No country for gay men

Multiple large representative national surveys say very few Hungarians are gay or bisexual

Sexual minorities have a prominent place in contemporary politics. Gay marriage, who gets to use which bathroom or what teachers are allowed to say about sexual orientation to kids in school are central elements of political campaigns and heated debates. Therefore, it is very important to know how many people are affected by this issue in the first place.

A dazzling array of survey data circulates on the Internet about the population proportion of sexual minorities, or, as they are often called, “LGBTQ” people. (I try to not use this acronym because there is not much these identity markers have in common if you depart from “intersectional” leftist ideas.) This survey by Gallup about how 7% of all Americans and 21% of all Gen Z members are “LGBTQ” keeps stirring up the Internet. But if you zoom in on the numbers, you can see that most of them (4% out of the 7% of the total sample and 15% out of the 21% of Gen Z) are bisexuals, and most of the growth between generations also comes from bisexuals and transgenders. Compared to baby boomers, Gen Z members are 2.5 times more likely to be gay, about 3 times more likely to be lesbian, but 21 times more likely to be either bisexual or transgender. Even after some growth, the rate of genuine homosexuality remains low.

This is in line with other data.

For example, in three waves of the the CDC’s National Survey of Family Growth the rates of self-reported homosexuality looked like this (the report for 2002/2006-2008 and for 2011-2013):

-          2002: 1.3% of all women; 2.3% of all men

-          2006-2008: 1.1% of all women, 1.7% of all men

-          2011-2015: 1.2% of all women, 1.8% of all men

For bisexuality:

-          2002: 2.8% of all women. 1.8% of all women

-          2006-2008: 3.5% of all women, 1.1% of all men

-          2011-15: 5.6% of all women, 1.8% of all men

This is of course all American data. Rahman et al (2020) published a paper in which they estimated the proportion of homosexuals and bisexuals internationally using a BBC internet survey. Rates of homosexuality varied between 1% (India and Turkey) and 8% (Germany), while rates of bisexuality varied between 3% (Turkey again) and 11% (Finland). Before you start making jokes about Turkish Chads and gay Western Europeans, note that this was a voluntary online survey, so the numbers not only don’t accurately estimate the frequency of non-heterosexuality, but they might be differently biased across countries. (The authors did not find anything though that would be significantly related to the rate of non-heterosexual orientation.)

None of these datasets surveyed Hungary though. I recently ran across multiple high-quality data sources which indicated very low rates in this country, at least on order of magnitude lower than not just the big-tent “LGBTQ” category, but than bona fide homosexuality in US surveys. Let’s look at them.

The first data source is Hungarostudy 2021. I wrote about this dataset recently. It is a nationally representative survey of 7000 people across the country who were visited in their households and given a questionnaire. They were asked to rate themselves as heterosexual, homosexual/lesbian, bisexual or asexual. The rates, also broken down by the standard 5-category age groups provided in the same look like this:

Population rates for all sexual minorities are very low, below 1%. We see trends familiar from American data: bisexuality is much more common among the young, homosexuality somewhat, but not overwhelmingly so. Interestingly, asexual orientation increases with age: I guess this is not flag-waving Tumblr asexuality, but rather older people interpreting this response as “I’m not sexually active anymore”. One could argue that people are hiding their sexual minority status and don’t answer the question because this is a conservative country: we will get back to this argument, but for now let’s just note that missing responses trend down, not up, with age and younger people report higher rates of sexual minority status, just like in other samples. If we assume that some proportion of “Missing data” responses are actually homosexuals or bisexuals in hiding, “Missing data” responses should show bisexual-like age trends, but they don’t. Also, the 0.3% population rate is indeed very low: we would need not one or two, but eight or ten hiding homosexuals for each visible one to match the American rates.

The second dataset is Youth Research, one I also recently wrote about. This is also a national survey of 8000 people visited in their households. Of the three waves, only the 2016 wave asked participants about sexual orientation, which they could report as heterosexual, homosexual, bisexual, asexual, abstaining or “Don’t know”. True to the name of the dataset, everybody here is under 30 years old – however, sampling goes down to 15. The results look like this:

The very low population rates are replicated: 0.2% report being homosexual and 0.7% report being bisexual. (There are fewer bisexuals here than in the age-matched Hungarostudy subgroup, but almost exactly as many homosexuals). It’s interesting to see how the “Abstaining” and “Don’t know” category vanishes with age, while other categories grow: presumably many 15-19 year olds don’t really know yet how to answer this question, but uncertainties resolve in their twenties, mostly as a heterosexual orientation, but sometimes not. The age range is very limited and maybe 2016 was too long ago but there is no trend for young people being less heterosexual overall.

The third sample is again Hungarostudy, but in a roundabout way. Let’s look at this recently published paper:

Kisfalusi et al 2022: Estimating the size of hard-to-reach subpopulations and segregation with the network scale-up method: lessons from a Hungarian survey

Abstract: In this paper, we provide an empirical, descriptive analysis of the social networks of Hungarian society and illustrate how the network scale-up method estimates the size of hard-to-reach subpopulations and segregation of social groups. Based on a representative survey of 7000 respondents from Hungary (HS2021), we first estimate the average size of the respondents’ personal networks. Then, we examine the social fault lines along various social groups and how accurately the network scale-up method estimates the size of these groups (e.g., unemployed, protesters, the Roma, Covid-infected). These estimates are then compared with data from other sources (census data, administrative data, surveys). Our results show that the network scale-up method estimates the size of visible social groups (e.g., the Roma, homeless people) quite well. The visibility of other social groups appears to be much lower. Social fault lines are greatest in the case of homeless people, protesters, and members of NGOs. Finally, we describe recent methodological advancements and summarize our suggestions for future research using this method.

Two out of two Hungarian sociology papers I have cited here now are authored by Dorottya Kisfalusi. I promise I’m not her but she is my favorite living Hungarian social scientist now. I think Hungarian social scientists often don’t seek international spotlight because they think that their findings only have local relevance. I disagree: it’s not great that so much of social science comes from US or UK samples, small countries are important for both original insights and replications, so I machine translated their paper to English and I also describe here what they did.

This paper uses a so-called scale-up method to estimate how many homosexuals (and other possibly hiding populations, such as homeless people, unemployed or Covid-infected) there are in Hungary. The way to do this is to ask respondents about their acquaintances. Everybody is asked how many people with certain characteristics they know, specifically:

1) people with certain given names

2) people who belong to potentially hidden groups (for example, people who are homeless, attempted suicide, had COVID, has been to a protest, or, most important to us, is gay or lesbian).

The rest is math.

From name registries we know exactly how many people have certain given names. We use this information for calibration: based on people’s responses we can estimate how big their social circle is. Somebody who knows a Milán (rare male name) likely knows a lot of people, somebody who doesn’t even know a single Sándor (very common name) probably doesn’t have much of a social circle. Given these estimated social circle sizes, we can ask: approximately how big does somebody’s social circle need to be for them to, on average, know one gay person? This number divided by one is the estimated population proportion of gay people. While you certainly can’t calculate this with just one person, you can get reasonable maximum likelihood estimates using the 8000 respondents in Hungarostudy. The advantage of this method is that it can reveal hidden populations. Well, not completely hidden ones. But if you are gay, hide this in a survey, but admit it to your friends, who in turn anonymously admit that they know you, you will be hidden from the survey but not from this method. You can read more papers like this here, here and here.

(Of course there are lots of ways this method can go wrong. What if people don’t know one of their friends is gay or don’t want to say? What if they can’t immediately recall how many of their friends are called Milán so we cannot correctly calibrate their social circle? What about the fact that people know other people similar to them so a gay person is much more likely to know other gays than somebody at random? Good points: this is why we are comparing this with actual surveys. Both can go wrong, but unlikely to do so in the same way.)

In this paper, they find that people on average have 170 acquaintances – the number is higher in men, in older people, among the more educated and in those living in larger cities, which looks reasonable. What happens when we estimate the number of hidden populations? This:

The weird formatting and odd phrasing are because this is an auto-translated table. The first column is the size of hidden populations using this fancy estimation method based on social circles. The middle column is something you should take with a grain of salt: it the number of people actually belonging to each category based on external data sources. How reliable these “actual numbers” are depends on which population we are looking at. For suicide attempts, it should be very reliable: these go on record. The 23% underestimate by others looks sensible: you would likely, but not certainly, know if some of your friends attempted suicide. For the proportion of the Roma, the census estimate is 3.5%, which is basically perfectly recovered from social networks.

For gays/lesbians, the assumed real number is the extrapolation of American data. This is not very good. As we saw before, this very dataset – together with another – established a much more reliable ground truth, with 0.2-0.3% of the population being homosexual. With a population somewhat below 10 million and these rates, 32 thousand gay Hungarians is exactly what we would expect. The social circle estimation method bullseyes right in the middle of what the surveys say, confirming a very low rate of homosexuality in the Hungarian population.

Why so few?

The obvious go-to answer for why there are so few sexual minority members in Hungary is that they are hiding because this is a conservative country. Hungary certainly is a conservative country and it is not my job to tell gay people how it is to live here, but I’m extremely skeptical of the idea that something like 80-90% of gay Hungarians don’t dare to admit this in an anonymous survey. We would need this much to explain our numbers which are an order of magnitude lower than comparable American data.

Homosexuality was decriminalized in Hungary in 1962. Civil unions have been legal since 2007. This is basically gay marriage, except for the name, the name change, and the possibility of adoption. This is one of the most liberal framework in Eastern Europe: only Estonia, Latvia, Czechia and Croatia have something similar. Budapest hosts a giant pride parade each year. A recently often circulated talking point about how Hungary is an anti-gay country is a law passed in 2021 which prohibits not only advocating, but even showing homosexuality to minors under 18 years old. (There are no sanctions mentioned in the law, so even though you can theoretically be found guilty of breaking this law, there is no consequence.) I have big reservations about this law, but I don’t see how it meaningfully limits the life of adult homosexuals, or minors for that matter, who are free to look up whatever they are interested in on the Internet or maintain private relationships with whoever they want to.

Liberal laws can of course hide a very intolerant society and political system. Let’s look at the 2023 report by ILGA for a report on incidents against sexual minorities in Hungary (page 73 and onward in the document) to see how bad the situation on the ground is. This is a sexual minority advocacy group funded by the European Union, the Dutch government and some American foundations. They are obviously not unbiased, which is great for us now, because we want the best shot: what are the worst incidents this group can report on?

Let’s start with “Bias-motivated speech”:

-          The Media Council imposed a 400,000 huf (950 euro) fine on Pesti TV for its transphobic broadcast in 2021.

-          Minister of Justice, Judit Varga said in an interview that Hungary would fight “gender madness”.

-          Upon the request of Háttér Society, the Commissioner for Fundamental Rights (CFR) launched proceedings against a deputy mayor of Győr, who said on social media that an LGBTQI-themed film should be “thrown in the trash”. The CFR rejected  Háttér’s petition, and the decision was upheld on judicial review.

-          A bus stop was tagged with a text calling for the “killing of gays” - a criminal investigation is ongoing.

Bias-motivated violence:

-          The perpetrator of a 2021 lesbophobic hate crime was sentenced to one year and eight months in prison.

- Háttér successfully challenged the police’s decision to not investigate far-right protestors who gave a nazi salute at the 2021 Pride march. The case is pending.

-          Áron Molnár, founder of the pro-democracy and human rights movement noÁr was sexually harassed by a well-known far-right activist at Pride. An investigation is pending.

-          In March, the Prosecutor General launched an investigation into a hate crime based on sexual orientation because a trans man and a woman were attacked on public transportation in Budapest.

The following claims are made based on this article by a Hungarian advocacy group:

-          A trans person was denied accommodation in a hotel this year.

-          Several hate crimes were committed this year, with the perpetrators in some cases referring to the 2021 ‘propaganda’ law as being “on their side”.

So the best a highly motivated and well-founded advocacy group could come up with were some isolated incidents by random idiots. Even in these, the stories are sometimes based on hearsay, the anti-gay motivation is often dubious, and by the group’s own admission police acts immediately on any incident, including just offensive language. I don’t think this is the social climate that would force 80-90% of gay people into hiding: we must look for the cause of low population proportions elsewhere.