30 Comments

Looks like high BMI correlates with a lot of people things related to class (income and spanking were especially noticeable)

Expand full comment

I always wonder why not showing a measure of spread (standard deviation, quantile in the plots). The graphs look very interesting but I am always dubious to drawing conclusions from average values, without understanding the spread and asymmetry of the empirical distribution.

Some shaded areas above and below of each line describing say, quantiles, remain very intuitive for people to understand and convey a lot more information than just the mean.

Expand full comment

That correlation between social status/wealth upbringing and responsibility upbringing though... oof!

Expand full comment

So women start thinking they're less attractive than others around BMI of 24 or 25, while men don't start thinking they're less attractive than others until obesity.

Expand full comment

Looks more like women think they are more attractive than others when they are really skinny and then rate themselves lower with every BMI point, while men rate themselves as less attractive as soon as they get overweight, around BMI 24-25, and then rate themselves as less attractive for each BMI point.

Expand full comment

looks like the "this is the graph for only people aged 18-22" graph got mislabeled if i'm reading right

these are fascinating though :0 am i seeing a strong correlation with men's mental health and hovering around the upper end of the healthy(?) BMI range?

Expand full comment

Sexual assault in childhood was fairly interesting but not sure if that just correlates to socioeconomic status growing up

Expand full comment

Very interesting presentation of data. Especially that recurring dip/bulge at the upper ranges of normal weight on the male side is intriguing.

The graph below the one called "Self rated attractiveness compared to others of the same demographic" has a confusing title. Shouldn't it also be called "Self rated attractiveness compared to others of the same demographic"?

Expand full comment

error bars

Expand full comment

n=570,000

Expand full comment

that implies a relative reduction in variance compared to smaller sample sizes; there’s still a variance.

Expand full comment

I don't know enough about statistics to know for sure, but elsewhere Aella has said the error bars would be so small one can completely ignore them.

Expand full comment

Then why are there weird notches in the graphs?

Expand full comment

The line graphs are weird and bumpy since she's visualizing discrete data points and doing connect-the-dots. It's not a line of best fit, it's just a series of straight lines between each data point.

Expand full comment
Aug 26, 2023·edited Aug 26, 2023

If the error bars are "too small to plot" then why are there clear peaks in "how many siblings did you grow up with?" and "I feel little concern for others"? Those look like noise to me...

Expand full comment

Aella is right! Her cleverness in statistical realms is a big reason why I read knowingless.

Expand full comment

Great results, I'd love to know how the non-Western world sees things

Expand full comment

Why is BMI the x-axis for every graph except polyamory? Kinda confusing

Expand full comment
Aug 24, 2023·edited Aug 24, 2023

It's the independent variable, I guess. But why polyamory's different, ask Aella. ;)

Expand full comment

Regarding the economic/social liberalness/conservativeness, it would be interesting to see per quadrant stats.

Expand full comment

An old researcher by the name of Sheldon came up with the somatotypes still in use today as Ectomorph, Mesomorph, and Endomorph.

Sheldon claimed that the lean ectomorphs were cerebrotonic - intellectual, private, intense, and (oddly) hypersexual. His corpulent endomorphs were viscerotonic - relaxed, good-humored, unambitious, and sociable in a friendly, submissive way. His musclebound, heavy-boned mesomorphs were somatotonic - restless, assertive, and not always good at knowing themselves.

Although BMI doesn't really distinguish mesomorphs well from the others, looking at this it occurs to me that the extreme ends of BMI are not bad proxies for Ectomorphs and Endomorphs. *Maybe* mesomorphs are in the middle, though that really needn't be the case. How do the personality items work?

"I am quiet around strangers?" Ectomorphs rather are.

"I sympathize with others' feelings?" Endomorphs do, a little.

"I have excellent ideas?" Ectomorphs *don't*.

"I shirk my duties?" Endomorphs do. (So do ectomorphs, I guess)

"I am relaxed most of the time?" Endomorphs *aren't.*

"I am high-powered, driven, successful?" Endomorphs and Ectomorphs aren't.

"I feel little concern for others?" Endomorphs do feel concern for others.

"I worry about things?" Endomorphs worry *more*

Overall not really a replication of any of Sheldon's ideas. Overall the low-mid BMI's seem confident and successful, while higher-BMI individuals just look to be higher in Emotionality (https://hexaco.org/scaledescriptions); fat is feminine.

Expand full comment
Oct 7, 2023·edited Oct 7, 2023

I got the impression Sheldon's stuff had been ditched with psychoanalysis, phlogiston, and alchemy. You could make the argument he made it before mass availability of high-calorie processed food, turning weight into a huge negative Conscientiousness indicator.

Also note a mid BMI could be muscular or just somewhat fat; there's no indicator of muscularity.

Expand full comment

The comparison between Sheldon's ideas and psychoanalysis etc. isn't unwarranted. But scientifically speaking Sheldon comes out looking somewhat better than Freud or Ge Hong, not necessarily because his ideas were on the mark, but because he developed objective methods for testing them using real, honest-to-goodness *numbers,* by golly:

http://www.innerexplorations.com/psytext/shel.htm

Better known researchers like Hans Eysenck also actually worked in Sheldon's tradition, largely by talking past Sheldon, however (who I understand was politically smelly) and focusing on the two factors of 1. body size and 2. Leptomorphy vs Eurymorphy. See https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2236697/pdf/jroyalcgprac00376-0053.pdf for an example. Though these papers aren't generally available online, they found a few modest correlations between these factors and measurable traits like IQ.

Unfortunately psychoanalysis is all over the place despite some fairly strong empirical refutations, while body type continues to show a handful of modest but significant correlations with psychology. For example, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3754851/ says about what you say it should: BMI is a small negative Conscientiousness factor.

Expand full comment
Oct 8, 2023·edited Oct 8, 2023

Eysenck was politically smelly too later on. Interestingly, he appears to have been an example of an antifascist racist (by modern standards)--he believed in the correlation of race with intelligence, but opposed the Nazis! (He was a quarter Jewish--his grandma had converted but died in a concentration camp.) He got into a bunch of other dubious ideas like the cancer-prone personality.

https://en.wikipedia.org/wiki/Hans_Eysenck

The Wikipedia article's been slanted left, of course, but from what little I can tell he was following the data and far-right people picked up on it.

As for Sheldon, yeah, I'd guess athletic people work out more and people with poor impulse control eat more and neurotic people with good impulse control do neither, so I would think his types had some validity at least at the point of development. The difference is now everyone looks more endomorphic so things might be different.

Expand full comment
Aug 23, 2023·edited Aug 23, 2023

How many people are in these these various "bins" of BMI (not sure the technical term)? You have some funky bumpiness around the 36-40 range, I'm just wondering if these are relatively small groups of data.

Expand full comment

Yeah, I had very similar feelings. I don't like drawing anywhere near general conclusions from this data given sampling bias, unless you're just drawing conclusions about specifically the type of people you're sampling (young white women who use tiktok and listen to kink podcast type of stuff), as well as the spread for people in each BMI category. I think the BMI ranges could probably have been combined a bit into broader categories (like <18, 18-22, 22-26, etc etc). Also, the graphs not being normalized on the y axis is somewhat infuriating lol

Expand full comment

Whoa, the change in male sexual submissiveness from low to mid BMI is neat and unexpected!

Expand full comment

Skinny subby femboys?

Expand full comment

“Rapist level” made me laugh.

Expand full comment