Matthew Yglesias and Vox On Guns, Bans, and Suicide, Part 1
This article deals heavily with suicide rates/methods. It’s pretty academic/dry in tone, but if this is something that you think might cause you to have issues or be in danger in any way, please use your best judgement deciding if you should read it. If you’ve somehow stumbled on this article because you are considering self-harm, please seek help. The National Suicide Helpline is 800-273-8255.
The data I’m working with on this is very messy and hard to interpret because it’s necessarily confounded by dozens of things. I very much welcome feedback on this - if my math is wrong or I’ve made logical errors, please point them out to me; I want to get better-not-worse at thinking about this subject.
As you are probably already aware, Matthew Yglesias is on Substack now. I don’t agree with him on much, and I agree with him absolutely on even less, but I think this is a very positive thing. He’s a good writer and I think he brings a lot to the Substack table in terms of ideological diversity, something that I think will help with the public image of the community as well as the quality of the work produced by it. Nothing is good unchallenged.
With that in mind I’ve been going over his recent posts to try to get a feel for what he’s trying to now that he’s been cut free of his editorial overlords, and I think the most interesting work he’s done so far is this article, which feels aimed at explaining to the cultural left why gun control has come to be considered a loser as a political issue and has as a consequence been receiving less focus from politicians on the left.
Despite the fact that I know that the previous paragraph is the focus of the article (which is a good piece, and you should read it), I’m of course going to ignore the general thesis of the article and drill down on two things he touched upon only briefly: Gun control’s effect on suicide rates and justifications for banning or restricting various things. The two subjects are related but not closely enough that they make sense to try to cram together, so I’m covering the former in this article, and the latter in an article to come.
The article’s basic stance as I read it is that guns are negative, perhaps immensely so, but that trying very hard to ban them is political poison and as such the Democratic Party at large should be cut some slack for not trying to ban them. In Matthew’s view, there’s other work to be done, and so long as guns are keeping the left’s candidates from winning, the left should be OK with them walking lightly around the issue of firearms - after all, some policy wins are better than no policy wins.
Part of why he thinks guns are bad is this:
Here’s the deal: There are about 40,000 firearms deaths per year in the United States and if you could make them go away that would be great.
But a majority of those deaths are suicides.
That “but” is because he goes on to talk about how reducing the suicide numbers requires not just stricter background checks and such, but generally reducing the amount of people who have guns at all as much as possible:
None of this is to deny that gun control laws could drastically reduce the incidence of firearms death. You might think that potential suicides would just substitute some other means of killing themselves, but research does not bare that out. If fewer guns were around, then fewer people would kill themselves.
What’s more: Gun enthusiasts are aware of this. So when progressives talk about the tragedy of gun deaths in America, it doesn’t matter if their actual proposal is a very mild tweak to background checks. When you define “the problem” as gun deaths, you are pushing toward a drastic solution that gun hobbyists don’t want, and they are highly motivated to vote against you.
Note that there’s two quotes there - there’s a lot that doesn’t have to do with suicide that I’m leaving out and I encourage you to read the article. The link in the first quote above leads to an article by Vox’s Dylan Matthews and is Yglesias’ main support for his statements on suicide and the role firearms plays in them. Specifically, here’s what it has to say about gun control’s effect on suicide:
Furthermore, there’s evidence that gun control can reduce suicide rates. A buyback program that wound up taking a fifth of Australia’s guns off the street wound up reducing firearm suicides by 74 percent without affecting non-firearm suicides. When the Israeli Defense Forces stopped letting soldiers bring their guns home over the weekend, suicides fell 40 percent, primarily due to a drop in firearm suicides committed on weekends. Firearm suicides are less common in US states that check if potential gun purchasers are mentally ill or criminal fugitives.
The IDF population referenced in the second link is about as non-typical as a population can be; I’m not bothering with it due to the fact that it doesn’t generalize particularly well. The third link references a study that finds a relationship between how vigorously different US states perform background checks and the associated suicide rates, but it failed to reach statistical significance in its findings and could not determine causation:
The findings of this study tentatively suggest that, by and large, more comprehensive background checks at the state level prior to firearm purchases may reduce firearm deaths. However, we strongly emphasize that these results are preliminary, and that more research is needed to establish whether these results are causal. Specifically, there are several underlying confounders that this study cannot account for.
So we are down two studies and left with only the first from the quote: an article examining the effects of Australian gun buybacks on the Australian suicide rate. The quick eyed among us have already noticed that the Vox article talks only about the firearm suicide rate. I’ll get to the specifics of why that doesn’t cover the whole picture, but that phrasing is being used because the author wants you to come to a conclusion that isn’t true.
The argument made regarding firearms as they relate to suicide is that while all suicides are negative, firearm suicides are perhaps particularly so because in the absence of a quick, reliable and maximally quick method of ending one’s life people might resist suicide in instances where the desire to kill themselves is an impulse rather than a persistent, lingering thing. The idea, then, is that by reducing access to guns you save lives; substitution is not expected to be absolute, so total suicides should drop.
Since our focus is supposedly on saving lives rather than getting rid of guns because we just don’t like them, the relevant metric here is total suicide rates, rather than firearm rates, because this gets around method-substitution and lets us know how much an intervention helped or didn’t help. By using firearm rates, the authors are letting you know that they themselves considered the total suicide rate to be insufficient in some way to prove their point - something between contradicting their goal or merely underwhelming.
In the case of Australia, it’s kind of both.
Australian Suicide Rates Are More Confusing Than Represented
For context, you should first know that Australian suicide rates, especially surrounding the years of the 1996 buyback, are weird. Page 5 of this paper (sorry for the archive version) details that in 1995, the overall all-methods suicide rate was .0125%. 1996 and 1997 were the years during which the buyback was enacted, so we can’t draw rely on their data very well, but the rates went up to .013% and .0146%, respectively. 1998 was still elevated at .0142%, 1999 at .013%, and only by 2000 did suicide rates return to 1995 levels at .0123%. Please note: these are not age-adjusted percentages, just suicides divided by population at the time.
Firearm deaths dropped pretty much every year during the same time period, with one anomalous jump in 1999. But if you were judging just by the numbers, it’s hard to see that the prediction that removing firearms from their owners reduces overall suicides holds much water. Vox’s source addresses this as follows:
Another paper (pdf) by Wang-Sheng Lee and Sandy Suardi, looks at the firearm death rates in Australia over time and found no "structural breaks" associated with the law. But Leigh and Neill note that, because of the large number of factors affecting gun violences, real changes due to the law could potentially not show up as "breaks."
"When policies have even modest lags, the structural breaktest can easily miss the effect," Hemenway explains. "It can also miss the effect of the policy that occurs over several years."
So to be clear: suicide rates went up around the time of the intervention. Hemenway, the expert quoted here, says that this is because of A. lags in the policy and B. because this policy might have taken affect over several years. But neither of those are possible! This policy didn’t happen over several years; the was during a one year period in 1996 and 1997. If taking away guns reduces total suicide rates, the individual effect happens the moment the gun leaves the hand of the owner. No lag is possible if what they’ve told us about suicide-by-impulse, gun ownership and total suicide rates is correct.
All that said, after 2000, suicide rates continued to drop, hitting a low of .0102% by 2004. At the all-time high 1997 rate, this would have meant about 850 more deaths; at an actual nationwide suicide count of 2095, that’s not nothing.
But it’s still murky; gun suicides were dropping, but so were other-causes suicides, at about the same pace. Remember here that the implication of the theory is that taking away guns lowers gun suicides without full substitution in non-firearm suicides, not that it will reduce non-firearm suicides themselves, since there’s no mechanism that makes sense by which that would happen.
The overall impression you get if you look at these numbers unmotivated is that the 1998-1999 suicide counts are counterintuitive to the “less guns less suicides” stance, and the rest of the numbers are minor changes that are hard to separate from drops in suicide rates both with and without guns.
US Suicide Rates Don’t Track Guns
Here’s a graph of household gun ownership rates, year by year, taken from here:
Or, to put it another way (same source):
Here’s US suicide rates over the 1999-2017 (taken from CDC):
So as US gun ownership dropped, our suicide rates jumped. You might note that by 2017, our suicide rate was 14 per 100000 (same source):
From 1999 through 2017, the age-adjusted suicide rate increased 33% from 10.5 per 100,000 standard population to 14.0 (Figure 1). The rate increased on average by about 1% per year from 1999 through 2006 and by 2% per year from 2006 through 2017.
What you might note here is our numbers look, at a glance, very comparable to Australia’s at the bookends of the 1990’s and now:
Rates began to rise in 1985 and fluctuated from 14.3 in 1987 to 11.9 in 1993 with a recent peak of 14.8 in 1997. This was followed by sustained declines over the early 2000s, with a low of 10.2 per 100,000 population in 2006.
After 2006, suicide rates began to rise. In 2019, the rate was 12.9 deaths per 100,000 population—down from a high of 13.2 in 2017.
So we have a situation where US gun ownership dropped, but suicide rates climbed substantially. The obvious explanation for this is that our suicide rates were affected by things other than gun ownership - the economy, lack of exercise, whatever - but as soon as you acknowledge this you get into trouble, because then you have to notice the double standard in play.
Where Australia’s suicide rate dropped, we are asked to ignore the downward trend in their non-firearm suicides and focus only on the firearm suicides lest we attribute their successes to other causes besides guns. Where the US suicide rate rose as gun ownership dropped, we are asked to ignore the other possible influences, lest we let guns off the hook. And we nearly converge at times, a lot closer than we are supposed to let on; Australia’s 13.2 per 100000 in 2017 is better than our 14, but by depressingly little.
Putting It All Together
I know this is article is a mess of data, but the data here is very messy. Let’s try to summarize a bit:
The Australian buyback’s effects on suicide rates are more confusing than they generally let on. They definitely reduced firearm suicides, as expected, but also coincided with an apparent overall downtrend in Australian suicide rates that make it hard to tell how many lives the buybacks actually saved.
The US has seen an increase in suicide rates that goes against what we’d expect considering our dropping gun ownership rates unless you consider other non-gun factors might be at play.
Around the time of the 1996 gun buybacks, our suicide rates were “competitive” or better than Australia’s. Recently, our suicide rates are worse, but not by much.
I want to be clear that none of this “proves” that reducing or eliminating private gun ownership in the US wouldn’t have an effect. In fact, it might have a massive effect far beyond what Australia’s numbers would indicate - a person in the US is far more likely (about 50%) to use a gun to harm themselves than was ever true in Australia during the time periods we’ve discussed here. It could be that this would be a massive, earthshaking difference.
Or it could be a drop in the bucket. In 2017, if we had Australia’s death rates instead of our own, the US would have saved about 2600 lives. That’s not nothing! But “Let us take every gun; we might get a 5% reduction in suicides” is much less impressive than “Let us take every gun; we might get a 76% reduction in suicides”, as Vox’s deceptive phrasing seems designed to convince people would be the case.
All this to say that the numbers here are a lot messier than Yglesias and Vox represent them to be. It’s understandable why - pulling all these charts and references took an enormous amount of time and it’s legitimately hard to spend this kind of effort on something you instinctively want to be true. But to the extent that it’s hard to say how much our suicide rates would go down based on what we know, it’s better to just say so rather than pretending another country’s messy numbers would generalize cleanly to our own.