Here's an article about a study where people of the opposite sex rated each other's attractiveness and where the rating average and distribution are much closer than on OkCupid, even though men are still generally rated a bit lower:
One caveat is that all the participants in this study were Ivy League students and graduates and therefore not completely representative of the general population. Prior to rating, they also had a brief in-person interaction with the person they were about to rate, which could have made the more agreeable people in the sample less inclined to give a harsh rating. Still, it confirms that photo ratings can be quite different from how people rate in real life.
The results are the same as for here. Women's ratings were higher, about 0.5 d (they have multiple raters per person, so their ratings are more reliable), and both distributions were pretty normal and somewhat above 'average'.
Good article — the world is benefiting much from your increased quant skills.
It would be interesting to see how people adjust their rating given knowledge of the actual average score. This also seems an area where social desirability bias will tend to impact results.
Is it really a white pill for some men to know that their problems are due to their behavior rather than their looks? How easy is it to change your behavior rather than looks? If you have money, the latter seems far easier due to the amount of and continuous progress in beauty products. Boosting your physical attractiveness seems easier than boosting your cognitive/psychological/psychometric/behavioral attractiveness. Didn't mean to be a killjoy. Great article. Sad that there seem to be no giant, comprehensive, longitudinal surveys of male and female attractiveness, where massive numbers of people across races, ethnicities, SES, etc. rate themselves and everyone else on countless different attractiveness scales across numerous points of time with and without make-up and all other beauty products. It would be an amazing study. The boost offered by expensive personal stylists and the like in the case of celebrities is probably huge, but currently unknown. The same with regular beauty products for regular people, especially women. Would be interesting to know whether female beauty products work on men as well. I assume that they won't work as well or even at all since it would strongly signal a gay vibe, but who knows. Maybe some do since there are so many different ones.
I don't think any of those things, except for having a good sense of humor, boost attractiveness, at least based on modern definitions of asshole-ness and handsy-ness. If you're talking about real asshole-ness and handsy-ness, then a tiny fraction of men are like that, and they can't change that anymore than they can change their IQ. Maybe they can pretend for a short time to be different, but it won't last long and you'll be miserable attempting to behave against your nature and repeatedly failing. How many people have you known who've had radical personality transformations of that kind? I admit maybe it is possible with the right structure/courses/program, but it is going to be expensive and a lifelong endeavor, I think.
People are very much more conscious about being judged when they are being interviewed by an actual person about being too shallow. For this reason, the OKCupid data is much more reliable as there is no interviewer that they are conscious of being judged by. And yes, dating sites are representative of the average person.
> The problem is mostly not your looks, but your behavior. You can improve this.
I'd prefer looks to be the problem. I could fix it via plastic surgery. Behavior, on the other hand... https://hotelconcierge.tumblr.com/post/140529495929/how-to-be-attractive this is very relatable. Except it's not about optimizing words; it's about being unable to get any satisfactory ones in real time. System 1 returns nothing. Yeah, maybe it's trainable...
> Thinking Fast and Slow picks questions that fit its conclusion. Yes, System 2 is superb at solving short, quantitative problems. But what if you have massive changing input and it’s not clear what variables are important? What you have information gaps and limited time to act? What if you don’t have any numbers at all? What if you don’t even know how to define the problem?
> System 2 makes you a bad actor, and nerds overuse System 2. The same type of thinking that makes nerds “interested in physics” is responsible for their klutziness, neurosis, and low status. And if I needed a word to describe this worldview, it would be “rationalism.”
> The post-industrial world needs rationality. It’s not a coincidence that the nerd archetype has grown with Moore’s Law. Rationality has made us successful—I am grateful for our advancements in engineering and medicine—but rationalism, as a worldview, demands a sacrifice.
> You’ll hate me, but your rationalism is inseparable from your anxiety. Anxiety asks only one question: “What am I supposed to do?” You know the rules with your friends, but you’re lost with strangers or in groups. So you freeze up and try to calculate the best option, but this never works, there are too many variables, and even if the punchline is good the delivery will be off. People invented small talk for this reason, but rationalism hates rituals: “This is pointless. We can do better.” Ironically, when you find a tribal artifact molded in plastic, painted in neon, and sold in the gift shop, you think: “This isn’t authentic at all.” Right, it’s trying so hard to please that it looks fake. Huh.
Interesting. And it appears to me that men's disadvantage is physical attractiveness is offset by John Archer's finding that in terms of mate choice preferences, "good looks" as a criteria for mate choice is skewed towards men: Cohen's d = 0.55. Men care more about looks than women. See "The Reality and Evolutionary Significance of Human Psychological Sex differences" by John Archer.
It's a shame all these debates still revolve around an old OKCupid dataset. It would be very interesting to see comparable data from tinder/hinge/bumble and to compare between these - anecdotally, I notice considerable differences in the kinds of women that use these apps
Isn’t the key problem the assumption that the male and female data have the same x-axis scales? Men and women are different. For all sorts of reasons women might be choosing a lower ranking for average.
You can easily remap the female results to a bilaterally symmetrical graph. Many “normally distributed” results in nature require such remapping.
1, Online dating is making up a larger proportion of share of the dating market, so the claim that only elite men with social skills pick up women at the bar may not be applicable if the large general public is mainly dating through such platforms with most couples meeting online; furthermore it may just be excluding the more extroverted men, which may have a higher success rate simply because of more social skills/risk-tasking dispositions
2. Attractiveness (physical) can only be inferred from images, and people do make positive assignments of value (emotionally) and cognitively. However men are more likely to tie the value to physical appearance (as noted by highest attractiveness ratings around the end of puberty - 16) with constant valuation metrics / highest response rates to women in their 20s<. Women may have other factors to select on, most notably the degree or extent of social, status and income mobility potential of the partner in addition to health metrics.
3. Since most men and women are dating in 20-30 range where copulation/fertility is most likely, data on 40>, the data correspondence can be skewed with ratings by 40+ being more experienced, jaded, etc.
4. Men can constantly select for neonatal features (i.e. Japan) which can lead to females being more attractive than males on average provided all other environmental factors are controlled for (i.e. not being surrounded by low satiety/excess calories from large proportioned-food in the Country). Sexual dimorphism can only maximize the differences so much (e.g. height) before males are also becoming increasingly directed towards the same sexual-preferences (i.e. taller with taller, more youth-looking longer with more youth-looking longer).
5. Scales are not equally heterogenous. People have an 'above-average bias'. In real life, the 'average' should be 2.5 but businesses are rated something like 3.8-4 on average, this applies to everything in life too, anime, dating. Women may or may not be as generous or harsh so if men judge on a scale of 1-5 but women judge on a scale of 1-4, the skew/distribution points may need to be normalized on a Gaussian curve, to see how much a marginal increase of 1 point of judgement score corresponds to 1 point of 1 point of physical attractiveness, that is the first 1-2 points can mean 60-70% of the variance of attractiveness (i.e. not deformity-looking) and the last 20-30% of variance can be the last point
(i.e. hotness factor).
6. Dating website studies in the past may not be as retrospectively applicable now, as there is a liberal-bias (I read somewhere) of the website owners eliminating conservatives/other groups, thus the attractiveness factor may be more variable than indicated
7. If one were to compare two groups of matches, and receive exposure to both, it could skew the data vs just a study of one-country one-data places of undergraduate students. i.e. Men preferring feminine , youthful, submissive types (as revealed by higher response rates to Asian women) and females preferring vice-versa (reverse selections), as this increases the supply of higher-quality mates from an appearance perspective; thus the samples may be non-comparative
8. If certain women do more cosmetic surgeries/use more beauty products and the average man does not, it could also skew the distributions of attractiveness;
But isn't a key difference that makes okcupid more relevant than many of these other sources that it was fairly consistently based on opposite-sex raters, which seems more relevant than what straight male raters think of men vs women, etc? I didn't click through to all of the studies you mentioned, but I would be fairly surprised if GSS ensures opposite-sex raters for example, but correct me if I'm wrong.
Profiles on OKCupid can be rated by any person. I don't think limiting to opposite-sex does much difference, as the correlations are strong across sexes. In any case, if you want only opposite-sex interviewers' ratings, I am sure you can just filter by the interviewer-sex.
Sure, anyone can rate on OKCupid, but given that it's a dating site I would guess the ratings are disproportionately from either opposite sex or gay people (i.e. are there really a lot of straight guys rating other guys on OKCupid). As far as the high correlation across sex, I haven't looked at the data, but if I'm understanding your claim, even a 100% correlation is consistent with female ratings of men being biased low compared to male-raters. But yeah, it could be interesting to filter by interviewer-sex and see if that explains much of the discrepancy.
I suspect women were rated higher than men till 75 cause female obesity rate is lower than male rate (at least in most Western countries), and women pay overall more attention to their apperance. (grooming, haircuts and so on)
Even if they're asked to rate only "physcial attractiveness" this could still skew the results
"This dataset is also used by the incel.wiki. Really you should click the link, it's an amazing collection."
It's an amazing collection of citations to datasets. It would be even more amazing if it were a collection of datasets.
According to Carver Mead, the reason astronomy has been so much more successful than physics in recent decades is the ready availability of the raw data to the public. This stands in stark contrast to the way the General Social Survey provides only "peep-hole" access through queries to the hidden datasets or, even worse, the way Raj Chetty is provided access to IRS data that no one else is provided even peephole access.
Has anyone in the "incel" community attempted to compile a corpora of _public_ datasets such as those cited in the incel wiki?
The General Social Survey data is completely public, as is the German GSS, and the part of the Add Health I used. Only the WLS is not public of the datasets I used here.
Incels are losers hypothesis proven once again. This is one of the reasons I think that most school programs should be heavily skewed towards some sports/military/shop work activity at least for the boys because the gynocentric education system we have now creates neurotic hikikomoris. It would be of no detriment to people that are 2+ SD in intelligence because they don't really need 12 years to prepare for college entrance and will equip the student population as a whole with a more robust, disciplined lifestyle and being fit and apt is just a great life skill in general (Even amongst the intelligentsia there are some dysgenic trends - most of the best minds of the 20th century were in shape active guys, Bohr and Fermi were even A-grade athletes, and today's scientists start converging more and more towards the frail nerd stereotype. This of course is purely anecdotal observation, I'd love to be corrected on it if someone actually has the data.)
I am not comparing them to today's athletes, we're comparing them to the scientists of today (you're argument actually makes them even more impressive considering the level of professional sport at that time, although the population in general was physically more fit than today.)
Here's an article about a study where people of the opposite sex rated each other's attractiveness and where the rating average and distribution are much closer than on OkCupid, even though men are still generally rated a bit lower:
https://www.lesswrong.com/posts/X3jz5mriJeWi2uLdF/how-subjective-is-attractiveness
One caveat is that all the participants in this study were Ivy League students and graduates and therefore not completely representative of the general population. Prior to rating, they also had a brief in-person interaction with the person they were about to rate, which could have made the more agreeable people in the sample less inclined to give a harsh rating. Still, it confirms that photo ratings can be quite different from how people rate in real life.
The results are the same as for here. Women's ratings were higher, about 0.5 d (they have multiple raters per person, so their ratings are more reliable), and both distributions were pretty normal and somewhat above 'average'.
Good article — the world is benefiting much from your increased quant skills.
It would be interesting to see how people adjust their rating given knowledge of the actual average score. This also seems an area where social desirability bias will tend to impact results.
Is it really a white pill for some men to know that their problems are due to their behavior rather than their looks? How easy is it to change your behavior rather than looks? If you have money, the latter seems far easier due to the amount of and continuous progress in beauty products. Boosting your physical attractiveness seems easier than boosting your cognitive/psychological/psychometric/behavioral attractiveness. Didn't mean to be a killjoy. Great article. Sad that there seem to be no giant, comprehensive, longitudinal surveys of male and female attractiveness, where massive numbers of people across races, ethnicities, SES, etc. rate themselves and everyone else on countless different attractiveness scales across numerous points of time with and without make-up and all other beauty products. It would be an amazing study. The boost offered by expensive personal stylists and the like in the case of celebrities is probably huge, but currently unknown. The same with regular beauty products for regular people, especially women. Would be interesting to know whether female beauty products work on men as well. I assume that they won't work as well or even at all since it would strongly signal a gay vibe, but who knows. Maybe some do since there are so many different ones.
I mean, you can just build a website yourself to collect this data.
Being less of an asshole, learning a joke or too, and learning to keep your hands to yourself is a lot less expensive than plastic surgery.
I don't think any of those things, except for having a good sense of humor, boost attractiveness, at least based on modern definitions of asshole-ness and handsy-ness. If you're talking about real asshole-ness and handsy-ness, then a tiny fraction of men are like that, and they can't change that anymore than they can change their IQ. Maybe they can pretend for a short time to be different, but it won't last long and you'll be miserable attempting to behave against your nature and repeatedly failing. How many people have you known who've had radical personality transformations of that kind? I admit maybe it is possible with the right structure/courses/program, but it is going to be expensive and a lifelong endeavor, I think.
People are very much more conscious about being judged when they are being interviewed by an actual person about being too shallow. For this reason, the OKCupid data is much more reliable as there is no interviewer that they are conscious of being judged by. And yes, dating sites are representative of the average person.
Sex ratio on dating app is more screwed in real life ,how is that representative?
> face tattoos
> piercings
do you have any evidence that these are associated with lower attractiveness in female POV?
I don't have any in particular, but I'm sure you can find some random studies. First one I found when looking about tattoos and attractiveness. https://twitter.com/KirkegaardEmil/status/1616686430906118144
> The problem is mostly not your looks, but your behavior. You can improve this.
I'd prefer looks to be the problem. I could fix it via plastic surgery. Behavior, on the other hand... https://hotelconcierge.tumblr.com/post/140529495929/how-to-be-attractive this is very relatable. Except it's not about optimizing words; it's about being unable to get any satisfactory ones in real time. System 1 returns nothing. Yeah, maybe it's trainable...
> Thinking Fast and Slow picks questions that fit its conclusion. Yes, System 2 is superb at solving short, quantitative problems. But what if you have massive changing input and it’s not clear what variables are important? What you have information gaps and limited time to act? What if you don’t have any numbers at all? What if you don’t even know how to define the problem?
> System 2 makes you a bad actor, and nerds overuse System 2. The same type of thinking that makes nerds “interested in physics” is responsible for their klutziness, neurosis, and low status. And if I needed a word to describe this worldview, it would be “rationalism.”
> The post-industrial world needs rationality. It’s not a coincidence that the nerd archetype has grown with Moore’s Law. Rationality has made us successful—I am grateful for our advancements in engineering and medicine—but rationalism, as a worldview, demands a sacrifice.
> You’ll hate me, but your rationalism is inseparable from your anxiety. Anxiety asks only one question: “What am I supposed to do?” You know the rules with your friends, but you’re lost with strangers or in groups. So you freeze up and try to calculate the best option, but this never works, there are too many variables, and even if the punchline is good the delivery will be off. People invented small talk for this reason, but rationalism hates rituals: “This is pointless. We can do better.” Ironically, when you find a tribal artifact molded in plastic, painted in neon, and sold in the gift shop, you think: “This isn’t authentic at all.” Right, it’s trying so hard to please that it looks fake. Huh.
Interesting. And it appears to me that men's disadvantage is physical attractiveness is offset by John Archer's finding that in terms of mate choice preferences, "good looks" as a criteria for mate choice is skewed towards men: Cohen's d = 0.55. Men care more about looks than women. See "The Reality and Evolutionary Significance of Human Psychological Sex differences" by John Archer.
It's a shame all these debates still revolve around an old OKCupid dataset. It would be very interesting to see comparable data from tinder/hinge/bumble and to compare between these - anecdotally, I notice considerable differences in the kinds of women that use these apps
Isn’t the key problem the assumption that the male and female data have the same x-axis scales? Men and women are different. For all sorts of reasons women might be choosing a lower ranking for average.
You can easily remap the female results to a bilaterally symmetrical graph. Many “normally distributed” results in nature require such remapping.
1, Online dating is making up a larger proportion of share of the dating market, so the claim that only elite men with social skills pick up women at the bar may not be applicable if the large general public is mainly dating through such platforms with most couples meeting online; furthermore it may just be excluding the more extroverted men, which may have a higher success rate simply because of more social skills/risk-tasking dispositions
2. Attractiveness (physical) can only be inferred from images, and people do make positive assignments of value (emotionally) and cognitively. However men are more likely to tie the value to physical appearance (as noted by highest attractiveness ratings around the end of puberty - 16) with constant valuation metrics / highest response rates to women in their 20s<. Women may have other factors to select on, most notably the degree or extent of social, status and income mobility potential of the partner in addition to health metrics.
3. Since most men and women are dating in 20-30 range where copulation/fertility is most likely, data on 40>, the data correspondence can be skewed with ratings by 40+ being more experienced, jaded, etc.
4. Men can constantly select for neonatal features (i.e. Japan) which can lead to females being more attractive than males on average provided all other environmental factors are controlled for (i.e. not being surrounded by low satiety/excess calories from large proportioned-food in the Country). Sexual dimorphism can only maximize the differences so much (e.g. height) before males are also becoming increasingly directed towards the same sexual-preferences (i.e. taller with taller, more youth-looking longer with more youth-looking longer).
5. Scales are not equally heterogenous. People have an 'above-average bias'. In real life, the 'average' should be 2.5 but businesses are rated something like 3.8-4 on average, this applies to everything in life too, anime, dating. Women may or may not be as generous or harsh so if men judge on a scale of 1-5 but women judge on a scale of 1-4, the skew/distribution points may need to be normalized on a Gaussian curve, to see how much a marginal increase of 1 point of judgement score corresponds to 1 point of 1 point of physical attractiveness, that is the first 1-2 points can mean 60-70% of the variance of attractiveness (i.e. not deformity-looking) and the last 20-30% of variance can be the last point
(i.e. hotness factor).
6. Dating website studies in the past may not be as retrospectively applicable now, as there is a liberal-bias (I read somewhere) of the website owners eliminating conservatives/other groups, thus the attractiveness factor may be more variable than indicated
7. If one were to compare two groups of matches, and receive exposure to both, it could skew the data vs just a study of one-country one-data places of undergraduate students. i.e. Men preferring feminine , youthful, submissive types (as revealed by higher response rates to Asian women) and females preferring vice-versa (reverse selections), as this increases the supply of higher-quality mates from an appearance perspective; thus the samples may be non-comparative
8. If certain women do more cosmetic surgeries/use more beauty products and the average man does not, it could also skew the distributions of attractiveness;
https://twitter.com/datepsych/status/1616397013318307841 will you respond to this?
Seems no particular reason to.
But isn't a key difference that makes okcupid more relevant than many of these other sources that it was fairly consistently based on opposite-sex raters, which seems more relevant than what straight male raters think of men vs women, etc? I didn't click through to all of the studies you mentioned, but I would be fairly surprised if GSS ensures opposite-sex raters for example, but correct me if I'm wrong.
Profiles on OKCupid can be rated by any person. I don't think limiting to opposite-sex does much difference, as the correlations are strong across sexes. In any case, if you want only opposite-sex interviewers' ratings, I am sure you can just filter by the interviewer-sex.
Sure, anyone can rate on OKCupid, but given that it's a dating site I would guess the ratings are disproportionately from either opposite sex or gay people (i.e. are there really a lot of straight guys rating other guys on OKCupid). As far as the high correlation across sex, I haven't looked at the data, but if I'm understanding your claim, even a 100% correlation is consistent with female ratings of men being biased low compared to male-raters. But yeah, it could be interesting to filter by interviewer-sex and see if that explains much of the discrepancy.
Data are public, so have at it and post results!
This made me laugh 😃
I suspect women were rated higher than men till 75 cause female obesity rate is lower than male rate (at least in most Western countries), and women pay overall more attention to their apperance. (grooming, haircuts and so on)
Even if they're asked to rate only "physcial attractiveness" this could still skew the results
"This dataset is also used by the incel.wiki. Really you should click the link, it's an amazing collection."
It's an amazing collection of citations to datasets. It would be even more amazing if it were a collection of datasets.
According to Carver Mead, the reason astronomy has been so much more successful than physics in recent decades is the ready availability of the raw data to the public. This stands in stark contrast to the way the General Social Survey provides only "peep-hole" access through queries to the hidden datasets or, even worse, the way Raj Chetty is provided access to IRS data that no one else is provided even peephole access.
Has anyone in the "incel" community attempted to compile a corpora of _public_ datasets such as those cited in the incel wiki?
https://incels.wiki/w/Scientific_Blackpill
The General Social Survey data is completely public, as is the German GSS, and the part of the Add Health I used. Only the WLS is not public of the datasets I used here.
Incels are losers hypothesis proven once again. This is one of the reasons I think that most school programs should be heavily skewed towards some sports/military/shop work activity at least for the boys because the gynocentric education system we have now creates neurotic hikikomoris. It would be of no detriment to people that are 2+ SD in intelligence because they don't really need 12 years to prepare for college entrance and will equip the student population as a whole with a more robust, disciplined lifestyle and being fit and apt is just a great life skill in general (Even amongst the intelligentsia there are some dysgenic trends - most of the best minds of the 20th century were in shape active guys, Bohr and Fermi were even A-grade athletes, and today's scientists start converging more and more towards the frail nerd stereotype. This of course is purely anecdotal observation, I'd love to be corrected on it if someone actually has the data.)
Terman and SMPY show that frail nerd stereotype is wrong. The smarter people are physically more healthy, taller etc.
Steve Hsu practices martial arts. xd
In Bohr's days the bar for A-grade athlete was much lower because sport was not at professionalized as it is now.
I am not comparing them to today's athletes, we're comparing them to the scientists of today (you're argument actually makes them even more impressive considering the level of professional sport at that time, although the population in general was physically more fit than today.)
Hsu is an outlier in many domains.