Are We Turning The Kids Gay?

Testing the Acquired Homosexuality Hypothesis

“They’re turning the freaking frogs gay,” Alex Jones, the crazy conspiracy theorist, once so famously said. Although it sounds insane, he was not completely wrong. Some scientists have shown that the presence of a popular herbicide in the environment could induce hermaphroditism in frogs. Well hey, even a broken clock is right twice a day. But what made the quote go viral, was its implication - if environmental influences could turn frogs ‘gay’, why couldn’t it do the same to humans? 

Recently the economist Bryan Caplan hypothesised exactly that. He defended the idea of ‘acquired homosexuality’. He argued that LGBT groups, with their current prestige, could evangelise, turning other people gay. Bryan Caplan agreeing with Alex Jones? Maybe he has gone insane?

Wikipedia claims acquired homosexuality is a “discredited idea” for which there is “no evidence”. So naturally, I’m inclined to take the idea very seriously. After all, Wikipedia’s impartiality and quality have been rapidly declining for controversial topics.

Caplan started his essay with some extraordinary data from Gallup polling - far more people are describing themselves as gay. Among men, 0.4% of the Traditionalists aka “the Silent Generation” considered themselves gay, but 2.5% of Zoomers do. That’s a 500% increase in the proportion of people who are gay! The increases in other forms of sexual orientation, bisexuality and lesbianism, are even higher.

This all begs the question, are we turning the kids gay? Or are they just more likely to come out of the closet, or even identify as queer when they are in fact not?

Caplan then follows the data with what I consider to be his four key points/arguments: 

1. The heritable influence on homosexuality (it’s around 30-40% in men) does not preclude there being any large societal effects that are changing over time.

2. If the increased rates of homosexuality were explained by reduced ‘closeting’ then we should not see such a difference between older and younger generations alive today.

3. Homosexuals are not reproducing and gay genes are surely ‘imploding’, suggesting any increase is likely caused by the environment. Or at least any massive increase in gay identification is then more likely to indicate an environmental effect making people homosexual.

4. The explosion in LGBT groups and the rising prestige of the movement allows and facilitates recruitment. 

Now I am quite sceptical of these arguments. In fact, I originally planned this blog post as a rebuttal, a ‘deboonking’ of Caplan’s position. For example, although the twin studies do find a substantial effect of the shared environment, they still use self-identification which may be impacted by closeting. But rehashing the whole debate is somewhat tedious - I recommend you read Michael Bailey et al’s review of the science around homosexuality if you want a careful review of the issue.

My key rebuttal to Caplan was going to be that if acquired homosexuality is true, then more people would be engaging in more gay sexual behaviour. This can be tested and falsified, which is what I expected to do. But the results didn’t exactly go the way I intended and as John Maynard Keynes said “When the facts change, I change my mind. What do you do?”

So with that, let’s get to the point and look at the data. 

How do we find out whether more people are becoming homosexual if we cannot trust them to tell us the truth? The answer is obvious. The gay comedian, Stephen Fry, once did an interview with an ‘ex-gay’ therapist who offered ‘gay conversion therapy’. In questioning the veracity of the therapist’s conversion he asked “when your close your eyes and masturbate, what images come into your head?”  We want to know what images people use for masturbation.

The trick to our test would be to use people’s internet search history. Ideally, we would just check whether more people are watching gay porn compared to straight porn over time using Google Trends. I compare two search terms “gay porn” and just “porn”. Google adds up the searches for both terms and then calculates the searches for “gay porn” as a percentage of searches for both terms. In America in the year 2004, gay porn represented 4% of searches that include the term porn. Over the past 12 months, the number is still 4%. Since the data we are given is in such a low resolution, rounded to the nearest percent we can’t tell what’s happened. It could be consistent with a substantial increase in the proportion of people who are homosexual, or a substantial decrease! 

We’ll have to try a method which is slightly more precise. What we can do is compare searches for “porn” and “gay porn” across different places. We would then want to see if positive social attitudes toward homosexuality then predicted more people choosing to watch gay porn. Now I’m not the first to do this. Seth Stephens-Davidowitz, the author of “Everybody Lies”, does the same across US states. In an interview with Vox, he says that “gay porn searches are about the same everywhere”, indicating there are about the same number of homosexuals in all states. 

Although Seth was not lying here, he was not quite telling the truth either. In his book he notes that “there are more gay porn searches in tolerant states compared to intolerant states”, but he claims differences are small, after giving a selected comparison between states. But if we look at the data available on his website “ finalgaydata3”, we see at the lowest extreme, 4.0% of porn searches are for gay porn in North Dakota compared to 6.4% in New York. That’s over a 50% difference in gay porn searches in two US states. 

In Seth’s dataset, he has a measure of support for gay marriage by US state, which correlates with searches for gay porn at 0.27 which is not nothing, but it is also not statistically significant. But when I download the data from Google Trends myself, I find a correlation of 0.31 (p < 0.05). So maybe something is going on?

Regardless, Seth claims the small differences he found could be accounted for by the internal migration of homosexuals away from homophobic states. However, I’d be somewhat surprised if internal migration alone could explain why, proportionally, searches for gay porn are +50% higher in New York than in North Dakota. But to be sure we should compare countries, where the effect of homophobia-induced migration is presumably smaller and the variation in attitudes to homosexuality should be a lot larger. Our measure of national gay acceptance comes from Pew Research. They found samples of people from different countries and asked “Should homosexuality be accepted by society?” The percentage of people who say yes to the question is our explanatory variable.

Unfortunately, there is a lot of measurement error because Google results won’t give us exact figures and measurement error tends to reduce the size of correlations. But still, the results are quite impressive. National acceptance of homosexuality correlated with the percentage of porn searches that are for gay porn at r = 0.55, p < 0.001. The data is very much consistent with the acquired homosexuality hypothesis! 

At Anglo Reaction, we are quite interested in average national cognitive ability. National IQ correlates at 0.47 with acceptance of homosexuality p < 0.01. Smarter nations are more tolerant. And so as we would expect, national IQ predicts searches for gay porn but only weakly.

But are we really sure this association is causal? But surely social acceptability is confounded with homosexuality being illegal? If it is illegal, many people who are gay just won’t search for the stuff or will use a VPN to avoid criminal sanctions. Well, the BBC has created a list of all the countries where homosexuality is illegal, so let’s control for the effect of legality

.

Making homosexuality illegal is associated with a reduction in searches for gay porn of 0.7%. Not too impressive. Homosexual acceptance still predicts gay porn searches after employing controls. Even when we control for national intelligence (a great proxy for economic and social development), the effect of homosexual acceptance remains strong. Our main model (column 1) implies that a society with 0 acceptance of homosexuality will use 2% of its porn searches on gay porn. A society with 100% acceptance will use 4.3% of its porn searches on gay porn. If this is causal, then perhaps acceptance of LGBT can double the number of homosexuals in a nation?

So far I have only discussed the critiques I can easily test. My biggest issue is that female porn watchers are far more likely than men to watch gay porn on Pornhub. Although women are much less likely to watch porn, women might be more likely to watch porn (relative to men) in liberal societies. And they are surely more likely to watch male-on-male porn in LGBT-friendly countries. That could certainly bias our estimated effect size upwards. 

Another big issue, which I mentioned in my last blog post on Google Trends, is what I call ‘linguistic confounds’. I have used searches of the term “gay porn” as a percentage of searches for the term “porn”. But these are English terms, different countries will use different search terms. What sort of people will use English search terms and will use Google, in non-English countries? And how does this affect our results? 

If anything I’d expect more liberals, and therefore more homosexuals, to be searching for English terms. If the effect of linguistic confounds was a big issue here, it would probably mean that the true effect size is even bigger. Further still, we have shown that acceptance of homosexuality predicts searches for gay porn more successfully than national intelligence does. If our results were due to some sort of linguistic confound related to culture or development then we wouldn’t expect our gay porn measure to be so robust to using this control variable. 

The last issue I’m going to raise is that maybe causality goes the other way. Perhaps some societies are more accepting of gays because they have more of them? But that raises the question - why would some societies have more homosexuals than others? A genetic explanation is not a compelling explanation. Within the USA, Blacks and Hispanics are probably more likely to be homosexual than Whites and Asians. If there were genetic group differences in homosexuality, then we’d expect the less intelligent, more intolerant nations to be more homosexual. And if the effect is cultural or environmental rather than genetic? Well then, congratulations! You have returned to the acquired homosexuality hypothesis.

Despite my efforts to falsify Caplan’s argument for acquired homosexuality, the data behaves just as his theory predicts. It’s plausible that Caplan could be correct - he certainly is not insane. So maybe we are turning the kids gay?

Extra content:

As I was writing this blog post I had a few gay friends remark that this measure of homosexuality is incorrect or at least not exactly what they mean when they say they are gay. The scientific literature generally has four ways of defining homosexuality: 

1. Behaviour - having sexual attractions with someone of the same sex

2. Sexual identity - one’s identification as being homosexual

3. One’s degree of sexual attraction to people of the same sex

4. Degree of physiological arousal to people of the same sex

Choosing to watch gay porn over straight porn tells us about your preference for people of your own sex, but it doesn’t exactly follow any of the above definitions. My measure seems to be closest to category 3, but if a man chooses to watch only straight porn, despite finding other men attractive or arousing he would be straight under my measure. But whatever, I don’t care. The ‘Stephen Fry’ definition of being homosexual is clearly very intuitive and obviously reasonable. I don’t think it ridiculous to say that a man who does not care for gay porn or gay sex, and very much prefers to be aroused by women, is straight even if he has the latent potential to become aroused by or attracted to men. If anything the whole debate around acquired homosexuality is about whether people who have a latent potential to be gay, can have their preferences changed by the environment.