There are already too many proposed causes for the secular decline in crime, but I can’t resist suggesting one more. A couple of months ago Nydwracu asked me whether it could be related to the secular decline in testosterone. The answer turns out to be “Maybe”.
This secular decline in testosterone is pretty dramatic. Our best source is A Population-Level Decline In Serum Testosterone Levels In American Men, which finds that from 1987 to 2004, average testosterone declined from 501 ng/dl to 391 ng/dl, with an even more dramatic decline in bioavailable levels of the hormone. That’s about minus 1% per year.
No one knows exactly why this is happening. Some people blame increasing obesity and decreasing tobacco use (wait? Smoking increases testosterone levels? THOSE TV COWBOYS WERE RIGHT ALL ALONG!). Other people have tried to adjust for these and found they don’t explain the entire effect, leading to a host of other theories. Recent scrutiny has focused on the role of feminizing chemicals in the water supply, probably a combination of industrial pollutants and discarded medications; the worst-affected areas are marked by an epidemic of transsexual fish (really).
(A quick aside – since these chemicals are gender-bending fish, frogs, and various other animals, could they be responsible for transgender in humans? This theory seems to still be in crackpot territory, but I don’t know why. Research shows that male-assigned-at-birth children exposed to diethylstilbestrol in the womb are more likely to become transgender than the general population. Other than that, there just seems to be one unpublished paper on the subject. Get to work, scientists!)
Annnnnyway, testosterone has been found to correlate a bit with violent crime. In a study of 692 male criminals, Dabbs et al found that those in prison for more violent crimes had higher testosterone than those in prison for nonviolent offenses. It’s hard to say exactly how much higher because they report their testosterone in a different way that doesn’t correlate to anyone else’s – I think part of it is that it’s salivary rather than serum testosterone but it’s still confusing even after I adjust for that. If we use relative rather than absolute, they do mention that 66% of inmates in the upper third of testosterone levels committed violent crimes compared to 46% in the lower third. High-T inmates were twice as likely to be in for murder as low-T inmates. Interestingly, testosterone was the highest risk factor for sex crimes, such as child molestation and (especially) rape – high-T inmates were four times as likely to be in for rape as low-T inmates. On the other hand, low-T inmates were about twice as likely to be in prison for drug offenses.
This “which criminals are worse” study is obviously not as good as an “are high-T people more likely to be criminals at all” study, but I can’t fin any of those with a good sample size. You can read a review of the research here.
According to the population decline study, testosterone levels declined about 110 ng/dL in 15 years. They don’t give me a standard deviation, but from this site I get one a bit less than 200. So testosterone declines by one standard deviation about 25 years? That means that a person in the top third of testosterone levels today would have been in the bottom third fifty years ago. Which – and I realize I’m doing all sorts of horrible things here to cover up my lack of actually useful data – if we extrapolate wildly from the results of these studies, we could sort of justify murder halving in about fifty years by falling testosterone alone.
The first problem with this is that we can’t really use data on prison inmates as representative of the population.
The second problem is that murder has halved in way less than fifty years. It seems to have halved between 1994 and 2004.
The third problem is that crime didn’t start falling until the early 1990s, but testosterone was falling since at least 1987 and probably earlier. This site, which doesn’t cite sources, says testosterone was higher in the 1940s, though they might be confusing that with “in men born in the 1940s, as studied in the 1980s”, which is of uncertain significance. Sperm count has been declining since the 30s, according to an article called Sperm Quality & Quantity Declining, Mounting Evidence Suggests
(it looks like somebody was not quite as virtuous as this Twitter user).
The fourth problem is that there’s contradictory evidence about whether testosterone is even falling at all, according to a a study that looked at the faces of Major League Baseball players of the past 120 years. This sort of makes sense – face width-height ratio is affected by testosterone (one reason women’s faces look different than men’s) and baseball players had standardized photographs taken of them for that time period. They find that, at least based on the face ratios, testosterone was increasing during that period, which would be interesting if it didn’t contradict everybody else. As it is, I suspect it just means baseball players were differently representative of the general population. For example, if baseball requires high testosterone, and scouts became better at selecting the highest-testosterone people over that period, that’d do it. Or if the nature of baseball changed to more of a “power game” rather than a “finesse game” (I think some people have said this) that’d do it too. Or if all baseball players suddenly started taking powerful testosterone-analogue chemicals at some point…hmmmmmmmmm…On the other hand – literally on the hand – we have the digit ratios of Lithuanians over 120 years. Someone in 1880 measured the length of Lithuanians’ fingers – which can be a proxy for testosterone levels – and then the experiment was repeated recently and the results compared. It did find the expected increase in testosterone, though no word on whether that was throughout the entire period or just concentrated in the past couple of years. So this sort of turned out to be a non-problem.
The fifth problem is that crime is dropping in women at the same rate as in men – women never really committed that many crimes, but now they’re committing fewer. Women do have some testosterone, so it’s possible that declining testosterone could affect female violence as well, but it wouldn’t be the first thing I expected. Also, I’m not sure if there are any secular trends in female testosterone levels, though I’d be fascinated to see data.
So overall while I like the approach of this hypothesis, I don’t think it gets the time window right. It would be a nice way to explain a gradual fifty-year decrease in violent crime starting in the 50s and continuing to the present day. Instead, we have a big spike in the 50s and a big drop in the 90s, which were not particularly abnormal in terms of testosterone decline.
This doesn’t really make sense to me. If testosterone is declining, it should cause a decrease in crime. One might argue that testosterone levels have been steadily operating behind the scenes causing very long term declines while other things account for the more visible short-term trends, but that seems like a cop-out.
I’d like to see studies comparing testosterone levels in violent criminals (both male and female) to those in the general population.
Also, we have cemeteries full of millions of dead people from every era of history, all carefully marked with what age they were when they died. Somebody needs to dig some of them up and measure their digit ratios – I assume you can still measure the digit ratio of bones, the overall length is still there. Then we can have a good answer for whether testosterone levels in men (and women) have been declining over time, when it started, and whether it’s been picking up recently. If it has been, the chance that it hasn’t had an important effect on our society worth exploring is pretty much nil.
I know, just once I want to get through an entire blog post without a call for disinterring the dead, but this is important.