Occasionally I see people complaining that they cannot find a rationalist girlfriend. Fortunately, we have a survey! And one of the variables is “Romance”, i.e., “have you met someone through the community that you wound up dating?” So I figured I would give some statistics-y help to the lovelorn.
(A Note on Methodology: I assigned “no” the value of 0, “yes” the value of 2 and “I didn’t meet them through the community, but they’re part of the community now” a value of 1. I understand this is debatable– as, in fact, rationalist girlfriends one did not meet through the community are still Rationalist Girlfriends– but we are looking for information on how to find a rationalist girlfriend, not a girlfriend you can convert to rationality. All correlations are significant at p = .01. Also, I ran my fishing expedition on Things That Have Numbers Attached. If you are curious about a particular bit of non-numbers data, ask in the comments and I’ll add it to the post.)
Interestingly, the number of years that one has been a rationalist is completely uncorrelated with one’s ability to get a rationalist romantic partner. That’s weird, because you’d think the longer you stuck around the more rationalists you might want to date you’d meet. It’s also basically uncorrelated with time spent on LW, but at this point level of participation in the rationalist community is basically uncorrelated with time spent on LW, so that doesn’t mean much.
The largest correlation is with number of current partners (Pearson correlation .372). Unfortunately, this item combines “is polyamorous” and “is capable of getting laid” into a single factor, which is not very helpful. When I included only people who prefer monogamy, the correlation went down to .123 (although it was still significant at p = .01). This suggests that, while ability to get laid plays a part in the strength of this correlation, a substantial amount of the correlation is that poly people are way more likely to get rationalist partners. The best thing you can do for your chance of getting a rationalist partner is to become poly.
Karma score is another reasonably strong correlation (.210), suggesting that rationalists are attracted to people who make awesome blog posts and contribute to the community. Similarly, donations to MIRI/CFAR are correlated with likelihood of getting a rationalist romantic partner (.123), as is charity donations in general, but the Pearson correlation is .089 so it might just be that people who donate to MIRI/CFAR tautologically also donate to charity. I am not suggesting a “donate to MIRI: get laid!” advertising campaign, but…
SAT Scores out of 1600 is correlated with likelihood of rationalist girlfriend (.180), but neither SAT Score out of 2400 or ACT Score is correlated, and anyway the sign on both of them is negative. I cannot explain this data. Rationalists… have a preference for people who don’t write essays on their standardized tests?
Supporting more immigration (.121) and a guaranteed basic income (.107) increase your chance. This may or may not be a reflection of liberals getting laid more; I’m not good enough at SPSS to do a factor analysis and rationalist-partner-finding is uncorrelated with support for things like taxes or a minimum wage. So right-libertarians, you may still have some hope (although neoreactionaries are screwed). Regardless, be sure not to talk about politics too much, because political interest is negatively correlated with finding a rationalist partner (-.093).
Probability of aliens in the galaxy (-.083), probability of religion (-.071), and probability of global catastrophic risk (-.116) are all negatively correlated with finding a rationalist partner. Okay, I get why we like atheists, but why do you apparently object to people who believe in aliens? I personally think UFO watching is a great first date.
Femininity on the Bem Sex Role Inventory (.106) predicts likelihood of finding a rationalist partner. I’ll talk about this more later. Number of older siblings and height are both negatively correlated with ability to find a romantic partner. 6’4″ youngest children may comfort themselves with the knowledge that the correlation isn’t very strong (-.093 and -.096, respectively).
Now, you may say I should rerun this without including women, on the grounds that any straight women who can’t get laid in a community that’s four-fifths male clearly has bigger problems that a blog post can help with. So I looked at heterosexual men (since queer men can presumably date each other).
Correlations that went away: probability of aliens, probability of religion, femininity on the Bem sex role inventory, height, charity. Immigration and political interest were still significant at the .05 level, so that’s probably just noise. Femininity and height are probably gender-related: women are more likely to have a rationalist partner than heterosexual men are. (Yes, I checked.) Charity is probably just that charity is correlated with MIRI/CFAR donations. However, I am pleased to note that women apparently are okay with UFO-watching as their first date.
Correlations that still exist: karma score (.267), number of current partners (.270), SAT scores out of 1600 (.166), probability of global catastrophic risk (-.149), basic income (.122), MIRI/CFAR donations (.103), older siblings (-.094), left hand digit ratio (.147).
New correlations: probability that the many worlds interpretation of quantum mechanics is true (.119), probability that we’re all living in a simulation (.091), and MIRI’s effectiveness (.092). MIRI’s effectiveness, how much one supports MIRI’s mission, probability that cryonics and anti-agathics work, probability that we’re living in a simulation, belief in a near Singularity, MIRI/CFAR donations, probability that MWI is true, belief in a basic income and global warming, and not believing in global catastrophic risk are all correlated. (Yes, not believing in global catastrophic risk. I don’t know.) I suspect there is an Agrees With Eliezer factor here, except I can’t actually do factor analysis, so I don’t know.
Nevertheless, judging by the new correlations, old correlations like MIRI/CFAR donations and probability of global catastrophic risk, and some of the ones significant at p = .05 (like anti-agathics and support for MIRI’s mission), I think if there is an Agrees With Eliezer Factor, then it is related to how likely one is to get a rationalist girlfriend. This may be a result of level of participation in the community, how determined one is to get a rationalist girlfriend, or rationalist women being really attracted to men who agree with Eliezer about everything. You might as well try agreeing with Eliezer about more things. Couldn’t hurt. I hear paleo diets are very tasty.
In short, here is my pickup advice for people who want a rationalist girlfriend:
1) Become poly.
2) Become high-status in the community, possibly via writing posts on Less Wrong.
3) If at all possible, do not be a straight man.
4) If you absolutely have to be a straight man, agree more with Eliezer.
5) Make really sure that you don’t take any standardized tests with a writing portion.
6) Travel back in time and interrupt your parents’ sex before they conceive your older siblings.
7) Probably save the UFO hunt for the third date, Mulder.
theunitofcaring said:
I’m surprised there’s not a correlation with attending real-life meetups, and with the gender breakdown of said real-life meetups, or was that data not collected? I would expect meetups with comparatively balanced numbers in the first place to produce more partners.
The convert-girlfriend-to-rationality strategy works much better (though admits itself of less statistical analysis.) SSC is a great gateway drug; depending on the temperament of the partner to be converted, rationalist tumblr could be great or terrible.
(The ‘locally high-status people get more girlfriends’ correlation is the only one which was not surprising to me.)
LikeLike
ozymandias said:
I didn’t check, because “have you attended a meetup?” is not a numeric variable. I converted it to one, though, along with the “physical interaction with the LW community” variable, and ding ding ding we have a winner. Meetups correlation .409, community participation correlation .548, significant at p = .001.
LikeLike
challquist said:
“SAT Scores out of 1600 is correlated with likelihood of rationalist girlfriend (.180), but neither SAT Score out of 2400 or ACT Score is correlated, and anyway the sign on both of them is negative. I cannot explain this data. Rationalists… have a preference for people who don’t write essays on their standardized tests?”
IDK about the ACT scores, but whether you have an SAT out of 1600 or 2400 is partly a reflection of how old you are. Benefits of intelligence for getting a girlfriend don’t kick in until you’re older, maybe?
LikeLiked by 1 person
srconstantin said:
or, just, “older people are more likely to have a girlfriend”. You could either fill in your SAT score out of 1600 or 2400. If this was calculated to fill in the un-answered SAT score as zero, then old people have a really low SAT score out of 2400 and young people have a really low SAT score out of 1600.
LikeLiked by 3 people
veronica d said:
Yeah. This one needs to be controlled for age.
LikeLike
ozymandias said:
I removed the younger half of LWers from the dataset and SAT Scores out of 1600 was still correlated with probability of finding a rationalist girlfriend, while SAT Score out of 2400 and ACT Score were still uncorrelated/possibly slightly negatively correlated. It’s not age.
LikeLiked by 1 person
endoself said:
What was the age cutoff here? People above a certain age would only need to take the SAT out of 2400 if they were entering university late, so you’d potentially end up with a very small sample, and one selected for unknown other traits.
LikeLike
ozymandias said:
26, because that’s the median age of rationalists.
LikeLike
PDV said:
26 is within a year or so of the switchover, as well, so that mostly checks out. I know for certain that no one who took the 1600 SAT at the usual time is under 24, and no one who took the 2400 SAT at the usual time is over 27 (personal memory; I took the 1600 in 7th grade for CTY, and the 2400 in 11th at the usual time; giving a full year of slack on each end to account for kindergarden-admissions slop gives those numbers).
LikeLiked by 1 person
queenshulamit said:
My brain sang the title of this post to the tune of Do You Want To Build A Snowman.
LikeLiked by 2 people
Vulture said:
‘It doesn’t have to be a girlfriend…’
LikeLiked by 3 people
stillnotking said:
Number six is just sound general advice, surely.
LikeLiked by 1 person
Emily said:
I think you might get some useful information from asking your female readers why they do or don’t date fellow rationalists, and if they do, how they choose among them. I would say this is an example of the odds being good but the goods being odd.
LikeLike
Hainish said:
Or even, asking female readers why they do/don’t identify as rationalists. (I’m thinking, if four-fifths of the community is male, there’s gotta be a reason.)
LikeLike
veronica d said:
On becoming poly to get a g/f, aren’t you making a correlation-ain’t-causation mistake here? (Or is it assumed everyone gets that and we should just take this as a fun exercise, cuz that’s cool also.)
Just some general musings:
I’ve never been to a LW meetup, so I don’t really know the social space, but from the forums I get the sense I would be all kinds of frustrated with the crowd, and (again judging from the forums) ’specially the dudes. But then, I probably would not count as a “rationalist girlfriend” (but I’ve read the sequences!). It’s just that — well, it’s hard enough to negotiate the trans stuff and the feminism stuff and the weirdbrain stuff and maybe rationalist folks would be better on some of that but I think they’d be way worse on other parts of it and the ones who were wrong would just be so goddam sure they were right cuz they thought about it real hard dontcha see.
And yes I know that is the opposite of how a rationalist is supposed to be — but really!
LikeLiked by 2 people
Kaminiwa said:
Totally anecdotal, but the Portland meetup didn’t seem to have any issues with trans or genderqueer. Conversely, Portland is just generally good at handling trans stuff.
LikeLike
veronica d said:
Without naming names, or (I hope) giving much away, there is a regular poster on one of the related forums (I’m being vague on purpose) who I know lives near me and has mentioned going to LW meetups. And like, this person is aggressively shitty on trans stuff. And sure, they’re just one person, and maybe face-to-face they’d shut their piehole, and actually I can be pretty intimidating so maybe they wouldn’t dare. But on the other hand, do I want to deal with that?
I quit a really amazing local writing group cuz one night a couple old cis dykes started discussing michfest, as in, what it was like this year, which fine whatever, but then this horrible little TERF-in-training started ranting about trans women and how unfair we were being. And like, I just SHUT DOWN AND COULD NOT DEAL.
But that was one kind of social setting where I’m really unbalanced (cis dykes do that to me) and maybe the LW crowd would bring out a different side but maybe not.
Mostly I hang with my own kind.
LikeLike
Bugmaster said:
> Mostly I hang with my own kind.
Reading stuff that people say on the internet rarely makes me sad, but this line did 😦
LikeLike
Doug S. said:
“Become poly” presumably includes “become willing to date a poly girl who has another boyfriend”…
LikeLiked by 1 person
PDV said:
In-person is much less aggressive in my experience. Lack of anonymity helps, of course, but also I believe the more socially incompetent people tend to avoid in-person meetups.
My city would tend to be above-average good, though.
LikeLike
veronica d said:
Oh, and this book has a lovely short explanation of Factor Analysis that is approachable, assuming your Linear Algebra is solid. If you understand Eigen-stuff, the SVD, and the QR decomp then you have what you need.
LikeLike
veronica d said:
Oh, the link: http://mitpress.mit.edu/books/introduction-machine-learning
LikeLike
fourthrootofunity said:
If you have your data in SPSS you really shouldn’t be doing a factor analysis by direct eigenvector computation unless you’re a mathematician or a masochist. And even then you might not — I’m a mathematician and I don’t do factor analysis by eigenvector computation; I use R’s dedicated function.
I’ve never used SPSS so I’m not 100% certain about this, but it looks like the walkthrough in the following link will get you an SPSS factor analysis without knowing more than how to get your data into SPSS in the first place (which you’ve presumably already done): http://www.ucdenver.edu/academics/colleges/nursing/Documents/PDF/FactorAnalysisHowTo.pdf
(content warning: the example data set is on IQ)
LikeLike
veronica d said:
Agreed. Use a built in function. That said, you should still understand *why* it works, which that book will help on. (Of course, many sources will, this is just one I found particularly helpful in getting the basics.)
There is an enormous difference between using a tool when you know how and why it works compared to shoving data into a black box and turning the crank.
LikeLiked by 1 person
fourthrootofunity said:
Actually, the walkthrough does give you some choices to make, which is probably confusing if you’ve never studied factor analysis. So here are my recommendations.
1. Every place it says something to the effect of “you might choose something else, but we will do [thing]”, you should do “[thing]”.
2. Leave “Extract” as “Eigenvalues over: 1” (This doesn’t matter much for how we are doing it.)
3. Use “Varimax” in the “Rotation” menu, because that’s what most people do.
Doing this will get you an answer; the paragraphs below are just to help you interpret it.
The Scree Plot is useful for significance, but they don’t tell you how to interpret it. Basically, there’s going to be a bunch of stuff on the right you can more or less draw a line through, and some things on the left that lie above that line. The ones that lie above the line are significant. (In their example they say both 1 and 2 are above the line; I’m not so sure about 2, but 1 is clearly above the line.)
The “Rotated Component Matrices” are what your factors are. If it didn’t show you everything that the Scree plot said was significant, rerun with a smaller “Eigenvalues over:” number in the “Extract” submenu.
Everything else should be in the walkthrough. And I’d happily answer questions if you try this and it doesn’t work (although I can only help with factor analysis itself, not SPSS, since I don’t have and haven’t used SPSS).
LikeLike
Nornagest said:
>SAT Scores out of 1600 is correlated with likelihood of rationalist girlfriend (.180), but neither SAT Score out of 2400 or ACT Score is correlated, and anyway the sign on both of them is negative. I cannot explain this data. Rationalists… have a preference for people who don’t write essays on their standardized tests?
This says to me “being smart is good, if you’re also mature”. People with SAT scores out of 1600 are older than people with SAT scores out of 2400, because the second test is more recent.
LikeLiked by 1 person
Nornagest said:
Whoops, I see someone has said this already. Pity there’s no edit function here.
LikeLike
Siggy said:
I hadn’t realized you worked on a survey for LessWrong. I help run an asexual community survey, so I irrationally believe that people who work on community surveys are cool. Although as an ace I’m obligated to tsk over the conflation of dating and getting laid.
The 2013 results say 119 people (7.3%) who had dated a LWer or someone who became a LWer. Unless the numbers are a lot bigger, I don’t understand how you can have such small but significant correlations. I’m also inclined to believe that strength of participation is the strongest factor, and all other correlations may be related to that. Were there really no more direct measures of participation levels?
LikeLike
ozymandias said:
Yep, I do! My boyfriend Scott runs the survey and makes me do the scutwork. 🙂 And yeah I know dating is not the same thing as sex. I’m dating an asexual. 🙂
The direct numerical measures of participation levels (time in community, time spent on Less Wrong) don’t correlate with anything interesting. The measures of participation in the physical community (“do you participate in the physical community?” “do you go to meetups?”) correlate way strongly, but they were non-numerical and I was too lazy to convert them into numbers until unitofcaring asked about it. 😛
LikeLike
Bugmaster said:
It sounds like the first three points (as well as arguably the last one) in your pickup advice summary are basically saying, “become more attractive in the conventional way, and expand your dating pool to more people”. It’s good advice or anyone, not just rationalists.
LikeLike
elijahlarmstrong said:
Unfortunate that the correlations (even the significant ones) are so low!
LikeLike
Bugmaster said:
I am not a scientist or a statistician, but still: isn’t P=0.01 kinda high ? At our company, we are used to dealing with 1e-5 at most, and even that is considered as “barely good enough to warrant further study”, and not “OMG let’s spend all of our budget on this now”.
LikeLike
ozymandias said:
p = .01 is a pretty high standard for the social sciences. (My extremely scientific method of selecting a p value was to choose one that gave me enough significant correlations for an interesting blog post and not so many it would be tiresome to read.)
LikeLike
Bugmaster said:
In that case, you might be interested in learning that divorce rate in Maine correlates with nationwide per-capita consumption of margarine 🙂
http://www.tylervigen.com/view_correlation?id=1703
LikeLiked by 1 person
elijahlarmstrong said:
The p-values are perfectly adequate, but the correlations are still low. Even if you follow the advice to the best of your ability, it will only slightly improve your chances (with the exception of poly).
LikeLike
Aceso Under Glass said:
Sounds like you work in something physicsish. In biology, p < 0.05 is considered sufficient.
LikeLike
polythrowawayname said:
(not using my regular handle to avoid tracking this back to paramours who value privacy)
I am not surprised that poly is the highest correlate to dating a rationalist (other than physical proximity, which is a good correlate for dating ANYONE). I was trying to collect my thoughts into abstract statement phrased int terms of things like P(poly|rationalist) and the number of edges associated in a node of a polyamorous romance graph, but I think an illustration will do better. My primary girlfriend, who is sufficiently in the lw memeplex to I think count, and I went to a LW meetup. We meet a couple that we knew from elsewhere there (Yay coincidences!). It came up that we were poly, and they said essentially “Oh! this updates our prior that poly is only for Eleizer and Scott and other weird internet people (meant with the utmost love, of couse). Cool!” And so we all started dating each other and 4 people had new rationalist bfs/gfs that they met at a meetup. I know at least a few of us made it on the survey 🙂
I dont know how typial this sort of thing is, but it seems to match stuff scott and alicorn have written. The “spend time with lw people –> higher acceptance of poly –> Higher number of partners –> partners likely to be people you spend time with” circle is pretty clearly causal
LikeLiked by 1 person
Lambert said:
What about the converse of ‘converted gf to rationalism’: ‘converted by gf to rationalism’. This would affect the results of the survey.
LikeLike
Airgap said:
Leaving the queers out for the moment, there are two kinds of guys we could be talking about here:
1. Guys who are currently dating a normal girl, but would like to trade her in for a rationalist. Dude, you should probably have a talk with her. Seriously.
2. Guys who are currently unable to get a normal girl to date them, and somehow think that among rationalist girls, their chances are better: What part of “20%” don’t you understand, moron? You’re clearly not innumerate, so what’s your excuse?
Realistically, all the guys interested in this are in group (2). The thing to do with them, in my opinion, is to stop enabling them to make up idiotic just-so stories about how being in vastly gender-imbalanced community is going to increase their levels of romantic satisfaction because the few girls around are more into bayesian statistics. Because they will make up those stories. And the stories will be idiotic. And your statistics are the kind of thing that enables them. We all know having a warm body to hold is way more important to your psychological well-being than whether you and your SO can talk about effective altruism at a featured LW post-quality level. You can always keep posting on LW. Tell your normal gf it’s porn or something.
Also, I’m going to go to a rationalist meetup and convince the prettiest girl there to go on a UFO hunt with me that very night because fuck you, Ozy (xoxo).
LikeLike
Pingback: Suggestion Saturday: January 10, 2015 | On The Other Hand