OkCupid had already figured this out back in Jan 2010 [1]:
> For women, a smile isn’t strictly better: she actually gets the most messages by flirting directly into the camera
> Men’s photos are most effective when they look away from the camera and don’t smile. Maybe women want a little mystery. What is he looking at? Slashdot? Or Engadget?
> We were sure [MySpace angle] pictures were lame. But we were so wrong. In terms of getting new messages, the MySpace shot is the single most effective photo type for women.
I always thought OkCupid's blog posts were very fun/interesting. I think they first one I read was about how quality of the picture matters (e.g. phone picture vs SLR...lighting...etc.)
I was disappointed to see just now that the latest entry is from April 2011 :\
My favorite is "Why You Should Never Pay For Online Dating" from 2010-04-07. It was deleted when they were bought by IAC. Thankfully, nothing is ever really deleted from the Internet:
Don't get excited, it's just a graph. But there is an asymptote that has been shifting lower as a demographics change and more people delay having children until basically never. Demographics vs. resources. For the majority, there's no longer as much social or familial obligation to continue uniform tradition.
Also, this market space is so crowded with higher-quality free apps, it will only get increasingly brutal.
(Btw, what's it like to be female and dating in China?)
I worked on the OkTrends team at OkCupid during the year before the acquisition, and I can assure you that that timing is definitely a coincidence.
The posts took a lot of work to create. When I was there, there were two and a half people working full-time[0] just on OkTrends (two engineers full time, and one founder part-time).
I can't give an average time per post, but to give you an idea, "The Real Stuff White People Like"[1] was predominantly my work[2], and it took almost two months, from start to finish.
If you're wondering why it took so long to create that, remember that we started each blog post as a blank slate - at most, we had a vague question that we wanted to explore, and it took several iteration for us to hit upon anything closely resembling the analysis that you'd end up seeing in one of the OkTrends reports.
It might be easy to create something if you have a specific destination in mind at the start, but the key to making OkTrends posts work was not having a specific result in mind - instead, letting the journey guide the process.
Anyway, I left to go back to school, and around that time, the other engineer on the data team started having to do more internal data work on top of the OkTrends research (he stayed at OkCupid for over a year after I left, but he's since moved on as well). That's what caused the slowdown in the posts during late 2010 to early 2011, not the sale to Match.com. As noted below, Christian is working on other projects as well, which have unfortunately left little time for OkTrends.
It's very unfortunate that the timing makes it look like Match.com put the kibosh on OkTrends, because that's not at all what happened, though I definitely see why people would make that mistake.
[0] By "half", I mean that we two engineers spent close to all of our time working on the data/stats, and the founder spent about half his time working with us on the projects and writing the posts.
Definitely good to know. It certainly looked the other way from an external viewpoint given the really close timing. That being said, I'd love it if they'd start up again it was always fascinating from a sociological perspective at what was working for people and what wasn't (It never helped me though, I ended up meeting my SO over IRC instead...)
You're not the first person to tell me that! A friend of mine met his husband on #gaynyc. I still love OkCupid, but I always idle in that chan now, just in case.... :)
It seems likely that at least one of the authors or reviewers saw the OkCupid post, possibly to the extent of influencing the research itself. Yet not being published in an peer-reviewed journal, it's not cited. The OkCupid data set is far more extensive and includes longer-term reactions, so it would be useful to see it in citable form. In lieu of a peer-reviewed paper, I think the blog post should be cited. [EDIT: Some communities use lack of peer review as an excuse for not citing. Of course it's "allowed".]
the controls are terrible. her hair isn't arranged the same in every pic and her breasts are more visible in the happy and pride pics than the neutral one.
all the pics are way too over-acted, the "pride" girl pic isn't a typical girl pose (it's super exaggerated for a guy, and even less realistic for a girl). these poses have little to do with real life. it's just ivory tower BS, incompetently done (quite apart from being so unrealistic, the bad controls are just incompetent even within the narrow field, just by basic scientific standards)
Plus, the guy looks down right creepy in the happy picture and he is showing off his muscles in the pride picture. Their shame poses are idiotic. They would have gotten much better results by cropping faces only.
I'm confused as to why this post is suddenly at the top of HN, it isn't a new article nor does it present a new idea. There are countless of studies that show that women are more attracted to men who seem mysterious/proud/narcissist/dangerous/the opposite of your minivan-driving soccer dad. Does this imply that the HN community is generally unaware of the difference in the male-female sexual preferences and thus finds this article mesmerizing? Is there nothing else as worthy of the front page? Am I just being cynical?
Here's a post from Heartiste (at the risk of losing my credibility) that talks about the subject.
This is highly anecdotal, but when I was younger (from ~5-15 years ago), I noticed I was far more likely to have people start fights* with me when I was in a good mood and enjoying myself. Yet this never seemed to happen when I walked around with a bit of an angry look in my eye, like I had been having a bad day and was just waiting for someone to start something. I even informally experimented with it, sometimes approaching people in a cheerful, happy way, and other times approaching them like I was in a really bad mood. Men seemed to treat me much better when I was faux-pissed, and much worse when I was in a good mood. The authors didn't experiment with this (that I could tell), but I'd hypothesize that males have the same reaction towards happy expressions in other males - they don't like them and it initiates aggressive tendencies.
*You might ask why I was getting into fights in the first place, and all I can say is that high school and college are rough and most of the population doesn't seem to be as sophisticated as the typical HN reader.
This always puzzled me.. when I have a great day, and walk around with a puffed chest smiling at people, women positively seem to ignore me. When have a bad day, or didn't wash my hair or something, it's the opposite, and I get eyed when I don't really want it. Once I walked around sulking, more or less staring at the ground, and a random young woman told me I was "very beautiful". I said thanks, but thought "WTF?! I look like shit."
Maybe people like it when they feel like they have to offer something to a person.
Disliking happy dudes seems natural enough. Imagine the inner monologue: "Oh ho ho, what are you so happy about? You smug bastard... projecting your mirth onto my group and me isn't going to work, see?" A grumpy man presents much less threat of trying to dominate the conversation with nonsense.
> In contrast, women over the age of 30 tended to rate shame- and happy- displaying men as equally attractive (and both more so than neutral).
happy guys finish ahead of control, not last... and the BS title is from the original paper.
that's over 30, but several of the graphs in the paper (not all) show happy guys beating the control. meanwhile i don't see anything clearly indicating happy guys lost to the control overall; it looks more the other way around. at the very least, it's not clear happy guys finish last and lose to the control.
EDIT: the abstract says:
> happiness was the most attractive female emotion expression, and one of the least attractive in males.
so they knew the title was BS and didn't repeat the claim in the abstract. in the abstract they use weasel words, in the title they intentionally lie to get more attention/views.
There are a variety of alternative explanations, consistent with the data. The conclusion in the title is just one of many possibilities they didn't differentiate between.
It could be, for example, that guy photos do best when they stand out instead of looking like every other guy. it could be that the best guy photo would be a happy smiling guy who stands out in a different way, but when you remove everything else from the photo and the choices are standing out or being happy, both good, then standing out is better.
this is a standard practice in pseudo "science" – get some data, make up a conclusion that doesn't contradict the data, say you have evidence for your conclusion (or say something stronger), don't carefully think about everything else you could have included instead and how to pick between them.
as usual with bad "science", there is no section titled "sources of error" or similar. nor is there a section covering alternative conclusions compatible with the data. if you aren't thinking about all the ways you could be wrong, it's not really science. (and if you think about them but don't publish that part, you're not publishing science)
What is with all the use of line graphs for showing categorical vs continuous data? Did the authors not know about bar graphs? It looks absurd and ridiculous.
Intuitively agreed at first. But then I imagined the main alternative of clustered columns/bars and now tend to think that line graphs can be grasped more quickly.
Any other alternative graph formats? Maybe just show the deviations from the neutral control?
I'm curious as to how accurately sexual attraction can be measured by looking at a picture of a person. I don't have a better proposal, barring actually filling a resort with cameras, inviting a bunch of single people to it, and tracking what happens. You could send in some actors who do shame, pride, etc and see how they do, but you'd need to control for their relative physical attractiveness. So I guess you do lots of groups, and the actors change their role every time. Or get identical twins. You'd have to be careful with ethics, though.
More scientifically: happy faces are more attractive. http://www.tandfonline.com/doi/abs/10.1080/02699931.2013.817... Note the study I reference only used faces, and used the same facial images, digitally altering them to alter the facial expressions. As opposed to Happy Guys Finish Last, which used different photos and included the upper body in the shots.
"Participants and procedure. In this study, 184 Canadian undergraduates (50% female; age 17–49 years, median 21; 52% Asian, 48% Caucasian)..."
It would be unwise to draw any conclusions from this study since it was preformed on such a non-random, non-representative slice of the general population.
The photos used in this test are terribly inconsistent.
Each photo has different lighting and distance from the camera.
The happy smile are those artificial smiles that people make when they are posing. A genuine sincere smile may rate completely differently.
The person raises their arms for the pride photo, and the posture and visible biceps in this photo may completely distort the perceived attractiveness.
Reis et al. (1990): American college students attributed smiling persons greater sincerity, sociability, and competence but less independence and masculinity.
I really can't understand why they didn't take into account the cultural bias tided to the different races. I would guess that Asian people are looking for different values in their partners than Caucasians. And even if I were wrong, I don't think that they should dismiss that proposition without even checking, or citing a relevant paper on the subject.
I know this is probably extremely simple statistics, but what are these terms?
They're metrics about the strength of the finding - depending on the statistical method you're using, you'll get different ones. They're using ANOVA (analysis of variance) here, which basically compares likelihood mean result of two cohorts is the same.
T(45) gives the t score, which is a measure of deviations from the mean, so 3.5 higher than you'd expect if the populations were the same.
p values is probability it occurred by chance(given equal populations), so 1% in this case, which is pretty strong.
I'm not positive on d, but I think it's a measure of variance within the groups.
Disclaimer: was an Econ major in college but haven't used anova in a long time, I am not a statistician, this is not statistical advice.
> For women, a smile isn’t strictly better: she actually gets the most messages by flirting directly into the camera
> Men’s photos are most effective when they look away from the camera and don’t smile. Maybe women want a little mystery. What is he looking at? Slashdot? Or Engadget?
> We were sure [MySpace angle] pictures were lame. But we were so wrong. In terms of getting new messages, the MySpace shot is the single most effective photo type for women.
[1] http://blog.okcupid.com/index.php/the-4-big-myths-of-profile...