Warning: what you are about to read is mostly speculation. Take with a grain of salt. Don’t trust my numbers blindly. Find your own data.

The suicide rate in the United States is twelve per one hundred thousand inhabitants. That means that every year, around 38,000 people kill themselves. About one hundred per day. Of course, not all suicides are equal. For example, some are the result of a carefully planned decision. Others are not so planned. Perhaps they are not spontaneous, but they are more preventable. I’m not going to claim expertise in the subject, most of what I know comes from conversations with my father; he was a psychiatrist who treated suicidal patients for decades.

Continuing the line of my recent posts, I decided to plot suicide rates vs. household gun ownership by US states:

Sources:

• Suicide rates by state.
• Percentage of households with guns by state.

The above chart surprised me. The correlation between the two variables is pretty high (0.6). What’s going on here? Here’s the first hypothesis that comes to my mind: suicidal people with easy access to guns are more likely to actually kill themselves.

I can imagine a scenario in which someone crosses an emotional threshold, grabs a gun from a drawer and kills himself (the majority are men). In a parallel world there is no gun in the drawer, this person would not be able to commit suicide so easily. Perhaps he would go out and drive his car into a wall, but this would give him more time to reconsider. The time elapsed between deciding to shoot oneself and being dead can be very short.

If you believe that these people should not die, then you could use this hypothesis to make the case for gun control. The majority of US gun deaths are in fact suicides, and firearm suicide is by far the most common method, so this is not an insignificant problem.

Of course there are other possible interpretations of the data. Maybe people in certain states are more predisposed to suicide AND to own guns? Occam’s razor would say no, but it’s worth a look. On one extreme we see states like Montana, Wyoming and Alaska. On the other, New York, New Jersey, Maryland. There are certainly demographic differences across these states: population density, income levels, etc. Could Montanans be three times more suicidal than New Yorkers due to those factors? I don’t know, and I can’t rule it out.

It’s also interesting to observe what happens across countries. I scraped Wikipedia’s List of countries by suicide rate and plotted that against Number of guns per capita by country:

In this case, the correlation is almost nonexistent (0.08). Of course, no other country comes close to the US in guns per capita so this is not a complete surprise. Also, the data is different: this is total guns per capita, not percentage of households with guns.

At this point, all I have is the question: would it be possible to prevent thousands of deaths every year by making guns more inaccessible to suicidal individuals? Would it be worth the effort? I wish I could offer answers to these questions, but I’m just one guy with some free time. All my data and “insights” come from Google, Wikipedia, and R. What data is out there that I could be missing?

Final note: this is a very sensitive topic. If you are going to comment on this post, please be reasonable and rational.

I decided to plot country murder rates against gross domestic product per capita, and also against Gini coefficients. I scraped the data from these sources (.csv files at the end of the post if you’d like to use them).

• GDP per capita (Wikipedia)

• Intentional Homicide Rates (Wikipedia)
• Gini coefficients (World Bank)

This is what Murders vs GDP per capita look like. The Pearson correlation in the chart (r) is -0.32.

Murder rates vs. Gini indexes. r = 0.5

This the same chart as above, restricted to countries in the Organization for Economic Co-operation and Development (OECD):

And the same chart, zoomed in (Mexico and Chile are outside):

Finally, the data. One caveat: if you look at the World Bank spreadsheet, you will see that not all countries have current Gini statistics. I used the most recent one for each country. These charts might look different if all countries were current.

• murders.csv
• gini.csv
• gdp.csv
• oecd.csv (from Mark Reid).

Now, do I dare draw any conclusions from these charts? We see the correlations, but what do they mean? I see many poor countries with high murder rates and high inequality. We could speculate that the problem for those countries is the failure of rule of law. It’s even more tempting to guess what’s happening in the last chart. Developed countries like the US make it easier for individuals to become very wealthy, and don’t offer a comparatively good safety net for the relatively poor. This incentivizes people to work hard, start businesses, avoid poverty. Obviously not everyone will succeed. Perhaps the price to pay for this model is a higher murder rate.

Of course it would not be serious to arrive at that conclusion from these hasty charts, but I can pose the question. Feel free to play with the data, and to suggest other variables to analyze.

Speaking in Newtown, CT yesterday, President Obama said:

We can’t tolerate this anymore.  These tragedies must end.  And to end them, we must change.  We will be told that the causes of such violence are complex, and that is true.  […]  Surely, we can do better than this.  If there is even one step we can take to save another child, or another parent, or another town, from the grief […] then surely we have an obligation to try.

It was a comforting speech for the victims of a tragedy, so it would be unfair to criticize the arguments from the point of view of logic. However, it is worth analyzing the issue of mass shootings as a problem that might be addressable with public policy.

I would start by measuring the magnitude of mass shootings as a problem. How does it compare to other issues such as preventable diseases, regular crime, terrorism? I searched for data, and found out that in the past 30 years, 543 people have been killed in 70 mass shootings. That’s an average of 18 deaths per year. For comparison, three times as many die from lightning strikes.

The New Republic article linked in the previous paragraph states “I can’t say exactly why mass shootings have become such a menace over the past few years, and especially in 2012.” Given the low numbers, it’s likely that it is just a random fluctuation without statistical significance.

To put things in perspective again, half a million Americans die every year from tobacco use. Two hundred thousand die from medical errors. Those numbers are large enough that it’s possible to track changes with statistical significance, and evaluate the effect of public policy. There must be a fair amount of low-hanging fruit. For example, it’s feasible that a 100% tax on the price of cigarettes would save thousands of lives ever year. Why is this not attempted? Probably because the special interest group that controls tobacco sales is powerful enough to stop it.

For mass killings, the numbers are already so low that the logical question would be: is it worth doing anything to try to reduce even more the chance of mass killings? What could be the undesired side effects of implementing policies to that effect? For example, let’s say that someone came up with a vaccine that guaranteed that a child who received it would never be a mass killer. However, one child in 100,000 dies from an adverse reaction to the vaccine. Clearly the vaccine itself would cause more deaths than mass killings, so it’s a net negative if we are trying to minimize unnecessary deaths.

At this point, I have to disagree with Barack Obama. I don’t think we have the obligation to try to reduce the incidence of mass killings because there are high chances that an intervention would be iatrogenic: the cure be worse than the disease. This is not a politically correct thing to say, so you won’t hear politicians say it. That doesn’t mean our legislators will do anything, of course. Mass killings are as inevitable as lightning deaths, and they will continue to be news precisely because they are infrequent and horrible.

Who knows, maybe doing nothing is the right thing. There are medical procedures that are not recommended anymore because they have potential complications, and they offer no measurable benefits when compared with inaction.

What makes matters more complicated is that mass shootings bring up the issue of gun ownership in the US. If this killing had been a bombing nobody would be talking about gun control. However, many people who normally don’t think about gun crime are emotionally moved by mass shootings. From a logical viewpoint, we should be more concerned with gun crime in general. If gun crime is a significant problem, then gun control could be a solution to that problem. Surely gun control would have side effects, but it’s likely that those side effects would not offset the gains.

So, is gun crime a problem? In the US there are about 3 gun homicides for every 100,000 inhabitants every year. That means about ten thousand people are shot to death in the country. For the average American, the odds of being murdered with a gun are 1,000 times higher than those of dying in a mass shooting. His/her odds of dying of cancer are “only” 60 times higher than those of being murdered with a gun, so the problem is not insignificant.

Let’s say that we believe that the cost of implementing gun control is less than the benefits. Perhaps we can save four thousand lives every year if we make it harder for criminals to obtain guns. More importantly, we can do it without taking any resources away from the fight against the main causes of death: cancer, heart disease, and accidents. How would we go about it?

The US is a very unique place when it comes to guns. As of 2009 there were 310 million non-military firearms in the country. It is possible to make it illegal to produce and buy new ones, but what do we do with the existing ones? What kinds of imbalances would be created if those who would only use guns to protect their property could not own them? What if most potential murderers kept their guns, and all the guns turned in (say, for cash or tax breaks) were the ones less likely to murder anyone? What kind of black markets might arise for guns and bullets?

I’m not even going to try to answer those questions, because they are extremely complex. I personally hate guns. I have never owned or even fired one. I wish they didn’t exist, but they do. However, believing that gun control would immediately save lives is magical thinking. It might work in the long run if implemented correctly for the US, but it when it comes to reducing murders it would not be a silver bullet (pun intended).

The other issue that many bring up when mass killings happen is mental illness. There is little question that those responsible for mass killings fit most definitions of “mentally ill.” However, they are a minuscule minority. At the same time, mental illness is a horrible condition that causes an enormous amount of suffering. It affects millions, and there is no question that it would be a good idea to address it through public policy. This might have the bonus of preventing the odd massacre in which the potential perpetrator could have been under treatment for a condition such as paranoid schizophrenia. However, not all sufferers of this condition would seek treatment. Norway has one of the best healthcare systems in the world, and that didn’t stop Anders Behring from killing 77 people. Some conditions are asymptomatic for a long time, and manifest themselves too quickly. “He seemed like such a nice, quiet guy. I don’t know why he flipped out.”

If there is one point I’d like to make with this long rant is that public policy should not be dictated by emotions. Minimizing unnecessary deaths and appeasing public opinion are different things. Most human beings do not understand concepts such as statistics or iatrogenics, so they will clamor for immediate feel-good action. I wish I lived in a world where people (or at least leaders) would always analyze issues rationally. Where they would act to maximize public good instead of their chances of being re-elected. All I can do is ask my readers to try to understand all sides of a delicate issue before forming an opinion, like I attempted to do in this post.

Discuss on Hacker News (please be civil!)

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.