Also, Coscott can put multiple polls in the same comment, saving on space.
I would do this myself, rather than hinting that Coscott should do it themselves, but I don’t know whether Coscott really means “biological” when they write “biological”, as opposed to meaning e.g. “genetic” or “innate”.
I don’t see how replacing biological with inate or genetic changes the question. Describe a position for which the word choice matters. Also, you can rewrite it if you like, and I will retract my version. I assume any way you phrase it will get me the data I want.
I don’t see how replacing biological with inate or genetic changes the question. Describe a position for which the word choice matters.
Someone might believe in the causal graph (male students implicitly expected by their peers, teachers and family to study sciences, while female students aren’t) → (female students having lower science knowledge) → (fewer women in the sciences).
If one thinks about it, this causal chain implies a biological difference between male students & female students. Knowledge is stored in brains, so a difference in knowledge implies a difference in brains, which would be a biological difference. But this biological difference wouldn’t necessarily be an innate or genetic difference. Were the causal graph correct and sufficiently complete, the male-female knowledge difference would be a biological but non-innate difference generated non-genetically. So someone believing in that causal graph would say “yes” if asked whether the difference were “biological”, but “no” if asked whether it were “genetic” or “innate”.
In the spirit of crude empiricism, I’ll give it a go anyway and see what happens.
Poll: what probabilities do you assign to the following statements?
Men generally have a larger variance than women in most traits. [pollid:585]
Because of genetic differences between men & women, men generally have a larger variance than women in most traits. [pollid:586]
Men vary more in intelligence than women. [pollid:587]
Because of genetic differences between men & women, men vary more in intelligence than women. [pollid:588]
Men vary more in intelligence than women, and that contributes non-negligibly to gender imbalance in the sciences. [pollid:589]
Because of genetic differences between men & women, men vary more in intelligence than women, and that genetically-driven difference in variance contributes non-negligibly to gender imbalance in the sciences. [pollid:590]
You need to be super clear and explicit about whether you want answers as percentages, or as values between 0 and 1! Otherwise you get an unexploitable mish-mash.
If I am reading the less wrong comment formatting page correctly, using poll:probability (with brackets) will force responses to be between 0 and 1, to avoid this. (And Upvotes to whoever put strict input limits on a free text field! As a person who handles raw data, I often don’t see this, and it should be encouraged when performed.)
It doesn’t avoid the problem if people want to vote with a percentage < 1%, and try to do so with a 0-100 value (e.g., .5 meaning .5% rather than 50%).
“Too much Patriarchy bullshit in the data to tell”.
There are very strong indicators that discrimination is going to hurt your career prospects in the sciences quite badly if you happen to be in possession of two X chromosomes. Number of widely cited papers needed in order to be awarded tenure, pay at a given level of qualifications, ect.. And even if you have not read those studies, the fact that this happens is blatantly obvious to anyone that is not both male and fairly oblivious.
This is quite sufficient to explain the disparity all on its own. Any underlying “biology” is utterly swamped in significance by this.
Also, the balance of probability is that any given woman you encounter in the sciences is at least one out of smarter, more stubborn or just flat out much harder working than her stature in the field would indicate. Often all 3. Getting anywhere against the weight of discrimination and discouragement requires you to be an outlier to start with.
You mean the bit where boys are more likely to get into academic trouble ect? Heck yes. dont mistake patriarchy for a system that actually helps men—it is a social pathology and not kind to either gender. At all. It contains ideas about identity and status that can and will fuck you. The bad boy trying to be cool by blowing of class and acting out against authority didn’t come up with that set of behaviors de novo.
But I wonder whether we have an anchoring problem here. I myself used round numbers and notice that the median is a round number and that the probabilities go down in steps of 0.05 (and the mean follows suit almost linearly).
If anything the compound probabilities should show more or less geometric progression.
Anchoring to one of the values and then just roughly correcting for the difference in phrasing will not work (i.e. don’t add any precision).
Do I notice this correctly? Can this be fixed? How?
Also, Coscott can put multiple polls in the same comment, saving on space.
I would do this myself, rather than hinting that Coscott should do it themselves, but I don’t know whether Coscott really means “biological” when they write “biological”, as opposed to meaning e.g. “genetic” or “innate”.
I don’t see how replacing biological with inate or genetic changes the question. Describe a position for which the word choice matters. Also, you can rewrite it if you like, and I will retract my version. I assume any way you phrase it will get me the data I want.
Someone might believe in the causal graph (male students implicitly expected by their peers, teachers and family to study sciences, while female students aren’t) → (female students having lower science knowledge) → (fewer women in the sciences).
If one thinks about it, this causal chain implies a biological difference between male students & female students. Knowledge is stored in brains, so a difference in knowledge implies a difference in brains, which would be a biological difference. But this biological difference wouldn’t necessarily be an innate or genetic difference. Were the causal graph correct and sufficiently complete, the male-female knowledge difference would be a biological but non-innate difference generated non-genetically. So someone believing in that causal graph would say “yes” if asked whether the difference were “biological”, but “no” if asked whether it were “genetic” or “innate”.
Although I question if that is worth it since several people already answered.
In the spirit of crude empiricism, I’ll give it a go anyway and see what happens.
Poll: what probabilities do you assign to the following statements?
Men generally have a larger variance than women in most traits. [pollid:585]
Because of genetic differences between men & women, men generally have a larger variance than women in most traits. [pollid:586]
Men vary more in intelligence than women. [pollid:587]
Because of genetic differences between men & women, men vary more in intelligence than women. [pollid:588]
Men vary more in intelligence than women, and that contributes non-negligibly to gender imbalance in the sciences. [pollid:589]
Because of genetic differences between men & women, men vary more in intelligence than women, and that genetically-driven difference in variance contributes non-negligibly to gender imbalance in the sciences. [pollid:590]
You haven’t learnt Yvain’s lesson!
You need to be super clear and explicit about whether you want answers as percentages, or as values between 0 and 1! Otherwise you get an unexploitable mish-mash.
If I am reading the less wrong comment formatting page correctly, using poll:probability (with brackets) will force responses to be between 0 and 1, to avoid this. (And Upvotes to whoever put strict input limits on a free text field! As a person who handles raw data, I often don’t see this, and it should be encouraged when performed.)
It doesn’t avoid the problem if people want to vote with a percentage < 1%, and try to do so with a 0-100 value (e.g., .5 meaning .5% rather than 50%).
Very good point. I had not thought of this earlier, but it is entirely correct.
This has taught me that I find it more intuitive to think in terms of conditional probabilities than marginal probabilities.
“Too much Patriarchy bullshit in the data to tell”.
There are very strong indicators that discrimination is going to hurt your career prospects in the sciences quite badly if you happen to be in possession of two X chromosomes. Number of widely cited papers needed in order to be awarded tenure, pay at a given level of qualifications, ect.. And even if you have not read those studies, the fact that this happens is blatantly obvious to anyone that is not both male and fairly oblivious.
This is quite sufficient to explain the disparity all on its own. Any underlying “biology” is utterly swamped in significance by this.
Also, the balance of probability is that any given woman you encounter in the sciences is at least one out of smarter, more stubborn or just flat out much harder working than her stature in the field would indicate. Often all 3. Getting anywhere against the weight of discrimination and discouragement requires you to be an outlier to start with.
Does Patriarchy explain the left tail too?
You mean the bit where boys are more likely to get into academic trouble ect? Heck yes. dont mistake patriarchy for a system that actually helps men—it is a social pathology and not kind to either gender. At all. It contains ideas about identity and status that can and will fuck you. The bad boy trying to be cool by blowing of class and acting out against authority didn’t come up with that set of behaviors de novo.
Seems that we have learned that P(A&B)<=P(A).
But I wonder whether we have an anchoring problem here. I myself used round numbers and notice that the median is a round number and that the probabilities go down in steps of 0.05 (and the mean follows suit almost linearly).
If anything the compound probabilities should show more or less geometric progression.
Anchoring to one of the values and then just roughly correcting for the difference in phrasing will not work (i.e. don’t add any precision).
Do I notice this correctly? Can this be fixed? How?
I rated the second question as more likely than the first because I think “most traits” means something different in the two questions.
OK, I guess I’ll keep mine up too to since anyone reading this far might as well just vote on both