Knowingless

Share this post

Which gender groups are more similar to each other?

aella.substack.com

Which gender groups are more similar to each other?

survey results

Aella
Jan 30
23
5
Share this post

Which gender groups are more similar to each other?

aella.substack.com

Alrighty you know the drill - my survey is at 500,000 people, with ~6,500 transwomen and ~17,000 transmen. Breakdown of the demographics who took it here, survey design methodology here, no you dont need a random sample jfc here, Raw averages here,

And as usual - please check my work. You are my peers; please review! If I got anything wrong I will update this blog post with a correction.

I asked about a bunch of stuff - big5 personality, childhood, mental illnesses, but mostly stuff related to sex - porn use and fetishes. I asked about age various fetishes began, about BMI, IQ, attachment style, where they got to the survey from, if they’re married, how ashamed they feel, how many sexual partners, etc. Overall there were around 1,100 datapoints gathered from this survey per person.

Knowingless is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

What happens when I check to see whose answers are more similar to whose, across the entire survey?

I took cis men’s average score for every numerical answer on the survey, and also cis women’s, and trans people’s, etc. I then filled in the gaps with averages and min-max normalization. Then I checked the correlation between the total answer sets for each group to all of the questions, which I figured should be a good proxy for “whose answers were more similar to whose”.

I excluded a small handful of questions about your gender identity, and questions only asked to one gender category. I did not do any thorough checking throughout the questions, so I might have missed some that would have been better off removed. But there were over 1k questions, so my guess is a few shouldn’t make a big difference to the final results?

Total correlation (1093 questions) (higher number means more similar):

Spreadsheet here, raw scores are in the second tab

Some quick observations:

In general, non-cis males (amab enbies and transwomen) had comparatively dissimilar answers with both cis women (.93) - and cis men (.94). Non-cis females (afab enbies and transmen), on the other hand, were much closer to cis women (.97) and much less similar to cis men (.90).

Non-cis answers tend to be similar to each other; in general, any given non-cis group will correlate more strongly with any other non-cis group, regardless of gender or birth sex. The only exception here is female enbies, who were more similar to cis women.

Enbies are very similar to their corresponding transgender group; male enbies answered about the same as transwomen, and female enbies about the same as transmen.

Is this rigorous?

Not sure what you mean by that, but this is more a “curious datapoint”; I didn’t go through and carefully select which questions I was testing for correlation. Answers might be thrown off if I happened to ask in much more detail about a subcategory of thing that genders are particularly different in (more questions would weight that section heavier overall).

I did kind of want to try this on a large question selection that wasn’t designed with testing gender differences like this in mind, to sort of approximate a general “If you wanna get an overview of someone with an emphasis on sexual preferences” similarity measure. Like, if you’re just gonna sit there and try to come up with a survey to get a really good grasp on someone’s personality and sexuality first, as opposed to designing it for gender testing. My guess is that a large portion of how we process gender isn’t conscious, and that something about testing a lot of things that aren’t, on their surface, obviously connected to gender, might get a more holistic snapshot? Like it’s something actually closer to a ‘random sample’ of question selection.

but wait


Speaking of random sample of question selection, I just realized I have my other dataset from my Chaos Survey! That’s a survey optimized for the *most* random set of questions! Questions were generated by asking people to submit any question at all, as long as it was answerable on a 1-7 spectrum. I did almost no filtration on which questions I included, and had people answer the questions entirely at random. Similarly to the kink survey, ended up with around 1100 questions.

(unlike the kink survey though, most people completed a random 10% of the question set; the kink survey had many questions presented as individual checkboxes, so the actual number of questions they went through was around 300)

Let’s check the correlations on the chaos survey!

I’m removing a bunch of questions that people generally skipped; these typically were ones that referred to obscure names or topics, and this narrowed the questions down to 657. I did this before looking at any results. On average, around 1600 people answered each question. Because sample was so low, I’m not going to look at enby and trans people as separate categories, but rather as just ‘non-cis.’

Cis male group was ~1270, cis female group was ~165, non-cis male group was ~84, and non-cis female group was ~37. Still low! And the population taking this survey was significantly different than the population taking my kink survey. But I’m curious to see if we’ll see the same similarity patterns emerge?

Raw data here in the raw tab where you can also see all the insane questions

Squinting, I think the trends seem to coincide? Non-cis males seem to be most similar to non-cis females. and almost equally dissimilar to both cis men and women. Non-cis females are more similar to cis women, cis men are the most similar to cis women, etc.

I’m still eyeballing it a little suspiciously - a total non-cis group of 121 answers per question is way lower than I’m used to working with! But this still makes me update slightly more in favor of the hypothesis that my original kink survey is getting at something pretty broad and possibly replicable in different domains.

Summary and Takeaways

In general, my data seems to suggest (and please always check my work yourself to see if you agree with my interpretation):

  1. In general, non-cis people are more similar to other non-cis people - regardless of gender or sex - than they are to cis people.

  2. This rule holds more true for biological males than biological females. Non-cis males are about equally dissimilar to men and women, but non-cis females are more similar to women than they are men

  3. Cis men and transmen had the most dissimilar scores of any group comparisons measured

  4. The highest similarity measured across groups was between nonbinary people and trans people of the corresponding biological sex.

Knowingless is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

5
Share this post

Which gender groups are more similar to each other?

aella.substack.com
5 Comments
Mark Carnegie
Jan 30

I think what you have done is fascinating overall. However I felt this post was a bit of a tangle and didn’t get the ‘so what’. Could you please think about a summary that either says: why what I found matters or why what I found interested me.

Thanks

Expand full comment
Reply
dianna
Jan 30

This is interesting! As a trans woman (whose data points are in your survey) it’s curious to know that as a group we’re really similar to AMAB Enbies across all of your questions. I suppose in a lot of ways that makes some sense.

Getting into the weeds, some of the really narrow numerical differences make the results hard to interpret though. Like when the lowest correlation you measure is 0.89 across such a huge question set, that’s hard to get an intuitive sense for what that means? Or what 0.99 across the data set really means?

What might be super interesting is to take that same matrix, but for certain sub-categories of questions or specific questions, measure *how different* the distribution of responses are. Like take a smattering of questions you think are interesting to consider, and then for each one look at how the distributions of answers fell across each gender and put a number to how distinct (or not) they are.

This might help the data interpretation be a little clearer, versus looking at calculated averages that may have very subtle differences.

From my cursory understanding of statistics and some wikipedia searching, the Bhattacharyya distance (or the coefficient if you think 0 to 1 is more intuitive than 0 to infinity) is useful here since your answers are already nicely binned into 0--7. It should be fairly easy to calculate as well, I think? And it spits out a nice number at the end.

You would double check your implementation here by looking at the question for self-identified gender. The diagonal on your matrix would be zeros and all other cells would be infinity. The interpretation is just that the distributions into buckets (M, F, MtF, FtM, Enby AMAB, Enby AFAB) are either perfectly identical or perfectly different, by definition.

Expand full comment
Reply
3 more comments…
TopNewCommunity

No posts

Ready for more?

© 2023 Aella
Privacy ∙ Terms ∙ Collection notice
Start WritingGet the app
Substack is the home for great writing