Homicides vs. Gun Ownership

I took the data and the script from @mdreid‘s blog post about Gun Deaths vs. Gun Ownership, and plotted Total Murder Rates (murders.csv) vs. Gun Ownership out of curiosity. The results are interesting, draw your own conclusions.

For all countries:



For OECD countries:



Zoomed in without Mexico and the US:



One caveat: the comparison between “countries” is somewhat arbitrary. The population range for the countries in this set spans two orders of magnitude: millions to hundreds of millions. Their ethnicities, geographies and densities are different as well. We must be careful in drawing conclusions when comparing small, relatively homogeneous countries such as Finland against a population with the size and heterogeneity of the US.

31 Replies to “Homicides vs. Gun Ownership”

    1. No, it wouldn’t be politically correct, since it would come out negative, just eyeballing the bulk of the datapoints. Mexico and USA will cancel each other out in the regression since they carve out almost a 90 degree angle with the origin.

      1. Assuming you’re using OECD countries and not the whole data set, then the relationship would clearly be positive. Mexico and the US would not cancel each other out; Mexico has very low leverage and it does not greatly affect the slope. Although Mexico is clearly an outlier, the US is also an outlier among OECD countries, but it’s difficult to see with the scaled used. Being a positive outlier means that there are significantly more gun deaths per capita than would be expected from the number of guns owned per capita. Simply seeing that if more people own guns, more people die from guns is not at all a surprise or interesting.
        The question is, “why do more people die from guns than expected”? In Mexico it’s probably because of drug cartels. In the US there are probably lots of reasons including class inequality and the fact that any freaking person in the country can get a gun almost instantly.

  1. These plots are specious.
    A valid plot is Guns per 100 People versus Murders-by-Guns per 100K People.
    What has murder by all possible means other than by guns have to do with gun ownership?
    Lies, damned lies and statistics.

    1. That plot was already done by Mark Reid. I wanted to see how total murders plotted against gun ownership. It’s a reasonable thesis that more guns could be correlated with more gun murders. It turns out not to be the case if you look at all countries, and I won’t go into plausible explanations.

      The murder rate for a society is more important than just the gun murder rate. If a country has a low gun murder rate but a high murder rate by other means, removing guns didn’t address the real problem: people killing other people by whatever means available. Of course this is not an assertion about any country in particular. It’s just a hypothesis.

      Plots are not specious, they are simply data. Trying to draw conclusions from those plots (which you’re doing) could be specious. Why don’t you think for yourself instead of repeating cliches mindlessly?

      1. At this point, it seems that an interesting next step would be to add additional demographic variables (pop. density, GDP per capita, etc), control for them and then examine statistically significant differences between the U.S. and other OECD nations. A scatter plot can’t tell us much. A linear regression is inappropriate. But maybe a t test would work here.

        1. I’d also like to see the following variables plotted against violence in general, and gun violence in particular: median age of population; % of population incarcerated; % of males, 15 – 25, without engaged, law-abiding father figures in their lives free in the general population; and wealth inequality across subgroups within nations.

          Disclaimer: Left of Center Perspective, NRA sucks.

    2. It is intuitive that countries with more guns are going to have more murders by gun than other countries. That isn’t particularly interesting. Countries with more cars will probably have more vehicular homicides…

    1. Hmmm…. Don’t want to spoil the detached, academic vibe here, but John Lott might not be the “go to guy” for the effect of guns on crime rates.

      His thesis of “more guns, less crime” was based on a survey he said he carried out. No one is quite sure what the survey contained because he lost it. Apparently he had a disk fail on him that contained the survey. Not only that, all his data from the survey was on the disk too. Bad luck eh?

      But that wasn’t the end of his bad luck. Normally you would get ethics approval from your institution for a research project like this which would contain the survey. Sadly, he forgot to get ethics approval. What rotten luck!

      But it didn’t end there. Some people who really wanted to know more about the survey publicly asked if those who’d been surveyed could come forward. Perhaps they could remember a bit about the survey and could recount their experiences. Gosh and darn-it! Wouldn’t you know that they didn’t want to publicise their unpleasant experience! None could be found!

      Some very cynical people asked if he had claimed expenses for printing his survey and mailing it out. Of course not! John wouldn’t want to put his institution to the expense so paid for it out of his own pocket. Sadly, he’s lost the receipts.

      Poor old John. Even more people began wondering if John had actually done a survey! After all, he had had an unfortunate incident that might have thrown his academic credibility in a bad light (google “Mary Rosh”). But John had the answer. In the best traditions of academic discourse, he sued them.

      Yes. John Lott. A reliable source on gun control.

    2. You can also probably get Dr. Lott’s “More Guns, Less Crime” from your local public library. No point in financially supporting a viewpoint you might, at least, initially disagree with, too much, before reading. Or, in supporting the concentration of wealth and power inherent in the corporate behemoth, Amazon.com.

      Disclaimer: Left of Center Perspective, the NRA sucks (although, you probably could have gathered that from my post 🙂 )

  2. I wonder how access to healthcare, equality, poverty, etc – the “general welfare” part of the constitution would correlate to murders per capita. Might be more interesting than a narrow focus on an individualist perspective of the 2nd amendment.

  3. I created the same graph independently. I should have just googled it. Well done! Some comments:

    The striking thing about this graph is that it appears as though an appropriate hyperbola drawn would bound all of the points. The equation for this line would be approximately y=500/x. I’m going to coin this the “Feeling Lucky” Limit. You will notice that there are no points that exist to the upper right of this hyperbola, and that there are not many either on that line or approaching it. You will notice that most nations lie very near the axes.

    Within a given population, most potential murderers will only attempt murder if they feel that they can get away with it (i.e. not be shot in the process). The higher percentage of people who are armed, the less the chances are for someone to be able to commit murder without being killed in the process. The two extremes are a completely disarmed country, where any thug who thinks he can get away with murder, will. The other extreme is an armed country where the risk of murder without dying in the process or triggering some sort of retaliation is unlikely.

    Phrases or sayings that come to mind: “An armed society is a polite society”; “Mutually Assured Destruction”; “The great equalizer”.

    Most countries cluster on either axis. The presence of a few guns is enough to dissuade most people from murdering, it appears. However, countries with a more risk averse population (or alternatively, groups of gangs that effectively lower the risk of a gang member getting killed, but do nothing to prevent a non-gang affiliated person being killed) will lie closer to that line.

    Evidently there are factors that stop people murdering in some disarmed countries. For example, an effective police force or an innately non-homicidal population. But it does seem as though the presence of guns acts as a limit to homicide, an armed society being one in which the possibility of being killed either in the act or through revenge is a distinct possibility, and hence a “polite” society.

    It’s also not surprising that the knee jerk response is to think that banning guns is going to reduce murder. There appears to be little evidence to support that position.
    Mutually Assured Destruction was/is counterintuitive as well. The fortunate thing with small arms is that if the mutually assured destruction thing fails to deter, the worst possible failure is on the order of the Norway Massacre (maybe 60 killed) and individual murders here and there, not the whole world being rendered unlivable.

    1. Isn’t easier to say there is no obvious connection between the availability of guns and the desire in a population to commit murder? I can’t see in these charts any indication that more guns equals less murders. A correlation like that would put the majority on the bottom right. There is probably a reason why the OECD countries extend slight more along the x axis (guns per 100p), while the poorer countries spread up the y axis (murders per 100kp). That reason is not revealed in these charts. The only conclusion here is that gun ownership simply doesn’t seem to factor into murder rates. I’d suggest looking at income inequality, education, socioeconomic mobility, etc.. I’d bet they are more likely to show a positive correlation than gun availability / murder rates.

      One more quick note – even without a demonstrated causal relationship between the availability of guns, and the impetus to murder, in the Newtown case, the access to guns to the perpetrator of that horrible crime, did absolutely enhance the damage that depraved individual was able to inflict on his victims. That kind of efficient violence is what guns are for. It’s what makes guns so powerful, so desirable, and so much more of a political issue than other means of murder.

      1. Random thought on the whole “murder” angle…is it possible that the U.S. with it’s fabulous emergency room system and quick response trauma units, keeps the actual “murder” rate relatively low, despite the enormous number of guns? In a bit of irony, it may be these gunshot wounds that forced U.S. trauma centers to evolve to meet the need to reduce these types of deaths.

        1. Never thought of that…very interesting. And, yes, while many Americans don’t have health care coverage, most have access to superlative emergency rooms, even in poor urban areas with high rates of gun violence.

          But, your point is troubling because of all violent crime, murder is the most reliable statistic. If we can’t trust that murder rates, collected by police and civil authorities, are an excellent indication of the prevalence of violent crime, what are we to do?

          Perhaps turn to hospital emergency rooms for data? Do many hospitals collect detailed information on causes of injury (violent: gunshot wound, knife, physical assault, or otherwise: car accident, fall, etc.)

      2. How about this thought – there is a high rate of gun murders in countries with lax gun laws and high rates of illegal gun ownership.

    2. “Within a given population, most potential murderers will only attempt murder if they feel that they can get away with it (i.e. not be shot in the process)”

      ” if they feel that they can get away with it.” Most homicides in the United States are gun homicides, I believe, and are committed by young (below the age of 30) urban African American males (if my assertions are wrong, please feel free to stop me), males with few engaged law-abiding father figures in their lives, males who live in neighborhoods awash in drugs, violent crime, and guns.

      From what I understand, “getting shot” or even dying may not be the worst fate perceived by these young, unaccountable-to-their-own-African American-community males: as in all honor societies, losing face or respect among your close male associates is often a fate worse than death.

      After all, knowing that young urban African American males may be armed doesn’t stop other young urban African American males from attacking them.

      For them, it’s not necessarily about getting away with it (whether that means not being shot or avoiding civil punishment (prison time, death penalty, fines, community service) in the broader community, it’s about saving face with your close male associates, your “brothers.”

  4. I’d like to correct the above:
    “The other extreme is an armed country where the risk of murder without dying in the process or triggering some sort of retaliation is unlikely.” should be:

    “The other extreme is an armed country where the risk of murder without dying in the process or triggering some sort of retaliation is prohibitive.”

  5. For me a more useful statistic might be deaths per gun owner. Many owners have many guns which may make the U.S. appear more gun happy, as the rest of the world does.

    1. Yes, you’re absolutely right. The United States may be awash in guns, but it may be primarily awash in some individuals who are building up large arsenals of weapons.

      But, as our local, state, and federal governments are not allowed to, for the most part, track individual gun sales with individual owners (many Americans are ever fearful that their government will confiscate their weapons based on gun ownership records, and reduce them to abject slavery), that’s data we’re not likely to collect.

      We might get some interesting data through conceal carry permits, though: most states require some kind of permit to conceal carry, and those permits sometimes require background checks, basic marksmanship skills, gun safety training.

      What are the rates of accidental death in the home, homicide, suicide, assault, gun-related or otherwise, among the most responsible of gun owners, those with conceal carry permits that meet accountable standards?

      Disclaimer: Left of Center Perspective, the NRA sucks

  6. As Diego Basch mentioned above, we may also want to look at violence in general, and gun violence in particular in accordance with population density and homogeneity across polities.

    Of special interest might be if a population subgroup that experienced severe oppression (slavery, forced confiscation of lands and resources) by another subgroup in society in recent history (within the past 500 years or so) was represented in the population, and how that correlated with societal violence.

    Disclaimer: Left of Center Perspective, the NRA sucks

  7. I’m quite curious about the relative number of gun related homicides in dating back to before the implementation of gun control laws. I haven’t found any data before 1975, but I just started looking when I stumbled onto your graphs.

Leave a Reply

Your email address will not be published. Required fields are marked *