Friday, July 27, 2012

Guest Post - On shootings and firearm statistics

The following was written by the Evil Dr. Pain and originally posted on facebook Wednesday, July 25th.

On shootings and firearm statistics

Following the recent atrocity in Aurora, the latest in a sadly growing list of gun-related attacks on groups of the public (for which the US is growing increasingly infamous), there has been the expected surge in arguments between the pro-gun and pro-gun-control lobbies. Regrettably, a large portion of this argument has either been from people voicing opinions in the same tired catch-phrases or variants thereon (the various versions of "guns don't kill people, people kill people", for example), usually without anything substantial to support their argument, or a neverending sequence of memes telling us the same things, disguised in cheap derivative humour. Sure, this is Facebook, and is a great vehicle for generic siliness (A little nonsense, now and then, is relished by the wisest men...), but when it comes to matters as grave as this, I hope we can put this aside and have a serious discussion about the issues.

The central question in discussing these issues must surely be: Just how dangerous are guns, on a societal level? We understand intuitively in individual cases - common sense tells us that a gun in responsible hands is no significant social danger, and a gun in the hands of the wrong person can be a terrible danger (these are the foundations of the arguments from the pro-gun and pro-gun-control camps, respectively) - but an answer requies not simply the consideration of the extreme cases, but the vast majority that lie between. The obvious thing to do is to look at some data.

Here, I'm focussing on gun crime only (ie excluding firearm accidents, suicides etc). Gun crime is quite a complicated issue, as it is tied to numerous variables, many of which hard to quantify (or even identify qualitatively). Consequently, the best this represents is a zeroth-order pass at considering gun crime statistics from a nation-by-nation standpoint.

The data

All data I've plotted here are sourced from the United Nations Office on Drugs and Crime and theSmall Arms Survey, and can be found at this article in the Guardian. The data are far from complete (many developing countries are missing, along with Russia and China), but cover a significant fraction of the world, including most of the developed world, and thus shed light on some general trends.

For all plots I generated, data points in blue are the entire "global" data set. Because many of the factors which influence gun crime (and crime in general) are social factors which are hard to quantify (as yet, I've not included any quantitative metric for them), I've labelled a selection of countries in each plot, in order for the viewer to illuminte the data with their own social and political knowledge. For each plot, I've included an inset of an unlabeled plot, so that the distribution of the data may be seen without the distraction of overlaid text. Furthermore, in an attempt to clarify the effects of the varibles plotted, I've taken a subset of the data (plotted in red), labelled "Western World". This selection of countries roughtly represents the "First World" (Western Europe, USA, Canada, Australasia, Japan and South Korea) - countries which are relatively similar in social and political terms (on a global scale). Plots of only these data are discussed following the discussion on global data.

Global data

Fig 1 - Firearm homicide rate vs number of firearms per capita

Fig 1 shows the firearm homicide rate (per 100,000 people, per year) against the prolificity of firearms (guns per 100 people). It is immediately clear is that there is no correlation between these two raw varibles. This is perhaps unsurprising, as any underlying correlation between the availablity of firearms and the homicide rate is bound to be lost in the numerous social factors impacting on crime rates. The extent to which the US is a global outlier is clear here; it has the highest number of guns per capita of any country, by several standard deviations from the norm. On a global scale, its firearm homicide rate is low (especially considering the prolificity of firearms), although the countries dominating the firearm homicide rate are not First-World countries, and most have significant social/political problems. The First-World countries in red have a vastly lower firearm homicide rate than the global average.

Fig 2 - Total homicide rate vs number of firearms per capita

Fig 2 shows the total homicide rate (per 100,000 people, per year) against the prolificity of firearms (guns per 100 people). There's is no real correlation between these two raw varibles, for the same reason as before. Again, the First World countries in red have a vastly lower homicide rate than the global average, and the US lies at between the First World and the remainder of the globe.

Fig 3 - Percentage of firearm homicides vs number of firearms per capita

Fig 3 shows the percentage of homicide that involved firearms against the prolificity of firearms. Surprisingly here, there is not obvious correlation between these variables, and in this case the first world countries and are more evenly ditributed with the rest of the globe. Overall, the fraction of firearms involved in homicidal behaviour seems independent of the availability of the firearms.

Fig 4 - Firearm homicide rate vs total homicide rate

Fig 4 indicates the strongest correlation in the data; the firearm homicide rate as a function of the total homicide rate. The correlation is positive, and is well desribed with a quadratic term (plot to follow), indicating that the more violent a country is, the more guns are involved in the murder rate.

Fig 5 - Percentage of firearm homicides vs total homicide rate

This is demonstrated in Fig 5, which plots the percentage of firearm homicides against the total homicide rate. Here, a group of very violent countries can be seen to have a separate grouping (distinct from much of the world, including the first-world countries) with distinct positive correlative trend. Curiously, in this parameter, space, the USA lies at the cusp between the most violent, troubled countries, and the rest of the world.

The 'First' (or 'Western') World data

In examining the first world countries (in the plots, labelled "Western World"), we are able to examine data for which the distribution of social and political factors is much smaller (all the countries herein are relatively wealthy democracies), more clearly illuminating any correlations between plotted variables. In all these plots, it can be seen how much of an outlier the USA is compared to the rest of the First World.

Fig 6 - Firearm homicide rate vs firearms per capita, for the First World

Fig 6 shows the firearm homicide rate against the number of guns per capita, for First-World countries. Unlike Fig 1, where any correlation was lost in the vast variation in social and political factors, here was can see a direct positive correlation. Though there are still considerable variations in the data, it can clearly be seen that in the democracies of the First World, having more guns directly correlates with more people using them to shoot people. 

Fig 7 - Firearm homicide rate vs total homicide rate, for the First World

Fig 7 shows the firearm homicide rate against the total homicide rate for the First World. Though here there is more scatter than for the global data (Fig 4, dominated by the more politically/socially troubled  countries), there is some evidence for a positive correlation between the firearm homicide rate and the total homidide rate(ie the more people are killing each other, the more they are shooting each other).

Fig 8 - Total homicide rate vs firearms per capita, for the First World

Fig 8 shows the total rate of homicides against the number of firearms per capita, for the First World. In these data (excluding the USA datum), it is difficult to say if there is any correlative effect at all between the total rate of homicides and the number of firearms per capita (though with the USA point, it is not difficult to imagine a quadratic trend).

Thoughts on the data

On examining the data, it is probable that, though the prolificity of firearms appears to have an impact on the number of firearm-related homicides in the First World, it probably has little impact on the total rate of homicides. That is, having fewer guns around does not make you safer from being killed, but it does make you safer from being shot.

If, however, the number of guns in the US is not the major factor driving the homicide rate, it clearly is true that the number (and nature) of guns in legal circulation directly enables the large-scale firearm massacres, of which Aurora was the latest. Of course, whilst these massacres are very high profile, they contribute very little to the total statistics. The Aurora death toll was about 0.08% of the homicide toll for the US in a typical year. Twice as many people are shot to death daily in the USA (and almost four times as many homicides in total are committed daily) as were murdered in Aurora. Despite the fact that these events contribute so little to the total homicide rate in the USA, it is necessary that they be so strongly focused upon. There is something disturbing about the complete lack of discimination involved in such acts. There is something disturbing about the sheer magnitude of death an injury. They are a canary for social health. 

Returning to the data above, it is clear that the USA is a considerable outlier with repect to the remainder of the First World in almost all parameters. The percentage of firearm-related homicides is the exception, in which the USA actually does statistically well compared to other nations, considering how many firearms are around in the USA.  However, the fact that the USA does considerably worse in terms of homicidal statistics compared to the rest of the First Worldsuggests a significant social factor, which drives its homicide rate (gun-related and otherwise).

The irony, then, of this study is that the focus in the wake of such attrocities should not be solely on firearm control (which of course could help limit the ease with which these major atrocities can be conducted; most sociopathic lunatics are hardly well set to make easy in-roads with black-market dealers), but on identifying and correcting the underlying social problems which stand in the way of the USA reaching its full potential as a civilised (meaning full of civilians, who are able to live without feeling the need to be armed) member of the First World.


  1. Thanks, Kelly, for a thorough and unsettling analysis. One from Harvard came to different conclusions, but still the effect of availability seems to be marginal.

    But for me, the freudian attachment many men have with their firearms is enough to give serious pause. The whole culture is so disturbing that I'd really rather it were nowhere near me and out of civilian precincts entirely.

  2. This presentation of the data was very informative. Thanks.

    I did not grow up around guns or gun culture and have little interest in them. However, I have zero interest in having any discussion about new gun control measures until we, as a nation, have two other discussions first: (1) privacy and (2) enforcement of that which is malum prohibitum.

    (1) Where do we start but with registration and data collection. Will the data be private or public? I’m not sure I’m comfortable with a service like Please Rob Me sharing who is armed and probably unarmed.

    (2) I am absolutely appalled by the increasing use of tactics, which I believe are malum in se, to enforce that which is malum prohibitium. For example:
    Malum prohibitium: suspected drug violation
    Malum in se: Kathryn Johnston killed during a botched no-knock raid
    Unauthorized possession of a firearm, like drug possession, is malum prohibitium. Given the recent history of drug enforcement, I’m not sure I’m comfortable with new restrictions on certain types of firearms. Even if I don’t have a weapon handy like Miss Johnston, my toddlers like playing near my front door and these guys are keen with their chainsaws.

  3. To both Burk and Anonymous, I totally agree that any gun control measures which stem from this tragedy (or this data) are treatments of the symptoms and not the root cause. We have to make a transition from being a gun culture to being a no-gun culture, and this will take time (and growing pains).
    It's the fact that we feel the need to have firearms - and fiercely justify that desire - that is telling of the real issue.

  4. Interesting stats. Perhaps for a given society, there is an optimum number of guns that should be in circulation to balance the needs of some to feel safe from their government, and the needs of others to be safe from fellow citizens. And mental health care could be more freely available. We often ignore the mental breakdowns in our midst. There were warning signs for this guy in Aurora.

    1. That's an interesting point - given a society and the people within it, there is probably some balance in the level to which the populace needs to be armed. We certainly seem to have a problem, in the US at least, with swinging from one extreme to the other, never considering the potential for middle ground.


Think carefully before you post. I reserve the right to moderate any comments posted to my blog.