Gender Census 2022: Worldwide Report

Contents

A warm welcome to the 2022 Gender Census report, and I’m excited to share it because we’ve got some ~drama~ for you this year:

  • There is a word that might be getting close to knocking nonbinary off the top spot in the identity word list!
  • How many new checkboxes next year?? (And a new system for choosing checkboxes.)
  • “Genders based on neurodivergence are all just kids on Tumblr” – are you sure about that?
  • Title anarchy in the UK.
  • A typed-in neopronoun is pulling ahead, and this is very unusual!

The survey took place for one month between 13th July and 13th August 2022, with 39,765 usable responses. It’s a community-based project that is not affiliated with any organisations, companies or academic institutions, so it was promoted entirely on social media and by word of mouth.

Before starting the survey, participants were required to check the following two boxes:

  • Yes, I confirm that I don’t really fit into just one of the two boxes of “always, solely and completely a woman/girl” or “always, solely and completely a man/boy”.
  • Yes, I understand that I can back out of the survey at any time before the end and my answers won’t be counted, and I understand that if I complete and submit the survey my anonymised response will be made publicly viewable.

You can read a much shorter summary of the main three questions here.

Q1. Identity words

This question is asked annually, and hasn’t changed much over the past several years. Participants were asked: Which of the following best describe(s) in English how you think of yourself? They had a choice of 31 identity words/phrases, plus none / I do not describe myself / “I’m just me” and questioning/unknown. Underneath the checkboxes there were 5 textboxes that invited participants to type extra words not otherwise listed. Checkboxes were randomised to reduce primacy and recency bias. This question was optional.

Checkboxes

Here’s the top 24, which is how many I could comfortably fit on a graph on my screen:

For a screen-reader-accessible list of identities and their popularity, please go to the Google Sheets spreadsheet of results for the identity words question.

Here’s our top 5:

  1. nonbinary: 63.9% (down 4.3%)
  2. queer: 54.6% (up 6.6%)
  3. trans: 38.2% (up 4.7%)
  4. gender non-conforming: 34.5% (up 1.6%)
  5. transgender: 33.9% (up 4.7%)

These words’ popularity relative to last year may be influenced by last year’s higher-than-usual proportion of participants aged 30 and under, but I haven’t been collecting information about age for long enough to know whether this is definitely the case yet. Here’s the top 10 words for the 30-and-unders compared to the 31-and-overs:

See more on the age differences in checkbox identities on the Google Sheet here.

Fun fact – the top 10 for ages 30 and under is exactly the same as last year, except that enby dropped from third place to sixth place. Nonbinary and enby have gotten less popular overall, and everything else in the top 10 has gotten more popular overall, but the actual words in the top 10 have remained the same even though they have moved around a bit. In related news, woman has disappeared from the top 10 among 31-and-overs, and it has been replaced by none / I do not describe myself / “I’m just me”.

Here’s our top 10 since 2015:

People who use screenreaders may want to see this graph on Google Sheets here.

It’s a little difficult to see what’s going on, but I can tell you what I’ve noticed. Enby had been rising consistently since it was added as a checkbox option in 2016, but this year it dips back down again by several places. Genderqueer had been sinking since 2015, but in the last two years it’s climbing again. Queer shot to second place when it was added to the checkbox list in 2019, it has overall been rising ever since, and I’m now starting to wonder whether it might become a contender for the top identity word in a few years’ time. (If that does happen, questions will have to be asked about whether it’d be suitable as an umbrella term for all genders beyond the F/M binary, since it is also used to describe sexual orientations and people of a non-straight nature and is therefore quite ambiguous in meaning without context.)

Some notes on the words added to and removed from the checkbox list this year and in recent years:

  • This year’s new checkbox entry, gendervoid, went from 1.3% of write-ins to 6.8% of submissions as a checkbox.
  • Lesbian and gay were added last year as checkbox identities, mostly on the back of lesbian write-ins, but somehow gay was selected as a checkbox way more often than lesbian. That trend continued this year, much to my bafflement. Why were lesbians writing it in but not gay people, and yet gay people are way more likely to choose it when it’s a checkbox?? Make it make sense! (It’s okay, you don’t have to make it make sense, I love the inexplicable.)
  • Like last year, there were several comments in the feedback box asking why femme was in the list but masc was not, so I thought I should mention for the benefit of those people that femme isn’t short for feminine – it’s an experience/expression of femininity in its own right, historically queer-specific. It is often contrasted with butch, which is more masculine, but is also commonly a queer female experience.
  • Like last year, I was also asked in the feedback box why bisexual was not listed in the identity list alongside lesbian and gay. I did think carefully about it when adding lesbian and gay, but decided against it because bisexual is typically considered a gender-neutral word, and therefore I can’t justify adding it automatically as a counterpart to lesbian and gay. As an update, bisexual has still not reached the 1% threshold to be added to the checkbox list this year; it’s up a little bit on last year, but still only 0.8% with more spelling variations considered.
  • Last year neutrois was entered by only 2.6% of participants, putting it under the 3% removal threshold, and this year it was typed in by 0.4% of participants, which confirms my suspicion from last year that the removal threshold is too low. (More on the method of choosing the checkbox list later.)
  • Nothing has dipped below the 3% checkbox list removal threshold this year. The closest is bigender, with 5.5%.
Write-ins

Content note: Some of the words in the write-ins listed here are often used as slurs against the LGBTQ+ community. There is also some discussion of prejudice towards neurodivergent people.

There were 14,622 unique textbox entries this year, almost twice as many as last year, which is pretty impressive considering there were several thousand fewer participants! That’s one unique new word for every 2.7 participants. 2,665 of those were entered more than once, which is more than last year but not twice as much more.

I made a modification to the checkbox term none/I do not describe myself this year, because people have been typing in terms that convey something very similar to what I was intending to convey with this checkbox. (Person, human, [my name], I’m just me, etc.) I changed it to none / I do not describe myself / “I’m just me”. That option got 4.4% last year, and this year it got 16.9%, which is too big an increase to put it down to natural variation, so I think something happened.

However, this year there was a huge jump in write-ins for identity, and not just the number of unique entries as I mentioned above. The top write-ins were also entered by more people, which as a person getting nervous about the length of the checkbox list is very intimidating. The established checkbox threshold is “over 1% in either the 30 and under group or the 31 and over group”. Here’s what the qualifying write-ins look like this year:

  • dyke: 3.16% overall
  • fag: 2.5% of <=30s
  • xenogender: 1.8% of <=30s
  • genderfuck + variants: 1.6% in both [genderfuck: 1.1% in both]
  • guy: 1.7% of <=30s
  • entries beginning with masc: 1.6% of <=30
  • faggot: 1.5% of <=30s
  • transsexual: 1.04% of <=30s
  • boygirl: 1.02% of <=30s
  • autigender + variants: 1.1% of >=31s

I am naturally very excited to add these words to the list of more well-known identity words, but using the current 1% checkbox threshold this would add 12 words to the checkbox list, taking the checkbox list to 45 words. (dyke, fag, faggot, xenogender, genderfuck, guy, masculine, feminine, transsexual, boygirl, girlboy, autigender.) The checkbox list was already getting unwieldy, and this is a 36% increase! Long story short, this is one of two reasons I need to reassess the addition/removal thresholds, which I’ll go into in the next subsection.

My next point of personal interest concerns gender words that typically come in at under 1% of write-ins, but it’s also part of the whole reason why I do this survey – testing common assumptions and prejudices. It’s the identity words denoting genders defined by the person’s relationship to their neurotype.

Some of this is really not pleasant to talk about, but here we go. If you read and talk about gender and neurodivergence online, you’ll probably be somewhat familiar with the stereotype that kids/teens online (especially on Tumblr) are making up genders about their mental health conditions and neurodivergence.

Every year when I’m sorting through the write-ins there are plenty of these identity words, and because of the inconsistency of spelling for write-ins and the tendency for participants to justify or define these words in the textboxes, it can be very difficult to count them. Having said that, the frequency draws my attention to enough that I make the effort to count them (roughly), just to check whether it’s getting close to 1% and is therefore necessary to count more accurately. Thanks to experience I think I did a better job of it more easily this year, and it looks something like this:

Folks who need to use a screenreader, head to cell S1 on this Google Sheet.

The colour-coding is my conditional formatting to let me know at-a-glance when something is over 0.5% (green) and over 1% (purple).

In summary, gender identities related to neurodivergence were entered more often by people in the older age group (aged 31 and older), not the younger age group (aged 30 and younger).

Here’s a breakdown, and please bear in mind that my definitions are informed by my limited incidental encountering online and may not be representative:

  • Autigender and variants (one’s experience of one’s gender is shaped or defined by one’s autistic experience) made the 1% threshold because people from the older age group wrote it in often enough.
  • Entries starting with neuro, most often neurogender (gender is shaped or defined by one’s neurodivergence) were typed in by 0.98% of participants aged 31 and over – so close to the 1% threshold!
  • Entries starting with plural, system and alter (gender is shaped or defined by one’s dissociative identity disorder or related neurotype) were entered more often by participants aged 31 and over. This one didn’t reach 1%, but it had so many different and inconsistent ways of being expressed that I’m really not confident I counted them all, and the automated system may not be suitable for counting them more precisely at all. The most popular iteration was only entered by a third of the entries that appeared in my search, so I don’t feel confident that I’ve observed a “correct”-ish term for it, but the most popular word was simply plural, followed by systemfluid.

The tendency for neurodivergent genders to be chosen by older participants is even more significant than it seems at first, because the older the participant, the fewer identity words they tended to choose/enter:

People with screenreaders may wish to head to cell O1 on the Google Sheet here.

It makes sense, when you think about it. For a lot of atypical neurotypes, the longer you exist, the more likely it is that you or someone else will notice that you’re neurodivergent. Like working out that you’re LGBTQ+, with experience comes self-understanding and coming out.

Although I haven’t done so, I imagine it might be interesting if someone were to use the survey data to break down the use of these terms by 5-year age groups instead of just two large age groups. I’d be curious to see if they generally increase with age. I would also be curious to see whether identities tied to mental illnesses become less common with age, as it is possible to recover partially or completely from mental illnesses, compared to neurodevelopmental neurotypes such as autism and ADHD that are not illnesses.

A sheepish note on survey design

Last year I noticed a big jump in the number of people censoring their words (e.g. the Q slur, f*g), thanks to automated moderation on social media.1 This makes it much harder to count the words accurately, because:

  1. Some characters often used to censor words can also be used in words that are intended to be written that way all the time, such as the asterisk in trans*. I would have no way of knowing which words have been censored and which have been written correctly.
  2. There is often no consistent way to censor a word, so one word could be censored by 5 different people in 5 different ways. For example: qu*er, qu**r, the Q slur, the Q word, qu33r.

(More detail on that in an older blog post here.)

To try to reduce this somewhat, I added some help text to the identity question: “Please do not censor your words unless that’s how you prefer to write them. Examples of censoring: fag → f*g, queer → the Q slur”

About three days into the survey I noticed that a lot of people were typing in fag and queer as write-ins. I did a bit of quick spreadsheetery and found that they were both in the top 3 write-ins, despite queer already being on the checkbox list. (The second most written in word was dyke, arguably a counterpart to fag.) I facepalmed so hard, you guys. I had accidentally caused bias just by reminding people that those words existed.

I hastily changed the help text so that the examples were not typically gender-related words at all: “Examples of censoring: banana → b*nana, apple → the A slur”

But there had already been 23,232 entries.2

So, I just want to reassure you that:

  1. I checked the percentages for before and after I corrected the example text, and dyke and fag are still typed in by over 1% of participants after the fix.
  2. I will account for this when calculating the checkbox list for next year.

After the fix there were 16,534 submissions, and the difference was pretty stark:

  • queer moved from #1 to #2 – from 6.9% to 1.5%.
  • fag moved from #2 to #7 – 3.1% to 1.1%.

Queer was typed into a textbox by 4.64% of participants overall, and I could hastily add it to the checkbox total but when you start to think about it it might actually not be as safe as you’d think. I’m not sure how many of the people who typed it in also checked the checkbox queer, more than half the queer write-ins were typed in after seeing some help text that reminded them of the word’s existence, etc.

All we can really say for sure is, this is a stellar example of the way question design can bias answers over the course of a survey and make a big difference to the outcome. Suffice to say, the help text will be fruit-themed from the start next year.

Overall

The most common number of identity terms chosen or entered was 4 (10.4%), but 5 and 6 was equally common overall. Here’s a graph that shows you the distribution:

See the table and graph here on Google Sheets.

It follows the usual trend, that younger people are choosing more words and older people are choosing fewer, but overall there’s not that big a difference.

The addition/removal thresholds

The ever-growing checkbox list concerns me. As the list grows it becomes harder to find the words you’re looking for. I can tell because (after I fixed the biasing example text) the second most common word written into a textbox was queer (242 people in the last ~16,500 participants) – which has already been a checkbox for several years.

I can also tell that the checkbox filter kindly coded by Andréa is working, because there isn’t another checkbox-duplicate until gay with 254 people typing it in over the whole survey, and then transmasc (222, over 0.5% of all participants), nonbinary (191), trans (152)… Before I added the filter, more of the top write-ins were already on the checkbox list, and this was not the case when the list was shorter.

But queer is the second most popular identity term in the checkboxes, it’s a very well-known word, there’s no way all of over 1% of participants are straight-up presuming it’s not on the list, even more so the less ambiguous nonbinary and trans! So I have to conclude that they couldn’t find it from an attentive skim through the checkboxes.

What with that and the addition of 12 new identity words to the list next year if I follow the current protocol, something’s gotta give.

I spoke with a mathematically-inclined friend about this, and I explained the situation. I also described The Checkbox Effect, where any word added as a checkbox is selected way more often than it was typed in last year. I told him I was increasing the add/remove thresholds, but he countered by suggesting the following method:

  1. Work out how much more often words are chosen as checkboxes compared to how often they are entered into textboxes when they’re not available as checkboxes, on average.
  2. Multiply all textbox terms by that factor. (So, for example, if a word typically goes from 50 people typing it in a textbox, to being checkbox-chosen by 100 people a year later, that factor would be 2. Multiplying a textbox term by 2 would “simulate” what its percentage would be if it were a checkbox.)
  3. Choose a maximum comfortable number of checkboxes, let’s call that x.
  4. Combine the checkbox statistics and the textbox-multiplied statistics, and rank that list.
  5. Choose the top x terms on that list to be the checkbox list for next year.

Using the “over 1% rule”, I’ve so far added 12 words to the checkbox list, and I can compare their numbers from before and after they’re added to the checkbox list, like so:

A table showing all terms added and their Checkbox Effect. Gender non-conforming, 24.3 factor increase. Queer, 14.9. Transmasculine, 12.9. Genderless, 12.4. Lesbian, 11.4. Femme, 10.5. Genderflux, 7.3. Demigirl, 7. Demiboy, 6.4. Gendervoid, 5.4. Butch, 5.3. Demigender, 3.6. Mean, 10.1. Median, 8.9. Mode, not possible. Minimum, 3.6.
The Checkbox Effect. This table can be viewed on the “data over time” spreadsheet on Google Sheets here.

If the number I choose as the factor is too high then the list of checkboxes would change more than it should from year to year, and I risk bumping some established and popular words from the list, making it harder to count them as textboxes. But if the number is too low, some checkbox words will look more popular than some of the type-ins actually are.

The next part is up to you. I’ve started a consultation to find out what survey-takers think of this system. I want to know whether you think it’s a good idea or not, and what a comfortable number of checkboxes is for you. Please do feel free to hop in and be opinionated! (Thank you in advance.)

Edit, 2022-09-04: The consultation has now closed, and you can see the outcome here.

Q2: Titles

Participants were asked: Supposing all title fields on forms were optional and write-your-own, what would you want yours to be in English? There were several specific titles to choose from, plus some hypotheticals, no title at all, “unknown”, and “I choose on the day depending on how I’m feeling”. This question was optional and answer options were randomised, but unlike the others it restricted people to one answer only, because on any given form asking for personal details only one title would be possible per person.

Here’s what we’ve got this year:

People who use screenreaders may want to see the graph on Google Sheets here.

Those of you familiar with last year’s title results may notice that the gap between “no title” and Mx is much bigger this year:

  1. No title at all: 38.6% (up 3.9%)
  2. Mx: 20.1% (down 4.6%)
  3. Mr: 9.4% (up 1.2%)
  4. Non-gendered prof/acad.: 8.4% (up 1.1%)
  5. Ms: 4.4% (down 0.5%)

And this graph confirms the trend in a very visual way:

People using screenreaders may find it helpful to view the data as a table on Google Sheets here.

When I noticed the trend last year I wondered whether it might be affected by the proportion of participants who are in the UK, where we looooove titles soooooo much. (It’s quite unusual to find a system where titles are optional, and even if you do the people operating within that system will look at your name and guess a title and use that. It’s just easier to choose the least bad title and grit your teeth.) So first I checked whether there was a lower proportion of Brits participating this year. There was not. Next I duplicated the spreadsheet and deleted all non-UK answers to see what the UK title stats looked like, and –

For the first time ever, Mx is lower than “no title” in the UK. For people who are not in the UK and/or are not me, this may not seem like a big deal, but I did a double-take. This is significant, and it’s exciting to see that kind of social shift happening in graphs in real time.

Anyway, aside from that, since it’s a bit hard to see what’s going on for anything other than Mx and “no title”, here’s that graph with those two removed:

Again, people with screenreaders will probably have an easier time with the table on Google Sheets.

Mr and “non-gendered professional or academic title” (that’s Dr, Rev, military titles, etc.) are both up, Ms and a hypothetical standard title for use specifically by binary-defying people are both down, most other things are a bit down, and “I choose on the day” continues to trend upwards. Nothing has risen to over 10%; despite its decline, Mx is still more than twice as popular as the next title on the list.

There is an “other” textbox if none of the radio button options fit you, and here’s the top 5 that were typed in:

  • M: 0.4% (164)
  • Sir: 0.2%
  • Mrs: 0.1%
  • Comrade: 0.1%
  • Mys: 0.1%

A special mention for Mistrum, which was next on the type-ins list, and was entered here and in the “a standard title that is used only by people other than men and women” textboxes 150 times (0.4%). Mistrum wasn’t typed in in 2021 at all, and to have a brand new title like this jump so close to the top is intriguing. I haven’t done enough research to talk about it more yet, but perhaps if it’s still around next year I’ll know more. My gut says this is one to watch.

Since we’ve found that Mx is generally considered gender-inclusive (anyone of any gender can use it), I am always wondering whether we yet have a standard title that generally denotes/expresses a nonbinary gender, in the same way that, for example, Ms generally denotes/expresses a female gender. For this reason, participants who choose “a standard title that is used only by people other than men and women” are taken to a separate question asking if they know of any. This year 1.6% of people chose that option. 79% of them left the ensuing textboxes blank, but here’s the top 5 from the titles that were proposed:

  1. Mx: 43 (6.6%)
  2. M: 15
  3. Ind: 7
  4. Person: 7
  5. Mistrum: 6
Q3: Pronouns

The pronouns question was split into two sections, as usual. The first invites you to choose from a checkbox list of 13 options. One of the checkboxes is “a pronoun set not listed here”, and if you choose that it takes you to a section where you can enter up to five new pronoun sets in detail. All questions about pronouns were optional, and answer options were randomised.

The initial icebreaker was: Supposing all pronouns were accepted by everyone without question and were easy to learn, which pronouns are you happy for people to use for you in English? Participants could choose as many of the checkboxes as they wanted. Checkboxes included various common and uncommon pronoun sets, plus options like “any”, “avoid pronouns”, “my pronouns vary” and “questioning or unknown”.

Checkboxes

The checkbox options were mostly the same as last year, but with some changes:

  • I changed the wording of the they/them option by removing the name “singular they”, to try and make its common usage a little clearer: They – they/them/their/theirs/themself (for referring to an individual, i.e. “they are a writer”)
  • Spivak (e/em) has always been on the list, but Elverson (ey/em) hadn’t gotten to the 1% threshold even in the 2021 survey. After doing a few polls I learned that many people may think that Elverson is a variant of Spivak, and one poll suggested that people may actually prefer Elverson over Spivak. It’s extremely unusual for me to add something to a checkbox list without it meeting the 1% threshold, but that one poll scared me into it.
  • People were expressing confusion about “mix it up”, which has been the wording from the very first survey in 2013, so I split it into “my pronouns vary depending on specific conditions” (e.g. for genderfluid people who want people to use pronouns that match their presentation) and “I want people to frequently change the pronouns they use for me” (e.g. for people who like several different pronoun sets equally).

They all fit on a graph, just about:

If you use a screenreader, you can see the table on Google Sheets here. (There is also a graph on there.)

The preference wasn’t as extreme as the Twitter poll suggested, but Elverson (ey/em) was indeed selected more often than Spivak (e/em), so that’s interesting.

The top 5 pronouns (or lack thereof) were:

  1. They – they/them/their/theirs/themself (for referring to an individual) – 75.7% (down 3.5%)
  2. He – he/him/his/his/himself – 40.4% (up 6.6%)
  3. She – she/her/her/hers/herself – 32.7% (up 1.%)
  4. It – it/it/its/its/itself – 16.2% (up 6.9%)
  5. Avoid pronouns / use name as pronoun – 11.1% (down 1.5%)

They/them was selected more than both he/him and she/her combined. 11.8% of people weren’t happy to be called they, he or she, which is an increase on last year. (It increased last year too, so it’s trending upwards a bit.)

Here’s how the pronouns are looking since 2015:

If you use a screenreader, it might be easier to look at this table or graph on the Google Sheet.

It’s not easy to untangle, so here’s the same again but with they/them, he/him and she/her removed:

Again, if you’re using a screenreader, it may be easier to decypher using the table or graph directly on Google Sheets.

I mentioned earlier that it/it is up about 7%, and on this graph you can put that into context – for a pronoun set other than the Big Three, that kind of increase is significant, and in this case it’s part of an upward trend. Very interesting! It’s also the first pronoun set outside of the Big Three to make it over 15%.

Because of this, I got curious about why (aside from it genuinely becoming more acceptable/popular) it might be doing so well, and because this year I have collected specific location information I can make this table showing whether people in particular countries like or dislike a particular pronoun set compared to the average:

This table here is unintelligible because of its width, but at least it looks like it might be interesting. If you want to see it more clearly I recommend viewing it on the public Google Sheet here. It did not at all help me to understand why it/it is increasing in popularity, but I thought it was pretty cool. 🤓

Here’s the top 10 for each age group:

This is another one that might be easier for people with screenreaders to view on the Google Sheet as data.

Last year the top 5 was fairly stable between age groups, but this year everything below they/them, he/him and she/her is a plate of spaghetti. Several pronoun sets dropped out of the 30-and-under top 10. It/it is three times more popular among the younger age group than the older age group. Questioning/unknown is new to the top 10 and I’m surprised to see it more common among the older age group.

Last year, they/them in the 31-and-over age group rose by almost 8%, bringing it almost level with they/them among the 30-and-under age group. This year that gap closes even further, and the difference is now negligible.

If you use a screenreader, looking at this graph on Google Sheets may be easier.

This year the average number of pronoun sets chosen was 2.1, slightly higher than last year – which makes sense, since Elverson (ey/em) had been added. Much like last year, people in the 31-and-over age group selected fewer pronoun sets overall.

Nothing got less than 3% in both age categories, so no checkbox options will be removed for next year.

Neopronouns

The most popular neopronoun overall was in the checkbox list: Xe – xe/xem/xyr/xyrs/xemself (9.9%, up 1.4% on last year). That makes sense, because anything in the checkbox list is selected a lot more often than anything that is written into a textbox. It’s been climbing for a few years now.

After that the checkbox options include:

  1. Fae – fae/faer/faer/faers/faeself: 6.1%
  2. Elverson – ey/em/eir/eirs/emself: 4.7%
  3. Ze – ze/hir/hir/hirs/hirself: 4.7%
  4. Spivak – e/em/eir/eirs/emself: 3.5%

And then we get to the write-ins. I’m not a neopronoun person myself, but I watch them with tense interest, like the bit at the end of Eurovision where they award the points.

When a participant selects the checkbox “a pronoun set not listed here”, they’re guided through a separate section where they can enter all 5 forms of up to 5 neopronoun sets. There are example sentences with fill-in-the-blanks, to help ensure that we get accurate information that can be counted easily.

Age-wise, people aged 30 and under were more than twice as likely to select “a pronoun set not listed here”, which is fairly typical for this question based on past surveys.

However, after that things start to get a bit less typical. Usually, when counted and sorted by subject/object (e.g. ze/zir), the list will probably have one set at the top with maybe 0.6% or so at the most, followed by one or two thousand sets, and the top several sets won’t have that much difference in popularity. To put it another way, there isn’t usually any one typed-in neopronoun set that is clearly taking the lead.

So last year’s top 10 looked like this:

Bar graph. ey/em: 0.59%. ae/aer: 0.44%. they/them: 0.38%. ve/ver: 0.26%. ze/zir: 0.22%. bun/bun: 0.22%. ne/nem: 0.20%. thon/thon: 0.19%. ze/zem: 0.18%. star/star: 0.17%.

But this year’s top 10 looks like this:

Bar graph. ae/aer: 0.70%. ze/zir: 0.28%. voi/void: 0.27%. they/them: 0.27%. thon/thon: 0.25%. star/star: 0.23%. void/void: 0.21%. ve/ver: 0.20%. ze/zem: 0.20%. ne/nem: 0.18%.

Ey/em (Elverson) is gone this year because it has been added to the checkbox list, which does change the shape of the graph a little bit. But aside from that, ae/aer is a very clear leader and is well over twice as popular as the next set. It’s still only at 0.7% so it doesn’t make the checkbox list, but still, I think it’s exciting. It’s not often we get to add a new pronoun to the checkbox list!

Here’s ae/aer‘s most popular set in a sentence:

I’m in a coffee shop with my friend Sam. Ae is buying aerself a coffee in aer reusable takeaway cup. “Is this your coffee?” the barista asks me, holding up Sam’s coffee. “No,” I reply, pointing to Sam, “it’s aers.”

More thoughts on the top 10:

  • Voi/void and void/void are new to the top 10 compared to last year, and I wonder if this is because gendervoid was recently added to the checkbox list in the identity question.
  • I suspect they/them isn’t down to over 100 people mistakenly typing a checkbox option into textboxes this time, because when I look at all 5 forms the most commonly entered set looks like this: they/them/their/theirs/themselves (plural verbs). That makes it “plural they”, i.e. they/them used to refer to two or more people.

2,470 unique pronoun sets were entered this year (based on subj/obj/reflex only), which is almost the same as last year even though there were 4,000 more participants in 2021. Of those 2,470 sets, 543 were entered more than once, which is one new pronoun set for every 16 participants.

Meta

Participants were asked for various bits and pieces of information that I thought it might be interesting to investigate, and most of the graphs and stuff that I got out of them can be found on this year’s public participation sheet that I was keeping up-to-date while the survey was open.

  • Age (grouped in 5-year increments)
  • Country
  • How you found out about the survey today
  • Whether you’ve taken the survey before

Asking respondents for a specific country was new. Usually I only ask whether someone lives in the UK or not, because I like to make a UK-specific report for my own activism. Last year I considered it and concluded that we have enough responses now that it’d be worth asking about people’s locations more specifically. Even though it adds a dimension of risk, reports can now be written up on several countries. The top 5 were:

  1. United States: 22,259
  2. United Kingdom: 4,527
  3. Canada: 2,607
  4. Germany: 1,789
  5. Australia: 1,474

Just to be on the safe side re: identifying information, any country that got under 10 responses (72 of them) had the country redacted. In all, 134 countries were represented in the survey, of which 63 had 10 or more respondents.3

People using screenreaders may have an easier time viewing the information on this Google Sheet.

In this survey, I think of age and referrer as being tied together. Different social networks etc. bring people in particular age groups. Discord and TikTok skew younger, Facebook and mailing lists skew older, and Twitter has the unusual quality of bringing in people from a comparatively broad range of ages.

As usual, it was a struggle to find participants over the age of 30. This year 14.4% of participants were 31 or older, which was better than last year and fairly consistent with previous years. This table helps me to see where I need to focus promotional attention in order to represent people in the older age groups, and what it told me this year is that younger people are leaving Facebook in droves…! It’s gotten bad enough that Facebook is down to #9 in the referral list, and as that’s one of the main ways older people find the survey, I finally bit the bullet and made a Facebook page for the gender census.

Two-thirds of participants came from the big three social networks: Tumblr, Discord and Twitter. 4.7% came from our own mailing list, which is higher than previous years, and I’m really relieved about that, because it’s the other way most older participants find out about the survey. Me and Andréa put a lot of time and confusion (and a fair bit of money) into moving the website onto its own server and self-hosting some mailing list software, which lets us have a mailing list with a lot of email addresses on it and then email everyone at a very low cost compared to typical hosted and paid services. It’s surreal to think that when I published last year’s report I was getting nervous about approaching TinyLetter’s 5,000 subscriber limit, and this year we’re at 7,000 and still climbing. It is a relief to be able to very affordably have an unlimited number of mailing list subscribers.

I remain open to suggestions for ways to get the word out among older and less internet-absorbed LGBTQ+ people. I still feel good about the flyers that supporters can print out and leave in LGBTQ+ community hubs or tuck into Etsy parcels all year round, and if you have any more ideas, please do email me: hello AT gendercensus DOT com.

This year, in response to a comment on social media, I used the timestamps provided by SmartSurvey to calculate how long it takes people to complete the survey. The graph says mostly 4-5 minutes, which is about what I was expecting.

17% of participants had taken part in the gender census in previous years, which was lower than I expected. I’m not sure what this data will be useful for. Perhaps, since the checkbox option order is fairly consistent year-to-year, it shows that these results are not simply the views of an insular community of gender census “fans” whose answers are heavily influenced by previous surveys?

Some of the questions this survey seeks to answer
  • What should the third gender option on forms be called? – Nonbinary was chosen by around two thirds of participants, and its use has been steady for many years. That leaves one in three who don’t identify as nonbinary, but there are no other comparably popular words in the top 10 that are unambiguous in meaning. I would recommend the third gender option be called nonbinary, and future surveys will monitor this issue closely.
  • Is there a standard neutral title yet? – Mx is far and away the most popular gender-neutral title, but at only about one fifth of participants. Internationally, participants are twice as likely to prefer to not have any title at all, and having no title was more popular even in the UK. Those designing forms collecting personal information must ensure that Mx is an option alongside Mr and Ms, but it is far more important (and increasingly important) that title fields in those forms are optional.
  • Is there a pronoun that every nonbinary person is happy with? – No. They/them is a safe bet if you’re not sure what to use, being chosen by just over 75% of participants steadily for six years, and it’s more than twice as popular as the pronoun that took second place. But that leaves one quarter of us who don’t want to be called they, and 12% of us don’t like to be called he, she or they, so it’s always good to check.
  • Is there any consensus on a gender-exclusive nonbinary pronoun? – No. There are no neopronouns gaining ground or being chosen/entered by over 10% of participants.
This year in review

The country question. This year for the first time I asked participants which country they live in, which will allow people to use the spreadsheet of responses to make their own reports about their own countries, for use in academia, activism and self-advocacy. While I can’t make reports for every country, now that there are so many participants taking part every year I can at least enable others to do so. (You can make a copy of a Google Sheet that lets you narrow results to up to four countries here.)

Participation. After last year’s 44,000 participants, I was pretty nervous about having to switch from Google Sheets to Excel. However, since then, Google Sheets have increased their cell limit from 5 million to 10 million, and there were only about 40,000 responses this year, so I thought I’d see how my new (crowdfunded) laptop handled it. It was a lot better! It was faster, and I only had to reload a crashed Google Sheet like 6 times, instead of maybe 50+. So, even though I feel slightly like a fraud for crowdfunding a whole laptop in order to use Excel and then using Google Sheets anyway, the new laptop has been a huge help. A big thank you to everyone who helped out with that, I appreciate it so much. 💜

Last year I put a graph in, so maybe I’ll do that here too?

People with screenreaders may find it helpful to go to cell A13 on the Google Sheet here.

Yeah, that was a good idea. 🤔

Survey design. For me, survey design is never perfect on the first go. It takes trial and error to find optimal question/answer formats, question and help text wording, question order, etc. This year I’d made a few modifications based on feedback from participants and they turned out well. I also made some educational mistakes that I can use to make better choices for next year!

Crowdfunding. Fees were comfortably covered by Patreon pledges! Great news. Any extra went towards Andréa, my trans and queer freelance coding friend in France, who helped a lot with transferring the site to a new host, setting up the mailing list, and making the survey more user-friendly with the search filter.

Thank you everyone for your support! If you would like to follow on Patreon (or add a pledge, every £1-per-year helps), you can visit it here.

And if you would like to thank Andréa for supporting me, her Amazon.fr wishlist can be found here.

What I’ll do differently next year

Survey design. I got a lot of helpful feedback in the feedback box and by email.

  • The method for choosing checkbox options will probably change from fixed percentage thresholds to a more dynamic system. (Take part in a consultation on that here.)
  • For the identity words that are not typically considered gender words in their own right (e.g. “lesbian (partially or completely in relation to gender)“), someone emailed me to suggest that I change it from “partially or completely in relation to gender” to just “in relation to gender”. My original wording was intended to convey inclusivity, but it is a bit convoluted so I’m going to trial changing the wording in this way just to see what happens.
  • I’m going to think more about the title question(s). Some of the help text is a bit confusing, or, one might say… unhelpful? And someone got in touch to point out that in some countries/languages it’s normal to use multiple titles simultaneously, e.g. Germany.
    • Are there any English-speaking cultures in which it is normal and common to be able to select more than one title on a form? Answers on a postcard to hello AT gendercensus DOT com.
  • I’m considering asking about familial language. (Mother, father, aunt, uncle, niece, nephew, etc.) In English a lot of family words are binary-gendered, and there are some family positions that don’t have a well-known standard gender-inclusive or nonbinary-exclusive term. A fair few people ask about this in the feedback box every year, and as it is directly related to binary-defying genders and language it is certainly within the scope of the survey. If Google Sheets have increased their cell limit and/or I’m capable of switching to Excel, I’d love to have a bash at this. (I’m sure the design for the first year will be a bit dodgy, but the second year will be better!) It’s not necessarily something I would need/want to ask about every year, but I’d like to at least ask until I feel confident that I’ve asked well and received helpful, representative answers.
Closing thoughts

I still think it’s amazing that so many of you are into sharing this information with me, and I’m always excited to hear about ways that you’re using the reports and the results spreadsheets too. I know I’m no expert in this statistics/social sciences stuff, but I still think the outcome of this project is pretty amazing, and I know for sure it helps me (and some of you, too) feel connected to a larger community, while giving people more tools to fight for equality and acceptance.

Every year it is so interesting, and I am humbled and touched that people are so enthusiastic about joining in with the survey and even contributing financially to my running costs and equipment. What an honour it is to work with you all! Thank you, from the bottom of my smushy nerdy heart.

Support me!

Thank you for reading! If you find this report and this project to be valuable and would like to give something back, you could pledge to support the survey financially on Patreon, or increase your chances of taking part in future surveys by following on Tumblr, Twitter, Facebook or the Fediverse, or subscribing to the mailing list. Alternatively, you could take a look at my Amazon wishlist! 🍫

Footnotes

To get back to where you were before, click the “back” button in your browser.

  1. In layperson’s terms, that’s when a social network automatically flags or removes posts that have uncensored LGBT-related words or slurs, and people using those words who don’t want to get their post/video taken down have to dodge the removal algorithm by making it unreadable by computers but still readable by humans, e.g. qu*er, le$bean.
  2. Oooo, a palindrome!
  3. I’m curious to know what you think of the “fewer than 10 respondents from a country → redact the country” rule. Email me! hello AT gendercensus DOT com.
Links to files

2022-08-22

email: hello@gendercensus.com