Women and men are different!
January 5, 2012 3:44 PM   Subscribe

This just in ... Men are different from women. OK, scratch that. More different than researchers had previously thought.
posted by anothermug (85 comments total) 14 users marked this as a favorite


 
It's a start.

Now do it again only with a significant number of transexuals, gays, lesbians, and other non-standard expressions of sex and gender from around the world as part of the study, and let's see how all that plays into the equation. THAT, to me, is where the truly interesting bits are going to lie.
posted by hippybear at 3:58 PM on January 5, 2012 [20 favorites]


Which fallacy is it where I point out that two of the three authors are from the University of Manchester Business School?
posted by Adventurer at 3:59 PM on January 5, 2012 [20 favorites]


"Meta-analysis" is statistics-speak for "almost certainly bogus." It's very, very difficult to accurately compare many different studies that were done in many different situations with many different factors, some of which will always remain unstated. And when you're just quoting the meta-analyses that other people have done, well... let me just say I'm ridiculously skeptical, and that is after reading just a few paragraphs here.
posted by koeselitz at 4:00 PM on January 5, 2012 [5 favorites]


I've been reading about this on Dienekes Anthropology Blog. I'm hoping that experts in Social Science Statistics can wax eloquent about the difference between multivariate and univariate analysis, perhaps "Explain Like I'm Five."

SUMMARY: overlap of only 10% between the male and female distributions. Even excluding the factor showing the largest univariate ES, the global effect size was D = 1.71 (24% overlap). These are extremely large differences by psychological standards.

Authors of competing theories are saying that the data is flawed-- after all, what red-blooded male would ascribe to themselves the attributes of WARMTH or SENSITIVITY???
posted by ohshenandoah at 4:00 PM on January 5, 2012 [1 favorite]


Seconding Bwithh. Supposed psychological differences between men and women are almost certainly a result of environment above biology. This is admittedly an intuitive judgement, but considering that the link is to an abstract and not the full paper, we really can't see for ourselves what data the researchers used. Does anyone here have access to the text?

Regardless, what is almost certain is that this study is going to be dragged out by light news programs and pop psychologists looking to sell books, regardless of its actual merits, and that is more than enough to make me start preemptively rolling my eyes.
posted by JHarris at 4:03 PM on January 5, 2012 [3 favorites]


"Meta-analysis" is statistics-speak for "almost certainly bogus."

I disagree, vehemently. Whilst those challenges you mention do indeed apply to metastudies, that's why they're so valuable and time-consuming - researchers are well aware of these difficulties.

However, we are in strong agreement on the merits of this particular paper...
posted by smoke at 4:03 PM on January 5, 2012 [11 favorites]


"Meta-analysis" is statistics-speak for "almost certainly bogus."

Maybe I'm misunderstanding you. THIS paper was meta-analysis? I thought the paper by Hyde, that they were refuting, was the meta-analytical one.
posted by Edgewise at 4:07 PM on January 5, 2012 [1 favorite]


Even if this is 100% true it would still fail as a predictive aid when dealing with individuals, which is unfortunate because too many people already expect all men and all women to act and think alike and are confused when they don't.
posted by The Card Cheat at 4:09 PM on January 5, 2012 [9 favorites]


Would someone mind doing a bit of a layman's translation of the methodology and findings? I have a reasonably robust understanding of quantitative analysis for a non-academic, but I was totally lost reading that.
posted by lunasol at 4:11 PM on January 5, 2012


Edgewise: “Maybe I'm misunderstanding you. THIS paper was meta-analysis? I thought the paper by Hyde, that they were refuting, was the meta-analytical one.”

It's kind of sideways; they're using meta-analyses to justify their objections to Hyde concerning whether it's kosher to use self-reporting. It feels like a fuzzy sort of collection of things that seem suspect to me, but I'm still reading.
posted by koeselitz at 4:12 PM on January 5, 2012


(Or if someone else has done that elsewhere, a link would be great.)
posted by lunasol at 4:12 PM on January 5, 2012


The interesting thing is that a lot of these results don't corroborate popular gender stereotypes. Women score higher on Apprehension (i.e., lack of self-confidence)? Not exactly news to women's rights advocates, who have been trying to raise awareness about "impostor syndrome" for years. Men score higher on Rule-Consciousness? But I thought that we were in the midst of a Boy Crisis, where girls are rewarded for sitting placidly in their seats and following orders, while their brothers are penalized by the female-dominated schools for their creative rebelliousness.

But based on the title of the article ("Mars and Venus") and of this FPP ("this just in") I somehow doubt that these will be the aspects of the research that people choose to focus on.
posted by Ralston McTodd at 4:14 PM on January 5, 2012 [4 favorites]


This is admittedly an intuitive judgement, but considering that the link is to an abstract and not the full paper, we really can't see for ourselves what data the researchers used. Does anyone here have access to the text?
Did you try the scrollbar?
posted by delmoi at 4:15 PM on January 5, 2012 [1 favorite]


It's tough, because I acknowledge that this must be a very difficult thing to study. I mean, I have reservations about self-reporting about personality traits; but how else would you test personality traits statistically? I'm not sure there's really another way.
posted by koeselitz at 4:19 PM on January 5, 2012 [2 favorites]


It's kind of sideways; they're using meta-analyses to justify their objections to Hyde concerning whether it's kosher to use self-reporting. It feels like a fuzzy sort of collection of things that seem suspect to me, but I'm still reading.

OK, so I'm guessing you think they are both bogus?

If nothing else, anothermug is gutsy to post this link to mefi, where it will will be sure to raise hackles. I'm agnostic on this particular paper, because I lack the ability to evaluate it. I am biased, though, because I already believe that men and women have significant innate personality differences.
posted by Edgewise at 4:21 PM on January 5, 2012 [2 favorites]


The multivariate effect size metric D absolutely makes sense. What I'm not sure about is whether it's appropriate to interpret the magnitude of D with the same heuristics used for Cohen's d.
posted by a snickering nuthatch at 4:22 PM on January 5, 2012


Did you try the scrollbar?

I did and...

Oh! Okay, nevermind, I saw the citations and was conditioned by my own paper writing to conclude that was the end of the document. Thanks.
posted by JHarris at 4:29 PM on January 5, 2012


Edgewise: “OK, so I'm guessing you think they are both bogus?”

I think it's called a 'knee-jerk reaction.' The kind where little forethought is involved.

sigh...
posted by koeselitz at 4:30 PM on January 5, 2012


. Supposed psychological differences between men and women are almost certainly a result of environment above biology

Why do you think this? It is demonstrably not true in most animal species where differences in behavior are hardwired. It doesnt mean we can all get along but I think its patently obvious that there are some key differences between male and female humans.
posted by fshgrl at 4:32 PM on January 5, 2012 [2 favorites]


Argh can't. We can't all get along. Stupid phone.
posted by fshgrl at 4:33 PM on January 5, 2012


JHarris: “Supposed psychological differences between men and women are almost certainly a result of environment above biology.”

I agree with you, but thankfully this paper doesn't seem to delve into the difference between "innate" and "socialized" sex differences, so I guess it doesn't really deal with that question. (As far as I can tell; I'm happy to be corrected if this is wrong.) It only seems to be about whether the differences are there, and whether they are significant. But there could well be significant socialized differences between men and women.

Probably the only criticism I am actually equipped to give of this paper is that it seems to only study the US; it would be revealing and even necessary, I think, to study the differences in many other countries simultaneously. But that's admittedly not much of a criticism.
posted by koeselitz at 4:37 PM on January 5, 2012 [2 favorites]


fshgrl: “Why do you think this? It is demonstrably not true in most animal species where differences in behavior are hardwired. It doesnt mean we can all get along but I think its patently obvious that there are some key differences between male and female humans.”

Well. To argue that there are significant innate differences between men and women, you would probably have to argue that men and women have certain innate characteristics that never change. And, given the diversity and variegation of history and the many societies we're aware of, that's a very difficult thing to argue.

I guess you could get around this by arguing that the sexes are amorphous but eternally locked in opposition, always different from each other even as they change separately. But I don't know how that could work.
posted by koeselitz at 4:40 PM on January 5, 2012 [3 favorites]


too much math: didn't read
posted by jonmc at 4:41 PM on January 5, 2012 [1 favorite]


I hear women have boobies.
posted by cmoj at 4:43 PM on January 5, 2012 [3 favorites]


Wow. If someone put a gun to my head and said, "figure out in 1,000,000 words how to say that women are different from men, " I couldn't do it. There are just some things that are too obvious to put into academic-speak.
posted by Melismata at 4:46 PM on January 5, 2012


Don't worry; one of our greatest minds is working on the problem.
posted by Crabby Appleton at 4:46 PM on January 5, 2012


I hear women have boobies.

"Girls have catchers mitts, boys have baseball bats...."
posted by jonmc at 4:47 PM on January 5, 2012


So I'm a Meyers Briggs ENFP (Sorry Ohshenandoah, but I like power tools and beating my friends with sticks if that makes you feel better.) and all I can think is, "So, what exactly are the units on Sensitivity?" I can only imagine how the Thinking Sensing types view this sort of thing.

I'd love to see a link the the questionnaire because I'm pretty sure we could debate the merits of the questions and what, exactly, they mean and their relative correlation to a person's warmth or sensitivity and what it is we mean by warmth and sensitivity longer than any of us would want to. Which as about how long this is going to make the pop-psychology rounds.
posted by Kid Charlemagne at 4:52 PM on January 5, 2012


I did a metanalysis using a python script, some webscraping, and the gender field on the profile page to figure out the differences between typical male and female MeFites:

Turns out, men comment like this, while women comment like this.
posted by mccarty.tim at 4:53 PM on January 5, 2012 [2 favorites]


jonmc: Girls have catcher's mitts, boys have baseball bats...

My favoritest sex ed film taught me that one.
posted by mccarty.tim at 4:56 PM on January 5, 2012 [1 favorite]


Melismata: “Wow. If someone put a gun to my head and said, ‘figure out in 1,000,000 words how to say that women are different from men, ’ I couldn't do it. There are just some things that are too obvious to put into academic-speak.”

So far in this thread, one of the things women and men seem to differ on is that they disagree on whether men and women are significantly different.
posted by koeselitz at 5:07 PM on January 5, 2012 [1 favorite]


wooh gender essentialism.
posted by wayland at 5:19 PM on January 5, 2012 [3 favorites]


Men are different from women.

Indeed. And according to Stephen Hawking they are also 'A Complete Mystery'.
posted by flapjax at midnite at 5:23 PM on January 5, 2012


men are enormous segmented biomechanical things of unknown provenance, women are a rapidly changing set of geometric phosphors emanating from every surface
posted by This, of course, alludes to you at 5:27 PM on January 5, 2012 [15 favorites]


Well. To argue that there are significant innate differences between men and women, you would probably have to argue that men and women have certain innate characteristics that never change.

That's a tremendous leap. "Present," "innate," and "immutable" are all different terms and concepts. There could be differences between men and women that are generally present, through socialization or something else or a combination of factors, not accounting for the differences between individuals or differences across societies. There could be innate differences which are hardwired in some way or another, but again, there's plenty of individual variation and interplay with other individuals and with society as a whole. And then there would be immutable traits, which as far as personality is concerned, would be relatively few and far between.
posted by Sticherbeast at 5:28 PM on January 5, 2012 [2 favorites]


Why should I care? Seriously, why?
posted by benito.strauss at 5:34 PM on January 5, 2012 [1 favorite]


Supposed psychological differences between men and women are almost certainly a result of environment above biology

I don't know if there are innate biological differences in the psychology of men and women but I do know that you have no real basis for your claim. You might in fact be correct that any differences are a result of environment, but if you are correct it is by accident and not reasoning.

I suspect there are some innate differences simply because there are unquestionably physical differences and it seems dubious to me that differences in psychology don't inherently follow from differences of a physical nature... but I don't know this to be the case. I definitely can't make any statements like "almost certainly". Neither can you.

I mean, there is no question that various hormones have a profound effect on behavior. If one sex or the other has more or less of a certain hormone on average, how could that not influence the median point on a curve related to that behavior?
posted by Justinian at 5:36 PM on January 5, 2012 [2 favorites]


When we raise a man and a woman in identical conditions and not under conditioning from birth from vast sociopolitical structures predicated on sex differentiation in our species and then examine the result, then we can talk about men being different from women as if we can tell how much is environment and how much is genetic.
posted by winna at 5:37 PM on January 5, 2012 [4 favorites]


I hear women have boobies.

Not necessarily. Said as the unfortunate owner of a perky pair of manboobs.
posted by MikeKD at 5:44 PM on January 5, 2012


Hm, this post dovetails nicely with the other one on the blue today about how results of research often get "less true" over time.

Like the fellow in "Diary of a Madman", sometimes I stay up late into the night reading certain Important Documents - though in my case these documents assert that Men Are Very Different From Women and Are Violent, Ambitious, Less Neurotic, Less Loving and Need More Young Sexual Partners Or They Will Die. And like the madman, after enough time staring at these things, I start to see the text that is written behind the text, and that text says "shut up and stop complaining about sexism, abuse, misogyny and bias, women, because science!"
posted by Frowner at 5:58 PM on January 5, 2012 [18 favorites]


If anyone's interested in the argument the paper is actually making as opposed to setting their flamethrowers against an army of strawmen, there's an interesting refutation of their methodology by Hyde (the guy whose hypothesis they're attacking) under the "Comments" tab. His is the very first comment: "The Distance Between North Dakota and South Dakota."
posted by yoink at 6:03 PM on January 5, 2012 [1 favorite]




(and really cool to boot)
posted by a snickering nuthatch at 6:14 PM on January 5, 2012


The multivariate effect size metric D absolutely makes sense. What I'm not sure about is whether it's appropriate to interpret the magnitude of D with the same heuristics used for Cohen's d.

Per Eq 1 and their R code, they're reporting D<-sqrt(rowSums((cohen_d%*%pooled_cov)*cohen_d)) [ie, sqrt((cohen_d%*%pooled_cov)%*%cohen_d)] where pooled_cov is the pooled correlation between personality traits for men & women. For small correlations D, will approximately be the magnitude of the vector of Cohen d's. Importantly, they don't seem to normalize by the number of traits, which makes me agree with you that it's not appropriate to compare D directly to mean(abs(d)) as they do, since each additional independent trait will necessarily increase D. I suspect this is also why their D with 15 traits is so much larger than those previously reported using only the Big-5 traits. And it makes their conclusion sorta silly -- a bit like saying "well, 11 feet to the left isn't far, 10 feet forward isn't far, 9 feet up isn't far... on average that's only 10 feet, but put it all together and now you're 17 feet away!"

In other words: they did the math and reported a provocative result, but if I'm understanding them correctly, they -- and their referees -- seem not to have thought very much about what their math means.
posted by Westringia F. at 6:20 PM on January 5, 2012 [4 favorites]


Even if this is 100% true it would still fail as a predictive aid when dealing with individuals, which is unfortunate because too many people already expect all men and all women to act and think alike and are confused when they don't.

I'm still struggling with the fact that not all humans think and act alike; never mind dividing them up into subgroups.
posted by Faint of Butt at 6:24 PM on January 5, 2012 [3 favorites]


since each additional independent trait will necessarily increase D

Yes, that's what makes me think intuitively this is not cool.

Quick clarification- does the expected value of the norm of a random vector of standard normals equal 1, or sqrt(#dimensions) (or something else I'm not foreseeing)? If the former, then the use of D as presented makes more sense; if the latter, it seems like it should be further divided by sqrt(#dimensions) before compared to particular values for Cohen's d.
posted by a snickering nuthatch at 6:32 PM on January 5, 2012


Jpfed: the latter -- sqrt(#dimensions).
posted by Westringia F. at 6:47 PM on January 5, 2012


I think its patently obvious that there are some key differences between male and female humans.

Now if we could agree on what those were, as far as personality, we'd have a hope in hell of studying it without biases.


I think it's more priorities than personality. Personality is so complex but the drive to mate with a person of a particular kind or to have babies by age 35, not to mention different circulating levels of testosterone are going to cause some pretty fundamental differences in how you go about life. You can see it in cats and dogs and cattle and goldfish too. Different priorities, different drives.

Trying to mix that with personality which is something we can only really describe with words (imprecise little buggers that they are) is probably doomed in terms of getting a definitive answer of any kind. Now, if you radio collared people and followed them around for years observing their behavior or bred a genetically homogenous line to use for lab based experiments you'd probably get better results [/biologist]
posted by fshgrl at 6:52 PM on January 5, 2012 [1 favorite]


When we raise a man and a woman in identical conditions and not under conditioning from birth from vast sociopolitical structures predicated on sex differentiation in our species and then examine the result, then we can talk about men being different from women as if we can tell how much is environment and how much is genetic.

No, then you could only talk about how that man is different from that woman.
posted by UbuRoivas at 6:54 PM on January 5, 2012 [6 favorites]


Based on what I've experienced in dealing with non-human primates, I would argue that there are greater personality differences within the sexes than between the sexes.

When primatologists first started visiting the fields and examining our cousins, they wrote a lot about male dominance hierarchies, but reported little on female behavior. As primatology became a more gender balanced field, it was discovered that females often have dominance hierarchies as well, with aggressive behavior that is more subtle than aggression between males.

It's also worth noting that economies highly play out how gender is expressed in both humans and non-human primates. When resources are scarce, males typically become dominant over females. When the resources are more abundant, then egalitarian or female dominant societies arise. It is easy to look at our current social structures and decide that males are more aggressive than females, but further within and between species analysis shows that culture contributes a huge amount to expression of aggression. Look at the difference within the US at rates of aggression between 'Cultures of honor' and cultures without.

It is also worth noting when examining this study that the majority of the respondents were white, and presumably share a similar culture.
posted by PrimateFan at 6:58 PM on January 5, 2012 [11 favorites]


Until there's no longer a stigma associated with men appearing "feminine" (or "effeminate"), until any coach who uses "girl" as a slur is considered ridiculous instead of normal, I don't see any way to not be deeply suspicious of any study like this or its methods, whether they involve self-reporting or assessment by others. The internet experiment wherein changing the apparent gender of a username (even from feminine to neutral) can get the same user very different feedback comes to mind immediately for me. As for self-reporting, we know that sometimes even women themselves can label their own actions "selfish" instead of "reasonable," "bossy" (for example) instead of "assertive," "slutty" instead of "virile," when men would (for lack of a reason to do otherwise) apply the less-compromised set of adjectives to their own identical behaviors, strictly on account of internalized ideas about gender. Could there possibly be anything else the original tests aren't compensating for, whether in the test-takers or in the people who devised the scenarios? I think a great deal of extra scrutiny is in order when you're purporting to demonstrate the immutable differences between half the people in America vs. the half whom were only granted the right to vote in 1920, on account of the majority of that first group didn't consider them fully human.
posted by Adventurer at 7:03 PM on January 5, 2012 [12 favorites]


Men are brought into this world via the birth canal, women are brought into this world via the birth canal.

On NASA's current budget, this is always on Earth.
posted by mccarty.tim at 7:05 PM on January 5, 2012 [3 favorites]


Men are brought into this world via the birth canal, women are brought into this world via the birth canal.

I was a cesarean.
posted by chrchr at 7:30 PM on January 5, 2012 [5 favorites]


Jesus Christ, we're all human beings. Variations between individuals are bigger than variations between groups, and that goes for the general groupings of those with XX and XY chromosomes, too. Sheesh.
posted by jokeefe at 7:40 PM on January 5, 2012 [5 favorites]


Said as the unfortunate owner of a perky pair of manboobs.

aka 'mannaries.'
posted by jonmc at 7:50 PM on January 5, 2012


All men have mammary glands, they just don't do much. Usually.
posted by LogicalDash at 8:03 PM on January 5, 2012



Thanks,
for the mammaries...

posted by flapjax at midnite at 8:26 PM on January 5, 2012 [1 favorite]


There's a large body of research into the psychiatric effects of testosterone, with most studies showing some increases in aggression and other personality changes. And if men produce on average about ten times as much testosterone, why wouldn't they also exhibit these personality differences?

I understand the desire from the feminist movement to show equality in all things, because it's easier to say "we deserve equality" when everything says we're "equal". But just because we may not be biologically equal doesn't mean we don't all deserve equality under the law.
posted by formless at 8:32 PM on January 5, 2012 [1 favorite]


As I understand it, aggression isn't one of the traits that personality inventories tend to measure.

I don't know enough about personality psychology to know why that is — whether there's a good reason to say "no, aggression really doesn't count as a personality trait," or whether it's just a historical accident or something.
posted by nebulawindphone at 8:44 PM on January 5, 2012


fshgrl: Argh can't. We can't all get along. Stupid phone.

If you can't get along with it, just get a new phone.
posted by Chuckles at 8:49 PM on January 5, 2012


Aggression is reasonably hard to measure using a personality test. A lot of the recent studies on it has used opportunities to "punish" other people and measuring the difference in time and severity, and there's some sign that it correlates with supplying more aggressive terms in an ambiguous letter-fill task. It's hard to link that up with what most people think of aggression, though, and right now I know of no evidence that someone who meets "aggression" characteristics in a standard laboratory experiment are any more aggressive outside of it or toward another human being.

Humans are remarkably complex, and gender socialization has been shown to begin at birth (studied using infants randomly assigned a gender, the differences being in how the infants were held and spoken to, and I believe there was a toddler study as well showing that adults intervened more quickly to help female-marked toddlers regardless of their actual gender, and left male-marked toddlers to struggle longer and often succeed on their own; the latter may not have had randomly-assigned genders for the toddlers).

The amount of individuals who defy or are annoyed by gender-biased assumptions, though, points to something more complex than socialization going on. As well, the effects of priming using non-chosen characteristics like gender and the associated stereotypes is both interesting and dismaying in terms of how large an effect stereotypes and expectations have on performance.

A more complicated look at gender could be interesting, and I'm all for more specific and accurate data, but one ongoing flaw of any study where subjects are grouped by an existing characteristic - like self-reported gender - no causal conclusions linking those characteristics and the variables being studied can ever be made.

One of the biggest problems in the general population is how much people mistake the correlational data of male/female comparison experiments with the causal data from other studies where subjects are randomly assigned to a group pre-experiment.
posted by Deoridhe at 9:42 PM on January 5, 2012 [7 favorites]



Thanks,
for the mammaries...


It's OK for you to sing. With me it's serious. Mammaries are all I have to cling to.
posted by howfar at 9:51 PM on January 5, 2012 [1 favorite]


A thread about differences between the sexes plastered over with an obnoxious derail about breasts? Seriously?
posted by koeselitz at 10:18 PM on January 5, 2012 [1 favorite]


Mod note: Yes, can we please cut out the kiddie stuff now?
posted by taz (staff) at 10:28 PM on January 5, 2012 [2 favorites]


"And if men produce on average about ten times as much testosterone, why wouldn't they also exhibit these personality differences?"

I don't know enough about the science of hormones to say whether or not this is true, but an easy answer would be that studies with the psychiatric effects of testosterone focus on abnormal increases or decreases. It's perfectly possible to have similar outcomes in different sexes with different net amounts if the baselines are different.
posted by klangklangston at 11:22 PM on January 5, 2012


For those having trouble interpreting the results, there's this helpful chart.
posted by Pyrogenesis at 12:08 AM on January 6, 2012 [3 favorites]


Why should I care? Seriously, why?

I suppose you mean these differences don't or shouldn't matter, and I can see that point. However, if you take a certain view of the world you will be disposed to argue that large personality differences may justify differential patterns of recruitment to certain professions and roles, including political ones. If you're of a different cast of mind on the other hand you may be disposed to argue that any personality differences need to be taken into account in ensuring that recruitment and election processes are not unfairly biased.

Either way, you'd want to know the facts, and even if you think in some third, basically neutral way you need to know about it because those other two sets of people are going to be trying to mess with your job and possibly your government.
posted by Segundus at 2:15 AM on January 6, 2012 [3 favorites]


Well. To argue that there are significant innate differences between men and women, you would probably have to argue that men and women have certain innate characteristics that never change

No, you wouldn't. Innate doesn't mean immutable. You might have a genetic diposition to being tall, but without proper nutrition, you're probably going to be short. You'll probably still be taller than someone with the same nutrition but without your genes, however, and people sharing your genes will probably be taller on average. Your argument is too reductive; you can't simplify the interaction between genes and envrionment to that level without losing signal.
posted by spaltavian at 6:40 AM on January 6, 2012 [1 favorite]


Why should I care? Seriously, why?

I...I don't know. Why care about anything?

I don't have much truck with the evo-psych crowd, and I don't have enough background knowledge to weigh in on one side or another of this meta-analysis, but if there did exist a thorough, credible study on general personality differences between genders, I would read it with great interest.
posted by Sticherbeast at 6:59 AM on January 6, 2012 [1 favorite]


"Girls have catchers mitts, boys have baseball bats...."

I wonder how the speaker thought baseball was played.
posted by Sticherbeast at 7:00 AM on January 6, 2012 [1 favorite]


Even if this is 100% true it would still fail as a predictive aid when dealing with individuals, which is unfortunate because too many people already expect all men and all women to act and think alike and are confused when they don't.

This is something that we fail to teach enough of.
posted by gjc at 7:12 AM on January 6, 2012 [2 favorites]


Well, if you throw enough sample size with the right tool you can get statistical significance out of almost everything, especially with tests designed to tease out difference. Yes, there are important gender differences. However, outside of professional sports and a few other places, the argument that a person's gonads and hormones are a better qualification than education and experience doesn't make a lot of sense. Looking at SAT scores for example the differences between men and women in math (31) is about the same as the difference between juniors vs. seniors (32), and religious vs. public schools (27). Having a parent with a 4-year college degree is good for approximately 62 points over parents with a HS education.

Discriminating against individuals on the basis of those small, aggregate differences will mean that you're wrong almost half the time.
posted by CBrachyrhynchos at 7:15 AM on January 6, 2012 [3 favorites]


"Girls have catchers mitts, boys have baseball bats...."

I wonder how the speaker thought baseball was played.


Indeed. Golf is a much better analogy. In bedroom golf, the goal is to get the club in while leaving the balls out.

(Unless the baseball metaphor is about domestic violence, of course.)
posted by Pyrogenesis at 7:20 AM on January 6, 2012


I was thinking about this article linked yesterday on the "decline effect" - the article talks about the fact that certain important studies can't be replicated very well, and that the replicability decreases over time, apparently even in cases where the experiments are re-run by the same people who did them initially.

One of the examples was about symmetry and mating - a bunch of studies in different fields showed - for a while - that symmetry was a good proxy for genetic health and that female creatures, including humans, preferred symmetrical mates. Apparently there was even a study showing that women enjoyed sex more if their partners had more symmetrical features. And then, per the article, all these results - these important results with depressing news about sex, desirability and gender - fell apart. It's not that the studies were full of lies; they just weren't replicable. Why? The article talks about all kinds of things, most of which revolve around "scientists see what they want to see" and "studies that give the desired results are more likely to be published and then more likely to be cited. The article even mentions that this is particularly true of fashionable topics.

Frankly, it's pretty goddamn fashionable to find serious business sciencey proof that women and men are very different and that these differences are innate and unchangeable. Why is this so fashionable? IME, because it dovetails with a lot of contemporary social anxieties and the needs of corporations - first, precisely because women, queers, gender non-conforming men and trans folks are messing up the usual understanding of what it means to be a gender and we're becoming more visible and less easy to exploit or abuse, so there's a LOT of social anxiety around gender; and second, because it's nice to be able to market a lot of crap to women based on 'science'. Oh, women genetically need to be recruited to these fields more than those, and you need to have pink recruiting brochures because women like to gather berries and berries are pink, etc etc. And fields that are really rough and tumble and kind of douche-y and male dominated - like analytic philosophy, frex - those fields only attract weird women who aren't typical of their gender, so we don't need to change a whole field just for those freaky-deaky outliers.

Anyway, I have a LOT of trouble believing research that basically shows precisely what mainstream people want to believe about gender and and precisely what lots of women, queers and trans folks are saying isn't true by their experience. The rewards for "proving" that the status quo is inevitable are always high.
posted by Frowner at 7:40 AM on January 6, 2012 [17 favorites]


For more data points contrary to the same old same old '80s BS, check out Women From Another Planet: Our Lives in the Universe of Autism. (I believe I found a reference to this book on the blue or the green).

The title is actually a riff on "Men Are From Mars..." but as you might expect, the collaborators don't much agree with the notion of gender as anything but a social construct.

Anyway, I have a LOT of trouble believing research that basically shows precisely what mainstream people want to believe about gender and and precisely what lots of women, queers and trans folks are saying isn't true by their experience. The rewards for "proving" that the status quo is inevitable are always high.
posted by Frowner at 9:40 on January 6 [3 favorites +] [!]


Shortly to be flagged as awesome.
posted by Currer Belfry at 9:17 AM on January 6, 2012 [1 favorite]


.... because those other two sets of people are going to be trying to mess with your job and possibly your government.

I guess I thought it was screamingly obvious that socialization by gender plays such a huge role in personality formation that any effects due to underlying sex are minor by comparison. It doesn't seem like this study addresses that issue (or did I miss it), and engaging with this study implicitly denies this.

if there did exist a thorough, credible study on general personality differences between genders, I would read it with great interest.

Ah. Maybe my mind is too willing to surrender to science. If I got deeply into such a study it would end up interfering with my ability to see people as who they specifically are. I would make a whole lot of assumptions based on the simple observable. (I'm not being sarcastic here; I'm not implying that you would do the same. I'm just explaining why I would avoid such a study. )
posted by benito.strauss at 9:27 AM on January 6, 2012 [1 favorite]


The rewards for "proving" that the status quo is inevitable are always high.

Worse than that: their numbers are inflated because they have used an inappropriate method. The article's contention (as summarized here) is that men and women are "more different than researchers had previously thought," but -- related to my & Jpfed's points above -- they have done it in such a way that simply considering more personality traits will necessarily increase their metric D... and they've got 15. Despite the fact that there are well known, well-established procedures for accounting for the number of variables in a model (likelihood ratio tests w/ appropriate degrees of freedom, AIC, BIC, &c.), they just let the small, insignificant univariate distances pile up.

Notice, also, there's not a p-value anywhere to be found. That's because they never actually compute the significance of the multivariate model. In fact, they're not even proposing a true multivariate model, but rather just combining the univariate d's. Worse, they're doing so in a way that "rewards" correlated traits, whereas a true multivariate model, selected parsimoniously, would remove collinearities.

The fact that they are committing such basic and profound errors and then have the hubris to then say that "the idea that there are only minor differences between the personality profiles of males and females should be rejected as based on inadequate methodology" is galling. The fact that it didn't raise suspicions in peer review is just fucking depressing.

--
Here's the basic jist of the D inflation:

For each of the 15 traits, they compute a "distance" (Cohen's d) between men and women, which is the difference between the average women's score for that trait and the average men's score divided by the pooled SD for men & women for that trait; that is, they're scaling the difference between the men & women by the spread of individual variation. The absolute values of those d's range from basically 0 to 1.34 (Table 1) with an average of about 0.3 on a scale where the assumed standard deviation is 1. Most of the d's simply aren't that big (the largest is for "sensitivity," and even that one would have p=0.09 when compared to a std normal).

Then they go to combine them all by computing the Mahalanobis distance, given in Eq 1. Like I explained above, if the traits are all independent (uncorrelated), that works out to sqrt(sum(d^2)) for all the d's -- it's like moving a blocks north, then b block east, so the total distance is sqrt(a^2+b^2) blocks. Fractional correlations can be thought of as going N then some angle NE, etc. What happens if two traits are perfectly correlated? That's like moving a blocks north, and then moving ANOTHER b blocks north: now you're a+b blocks away, farther than in the independent example. Perfectly anti-correlated is the reverse -- going north then south -- but since you'd expect anticorrelated traits to have opposite signed differences (ie, if trait A goes up as trait B goes down and men are "more" trait-A than women, we expect they'll also be less trait B), this means that the anticorrelated ones are more of a "a blocks N and NEGATIVE b blocks S, ie, b blocks N" situation.

That is: with each additional trait you increase D. That's something they've done 15 times (vs just once for the little d's, or 5 for the big-5 D's they report), yet haven't accounted for anywhere. If they duplicate each of their traits and recompute it with all 30, they could trumpet an even bigger D from the rooftops; you'd say, "hey, no fair, that's inflating it" -- and yet that's pretty much what's happening here.

posted by Westringia F. at 9:33 AM on January 6, 2012 [17 favorites]


Well. To argue that there are significant innate differences between men and women, you would probably have to argue that men and women have certain innate characteristics that never change

No, you wouldn't. Innate doesn't mean immutable. You might have a genetic diposition to being tall, but without proper nutrition, you're probably going to be short. You'll probably still be taller than someone with the same nutrition but without your genes, however, and people sharing your genes will probably be taller on average. Your argument is too reductive; you can't simplify the interaction between genes and envrionment to that level without losing signal.


Also worth noting that the biggest (only significant?) difference between male and females are the sexual organs and hormones, which are certainly not immutable. Male babies receive large boosts of testosterone at certain stages of development.

Nonetheless, also just Nthing all those who agree gender is primarily a social construct and that difference within groups are far, far greater than differences between groups.

Great breakdown of the math, WF. Thank you.
posted by mrgrimm at 9:43 AM on January 6, 2012 [1 favorite]


Sorry, but if they are basing their results on a questionnaire given to undergrads, then I suggest taking any of their supposed results with a grain of salt.

Studies of this kind mostly test for selection bias.
posted by clvrmnky at 9:56 AM on January 6, 2012 [1 favorite]


Imagine a Venn diagram with only 10% overlap. The research is saying that only 10% of personality attributes are shared between men and women!!!

Hyde states that "Del Giudice and colleagues have focused on one particular topic, personality, and, using a technique to maximize differences between groups, they have found a large gender difference on a single, undefined dimension of personality."

STRAW MAN ARGUMENT FOLLOWS

I imagine a graph, with a red pin representing the average Agreeableness score for women, and a blue pin for the average Agreeableness score for men. Exaggerating and simplifying the analysis by Westringia F, I will now calculate the average score for Extraversion for women and men, and move the two pins further apart. I will move the pins on the graph for each of the Big Five factors of personality: Neuroticism, Extraversion, Openness to Experience, Agreeableness, and Conscientiousness.

Since research agrees that personality consists of the aggregate of the Big Five Factors, my two pins now identify the difference in personality between men and women.

END OF STRAW MAN

With a tip o' the hat to Kid Charlemagne-- men would identify themselves as "astute" or "observant," adding up to a Sensitivity score. Men would admit to being gregarious and part of a jolly social crowd, which would contribute to a Warmth score.
posted by ohshenandoah at 10:40 AM on January 6, 2012


I'm no scientist but even I can pick up prime examples of daftness on a quick skim e.g. "Another possible objection is that our findings are based on self-reported personality, and may be inflated by gender-stereotypical or socially desirable responding. . . . First, meta-analytic evidence shows that sex differences in aggression (a highly sex-typed behavior) are very similar when assessed by observation and self-reports, and even stronger when measured by peer-reports [70].

I don't understand why you think that's "daft." The point about how observation and self-reporting yield similar results seems like a reasonable argument that the self-reporting is reliable. I still don't think self-reporting is reliable, but I understand their argument. Anyway, even if the self-reporting were definitely reliable, that wouldn't mean they had isolated the biological causes of the behavior, as other commenters have pointed out.

Also, I don't see their point about how the differences are stronger when measured by peer reports. It's not obvious to me that peer reporting isn't even more skewed by gender stereotyping. For instance, I might be more likely to explain my own behavior in terms of my unique, nonstereotypical traits because I have first-person insight into my own quirky tendencies in spite of being male. In contrast, if I'm describing the behavior of another man who's relatively unfamiliar to me, I might be more likely to lazily fall back on male sterotypes.

Beyond the specific findings and arguments, I'd like to question why so many people feel so strongly about the issue of how much gender personality/behavior is or isn't biologically rooted. It seems to me that many people would prefer to believe there's very little to no biological origin of gender differences in personality/behavior. Why? Maybe because it might seem that to the extent these differences are "cultural" rather than "biological," they're fluid and malleable; we can potentially break free of stereotypes. But if I'm right that that's the underlying concern, that seems mistaken to me for several reasons:

(1) Biological explanations aren't mutually exclusive with cultural explanations, and vice versa. You can describe "social constructs" but also understand how they come from people's physiological makeup. For instance, there clearly are many, many social constructs that encourage society to stereotype men as aggressive and encourage men to behave that way. At the same time, it may be also true that men's brains are different in some way that tends to cause them to be more violence. I'm not saying I subscribe to that view, just that a cultural explanation doesn't automatically refute a biological explanation.

(2) A biological explanation doesn't preclude the potential to transcend stereotypes. For instance, I believe there are biological causes of men being more violent than women, but I am also optimistic that we can work on reducing the amount of violence (by men or women but especially men). We're constantly fighting against people's natural leaning to violence (through parenting, education, policing, etc.), and that's a perfectly worthy endeavor.

(3) The converse of point 2: the fact that a gender difference is a result of "social construction" doesn't make it particularly easy to undo. Many social norms are very firmly rooted. Whether they're more or less difficult to overthrow than biological tendencies is an open question.
posted by John Cohen at 10:41 AM on January 6, 2012 [3 favorites]


Westringia - About the "inflated D" issue: I can understand how adding more traits they can make 'd' bigger. But don't they have to convert that 'd' to some sort of P-value to see if the difference is statistically significant?

To do that they need to check against the distribution of 'd' under a null hypothesis of d = 0. And I'd expect that distribution to depend on some sort of degrees of freedom (which would increase with the number of traits) — just like a raw chi-square value has little meaning without knowing the d.f. for the statistic.
posted by benito.strauss at 12:04 PM on January 6, 2012


About the "inflated D" issue: I can understand how adding more traits they can make 'd' bigger.

Take care to distinguish little d (Cohen's d) from big D (the multivariate measure). Each trait gets its own (little) d; the whole set of traits gets a big D.

But don't they have to convert that 'd' to some sort of P-value to see if the difference is statistically significant?

I don't think people usually test d (or D) for statistical significance. I thought d was usually calculated after statistically significant differences had already been verified, to determine whether the effect is actually large (you can get statistical significance despite a small effect if you have enough subjects).

There is a set of very rough guidelines people use (set down by Cohen) to guide their judgments of what is a small/medium/large effect that applies to (little) d. I don't know whether a comparable set of guidelines exists for (big) D, but if it did, I would expect it to involve first dividing D by the square root of the number of dimensions to combat the inflation.
posted by a snickering nuthatch at 12:36 PM on January 6, 2012 [2 favorites]


Take care to distinguish little d (Cohen's d) from big D (the multivariate measure).

Oops. My bad.
posted by benito.strauss at 1:51 PM on January 6, 2012


Anthropologist Agustin Fuentes outlines some serious methodological and conceptual problems with the OP study in this SciAm blog post: Get Over It: Men and Women Are from the Same Planet.
posted by col_pogo at 1:59 AM on January 26, 2012


« Older ...And you thought that you loved cat videos   |   Imagine there's no people Newer »


This thread has been archived and is closed to new comments