As originally posted in the Aurora shooting thread, I found two interesting lists: guns per capita for a bunch of countries and firearm-related homicides per capita by country. It occurred to me that seeking a measure of the correlation between these two sets of numbers might reveal something interesting about the need for greater gun control. If the number of guns present in a society correlated strongly with the number of gun homicides, that would be a good indication that gun control, if successful in reducing the total number of firearms per capita, might result in a reduction in gun homicides.

Correlation is not causation, but direct causation isn't really the issue here. No one is going to argue that a gun, in and of itself, causes a homicide. The causal presumption is that when the circumstances for a violent attack exist, the ready availability of a gun, which is obviously related to the number of guns in existence, will increase the likelihood of said attack being made with a gun, and thus being, on average, more lethal than non-gun attacks.

No one would argue that the relative availability of guns would be the only factor influencing the number of gun-related homicides in a nation, nor even, I would think, that availability was an overwhelmingly central and critical factor. But a positive correlation would, I think, constitute a prima facie case for seeing availability as a factor. It would be important to know if it's a factor because among all possible and likely factors, some will be easier than others to influence if and as we try to bring down gun homicide rates. It might be harder to change various cultural factors than to change gun control laws.

On the other hand, if we find that among the nations of the world there is little or no correlation between the number of guns per capita in the society and the rate per capita of shooting deaths, we might very well conclude that trying to limit the number of total guns via legislation/enforcement is a blind alley - the wrong solution.

Do I have you all on tenterhooks, waiting breathlessly to see that correlation coefficient?
First the raw data, in a new post, so one or more of you can check my math.

Okay; tables really suck but this isn't too bad. The first number is the gun-related homicide rate per 100,000 population per year. The second number is the gun ownership rate - guns per 100 residents in 2007.

Argentina 2.11 - - - 10.20
Australia 0.44 - - - 15.00
Austria 0.42 - - - 30.40
Azerbaijan 1.47 - - - 3.50
Barbados 3.00 - - - 7.80
Belarus 0.38 - - - 7.30
Belgium 0.60 - - - 17.20
Brazil 10.58 - - - 8.00
Bulgaria 0.77 - - - 6.20
Canada 0.76 - - - 30.80
Chile 0.06 - - - 10.70
Colombia 51.77 - - - 5.90
Costa Rica 3.32 - - - 9.90
Czech Republic 1.77 - - - 16.30
Denmark 0.23 - - - 12.00
El Salvador 50.36 - - - 5.80
England/ Wales 0.07 - - - 6.20
Estonia 8.07 - - - 9.20
Finland 0.86 - - - 32.00
France 0.44 - - - 31.20
Germany 0.22 - - - 30.30
Greece 0.59 - - - 22.50
Guatemala 38.52 - - - 13.10
Honduras 46.70 - - - 6.20
Hungary 0.23 - - - 5.50
India 0.93 - - - 4.20
Ireland 0.03 - - - 8.60
Israel 0.72 - - - 7.30
Italy 1.66 - - - 11.90
Jamaica 47.44 - - - 8.10
Japan 0.02 - - - 0.60
Kuwait 0.36 - - - 24.80
Latvia 1.20 - - - 19.00
Lithuania 0.46 - - - 0.70
Mauritius 0.00 - - - 14.70
Mexico 9.88 - - - 15.00
Moldova 0.47 - - - 7.10
Netherlands 0.36 - - - 3.90
New Zealand 0.17 - - - 22.60
Nicaragua 7.14 - - - 7.70
Northern Ireland 5.24 - - - 21.90
Norway 0.30 - - - 31.30
Panama 12.92 - - - 21.70
Paraguay 7.35 - - - 17.00
Peru 1.87 - - - 18.80
Philippines 9.46 - - - 4.70
Poland 0.29 - - - 1.30
Portugal 1.28 - - - 8.50
Qatar 0.18 - - - 19.20
Republic of Macedonia 1.28 - - - 24.10
Scotland 0.19 - - - 5.50
Singapore 0.07 - - - 0.50
Slovakia 2.17 - - - 8.30
Slovenia 0.35 - - - 13.50
South Africa 74.57 - - - 12.70
South Korea 0.04 - - - 1.10
Spain 0.21 - - - 10.40
Swaziland 37.16 - - - 6.40
Sweden 0.18 - - - 31.60
Switzerland 0.58 - - - 45.70
Taiwan 0.13 - - - 4.60
Ukraine 0.35 - - - 6.60
Uruguay 3.24 - - - 31.80
Zimbabwe 4.75 - - - 4.60
United States 4.14 - - - 88.80

Now for the result. Drumroll please....

I was very surprised to find that the correlation wasn't just weak, it was actually negative. Minus 0.17. Weakly negative, for sure, but negative. [Note: I did not include US data in this computation. With USA included the Pearson Correlation Coefficient is -0.14... not much different.]

What this means, as best I can interpret it, is that our problem in the USA isn't the total number of guns we have. That's great news in a way, because if total number of guns was the problem, we'd have a heck of a time trying to reduce ourselves from the lonesome pinnacle of that list. Of course, when it comes to the distribution of guns within a population, total number of weapons isn't the only issue. If one very rich person owned all 200+ million guns, we'd have extremely few gun homicides (I'd hope). "Gun Control" need not only refer to limiting total sales. It can refer to the types of weapons that can be sold, and it can refer to the limitation of to whom they may be sold.

According to Wikipedia HERE, most gun homicides by far are committed with handguns, not assault rifles or anything else very fancy. And very few of these guns are bought retail or at gun shows (where gun laws get enforced); more often by far they're gotten second-hand outside the gun control enforcement regime. The vast majority of gun murderers are young, and are disproportionately from inner cities. Most (I think) are involved in gangs. "Research has found some policies such as gun "buy-back" programs are ineffective, while Boston's Operation Ceasefire, a gang violence abatement and intervention strategy, has been effective."

Maybe, just maybe, gun control isn't the issue. Maybe if we could reduce gang activity our gun-related homicide rate would fall into line with peer nations.

The problem is, gang control isn't as sexy, or as easy a topic. Few of us are going to think we're experts. And it's not an issue where we can have strictly partisan debates. We'd have to actually get together, use our heads, and rely on evidence and the methods of social science.

It's hopeless.

Thanks for the table Purple--it is very interesting. I took a look at France and Germany in particular--they have pretty high gun ownership totals but low gun deaths. Let me suggest four commonalities in these countries: (1) They are wealthy and even if wealth is stratified there are strong safety nets, (2) they are for the most part ethnically homogenous, and (3) gun ownership is heavily regulated with gun ownership limited to state approved activities like hunting and sporting, (4) overall homicides have declined in those countries to something like 1 per 100,000 people.

Purple has shown a negative correlation between the total amount of guns and gun deaths. But I am not sure that we can equate countries that have strict gun control laws like Germany and France with those with more of a free-for-all system like the United States. While gun ownership in Germany and France is high it is also heavily regulated and presumably it is less likely that a particular gun will get into the hands of an unstable person or a criminal.

The example gun owners use is Switzerland, but Switzherland ties gun ownership to defense of the country. Apart from that, it has all of the elements that have led to low gun deaths in Germany and France.

The U.S. a non-homogenous population, wealth is stratified, weak gun control laws (and btw the fact that state and local gun control laws have failed means nothing--people can go to nearby cities and states and get guns so these laws have no effect)., and higher levels of violence than Western Europe. I suspect if we worked on reducing wealth dispartity, and heavily regulated guns we would see gun-related homicides fall. But a free-for-all economic system with some groups being in and some out combined with no gun control, well, what do you expect

Interesting, although there are several factors that could be taken into account.

Firstly, you (Purple) have already observed that there' a financial element - wealthy countries tend to have higher gun ownership, and they also tend to have lower general rates of crime.

Secondly, total gun ownership masks the kind of guns and restrictions on their use/storage etc. Also how many weapons are legally held compared to being illegally held.

Thirdly, there are different 'kinds' of gun homicide. Some are related to civil unrest. Some are due to gang violence. Some are more straightforward crime. Others are not criminal (accidents, self-defence) and then there's suicide.

I think if the comparison above was taken using some filters or controls for some of the above, it would perhaps shine a different light.

Would you be able to run two separate regressions: one for developed countries, and one for developing countries?

danivon wrote:Others are not criminal (accidents, self-defence) and then there's suicide.

Clearly you didn't look at the source web pages. Suicide and accidents are not included. Self-defense might be, but I really don't think it would amount to much.

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

GDP per capita: Had to eliminate Great Britain from the list because I couldn't match GDP to the gun data for the separate pieces of Scotland, N. Ireland. Eliminated Zimbabwe because of ridiculously low GDP outlier would have messed up my analysis. Deleted Qatar because GDP per cap. is absurdly high.

Took lowest GDP (the socialist paradise of Nicaragua) and normalized to 1 - carried through all. Then took square root. This yielded "normalized" GDP per capita numbers from 1 to 4.3. Divided gun ownership by this to try to remove the effect of affluence on ownership (without going overboard).

The new Pearson: positive 0.22. This is still way too low to be statistically significant. The number of guns and the number of gun homicides are not significantly correlated.

Note: GDP data taken from CIA table HERE.

New data table, sorted by gun homicide rate in first column; second col. is guns per capita and third is GDP per capita:

Mauritius - - - 0.00 - - - 14.70 - - - 15,000
Japan - - - 0.02 - - - 0.60 - - - 34,300
Ireland - - - 0.03 - - - 8.60 - - - 39,500
South Korea - - - 0.04 - - - 1.10 - - - 31,700
Chile - - - 0.06 - - - 10.70 - - - 16,100
Singapore - - - 0.07 - - - 0.50 - - - 59,900
Taiwan - - - 0.13 - - - 4.60 - - - 37,900
New Zealand - - - 0.17 - - - 22.60 - - - 27,900
Sweden - - - 0.18 - - - 31.60 - - - 40,600
Spain - - - 0.21 - - - 10.40 - - - 30,600
Germany - - - 0.22 - - - 30.30 - - - 37,900
Hungary - - - 0.23 - - - 5.50 - - - 19,600
Denmark - - - 0.23 - - - 12.00 - - - 40,200
Poland - - - 0.29 - - - 1.30 - - - 20,100
Norway - - - 0.30 - - - 31.30 - - - 53,300
Ukraine - - - 0.35 - - - 6.60 - - - 7,200
Slovenia - - - 0.35 - - - 13.50 - - - 29,100
Kuwait - - - 0.36 - - - 24.80 - - - 40,700
Netherlands - - - 0.36 - - - 3.90 - - - 42,300
Belarus - - - 0.38 - - - 7.30 - - - 14,900
Austria - - - 0.42 - - - 30.40 - - - 41,700
France - - - 0.44 - - - 31.20 - - - 35,000
Australia - - - 0.44 - - - 15.00 - - - 40,800
Lithuania - - - 0.46 - - - 0.70 - - - 18,700
Moldova - - - 0.47 - - - 7.10 - - - 3,400
Switzerland - - - 0.58 - - - 45.70 - - - 43,400
Greece - - - 0.59 - - - 22.50 - - - 27,600
Belgium - - - 0.60 - - - 17.20 - - - 37,600
Israel - - - 0.72 - - - 7.30 - - - 31,000
Canada - - - 0.76 - - - 30.80 - - - 40,300
Bulgaria - - - 0.77 - - - 6.20 - - - 13,500
Finland - - - 0.86 - - - 32.00 - - - 38,300
India - - - 0.93 - - - 4.20 - - - 3,700
Latvia - - - 1.20 - - - 19.00 - - - 15,400
Republic of Macedonia - - - 1.28 - - - 24.10 - - - 10,400
Portugal - - - 1.28 - - - 8.50 - - - 23,200
Azerbaijan - - - 1.47 - - - 3.50 - - - 10,200
Italy - - - 1.66 - - - 11.90 - - - 30,100
Czech Republic - - - 1.77 - - - 16.30 - - - 25,900
Peru - - - 1.87 - - - 18.80 - - - 10,000
Argentina - - - 2.11 - - - 10.20 - - - 17,400
Slovakia - - - 2.17 - - - 8.30 - - - 23,400
Barbados - - - 3.00 - - - 7.80 - - - 23,600
Uruguay - - - 3.24 - - - 31.80 - - - 15,400
Costa Rica - - - 3.32 - - - 9.90 - - - 11,500
United States - - - 4.14 - - - 88.80 - - - 48,100
Nicaragua - - - 7.14 - - - 7.70 - - - 3,200
Paraguay - - - 7.35 - - - 17.00 - - - 5,500
Estonia - - - 8.07 - - - 9.20 - - - 20,200
Philippines - - - 9.46 - - - 4.70 - - - 4,100
Mexico - - - 9.88 - - - 15.00 - - - 15,100
Brazil - - - 10.58 - - - 8.00 - - - 11,600
Panama - - - 12.92 - - - 21.70 - - - 13,600
Swaziland - - - 37.16 - - - 6.40 - - - 5,200
Guatemala - - - 38.52 - - - 13.10 - - - 5,000
Honduras - - - 46.70 - - - 6.20 - - - 4,300
Jamaica - - - 47.44 - - - 8.10 - - - 9,000
El Salvador - - - 50.36 - - - 5.80 - - - 7,600
Colombia - - - 51.77 - - - 5.90 - - - 10,100
South Africa - - - 74.57 - - - 12.70 - - - 11,000

Purple wrote:

danivon wrote:Others are not criminal (accidents, self-defence) and then there's suicide.

Clearly you didn't look at the source web pages. Suicide and accidents are not included. Self-defense might be, but I really don't think it would amount to much.

If it does not include them then it still doesn't differentiate between them. A line has been drawn and all deaths by gunshot are either within it or without it.

Accident rates would be interesting to compare, as a high rate of accidental shootings would indicate a difference in terms of how gun owners store and secure their weapons, how well trained they are etc.

The new Pearson: positive 0.22. This is still way too low to be statistically significant. The number of guns and the number of gun homicides are not significantly correlated.

Note: GDP data taken from CIA table HERE.

That isn't quite what was suggested. The request was to split 'developed' from 'developing' nations.

Personally, I would want to see a comparison of overall crime rates, violent crime rates and gun crime rates, to cover the question of whether that is also a major factor.

Thanks.

Do more gun homicides correlate with more or fewer homicides with other means. In other words, if people do not carry guns, are they more or less likely to kill using other weapons?

(cross posted with Danivon)

I'm off on vacation for a while - do your own math.