Gender Census 2025: Worldwide Report

---

8,660 words, ~45 minute read.

Contents

• Q1: Identity words
  • Checkboxes
  • Write-ins
  • Overall
  • Generating the checkbox list
• Q2: Titles
  • Titles (with name)
  • Titles (without name)
• Q3: Pronouns
  • Checkboxes
  • Neopronouns
  • Lack of nuance in pronoun data
  • Pronoun checkbox selection method
• Q4: Family/relationship terms
  • The textbox question: sibling’s child
  • The checkbox question: sibling
• Meta
• Some of the questions this survey seeks to answer
• This year in review
  • Limitations
  • What I’ll do differently next year
• Closing thoughts
• Corrections
• Support me!
• Links to files

---

Welcome to the 2025 Gender Census report, with more words, more graphs and more genders than ever before! This was the twelfth annual survey, so probably the graphs are also wider than ever before.

The survey was open for 32 days, opening on 30th July 2025 and closing on 30th August 2025. There were 43,096 usable responses.

The usual disclaimer: The Gender Census is a community-based project run by an enthusiastic amateur volunteer, and it is not affiliated with any organisations, companies or academic institutions. It was promoted entirely on social media and by word of mouth.

Before starting the survey, participants were required to check the following three 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.
• I understand that if I complete and submit the survey my anonymised response will be made publicly viewable.

You can see a summary page of the big four questions here.

---

Q1: Identity words

Participants were asked: Which of the following best describe(s) in English how you think of yourself? There were 18 identity word/phrase checkboxes, plus none / I do not describe myself and questioning or unknown. Checkboxes were randomised to reduce primacy and recency bias. This question was optional, and participants could select as many checkboxes as they wanted to.

Underneath the checkboxes there were 10 textboxes that invited participants to type extra words not otherwise listed.

Checkboxes

Content note: Some of the words in the checkbox list are often used as slurs against the LGBTQ+ community.

Here’s our top 5 this year, and how they compare to last year:

1. nonbinary: 61.7% (up 1.3%)
2. queer: 56.1% (up 2.5%)
3. trans: 46.5% (up 1.8%)
4. transgender: 41.1% (up 2.3%)
5. a person / human / [my name] / “I’m just me”: 39.7% (up 0.6%)

The only order difference compared to last year is transgender has swapped places with a person / human / [my name] / “I’m just me”. They were within 1% of each other last year so it didn’t take much.

Here’s something more visual, with a few more identities:

But, as usual, these results are heavily influenced by age – 75% of participants were aged 30 or younger. So there is a little difference between the top 10 terms from the 30-and-unders compared to the 31-and-overs:

Something that makes me go “whattttt!” is, enby this year is equally popular in both broad age categories. I’m interested in this because it was originally coined as an alternative to “boy/girl” for words like “boyfriend/girlfriend”, which suggests it might be a diminutive term (e.g. for referring to children), and so I have investigated once or twice what the age trends were for enby specifically, and generally found that younger people were identifying with it more, but this was shifting. This year there’s only 1.2% difference between the 30-and-unders and the 31-and-olders, and in fact it is a little higher in the older age category and has been for two years! So I’m calling it, it’s official – enby isn’t currently being used only for/by kids.

I have made a graph, because of course I have:

It’s always excellent to see the “top 10 since 2015” graph growing:

For most of those terms it’s pretty tangled, so here’s some individual bar graphs as well:

Here’s my insights from swimming around in it all:

• I was a bit worried about nonbinary (dark blue, at the top) last year as it was at its record lowest, but it’s up again this year. Having said that…
• Queer (red) went straight to second place in 2019 when it moved from the textboxes to the checkboxes, and the gap between it and nonbinary has been closing very slowly ever since. Last year the gap was 6.8%, and now it’s 5.6%. It’s so gradual that the two words are almost parallel so it is really hard to know whether they might switch places.
• Enby has definitely peaked! Its highest was in 2021 with 37%; now it’s down to 30% and that’s part of a consistent downward trend over several years.

Write-ins

This year there were 11,924 unique textbox entries, of which 2,313 (about 1 in 5) were typed in more than once. There was one new word entered for every 3.6 respondents, which is about the same as last year.

The top 10 write-ins list tends to change around a lot, because sometimes things get moved to the checkbox list, sometimes they’re aggregates because I noticed something was popular but being written in a lot of different ways, stuff like that. Here’s the top 10 this year:

1. butch: 2.5% [1,096]
2. bigender: 2.5%
3. creature: 2.3%
4. demigirl: 1.5%
5. thing: 1.5%
6. genderfuck, and variations thereupon: 1.4%
7. tran(s)sexual: 1.3%
8. lesbian: 1.2%
9. xenogender: 1.1%
10. guy: 1.0%

Honourable mention to entries starting with [a] man and [a] woman, for being extremely balanced:

• starting with man/a man: 480 (1.1%), 135 unique entries
• starting with woman/a woman: 472 (1.1%), 136 unique entries

Looking closer, they were extremely varied, with some woman* entries being things like “woman lite” and “woman-aligned”, and some man* entries being “manthing” and “manwoman”. The suffixes “-adjacent” and “-ish” were popular for both woman and man write-ins.

This year I reduced the number of identity textboxes from 20 to 10. Last year less than 1% of people used more than 10 of them, and this year less than 1% used more than eight of them – and very few people complained about it in the feedback box. So I think this was a good move.

Overall

The average number of identity words chosen/entered was 6.1, a little higher than last year despite the reduced number of textboxes. Most people (25%) went with 4 or 5 words.

Every year the older age groups tend to choose/enter fewer identity words, and this trend continues into 2025:

Generating the checkbox list

Content note: This section refers to words that are sometimes used as slurs against LGBTQ+ people.

This might be more information that most people need, but I like to include details about the system for the checkbox list for transparency. It’s designed to generate a good manageable list of terms that are representative of participants from a broad age range.

The following terms were new additions to the checkbox list this year thanks to this system:

• autigender – its first time on the checkbox list, it got 5.5% this year, compared to 0.6% last year as a textbox entry. That makes its multiplier 9.2x, which is higher than average.
• dyke – 4.5% as a textbox entry last year, and 14.2% this year as a checkbox, giving it a multiplier of 3.2x (lower than average). This is its second time on the checkbox list; its first was in 2023, so it has bounced off and back on again.
• tranny – also new to the checkbox list, it got 1.8% in the textboxes last year and 11.3% as a checkbox this year. This gives it a multiplier of 6.3x, which is about average.

The following were removed from the checkbox list this year:

• demiboy – 6% as a checkbox last year, and dropped to 1% as a textbox this year (6x).
• demigirl – 7.1% as a checkbox last year, and 1.5% as a textbox this year (4.7x).
• butch – last year’s checkbox got 7.9%, and this year as a textbox it got 2.3% (3.4x).
• bigender – this one has bounced on and off the checkbox list a couple of times now. Last year as a checkbox it got 5.7%, and this year as a checkbox it still got 2.5%, giving it a very low multiplier of 2.3x.

Despite demiboy and demigirl being lost from the checkbox list this year, all the demi- words did fairly well in the textboxes in 2025, and were in the top 10 of several age groups.

Terms seem to be prone to hopping on and off the lower end of the checkbox list when they have a lower-than-average multiplier in both directions. It means that they get typed in a lot, so the average multiplier gives them a bigger boost, and then when there isn’t so big an increase something else can bump it back off the list more easily a year later. I am not sure yet if this is a bad thing.

In order to see if I could fairly reduce the “yoyo” effect I tested out a couple of slight changes to the system. I tried using the lowest multiplier found so far (bigender, 2.5x in 2024). I also tried using the average multiplier for any term unless it had a previously recorded multiplier, and then I used that instead. Both options meant that those low-multiplier terms never made it back onto the checkbox list, and trendy new words took their place instead. So these systems would keep established but less commonly typed in words off the checkbox list, and provide an ever-changing supply of 2-3 less well-known and less established words on the list for one year each only. I’m not sure this is an improvement; I would rather have a steady circulation of well-established and recognisable terms than a ever-changing supply of unfamiliar but trendy terms. Another option might be to simply shorten the checkbox list; if 2-3 words change every year and that leaves about 17-18 options that are really steadily popular, perhaps a list that is only 18 items long will provide more consistent data for the terms that would otherwise be switching between checkbox and textbox every year. I will keep thinking about this. More years of words like bigender and the demis yoyoing on and off the checkbox list might provide some helpful context.

On to next year’s list. We apply the median multiplier (the amount a textbox typically increases by when it becomes a checkbox) to all textbox entries, to simulate what might happen next year. This year the median multiplier is 6.4x. We take the top terms from each age group until we have a list 20 items long. This gives us:

• a person / human / [my name] / “I’m just me”
• agender
• binary
• butch
• cisgender
• demiboy
• demigirl
• enby
• fag
• gender non-conforming
• genderfluid/fluid gender
• genderqueer
• nonbinary
• none / I do not describe myself
• queer (in relation to gender)
• questioning or unknown
• trans
• transfeminine
• transgender
• transmasculine

I’ve bolded the new words to make them easier to spot. Let’s zoom in! They are all low-multiplier “yoyo” words with a tendency to hop on and off the checkbox list:

• butch – this one was typed in by 2.5% of participants (1,096). It makes it onto the list because it got entered often enough by two age groups – the 21-30 crowd.
• demigirl – typed in by 1.5% of all participants (648). This one makes it in thanks to the 41-45 age group.
• demiboy – it was typed in by 1% of participants (413), and didn’t make it high enough in any age group, but I’m adding it for fairness because it’s in a set with demigirl.

We lose all three words that were added this year, which were autigender, tranny, and even dyke, which got 14.2% as a checkbox this year! Dyke has joined and then left the checkbox list twice now.

Here’s how those three fared among the age groups this year, with very small age samples removed to prevent outliers skewing the graph:

Dyke and tranny have historically been used as slurs against queer people. Both show hints that they are being “reclaimed”, which is to say, as being gender non-conforming becomes more “socially acceptable”, younger people who haven’t had a term used against them violently or as a slur are more comfortable using it for themselves in a positive way. On the contrary, autigender is a new word, created by the community claiming it for itself, and is therefore similarly popular across all age ranges.

On the topic of generating the checkbox list, I’ve written up a (comparatively) short blog post about some terms that people frequently ask me to add to the checkbox list here.

---

Q2: Titles

Titles (with name)

Participants were asked: When someone writes “Dear [your name]” at the top of an addressed letter/email, what title would you want someone to use when writing to you, if any? There were several specific titles to choose from, plus some hypotheticals, no title at all, unknown, I choose on the day depending on how I’m feeling – and a title not listed here, which sends you off to some textboxes to type your preference in. All of these were presented as radio buttons (one answer only) to mimic the “filling in a form” context, and the order was randomised to prevent primacy and recency bias.

Here’s what the top 5 looks like in 2025:

1. No title at all: 42.6% (up 0.4%)
2. Mx: 14.7% (down 2.7%)
3. Mr: 11.0% (down 0.3%)
4. Non-gendered professional, academic, religious, military or nobility title: 8.9% (down 0.3%)
5. Ms: 5.6% (down 0.2%)

These are the same titles in the same order as last year. In fact, the top 5 titles have been the same titles in the same order since 2020.

In general, everyone prefered no title, but younger people were mostly a little more likely than older people to pick gendered titles, and older people were mostly a little more likely than younger people to pick gender-neutral options:

Here’s our graph of ongoing trends:

Our blue line at the top there is no title at all, which is continuing to climb. And Mx is continuing to dip, the gap is still closing between that and Mr in third place. Having said that, Mx is still the most popular specific title. Mr, Ms and Miss haven’t moved since last year, and nor has non-gendered professional, academic, religious, military or nobility title.

I’ve also got a bar graph I can show you, which I think makes a bit more sense if you’re looking into movement of individual titles:

This year saw the addition of a new title to the checkbox list based on the textbox entries from last year. The new title is M, pronounced “em”. Last year, most people entering that title (41.7%) said that it was a gender-inclusive title that anyone of any gender can use. Last year it was typed in by 0.3% of participants, and this year the box was checked by 2.7%. This gives a checkbox multiplier of 9x – more on that later.

3.4% chose a title not listed here, and were taken to a secret write-in title section. This percentage is a little higher than last year. One title was far and away the most common: Mg, which most people entering it said should be pronounced “mage”. In terms of gender it was very close – 38.2% said it could express any nonbinary gender (making it a “gender-exclusive” title), and 36.1% said it was a gender-inclusive title that could be used by anyone.

Mg was entered by half of the people who chose a title not listed here, and by 1.7% of participants overall, which is huge for a write-in title. Its popularity this year seems to be because of a Tumblr blog post that went viral. It was written just over a year ago, on 15th May 2024, by apolloendymion on Tumblr. (Here’s a reblog with more reasoning a couple of days later, and here’s a more neutral blog design to read it on.) At the time of writing it has about 70,000 reblogs, which is a lot. This title has travelled a very long way in a very short time!

Speaking of titles that show up in the Gender Census after going viral on Tumblr, I was surprised to note that Mm (mistrum), coming second last year with 9% of not listed participants, is now down to fourth position with 2.4% of not listed participants.

In fact, the new #2 of write-in titles was Mrs, entered by 0.2% of all participants (96) and 6.3% of not listed participants. It came higher than Ind, a checkbox option that was removed last year. This title is very well established and I’m sure the vast majority of English-speakers will be familiar with it, but just in case: Mrs is pronounced “missus”, and it’s the traditional (feminine) title for a married woman. It has previously been on the checkbox list, but was removed for not being chosen often enough. It does tend to get entered more by the older age groups, presumably because the longer you live, the more likely you are to have gotten married – so perhaps the increasing proportion of people over the age of 30 in the Gender Census is bringing it back?

Which takes us smoothly into…

Choosing the titles (with name) list

Last year I decided to start recording multipliers for these titles so that I can automatically calculate what should be on the titles checkbox list. This year M was added and its multiplier was 9x, which gives us a good starting point. (See the titles multiplier sheet here.)

I adapted the checkbox list choosing system for titles this year, which gives us the following for next year:

• Mg (mage)
• Mr (mister)
• Mrs (missus)
• Ms (məz)
• Mx (məx)
• Gendered professional, academic, religious, military or nobility title
• Non-gendered professional, academic, religious, military or nobility title
• I choose my title on the day depending on how I’m feeling, even for long-term things like bank accounts
• No title at all
• Questioning or unknown
• A title not listed here [and then the participant gets sent to the textbox section]

We lose M and Miss, which get bumped off the list by Mg and Mrs.

• Mg makes it onto the list in every age group between 11 and 45. It’s so “in” this year that it scored higher than everything except no title in 3 of those 7 age groups.
• Mrs is included because it was entered often enough by the 36-40, 46-50 and 51-55 age groups.

Titles (without name)

Participants were asked: When a stranger addresses you and they don’t know your name, what title(s) would you want someone to use, if any? This question is here because of the fairly consistent question from USians, “what’s the alternative to sir/ma’am?”

Participants were able to choose as many options as they wanted, and answer options were presented in a randomised order to prevent primacy and recency bias. The question was optional.

Here’s our international results from this year:

This graph looks really similar to last year, because there has been very little change – mostly under 1% for each title. Here’s our top 5:

1. No title at all: 66.9% (up 2.0%)
2. Friend: 35.9% (up 0.9%)
3. Sir: 32.4% (down 0.8%)
4. Comrade: 23.5% (down 0.6%)
5. Mx: 18.1% (down 2.5%)

The long term trends spreadsheet for titles-without-name can be found here, and it’s still only 3 years old so it’s not very exciting yet. Look how narrow the graph is!

I really like the many-small-bar-graphs format for this kind of thing, because it shows how consistent things tend to be year-on-year:

Comrade was so low in 2023 because it wasn’t a checkbox yet, but aside from that it’s very close and the order hasn’t changed much. (Maybe that’s not so strange, since it’s only been three years and the sample is so big.)

In both 2023 and 2024, dude got 2.1% in the textboxes. This year it was top of the textboxes again with 2.0%. This is impressively consistent, dudes.

I was curious about differences across the two broad age groups, and so…

Once again, younger people were a little more likely than older people to favour gendered titles.

No title at all is so far ahead, with two thirds of respondents preferring people simply say “excuse me” or “hey you”, judging by the textbox entries. But even so, cultural attachment to sir/ma’am varies from country to country and even varies a lot regionally within countries. Whether or not you can simply opt out without seeming rude will be extremely context dependent. Those of you looking for another option can make a copy of the results spreadsheet and mess around with it to show only results for your country, which may provide more relevant information for your situation.

Having said that, here are the options that got over 1% in the textboxes:

1. dude: 2.0%
2. mate: 1.2%
3. buddy: 1.2%
4. mage: 1.0% (as in Mg, the new formal title – see above.)

I have no plans to implement a checkbox choosing method at the moment, because there’s not much movement in the checkbox options and none of the top textbox entries are particularly formal.

---

Q3: Pronouns

Our opening pronoun question is: 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 15 randomised checkboxes as they wanted. Checkboxes included 11 common and uncommon pronoun sets, plus non-pronoun options like “any”, “avoid pronouns”, and “questioning or unknown”.

One of the checkboxes is “a pronoun set not listed here”, and if you choose that it takes you to a second section where you can enter up to five new pronoun sets in detail.

All questions about pronouns were optional.

Checkboxes

Here’s our usual bar graph:

The order of the top 5 pronouns (or lack thereof) has been the same since 2022, with only very slight changes to the percentages:

1. They – they/them/their/theirs/themself: 75.0% (down 0.5%)
2. He – he/him/his/his/himself: 40.6% (down 1.4%)
3. She – she/her/her/hers/herself: 34.1% (down 1.9%)
4. It – it/it/its/its/itself: 22.8% (up 2.5%)
5. Avoid pronouns / use name as pronoun: 14.4% (up 0.5%)

8.9% of participants didn’t want to be called they, he or she this year, which is about the same as last year.

Pronoun trends since 2015 are looking fairly consistent in the upper reaches:

This graph contains only pronouns for which we have at least the last 3 years of data, hence the absence of the pronouns that were new this year.

That blue line at the top is they/them, and the lines below it are he/him (red) and she/her (yellow). They/them has been fairly consistent over time, fluctuating by only about 6% since 2015, and never dipping below 74%. The fourth line, dark green, is it/it, which has been steadily climbing since around 2021, and it climbs again this year.

Several smaller bar graphs show the trends for the less popular pronoun sets more clearly:

Here’s the top 10 for each age group:

I’ve mentioned before the tendency for older age groups to choose/enter fewer terms than younger age groups, and that is the case with pronouns too. Almost everything in the aged 31 and over top 10 was selected less often than in the aged 30 and under top 10, with two exceptions. They/them and she/her are both chosen more often among older age groups, for reasons unknown.

I am particularly interested in they/them as it’s the most popular pronoun set, so here’s some more investigating. We don’t have many years of data to go on here, but the gap between they/them and age group is growing. Last year the 31 and over group favoured they/them 5.6 percentage points more than the 30 and under group, and this year the difference has risen to 6.2 percentage points:

I made this graph to try to show how the older age groups are getting into they/them and the younger age groups are less likely to pick it than previous years, like a wave motion:

Even so, a preference for they/them is higher overall and more even across age groups than it used to be.

Despite two thirds of participants being new to the survey this year, the “number of pronouns each, split by age group” graph is again virtually identical to previous years:

[Click here for the 2023 data on Google Sheets. Click here for the 2024 data on Google Sheets.]

Most people aged 30-and-under chose two pronoun sets, and most people 31-and-over chose one. The average (mean) number of pronoun sets chosen overall has been 2.2 for three years running now:

Neopronouns

Here’s our checkbox neopronouns this year:

1. Xe – xe/xem/xyr/xyrs/xemself: 8.8% (no change)
2. Fae – fae/faer/faer/faers/faeself: 6.2% (no change)
3. Ze/zir – ze/zir/zir/zirs/zirself: 5.7% (up 0.7%)
4. Elverson – ey/em/eir/eirs/emself: 4.4% (up 0.1%)
5. Spivak – e/em/eir/eirs/emself: 3.9% (up 0.3%)
6. Star – star/star/stars/stars/starself: 3.5% (new this year)
7. Thon – thon/thon/thons/thons/thonself: 2.0% (new this year)

None of them were selected more often than any of the established pronouns (they/them, he/him, she/her or it/it), but 32.9% selected or entered any neopronoun. A unique neopronoun set was typed into the textboxes for every 19 respondents, which is about the same as last year.

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.

9.0% of participants entered at least one neopronoun set this year, which is a bit higher than last year. 2,259 neopronoun sets were entered in total (based on subject/object/reflexive only), of which 499 were entered more than once. Participants aged 30 or younger were a little over twice as likely to type in a neopronoun than those aged 31 or older, which is consistent with previous years.

The top two write-in neopronouns were the two that were removed from the checkbox list following last year’s survey:

1. ae/aer/aer/aers/aerself (singular verbs) – when counted by the first two forms only (subject and object, in this case ae/aer), it was entered by 0.53% of respondents, which is actually pretty high for a textbox pronoun. Last year as a checkbox it got 3.8%.
2. ze/hir/hir/hirs/hirself (singular verbs) – when counted by the first two forms only, it was entered by 0.46% of participants. Last year as a checkbox it got 3.2%.

Here’s ae/aer in use, to give you a feel for it:

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. I’ll take it to aer.”

Lack of nuance in pronoun data

In 2023 I took out some checkbox options from the pronoun list that were for how people should select which pronouns to use, because they didn’t fit the scope of the question. In 2024 I attempted to add a question to gather information about how people want others to choose which pronouns to use. (For example, whether people should change pronouns randomly, or whether pronouns change depending on specific conditions.) I had to abandon this question because preferences were so nuanced that I was unable to present it coherently in the report. I concluded that people’s preferences in this area are simply extremely individual.

This year I did get a fair number of people asking (in the feedback box, by email and otherwise) how to answer the pronoun question if their pronouns vary. I had to reply to each person and say, “check the boxes for the pronouns you are comfortable for people to use for you, even if those pronouns are only sometimes appropriate.” It seems that for many people, their answers feel incomplete if they’re not able to explain the nuance that the pronoun question isn’t able to collect. I will, as always, keep thinking about this. At the moment I think the answer is that the Gender Census survey simply doesn’t and can’t collect that kind of nuance, and if it is valuable enough someone else may have to do that further research. (And perhaps, if a lot of people are asking me for advice about it when filling in the survey, I should add some help text. But there is already so much help text!)

Anecdotally, the most common reason people raised this issue this year was because they prefer “mirror” pronouns. This means that they want whoever is speaking about them to refer to them with the pronouns that the speaker prefers for themself. So if I, a they/them person, were to speak to Andréa (she/her) about someone who prefers “mirror” pronouns, I would refer to them using they/them. And Andréa would respond and refer to them with she/her pronouns. (I asked online how a person who prefers mirror pronouns would refer to someone else who also prefers mirror pronouns, and someone replied to say that people with mirror pronouns often have a “fallback” pronoun.)

Pronoun checkbox selection method

We’ve got data for four pronoun sets being moved from textboxes to checkboxes now, and they give us a median multiplier of 8.7. (Pronoun multiplier data can be found here.)

Applying this to the write-ins, taking the same number of top pronouns from each age category, and excluding anything entered only once in that age category, gives us a list of pronouns either 14 or 16 checkboxes long. This is unsatisfying because the goal for the length of the list is 15, but that kind of thing happens with the identity words checkbox list sometimes too. I will let the checkbox list be 16 items long this time, because…….. science?

• Any
• They – they/them/their/theirs/themself (for referring to an individual, e.g. “they are a writer”)
• He – he/him/his/his/himself
• She – she/her/her/hers/herself
• It – it/it/its/its/itself
• Fae – fae/faer/faer/faers/faeself
• Person – person/per/per/pers/perself
• Sie/hir – sie/hir/hir/hirs/hirself
• Xe – xe/xem/xyr/xyrs/xemself
• Ze/hir – ze/hir/hir/hirs/hirself
• Ze/zir – ze/zir/zir/zirs/zirself
• Avoid pronouns / use name as pronoun
• Questioning or unknown
• A pronoun set not listed here

Here’s some info about the new sets and how they come to be here:

• Person – person/per/per/pers/perself: On the checkbox list because it was written in by only two people in the 61-65 age group!
• Sie/hir – sie/hir/hir/hirs/hirself: This set goes back to the 1990s of the internet, so I am very glad to add this set and get some up-to-date stats on it. Again, on the checkbox list due to being written in by only two people in the 61-65 age group.
• Ze/hir – ze/hir/hir/hirs/hirself: This has been on the checkbox list before, but got usurped by some nounself pronouns. I’m glad to see them again, because they are well-established, going back to the early 2000s. It’s on the list thanks to two people, this time from the 51-55 age group.

Lost from the list next year will be two very well-established sets, and two sets that were just added this year:

• Spivak – e/em/eir/eirs/emself: Created in 1982 by Michael Spivak, and popularised by LambdaMOO. Today I learned that it was originally capitalised the way “I” (first person pronoun) is capitalised, but I lowercase everything to count it so that may get lost along the way. It was chosen by 3.9% of participants this year.
• Elverson – ey/em/eir/eirs/emself: Often mistaken for Spivak, with only one letter difference. It’s actually way older, from 1975! And named after the creator, Christine M Elverson. This checkbox was selected by 4.4% of participants this year.
• Thon – thon/thon/thons/thons/thonself: Added last year, and one of the oldest on record (1858, wowza). Because of that it’s a bit famous, so I wonder if it’ll be back. When it became a checkbox it was selected 7.1x as often, with 2% this year.
• Star – star/star/stars/stars/starself: Added last year, this is the first “nounself” pronoun set to make it into the checkbox list, and I’m sure it won’t be the last. (I wonder if person/per counts as a nounself pronoun?) When star/star was added as a checkbox it was chosen 10.3x as often, by 3.5% of participants.

If it seems a bit strange to have popular and decades-old neopronouns be pushed off the list and replaced by pronouns that only two people in one age group wrote in, I get that! It is weird, isn’t it?? But I think maybe it’s okay. This system is intended to ensure that checkbox lists contain words that are familiar to a wide range of ages. That’s because I don’t want anyone to open the survey, see nothing relatable, and close it again. Very early on in the survey I got feedback on social media from older people telling me that they didn’t see themselves in the answer options and concluded that the survey just wasn’t for them, and so they closed the tab. If a group isn’t represented then members of that group won’t participate, and so the group continues to not be represented, right? (I worry about this being affected by demographics other than age, too, but I haven’t worked out how to deal with that yet.)

The older age groups are less likely to use neopronouns, and those groups are also smaller, so anything that does get typed in carries a lot of weight. My only question is whether a pronoun set should be entered by more people within an age group before it’s considered for inclusion, and I will keep reconsidering that. But for the time being I’d say that the system is working as intended, because the sets we’re adding are established neopronouns that I think people in older age groups are more likely to recognise. And over the last several years, more and more people over the age of 30 are participating. I’m not sure if it’s a direct-cause-and-effect situation, and chances are we’ll never know for sure, but I am very glad of it.

---

Q4: Family/relationship terms

In 2023 I started working towards a many-years-long journey of investigating preferences in terms that we use to describe our long term relationships to each other. That’s words like aunt/uncle, girlfriend/boyfriend, etc.

Each year I ask two optional questions:

1. The textbox question. This will ask about terms for a particular relationship for the first time, with several textboxes to type your answer. No checkboxes.
2. The checkbox question. This will be a repeat of last year’s textbox question, with answers provided as checkboxes that are informed by last year’s answers. Textboxes also provided.

In this way we will gradually build up data about language for various social/family relationships in a reasonably fair way, without adding 10 new questions for people to get through. The theory is that year one gathers lots of information, and year two tidies it all up a bit – so we get some reliable data for one relationship type per year.

The finalised lists of terms will live on this spreadsheet.

For both of these questions, since I have to avoid putting words that people might choose in the question itself (to avoid prompting people’s answers), I have to describe the relationship in quite a roundabout way. This was causing some confusion towards the start of the survey, so I added some handy diagrams a few hours in, which seemed to help a lot. Since the questions’ exact wording didn’t change and it was after such a short amount of time, I don’t think that has affected the results much.

The textbox question: sibling’s child (niece/nephew)

The survey asked, Imagine your parent’s sibling is introducing you to someone in English. Which word(s) would they ideally use to describe your family relationship to them? E.g. “This is [name], my ____________.” (We’re going for niece/nephew equivalents.)

This is the first time this question has been asked. Answers were in the form of five textboxes, and the question was optional. Participants were also given the following additional guidance:

• If your parent(s) don’t have siblings, or if they do but they don’t speak English, answer hypothetically.
• It’s okay to enter non-English words that you would like your parent(s)’ sibling(s) to use while speaking English.

Here’s our top 10:

• nephew: 16.5% [7,122]
• nibling or nibbling: 14.9%
• niece or neice: 14.5%
• [sibling’s] child: 9.0%
• [sibling’s] kid: 7.8%
• relative: 2.6%
• [my] family or family member: 1.5%
• cousin: 1.0%
• kin: 0.8%
• niephew or neiphew: 0.4%

Here they are as a bar graph:

The checkbox question: sibling

The survey asked, Imagine another child of your parents is introducing you to someone in English. Which word(s) would they ideally use to describe your family relationship to them? E.g. “This is [name], my ____________.” It then provided 12 terms, plus questioning or unknown. These options were informed by the textbox responses to this question last year. The question was optional, participants could choose as many as they wanted, and there were five textboxes provided below for those whose words were not all on the checkbox list.

One point of confusion was the inclusion of the term nibling in the checkbox list. It’s a term coined in the 1950s as a gender-neutral alternative to niece/nephew (like, niece/nephew + sibling). This was entered often enough in the textboxes last year (0.2%) that it qualified for the checkboxes this year, and I am not sure why.

Option	This year	Last year	Increase
sibling	83.5% [35,998]	67.2% [32,678]	1.2x
sister	27.2%	20.5%	1.3x
brother	26.8%	21.2%	1.3x
family	17.2%	0.2%	86.0x
sib	17.1%	3.3%	5.2x
bro	16.1%	1.3%	12.4x
relative	13.3%	0.3%	44.3x
kin	12.5%	0.3%	41.7x
sis	10.8%	1.0%	10.8x
friend	6.2%	0.2%	31.0x
questioning or unknown	3.8%
nibling	3.8%	0.2%	19.0x
sibster	3.4%	0.1%	34.0x

Here’s the top 10 as a bar graph:

It’s interesting that the percentages for “brother” and “sister” are so close, and because people could choose any number of terms I was curious about how many people chose just one and how many people chose both. Here’s the answer:

• Chose only “sister”: 21.4% (78% of people choosing “sister”)
• Chose only “brother”: 21.1% (79% of people choosing “brother”)
• Chose both “brother” and “sister”: 5.8%

So it’s still incredibly close!

As usual, there are cultural variations in terms people use for family members. If you’re investigating gender-neutral terms and you want to prioritise regionally relevant terms, I recommend downloading the spreadsheet of results and cherry-picking the data for your country (or whichever specific country you have in mind).

---

Meta

The last page of the survey asks participants for various bits of unrelated information, some of which help me to design a better survey and promote it effectively (age, feedback, referrer), and some of which make the data more useful for others (age, country).

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

Many graphs and summaries based on these questions have been available throughout the survey’s duration, and you can view this year’s public participation sheet here.

The top 5 countries represented by number of participants were:

1. United States: 22,740 [52.8%]
2. United Kingdom: 4,695
3. Canada: 2,824
4. Germany: 2,401
5. Australia: 1,987

The top 5 this year are the same countries in the same order as last year. Every year I work hard to make sure places outside of the USA are as well-represented as possible; the USA percentage is a little lower this year, but still over half. 69 countries got 10 or more responses, and any country with under 10 responses (of which there were 62) has had the country redacted to ensure participants’ privacy and safety.

Here it is as a bar graph:

Collecting participants’ country of residence allows researchers (and anyone else) to download the spreadsheet of results and process the data for just one country (or a handful of countries). I want to enable people to make a report based on just one country, for two main reasons:

• National projects or forms can find out which identities, pronouns, titles, etc. are most important to include.
• Nonbinary and otherwise gender-divergent people can find common language for themselves easily when coming out.

As usual, Tumblr was the main source of participants at 43%. This is a little higher than last year, and I think that might be because a few people “blazed” (paid to promote) some blog posts while the survey was open – in particular, one post that was appealing for people from less well represented countries to participate and share the survey, which worked quite well! Thank you, kind Tumblr people. <3

There were more people on the mailing list than last year, and the percentage coming from there this year was 9.1%, which is 1.8 percentage points higher than last year – good stuff! Also, there were some weird technical issues on my IP address that might have interfered with mailing list emails getting through, so me and Andréa are going to try setting something up that might fix that for next year. (Andréa is endlessly patient with my shenanigans and there is always something technical that neither of us can predict! If you wish to express your support for her support, please feel free to visit her online wishlist.)

Since so many people come from Tumblr and the USA, I’ve been trying to diversify my social media presence in the hope of reaching more countries. There are so many places I’m posting about the Gender Census now that I’ve got a whole page just for that! (Check it out here if you want to make sure you don’t miss next year’s survey.) The Gender Census website used to be hosted on Tumblr and I have always been better at posting and promoting there, so it makes sense that I would have a big followership there. It will probably take a few years to get up to speed with other sites/platforms and build a following, but it’s important to me that this survey reaches a diverse range of people in many countries, and it will 100% be worth it.

For a few years I’ve been grouping participants very vaguely by “30 and under” and “31 and over” as a rough guide to how well the survey is reaching participants of all ages. Last year it was up on the previous by 6 percentage points, and this year it was up again by another 3 to 24.3%. This year I got around to calculating the average age of participants based on the year groups, and for 2025 that’s 26.2. Then I added that calculation to previous years’ spreadsheets, and plotted them:

I’m so glad to see the average age rising! Average age seems to be going up by about half a year per annual survey.

As we know, getting the word out to new people is really important. Special shout-outs to:

• Tuck Woodstock, who mentioned the Gender Census on their podcast Gender Reveal, which brought a good flurry of people in
• Navi, who kindly stuck a poster up in a unisex toilet in Austria
• The mystery pixel artist who left the Gender Census URL written in Australia on wplace.live

Imaginitive sharing like this really does make a big difference, bringing in more different kinds of people, and I love it so much. Additionally, “local support group” was up 24% this year (291 people), even though participation overall was down 11%, so that is really positive.

About a third of people had done the survey before, which is a little higher than last year:

---

Some of the questions this survey seeks to answer

• What should the third gender option on forms be called? – Nonbinary is the most commonly used, holding steady at around 60%. It’s far from universal, but there are no other comparably popular words in the top 10 that are unambiguous in meaning. I would recommend the third gender option on forms be called nonbinary.
• Is there a standard neutral title yet? – No. Internationally, participants generally prefer to not have any title at all. 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 standard gender-neutral way to respectfully address superiors and strangers (sir/ma’am) yet? With only three years of data it’s too soon to say for sure. Avoiding sir/ma’am altogether has been the most popular suggestion so far by a very long way for three years running, and the next most popular is friend. Counterintuitively, older people are more likely to prefer avoiding sir/ma’am.
• 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 around 75% of participants since the first survey over a decade ago. But that leaves one quarter of us who don’t want to be called they, and 9% of us consistently don’t like to be called he, she or they, so it’s often good to check.
• Is there any consensus on a “nonbinary pronoun”? – No. About 1 in 11 participants were happy with xe/xem/xyr/xyrs/xemself (singular verbs), but its popularity can fluctuate.

---

This year in review

Technical issues. The survey shortlink and the website were blocked by malware plugins because someone else on the same IP as gendercensus.com is, well, doing a lot of malware. I contacted my webhost about it several times, but I think maybe webhosts just don’t care when people are doing malware? So, me and my tech witch Andréa have a plan to deal with that once I’m back in the inter-survey lull.

Increased social media presence. Much as I love Tumblr’s receptive audience and super-effective reblog feature, I need to put more promotional work into other social networks. Over half of participants are from the USA, and I would love to boost other countries’ representation in the results.

Crowdfunding. Patreon supporters have covered costs this year, with some left over for me to pay Andréa for her sourcery, which I really appreciate! There are a lot of significant costs involved with this project, and it is a pleasure to be able to pay Andréa for her work. You can visit the Gender Census Patreon project page here, and you can buy Andréa fancy things to say thank you here.

Limitations

Any research has limitations, and this is no exception.

• I am an amateur. Ultimately, I have no relevant training, professional experience or academic qualifications of any kind, and I work (mostly) alone. My work has not undergone ethics review or peer review. (I am happy to do it anyway, because I hope it can inform and support people who can publish research that is more scientifically robust.)
• Statistical data lacks depth and nuance. We have some basic information about how many people like particular terms, but often people have a lot of contextual stuff that affects their language preferences in ways this survey doesn’t capture. For example, many people have pronouns they prefer at work and different pronouns they prefer with close friends, or they want people to use pronouns that match their gender presentation, or they only feel comfortable with other queer people using it/it pronouns for them, or you can call them anything except [specific gendered word they strongly dislike].
• Participants are self-selecting via social media. The survey is spread almost entirely by word-of-mouth, mainly online via social media. Almost half of respondents came from Tumblr. All of the above shapes the demographic profile of the sample, which is overwhelmingly young and USA-based. Results are heavily weighted to reflect white, affluent, Western, young, and Extremely Online perspectives.
• Checkboxes are chosen more often. This means that textbox entries should be considered separately from checkbox entries for analysis.

What I’ll do differently next year

• I’ll try to restore the “in an ideal world” vibe to the titles-with-name question, as a couple of people said in the feedback box that they struggled with the ambiguity of the question wording. (It changed three surveys ago when I added the titles-without-name question.)
• I’ll include diagrams with the family/relationship terms question from the start.

---

Closing thoughts

I just really love doing this project, and I know I always say this but I am very grateful to everyone who participates and supports in other ways, because I really love diving in and swimming around in it all. I really hope it helps people, not just in terms of research about us, but also in direct and indirect real-world impacts like helping us to feel seen and accepted and celebrated and connected.

I really appreciate the support and brains of Andréa and Avery.

---

Corrections

• 2025-09-14: Removed reference to Patreon forcing me off the “per creation” funding model in November, because someone just told me that there is no longer a “per creation → per month” changeover deadline, woohoo!
• 2025-09-21: This report used to say that Mg was invented by the person whose blog post went viral, but now it just says that it became popular because that person’s blog post went viral.
• 2026-02-15:
  • The original report and spreadsheet used an incorrect total to calculate the percentages for “title without name” age group comparisons. The proportions are unchanged, but the spreadsheet table and the graph here (and its alt text) have now been corrected.
  • The original report and spreadsheet used an incorrect total to calculate the ‘neither he, she, nor they’, ‘only “a pronoun set not listed here”‘ and ‘selected/entered any checkbox neopronoun’ percentages. This has now been corrected here and in the spreadsheet.
  • In the pronoun section, the bullet point on Spivak pronouns said that the pronoun set was coined by Spivak in 1990, which was in a book called “The Joy of TeX”, but the book was actually published in 1982. I’ve fixed this in the report, and pointed the link to a more reliable source. This bullet point also previously said that the pronoun set is typically capitalised the way “I” is capitalised, but current capitalisation habits are varied and unquantified, so I changed it to say that the pronoun set was originally capitalised.

---

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, Facebook, Bluesky, Instagram, the Fediverse, or the mailing list – or somewhere else! Alternatively, you could buy something from the shop (zines, fidgets and knitwear made by my own hands) or take a look at my Amazon wishlist. 🍫

---

Links to files

• Unprocessed responses
• Processed responses – and I should warn you, this spreadsheet is LARGE (50 sheet-tabs) and may take a long time to sort itself out. Open it and then go and put the kettle on.
• All identities by all age groups spreadsheet
• Family/relationship words spreadsheet
• 2013-2025 trends

---

2025-09-14