tl;dr is that being conventionally pretty helps a lot :-P
That’s not the tl:dr of this analysis at all. there’s Betty-Veronica type trade off effects, there’s individual differences, there are mechanisms of assortive mating...i can’t speak to the validity/certainty of the statistics but if the conclusions are valid there is a publishable amount of juicy stuff in here.
i can’t speak to the validity/certainty of the statistics
I realized that I should have gone into more detail in the way of justification, and added some material on this in the “What this means in tangible terms.” I still have to post the code that I wrote for this part so that my analysis is transparent.
I can’t speak to the validity because my quantitative skills are untrained to the point that I have to spend an entire day learning in order to pick up enough background to distinguish good math from bad math for anything more complex than a p-value, and until this changes (and I fully intend to change it) I have to just hope that the equation-using overlords know what they are doing.
I definitely wasn’t saying your math wasn’t well defined enough! I have no opinions on that particular matter, because, see above. It’s not the fault of your writing at all—you’d probably have to write a textbook to get me completely up to speed here. I’m generally sharp enough to know what you are trying to do with the math, but not nearly knowledgeable enough to know whether or not it is correct.
I very much enjoyed the results though! (Especialy the trade-off thing...it fits together with a lot of the other literature on prioritizing in long term and short term mating, and I don’t think anyone has ever supported that particular theory using this particular method. Kinda wish we had data on menstrual cycle phase to go with this :P If you’re planning on publishing you might want to cite stuff within that whole body of work.
I found it all quite understandable, clear, and interesting =) Totally disagree with shminux on the “dry” part.
Although, an abstract up-front would probably help reach more readers..
Don’t worry, I didn’t make my changes in response to your comment specifically – I just realized that the section that I changed was weak because I was unnecessarily vague in my comments along the lines of “here is the data, but the effect size is smaller than it looks because of regression to the mean” – I could instead just show the degree to which past behavior predicted future behavior.
I fully intend to change it
Anything that I might be able to help with? I have a background is in math education. You might find some of the pointers in the Cognito Mentoring math and statistics learning recommendations to be useful.
I have to just hope that the equation-using overlords know what they are doing.
My experience has been that while statistics sometimes requires substantive mathematical knowledge, intuition built from exposure to data and experience with thinking about it gets you ~80% of the way to a full understanding. On the flip side, mathematical knowledge without exposure to real world data doesn’t get one very far...
Especialy the trade-off thing...it fits together with a lot of the other literature on prioritizing in long term and short term mating
It’s certainly natural to hypothesize that the differences came from some people looking for short term mates and others looking for long term mates, but I’m not sure that that’s what was going on. You’ll see what I mean when I show the demographic correlates. I think that your reference to assortative mating is highly relevant.
Kinda wish we had data on menstrual cycle phase to go with this :P
There was in fact more variation with respect to preference on the tradeoff dimension amongst women than amongst men. The comparison isn’t an apples to apples one, because e.g. men found women more attractive than women found men. The difference isn’t very large: 0.42 standard deviations vs. 0.36 standard deviations.
If you’re planning on publishing you might want to cite stuff within that whole body of work.
I need to talk with psychology researchers to get a better sense for how what I’ve done fits in with the literature. In the past, I read some papers on the subject out of casual interest, but I haven’t done a deep dive.
I would guess that some of the phenomena in the data are in fact previously unknown to academic psychology. The dataset is unusual in a number of ways: the sample size is large (~500 people), the data was collected in a real world context where people were actually looking to find partners as opposed to a lab setting, and many features were collected. Moreover, even to the extent that there exist equally rich datasets, they’re not in the public domain (sometimes for privacy reasons, sometimes because the researchers aren’t motivated), so nobody but the researchers would be able to analyze them. Fisman and Iyengar were unusually shrewd in their choice of experimental design, and unusually generous in going out of their way to anonymize and publish their data.
Of course, many people are familiar with the phenomena at an informal level, through observations of peers and personal experience, but the effect sizes may not have been quantified.
I found it all quite understandable, clear, and interesting =)
Thanks :-)
Although, an abstract up-front would probably help reach more readers..
Yes, several people have said this. What do you see as the key points of the article? It may seem as though I should know better than you :P, but I’m not sure how best to summarize the situation, which seems to have irreducible complexity, although impressions of this type often turn out to be mistaken.
What do you see as the key points of the article? / How is my new sumarry
Here’s what my “abstract” would be—apologies if the parts describing math are wrong.
“Speed Dating participants rated each other for attractiveness, fun, ambition, intelligence sincerity, and likability prior to choosing matches. While all participants were more likely to choose partners that others had rated highly, a principle component analysis revealed that a major source of variation was the degree to which participants differed in prioritizing ambition, intelligence, and sincerity vs attractiveness, fun, and likability. Further analysis suggests that this effect may be driven by… [demographic correlates from next posts, assortive mating, etc]
Our writing styles differ but your new summary seems to cover the same points, so we do agree on what they are! So much for irreducible complexity :P
I need to talk with psychology researchers to get a better sense for how what I’ve done fits in with the literature. In the past, I read some papers on the subject out of casual interest, but I haven’t done a deep dive.
I’m to be a psych or neuroscience graduate student fairly soon (depending on where I get accepted). Mate choice isn’t something I know a ton about and there are of course many, many better people to talk to than me, but I know enough that I could easily locate and understand the relevant literature. I’d be quite happy to collaborate if you are interested!
Anything that I might be able to help with? I have a background is in math education.
I’m not sure—there are a lot of unknown unknowns! Thank you for the links.
I’ve got a decent foundation in introductory calc and stats (by which I mean, I successfully memorized the relevant equations with a reasonable intuition about how they work) but the math bug didn’t really bite me until I took a proof based class on logic and set theory,finding it intuitive and fun. So, I thought I’d keep going in math, but I found myself seriously falling behind in subjects like linear algebra and never completed the major. (Got the certificate at least :D)
My impression is that I’ve got the fluid intuition for it but I start fall behind when cumulative crystallized knowledge requirements start increasing because I tend to forget things more quickly than others and rely on re-derivation a bit too much.
Statistics is probably going to be the most useful thing for me to learn, given my career choices—although I suspect I automatically gravitate towards less applied, proofy theory things more. (I’m probably too early in my math progression too early to know that for sure, but it certainly seemed like math which others found hard were easy for me, while some types of math which others found easy were really hard for me)
One sees that past a certain point, the low group is not responsive to increasing sincerity and intelligence, whereas the high group is.
^ from the article. Is this right? It seems like that aught to be reversed (low group is responsive, high group is not responsive)?
It’s an exercise in statistics. If you’re interested in the topic, the tl;dr is that being conventionally pretty helps a lot :-P
That’s not the tl:dr of this analysis at all. there’s Betty-Veronica type trade off effects, there’s individual differences, there are mechanisms of assortive mating...i can’t speak to the validity/certainty of the statistics but if the conclusions are valid there is a publishable amount of juicy stuff in here.
I realized that I should have gone into more detail in the way of justification, and added some material on this in the “What this means in tangible terms.” I still have to post the code that I wrote for this part so that my analysis is transparent.
Oh no that’s not what I meant!
I can’t speak to the validity because my quantitative skills are untrained to the point that I have to spend an entire day learning in order to pick up enough background to distinguish good math from bad math for anything more complex than a p-value, and until this changes (and I fully intend to change it) I have to just hope that the equation-using overlords know what they are doing.
I definitely wasn’t saying your math wasn’t well defined enough! I have no opinions on that particular matter, because, see above. It’s not the fault of your writing at all—you’d probably have to write a textbook to get me completely up to speed here. I’m generally sharp enough to know what you are trying to do with the math, but not nearly knowledgeable enough to know whether or not it is correct.
I very much enjoyed the results though! (Especialy the trade-off thing...it fits together with a lot of the other literature on prioritizing in long term and short term mating, and I don’t think anyone has ever supported that particular theory using this particular method. Kinda wish we had data on menstrual cycle phase to go with this :P If you’re planning on publishing you might want to cite stuff within that whole body of work.
I found it all quite understandable, clear, and interesting =) Totally disagree with shminux on the “dry” part. Although, an abstract up-front would probably help reach more readers..
Don’t worry, I didn’t make my changes in response to your comment specifically – I just realized that the section that I changed was weak because I was unnecessarily vague in my comments along the lines of “here is the data, but the effect size is smaller than it looks because of regression to the mean” – I could instead just show the degree to which past behavior predicted future behavior.
Anything that I might be able to help with? I have a background is in math education. You might find some of the pointers in the Cognito Mentoring math and statistics learning recommendations to be useful.
My experience has been that while statistics sometimes requires substantive mathematical knowledge, intuition built from exposure to data and experience with thinking about it gets you ~80% of the way to a full understanding. On the flip side, mathematical knowledge without exposure to real world data doesn’t get one very far...
It’s certainly natural to hypothesize that the differences came from some people looking for short term mates and others looking for long term mates, but I’m not sure that that’s what was going on. You’ll see what I mean when I show the demographic correlates. I think that your reference to assortative mating is highly relevant.
There was in fact more variation with respect to preference on the tradeoff dimension amongst women than amongst men. The comparison isn’t an apples to apples one, because e.g. men found women more attractive than women found men. The difference isn’t very large: 0.42 standard deviations vs. 0.36 standard deviations.
I need to talk with psychology researchers to get a better sense for how what I’ve done fits in with the literature. In the past, I read some papers on the subject out of casual interest, but I haven’t done a deep dive.
I would guess that some of the phenomena in the data are in fact previously unknown to academic psychology. The dataset is unusual in a number of ways: the sample size is large (~500 people), the data was collected in a real world context where people were actually looking to find partners as opposed to a lab setting, and many features were collected. Moreover, even to the extent that there exist equally rich datasets, they’re not in the public domain (sometimes for privacy reasons, sometimes because the researchers aren’t motivated), so nobody but the researchers would be able to analyze them. Fisman and Iyengar were unusually shrewd in their choice of experimental design, and unusually generous in going out of their way to anonymize and publish their data.
Of course, many people are familiar with the phenomena at an informal level, through observations of peers and personal experience, but the effect sizes may not have been quantified.
Thanks :-)
Yes, several people have said this. What do you see as the key points of the article? It may seem as though I should know better than you :P, but I’m not sure how best to summarize the situation, which seems to have irreducible complexity, although impressions of this type often turn out to be mistaken.
How is my new summary?
Here’s what my “abstract” would be—apologies if the parts describing math are wrong.
“Speed Dating participants rated each other for attractiveness, fun, ambition, intelligence sincerity, and likability prior to choosing matches. While all participants were more likely to choose partners that others had rated highly, a principle component analysis revealed that a major source of variation was the degree to which participants differed in prioritizing ambition, intelligence, and sincerity vs attractiveness, fun, and likability. Further analysis suggests that this effect may be driven by… [demographic correlates from next posts, assortive mating, etc]
Our writing styles differ but your new summary seems to cover the same points, so we do agree on what they are! So much for irreducible complexity :P
I’m to be a psych or neuroscience graduate student fairly soon (depending on where I get accepted). Mate choice isn’t something I know a ton about and there are of course many, many better people to talk to than me, but I know enough that I could easily locate and understand the relevant literature. I’d be quite happy to collaborate if you are interested!
I’m not sure—there are a lot of unknown unknowns! Thank you for the links.
I’ve got a decent foundation in introductory calc and stats (by which I mean, I successfully memorized the relevant equations with a reasonable intuition about how they work) but the math bug didn’t really bite me until I took a proof based class on logic and set theory,finding it intuitive and fun. So, I thought I’d keep going in math, but I found myself seriously falling behind in subjects like linear algebra and never completed the major. (Got the certificate at least :D)
My impression is that I’ve got the fluid intuition for it but I start fall behind when cumulative crystallized knowledge requirements start increasing because I tend to forget things more quickly than others and rely on re-derivation a bit too much.
Statistics is probably going to be the most useful thing for me to learn, given my career choices—although I suspect I automatically gravitate towards less applied, proofy theory things more. (I’m probably too early in my math progression too early to know that for sure, but it certainly seemed like math which others found hard were easy for me, while some types of math which others found easy were really hard for me)
^ from the article. Is this right? It seems like that aught to be reversed (low group is responsive, high group is not responsive)?
Yes, this was an oversight – thanks for correcting it.
As for everything else, how about we switch over to email? I’d be happy to hear from you – you can reach me at jsinick@gmail.com.