not comparable at all...

I grew up in a hunting culture, in which owning a gun and hunting were a right of passage. I was given my first gun, a single shot 20 gauge, when I was about eight or nine, and later given a 20 gauge automatic (three rounds) and a rifle. We had a gun case in our home, as did many of my friends. We had access to very good hunting (deer, turkey, quail, dove) on private property, and hunting trips were as much a social experience as a hunting experience. That was a long time ago, and years ago I gave my single shot to my great nephew and my other guns to friends who continue to live in a hunting culture. I no more want a gun than a motor cycle, as the risk of owning one is far greater than the benefit I might derive from it. But that doesn't mean I oppose guns or gun ownership. While I no longer hunt, I fish, and fish often - my home is on a salt marsh creek - and many of my low country friends own guns for hunting. What I don't understand is the almost unregulated sale of weapons, guns that are intended for hunting people not game. I can't recall seeing one of these weapons when I was growing up. Now, guns for hunting people are advertised on television and in the newspaper, and I wouldn't be surprised if their sale exceeded that of shotguns and rifles intended for hunting game. The argument for weapons intended for hunting people is that they are needed for protection. Protection against whom, a deranged mass killer in a mall, school, or night club? Fear causes people to act irrationally, and politicians exploit that fear for their own gain. In the South "open carry" laws have become commonplace, the places allowed to carry ever expanding, to schools, churches, bars, and some politicians encourage people to carry weapons for hunting people to public places just in case a deranged killer is present. That advice is as deranged as the deranged killer. Geraldo Rivera said the victims in the Orlando night club were at fault for not attacking the killer, for not sacrificing their own lives for the greater good. Rivera is deranged. Donald Trump said President Obama is at fault for not intervening in Syria, overthrowing the Assad government, and empowering the Sunni Muslim extremists who are waging war against the Shiite (Alawite) government in Syria and carrying out mass killings in Syria, Iraq, and Europe. Trump is deranged. Fear makes people act irrationally. The data presented in this post are supposed to show that the number of guns isn't the greater risk of a mass killing; rather, it's the purpose for which the guns are used that is the greater risk. I would modify that with this: it's not the number of guns but the type of guns that are the greater risk, guns intended for hunting people a far greater risk of a mass killing than guns intended for hunting game.

I no more want a gun than a motor cycle, as the risk of owning one is far greater than the benefit I might derive from it.

What risk? You don't seem to be either a reckless or depressed individual.

In the past 50 years, a little less than 1,000 people have been killed in the U.S. in mass killings (defined as four or more deaths, excluding the shooter). There are about 10,000 suicide deaths by firearm each year in the U.S., and about 15,000 unintentional, non-fatal firearm injuries each year in the U.S. There are almost 5,000 deaths a year from motor cycle accidents in the U.S., and over 92,000 non-fatal injuries a year from motor cycle accidents in the U.S. Over a 50 year span, that would total about 500,000 suicide deaths by firearm, 750,000 unintentional, non-fatal firearm injuries, 250,000 deaths from motor cycle accidents, and 4,600,000 non-fatal injuries from motor cycle accidents.

Mass killings by firearm. Implicit but clarified.

What is this replying to?

Clarifying the statistic that about 1,000 people have been killed in the past 50 years in mass killings - mass killings by firearm. Mass killing by other means (such as jets flying into buildings) aren't included in the statistic.

I asked you what risks you faced by owning a gun, given that you don't seem reckless or depressed. And if you are in terminal health and decide to pull the trigger before you end up in hospice, that seems like a non-systemic risk for gun ownership.

Under an Australian style "category" system a single or double barreled shotgun would be an easier "get." Just say you want to hunt, or shoot skeet. A removable magazine rifle should be a much harder "get." Why do you want it? A non-crazy reason please.

You had an automatic shotgun as a chikd and give guns to children but complain about semi-automatic rifles owned by adults withiut criminal backgrounds? Uhh, what?

So, guns aren't the bottleneck in terms of problems in the US, meaning that we don't have a shortage of guns. ie if I cut the gun supply, there's slack to be picked up, making that fix less then useful.

This is like asking

How many

Idiots per cubicle.

Only an idiot would frame it this way.

If we just had more guns in the hands of everyone,

The statistics would get even better!

Per gun might not be the right measure.

Theoretically having *1000 more guns will make your number close to zero.....

Still an interesting point.

From today's NYT:

"To give you a sense of how unusual America’s gun violence problem is, consider the daily death toll compared with other Western democracies. The chart below imagines that the populations of those countries were the same as the population of the United States.

No Other Rich Western Country Comes Close
Shown are Western countries that have G.D.P. per capita over $25,000 and that make statistics on gun homicides available.
Sources: Small Arms Survey (2007–12 average); World Bank
International comparisons help highlight how exceptional the United States: In a nation where the right to bear arms is cherished by much of the population, gun homicides are a significant public health concern. For men 15 to 29, they are the third-leading cause of death, after accidents and suicides. In other high-income countries, gun homicides are unusual events. Last year’s Paris attacks killed 130 people, which is nearly as many as die from gun homicides in all of France in a typical year. But even if France had a mass shooting as deadly as the Paris attacks every month, its annual rate of gun homicide death would be lower than that in the United States."

I would suggest breaking the US down into individual state numbers, and you'll see that these are even more dramatically different. There are US states awash in guns with extremely low ratios along these lines (in New England and in the West), and US states with very high ratios (mostly in the South.) (And yes, the states with high ratios also have high rates among whites, it's not just a racial artifact.)

Why stop there John.

If you played with the statistics countries which banned guns would have the highest rate of gun deaths per gun.

Is this really a website that invites critical thinking. We both know that if you have a high rate of homicides by guns per 1000 population (that is, a high numerator), what you do to lie with statistics is invert the numerator, making it the denominator, and put whatever number you want on top.

This magic trick works because, with a slight of hand, your eye came off the pea under the shell: the number of gun homicide deaths per thousand.


Except that there are a lot of states with European levels of murders. The disparity is amazing.

In any case, isn't the naive interpretation of these graphs pro gun control, arguing that US violence would be reduced by reducing guns? I think that the graphs are misleading because of aggregating states, but without disaggregation it's pro gun control.

I am ashamed of all the mood affiliation here among the comments, assuming conclusions not in the post.

John, see immediately below where I decompose by state the statistics on gun death per 100k.

I've done some scatterplots of gun ownership rate (not raw numbers of firearms) vs firearm deaths by state, and nationally, there is a relationship between higher ownership rate and more deaths.

However, looking within regions, the relationship is flat within the northeast (unless you exclude Delaware (high deaths) and Massachusetts (low deaths) as outliers). The midwest also shows no relationship (but you can massage this depending on whether you want to exclude Missouri and North Dakota).

If you exclude the 3 coastal states, the west shows a slight increasing slope. And I note that you can get rid of lots of relationships if you start chopping data into finer and finer slices. So we shouldn't reject the national correlation just because the northeast and midwest may not have it.

But the relationship is as strong as ever among the southern states: higher gun ownership rates correlate with higher gun death rates. Florida is actually very close to the national median in terms of both gun ownership and gun death rates. But it's closer to the bottom among the southern states.

To build the scatterplots, just use Bill's source below, then add data from Kalesan and Keyes 2015 ( and if you really want to look at raw firearms per capita, you can get that data from

John, Let's look at the facts. Rather than homicides per gun (where you say the South stands out), let's look at the statistics of deaths from guns per 100,000 and see where the South or other states awash in guns stand out relative to all states

Here are the list of states which exceed the national average of 10.3 deaths due to injury by firearms per 100,000 and let's see how the South, awash in gun states and all other states stand out relative to the average (i.e., this is a list of states which exceed the national average of 10.3) :

Alabama (16.9); Alaska (19.2); Arizona (13.5); Arkansa (16.6); Colorado (12.2); Delaware (11.1); DC (11.7); Florida (11.5); Georgia (13.7); Idaho (13.2); Indiana (12.4) ; Kansas (11.3); Kentucky (13.9); Louisiana (19); Michigan (11.1); Mississippi (18.3); Missouri (15.3); Montana (16.1); Nevada (14.8); North Carolina (11.8); North Dakota (12.3); Oklahoma (15.7); Oregon (11.7); Pennsylvania (10.5); South Carolina 15.5); Texas (10.7); Utah (12.3); West Virginia (14.6); Wyoming (16.2).

So, let's look at those east coast states you were pointing to with low gun ratios: Connecticut (5.0);; Maryland (9.0); Massachusetts (3.2); New Jersey (5.3); New York (4.2); Rhode Island (3.0).

Here is a link to the source of deaths from firearms by state:

My vague impression is that Catholic states don't do suicide much. For example, Rhode Island is quite Catholic, kind of screwed up and downscale, and has a very low gun death rate.

In contrast, the northwest quadrant of the country has a higher suicide rate among whites, perhaps because it's pretty ethnically Protestant and not very religious.

For example, Rhode Island is quite Catholic, kind of screwed up and downscale, and has a very low gun death rate.

It isn't. Personal income per capita in Rhode Island is about 5% higher than national means. It is unusual in that nearly 90% of the population lives within a metropolitan commuter belt, but the share actually living in dense settlements over 50,000 in population is not radically higher than the national mean (approx 70% v. 65%). Rhode Island has an abnormal quantum of corruption, an abnormally influential Siciliante mob (or it did at one time), and a large population of boutique ethnics (French Canadians and Portuguese). It's also the state with the highest share of nominal Catholics (though Providence is not a notably vigorous diocese). At one time, it also had nearly the lowest divorce rate in the country. It's a curious place, but not downscale.

Western states have a great deal more demographic churn. That sort of thing is correlated with problems which seem to be exacerbated by weak social networks. Pennsylvania and Louisiana are largely populated with natives (80% or more). Not so Alaska or Nevada. I think Norval Glenn published some academic articles on this ca. 1971, making use of Canadian data. The divorce rate in British Columbia was many multiples of that in the Maritime provinces, and that peaked his interest.

Exactly right. For example, MA is low, while OR tops the list.

Counting suicides is absurd. It is not "violent" in the sense that you are hurting anyone but yourself. It is also not something that you can simply infer that removing guns would solve (look at sky high suicide rates in Japan, Korea). Using numbers without suicide tells a completely different story. Utah's numbers, for instance, fall to low European levels.

Including suicides is dishonest when you support the right to commit suicide like a good liberal.

It's funny to do such graphs in response to a mass shooting, because on average there are no mass shootings. That's the great justification for keeping ARs and chest packs full of magazines. On average, they are never used. The median number of mass shootings per "rig" is zero trailed by a long stream of zeros.

So AR's for everyone.

As I've noted before in these debates, I'm sure lots of German Jews and Middle Eastern Christians had wished they'd kept more firepower.

The frustrating thing for me is not that people think they have a right to be paramilitary armed, that's crazy, but not frustrating.

The frustrating thing is that people won't admit that 50 nightclub deaths are part and parcel of that decision, that cultural value.

Because if everyone has the right to be a one-man army, to transport kit in the trunk of one's car, then everyone has the option to just do it.

Part and parcel.

"Because if everyone has the right to be a one-man army, to transport kit in the trunk of one’s car, then everyone has the option to just do it."

And another part is that everyone already has the option anyway, whether legal or illegal. Think about the Bataclan, and then look at French gun control laws (let alone controls on possession of grenades and other devices.)

You could almost certainly figure out where to get ammonium nitrate and diesel fuel. That doesn't mean you're likely to follow in Timothy McVeigh's footsteps.

I think the laws shape outcomes in two ways. One is that if it is illegal to have a trunk full of paramilitary gear, that is an easy arrest. You don't have to send someone on their way fully kitted.

But the second is the behavioral view. It would say that different views of death have different psychological responses. Perhaps someone is suicidal. There is not a gun in the house, but there is a radio. Would they really be equally ready to get in the bath and drop in the radio? I think the answer is no. By that I don't mean that no people would choose the radio, but that that there would be different rates of uptake, and that matters. This is supported by the data, that houses without guns suicide less.

The same goes with mass murder. Yes, someone could do it with gasoline plus whatever, but would they really hold the two approaches with equal fascination? I think not. This too is supported by the data, that people without guns kill less.


I have some very bad news for you: small arms are prolific, all over the world.

Seriously, go and die in a ditch. Jews fought back all over Europe. And they were killed. The Wermacht was destroying professional armies and you think a group of civilians with hunting rifles would have been able to do anything? You are an anti-semitic POS.

Yes, because they actually did do something. As delaying actions and to so confusion weapons proved highly useful. For instance, at Sobibor more Jews escaped during the armed uprising that through any other means. Likewise at Treblinka and in Warsaw, men with guns can buy time and space to get a few people out.

Or look at the American South, guns were one of the few effective counters to the Klan. During the 1919 violence I had relatives who survived by taking up arms and standing down mobs with rifles.

A minority population will never win against the organs of state in an existential struggle, but firearms can raise the cost to prohibitive levels that the majority may simply be unwilling to bear. It can certainly buy time to either organize more fully or to get the heck out of Dodge.

They would have fared a lot better than if they did nothing. Id rather die in the warsaw ghetto uprising than in a gas chamber any day.

"The Wermacht was destroying professional armies and you think a group of civilians with hunting rifles would have been able to do anything?" - When your blood pressure comes down a bit, and if you actually care about history: Money quote: "Switzerland was at risk of being invaded by Germany during World War II but was spared, historians say, because every Swiss man was armed and trained to shoot." Armed civilians may not win pitched battles against armored divisions, but can make an occupation next to impossible.

You are not far wrong, and I wouldn't normally fault you, but since you said "care about history" ...

An active Reserve system is different than people who do no training. It is different than the political "militias" we have in the US:

My Swiss friends tell me that you get basic training in the Reserve, and then a gun and a sealed block of ammunition for your home that you may not open, under penalty of law, until official call-up is made. I think US gun enthusiasts would not live under those oppressive rules.

Your point is absolutely correct, but (I think) somewhat orthogonal to mine. Yes, active reserve is way more effective than untrained militia, at a ratio of... 1-to-4, maybe?...1-to-6?.... But Switzerland is small-ish. What Switzerland deterred with 2 million active reserve trained men, France could have matched with 12 million armed, but under-trained "militia". Yet, Hitler invaded France, but not Switzerland.

My point is: "civilians with hunting rifles" would be no match for the Wehrmacht (per efcdons) is wrong. The armor would run right over, just as the US did in Iraq. Governing the conquered territories would have been problematic. The US Revolutionary War bolsters both your point and mine.

I think that it would be better to choose countries strictly by income rather than defining Westernized in a way that excludes Latin America. It is basically selecting for low violence.

Here's the problem. A country like Venezuela or most of Africa with very high rates of gun homicide has serious social problems and issues related to low intelligence and education that make them dangerous places. If suddenly all the guns in these places were removed the same numbers of death would happen but by machetes and clubs. On the other side of the coin what this kind of evaluation misses entirely is the safety of those people not prone to violence. For them guns SAVE lives. Just the presence of a gun in the hands of a law abiding person or potential victim saves lives. Probably if you took the guns away total homicides would go up in 90% of the countries.

Remembering the genocide in Rwanda in 1994 imagine if every household had one gun. Most of the million or so people who died in this tragedy were innocents, the young, women and old people. There very kinds of people who cannot fend off an attack from thugs using a machete or clubs.

The actual stats in America are that gun owners die more than non-owners.

One of the more tragic ways this happens is that the family owns guns, kids hit difficult adolescence, choose suicide. In that case it is not the purchaser who suicides.

Tragic, yes, but clearly not the same sort of problem as gun homicide. If the kid chooses any other form of suicide, do you call for 'control' or banning of the implement? Rope control? Pill control? Razor control?

Suicide in the U.S. is about 10 per 100,000. In Japan it is 20 per 100,000. Most Japanese suicides don't involve a gun. Yet your point seems to be that the presence of a gun makes suicides more likely. Hmmmm! maybe you should rethink that theory.

Neither of you really addressed the data. For the culture that is America, with gun ownership as the marginal change, suicide (and homicide) increases. Yes they "could" set themselves on fire. For some reason they do not at the same rate.

"the culture that is America"!!! Was that supposed to be insulting?

America is way down the list on both homicides and suicides. In fact countries where guns are totally banned have greater rates of both than America does. For what it's worth most homicides and suicides in America are related to drugs either users or sellers of drugs. If you exclude drug (and alcohol) related homicides the homicide rate in America is lower than Canada, Japan or countries in Western Europe. So if you want to understand the problem it is NOT guns it is drugs.

No, that was meant to be concise and scientific. American gun-owning and non-owning families are much more similar than families in other countries. Therefore the comparison between American gun-owning and non-owning families is much more applicable.

If we want to go deeper than that, I agree that differences is local culture matter. This study tries to control for a LOT of those factors:

The conclusion:

"We observed a robust correlation between higher levels of gun ownership and higher firearm homicide rates. Although we could not determine causation, we found that states with higher rates of gun ownership had disproportionately large numbers of deaths from firearm-related homicides."

I have to ask you SweetPea, did you not know? Or did you not want to know?

You are either intentionally missing the point or your biases are so strong you cannot see the point.

There is a strong anti-gun lobby and they will stop at nothing to spin the story. Statistics to them is a golden opportunity to spin. The simple fact remains that when you examine ONLY the gun owners in America who are of European descent our gun homicide rate is extremely low, lower even than Canada. Additionally this group is the largest majority of Americans who own guns, perhaps 220 million people with 250 million guns. IF as you claim, it was simply the presence of guns then the statistics would not vary so much. But these simple truths mean nothing to those who will use anything to thwart the constitution. There is a reason why the left wants to ban guns and confiscate guns. The same reason that Lenin and Hitler banned guns. It is difficult to impossible to dominate a free citizenry that is armed.

Last point: If the leftist gun grabbers were honest they would cite the statistics that prove gun violence in America is caused by drugs and thugs. A culture that is allowed in certain left leaning cities because it suits their agenda.

This can only work if you know how many guns there are in a country. You don't. In all probability, the more restrictive the gun laws are in a country, the less accurately anyone knows how many guns there are.

Poor argument. Almost everything statistics-wise discussed on this blog has the issue you describe.

Extremely poor rebuttal.

Some statistics are better than others. Homicides generally have pretty good statistics, as do things that are legal and open. While it's bad to simply throw up one's hands and say that "statistics are always inaccurate, so don't depend on them," (the strawman(?) you attack) it's even worse to simply throw up one's hands and say "statistics are always inaccurate, so just trust whatever terrible statistics we have."

Rather than that absurd dilemma, it's surely worth determining which statistics are more or less accurate.

Dearieme has actually made a very good argument.

You have to violate a country's gun law first, and then you are able to do a killing with it. Violating the gun law in the first place is the purposeful act, undertaken with legal consequences.

Probability of A given B.

China has a surprising number of illegally possessed firearms. Sometimes they do an amnesty where people can bring in guns, and you'd be surprised at the weaponry that turns up. People also make homemade guns and cannons. Mass murder is sometimes done with things like poison (a teacher gave all her students laced Kool-aid) or with dynamite (lots of construction and demolition going on).

What are the thoughts on requiring liability insurance (though I feel I can probably guess)? It seems to me that a well thought out system (with gun dealers very much included, and policies only covering excess liability for criminal acts) would put a fair price on the likelihood that a given gun will cause harm.

Homeowners have liability insurance that covers negligent homicides by firearm. Many, I don't know if it's most, apartment rentals require renters to carry liability coverage as well. (Among other things, it's how you tax out the riff-raff.) No insurer will cover intentional acts.

Gun dealers: should you be liable because someone used a legal product in an illegal manner? Hammers? Chef's knives? This might actually be a good idea: gun dealers might be more reluctant to sell to amped-up, young Afghan males and rap music producers. Oh crap, I just violated civil rights law ...

Why wouldn't insurers cover an intentional act? The benefits would only be able to go to the victim and the non-monetary consequences are high, so the moral hazard is minimal. And, as people have pointed out the vast majority of guns are not used to shoot people, so the payout would be rare.

Dealers would only be liable until a gun went on someone else's insurance. So if you check someone's insurance and give them a gun, then no problem. If you "lose" a few guns that end up in a robbery gone wrong, you will still be liable.

I won't touch the racist comments, but presumably insurance rates would vary dramatically (and possibly prohibitively) for long arms vs. pistols, people who've completed a safety class, and people with criminal records.

Moral hazard.

This is at best useless, at worst deliberately misleading. I did a little blog post in reply.

There is a pretty good reason why "rough drafts" aren't deposited into arXiv. "well, gee, I know my data is bad, but let's pretend it isn't and spend time discussing it" doesn't impress me much.
Is there an ignorance index? Is there an "unhappiness" index? or a "hate" index? There is obviously the street light problem going on here. What does GDP have to do with people in the black, grey, underground economy or those who have been marginalized?

The long term trend towards increasingly lower levels of violence in the US was sharply reversed in 2015, (the graphs mentioned here are from 2012 data) when the Obama Administration launched a campaign of wholesale harassment against local law enforcement.

Right. So, this is a Look, Squirrel moment. It's especially an attempt by all who are good and holly to distract from the Muslim aspect of this Muslim terrorism.

Two incidents on the same day. Two AR-15s. (Actually four AR-15s for the two men?) One was stopped and one wasn't.

In conclusion, the results are not decisive so we shouldn't do anything.

It's stunning that the NRA and the politicians they own are totally unwilling to even try a policy change. If you think that mass shootings are an acceptable price to pay for freedom, at least have the courage to go on TV and say so. Justify your inaction instead of kicking the can down the road indefinitely.

They therefore claim that consistent above-market returns are “luck” and will regress to the mean over a sufficiently lengthy time horizon. zincirli caraskal

Comments for this post are closed