It comes up a lot in discussions of women in computer science, women who write code, women in open source. Eventually, someone brings up the fact that women score slightly lower on math tests. Clearly, they claim, this biological inferiority must explain why there are fewer women in math heavy fields.
It sounds like a compelling reason, and it gets a lot of play. Except, you know what? It’s a lie.
I’m a mathematician. I’ve looked at those numbers, I’ve read some papers. The research into biologically-linked ability is fascinating, but it simply isn’t significant enough to explain the huge gender gap we see in the real world. I used to do this presentation on the back of a napkin for people who tried to spout this misconception to my face, and I finally put it online:
Love it? Hate it? Learn something? Catch the Mathnet reference? Let me know.
Geek Feminism Blog 
I think it’s a nice presentation. I was already of the opinion that biology doesn’t explain the difference, and being a programmer I’m also aware that the amount of math required for the majority of programming jobs is quite minimal.
However, the graph was not what I expected. I thought there would be a steeper curve for women and a more flattened curve for men, because I remember reading somewhere that men tend more to the extremes than women do (ie there are more very smart men than very smart women, but also more very dumb men than very dumb women). Of course that wouldn’t have affected the outcome given the rather low level of math ability required for programming, but it would have explained the difference in the number of programmers if the math requirement was high. Now you’ve made me want to search for this study and read it myself, rather than relying on second-hand media reports, which so often get things wrong.
There are actually differences in the curves, but it complicates the explanation a fair bit without changing the results, so I went with the simpler graphs — it’s a simplified explanation as is.
I do recommend you check out the study I cited, because it’s quite interesting and talks about how the graphs change based on age… but it doesn’t have any of the graphs that show the flatter curves, so you’ll have to search a bit further afield.
Short version of the longer explanation: Yes, the curves for men are flatter, which means there’s a few more geniuses and a few more imbeciles. But frankly, most people at that genius level of math won’t go in to CS, so it’s pretty much irrelevant to the argument — you wind up having to chop off the tail to get anything resembling accurate. And at that part of the tail, you’re talking so few people that it’s a tiny percent difference anyhow.
(There’s a study that talks about that, but I couldn’t find it online, either because I misremembered the keywords or because it simply isn’t online — psych and biology papers are often not freely available.)
Someone pointed me at some interesting blog commentary on the women in Open Source numbers some time back:
This should not be taken to read “actually, programming and physics [to which he’s comparing it] aren’t actually all that hard after all, so women could do them.” But the resistance to women does seem related to this point: our field requires such rarefied intelligence results that you can’t look at the big bit of the bell curve to judge us: how many women are in that top 0.00001%, do tell? (In reality my understanding is that ‘even’ mathematics is accessible to women and men who are ‘merely’ quite intelligent.)
I believe there’s actually significant evidence that IQ is a terrible predictor of future performance, but I didn’t even bother touching that since it’s a little more contentious.
I tend to avoid using the term “IQ” for that reason, although “intelligence results” is starting to shade in that direction.
somputer science? ;)
also just to note, Im a software engineer. i *suck* at maths. i have to google what the squiggly lines mean in papers i read that are even slightly mathematical in nature.
I was also under the impression that the means for for the two curves are pretty close as shown in the graph, but the stdev for males was much wider resulting in more males with mental retardation and also with extreme ability such as Carl Friedrich Gauss and Kim Peek (the basis of the character in Rain Main). As the group under question becomes more intellectually demanding, the gender gap would widen and become widest at the end such as Nobel laureates in Chemisty, Physics, Economics or the Field’s Medal.
I’ve always thought that the reason girls are better at math in elementary school but boys are better by high school was the difficulty involved in the subjects taught to those groups. It’s pretty easy to teach anyone basic arithmetic and algebra as long as they’re will to put some effort into it and spend an hour a day learning it. It’s much more difficult to teach calculus or quantum mechanics to someone who doesn’t have an a strong enough interest to stare at numbers on a page for hours on end trying to make a tiny bit of progress in a single problem.
I also think practical CS (e.g. Software Engineering or CompEng) require the same level of obsession. In my college, there were fewer women in SE than CS, and CE had the widest gender gap. Similarly, I find that there are fewer women in computer related professions as the level of obsession required increases. The gender gap exists in IT, but it becomes wider as you progress into QA automation, web development, applications programming, and systems engineering.
I tend to agree with the view that there is a link between autistic spectrum disorder and being a good programmer, though my agreement is based on personal experience and anecdotal evidence. I think men and women are much closer in ability for undergrad level computer fields than the enrollment and matriculation figures suggest, but I think men have more of an inherent obsession than women which is why there are more of us in college educated computer fields and also in self taught computer professions. Though, my experiences are limited to Computer Engineering rather than pure CS so the reasons might be different over there.
This is a commonly-cited fallback position when confronted with the idea that men and women just aren’t very different when it comes to the cognitive capacities required for technical fields: that instead we can account for the difference by observing differences in other kinds of personality and emotional traits that make women less well-disposed to succeed in these fields.
I have a difficult time believing this considering the prevalence of hobbies like knitting among women. It strikes me (as a non-knitter) that it requires a certain amount of “programmatic” thinking and planning ahead to do it well, as well as a certain amount of persistence.
I have certainly never observed that QA automation, web development, applications programming, and systems engineering required obsession. The few who start new companies or invent new technologies in those areas are usually obsessed with their projects but the vast majority of practitioners in those areas are doing routine work.
And the requirement of math in programming (assumed by many) has always struck me as odd as well as irrelevant to the question of whether it makes programming less attractive to women. For example, women are now the majority in accounting so they’re clearly not averse to using some math skills there.
I suspect that part of the reason that CS students and open source communities have so few women is the negative feedback loop where an early shortage of women led to new women feeling conspicuous and uncomfortable joining groups with overwhelming majorities of men.
Only part of the reason, though. I used to assume it was more coincidental than not that my own area of open source had so few who are not young, white, and male but the Aimonetti incident and the reaction to it showed me that there is a fair amount of active resistance to diversity.
just two very short notes:
when you do maths or physics, you very rarely look at actual numbers, you mostly look at formulae.
I’m a physicist by training; my starting year for physics at university had 10% women, and every single one of them was at a all-girls school at least for a large part of their pre-university schooling. Less than 5% of pupils in my school-finishing year went to a gender-segregated school. What does that tell you?
spz:
Such is the difference between arithmetic and math. I rather prefer arithmetic. If I have to learn math, the teacher had better be prepared to show me a proof for everything they teach (I just cannot make heads & tails of math without a proof).
I liked the slideshow.
I’m taking classes at a community college, some of which are to prepare for professional certifications, some of which are to prepare for transfer to a four-year school for a CS degree. (I’m in early middle age, and tired of the jobs that dropout English majors can get.) I’ve been finding that most of the mathematics courses have relatively even numbers of women and men. Not so the tech and programming classes, in which there are very few women — there’s a nine-to-one ratio at best.
It’s most certainly not any need for mathematical skill at issue, since the tech and programming classes have called for nothing more than basic arithmetic.
I don’t think it’s that programming takes obsessive levels of discipline, either. First semester calculus, I found, required about ten times the study time of all my other classes combined — and that class had a majority of women as students. (By the middle of the semester, most of the men had dropped out.) I’ve perhaps needed to spend two hours, total, study time for an introduction to C++ class, a class in which there are about two dozen men and three women. Perhaps “obsession” comes in later, but if so, it comes in well after it’s been established that most of the students are men.
On the other hand, *everyone* has to take a few math classes to get an Associates or Bachelors degree, while not everyone has to take a computer class.
I rather doubt obsession comes in later. I think obsession is what gets people into post-secondary level CS classes (well, that or “if I learn Java I can get a good job!”). Somewhere back around age 12.
I could have completed an English degree without taking a math class after high school. It wasn’t a requirement at all. I think that’s the norm for humanities majors.
Wow. At my school, as a Japanese major I would’ve needed I think it was 6 credits math & 6 credits science. I changed to International Affairs major (with Japanese as 2nd major) because it had lower math & science requirements (6 credits of the two, total).
I ended up shooting myself in the foot and becoming a computer science major, where I need 1 year (6 credits) Calculus, 1 year of a natural science (8 credits), 1 semester of a different natural science (4 credits), and 3 credits of a math/science elective.
“I ended up shooting myself in the foot and becoming a computer science major, where I need 1 year (6 credits) Calculus, 1 year of a natural science (8 credits), 1 semester of a different natural science (4 credits), and 3 credits of a math/science elective.”
Huh, interesting how this varies across the world. When I was studying CS, in our first two years we had to take a semester of Calculus, a semester of Linear Algebra, a semester of Differential Equations and Numerics, a semester of Linear Algebra II, a semester of Mathematical Logic, a semester of Probability Theory and Statistics and one other course I can’t remember (that’s all 4 hours of lectures + 4 hours of other stuff per course) + 2 courses in a science or economics. In the first two years, half of the courses were Maths and it was usually the reason why students chose to study something else, because they couldn’t handle the Maths requirements.
Women made up 10% of CS graduates at my university back then :/ I suppose it was even more frightening when you were socially conditioned to think Maths is harder for you then for the men.
Yvi:
Oh, we have to take statistics too. I didn’t count that as a math, though I guess it is… (drat!)
I studied IT at university, not CS (a mistake, in retrospect, but it had a scholarship that would let me support myself). We had to have calculus already before starting (it was a requirement when applying, that that particular form of math had to be among your 4 best subjects in year 12), but during the course itself the only maths we did was financial maths (compulsory) and I also took an elective in operations research.
Thanks for the concise and funny slideshow!
I would have commented over there in response to the comment “where does the difference come from, then?” but didn’t want to create an account there.
I’ll say it here, though. The other commenter was saying that if they’re not significantly better at math, then perhaps men are more interested in logic puzzles. It’s been my experience, as a woman math major with lots of geeky female friends, that while many women /love/ logic puzzles and are drawn to logical systems, a smaller percentage of them pursue careers that use those skills.
I’m sort of reluctant to state the obvious, but in light of that other comment, I think it needs to be said:the reason for the skew is that the structures (clubs, mentorships, internships, and job cultures) all strongly favor men and are moderately to severely hostile to women.
I think it also bears pointing out that for whatever X in “men are better at/more interested in X than women” (whether it be logic puzzles, tinkering, staying up all night drinking Mountain Dew, or whatever) we’re *still* talking about a bell curve like the one in Terri’s slide 17, or maybe even slide 20. You (generic you) can’t just replace “maths” with “puzzle-solving” and think you’ve come up with a brilliant new argument.
I mostly maintain that there are lots of potential reasons for the skew — Being that this is about people, it’s highly unlikely that any single factor explains the majority of the results.
Also, we’ve been talking a lot (in this discussion and elsewhere) about things that boys are good at, but it could equally be affected by things boys are bad at: maybe it’s that CS is less harsh on people with poor language skills, and since boys score lower on language tests, they gravitate towards careers that won’t penalize them as heavily? Maybe because there are more autistic boys who have little-to-no interpersonal skills, they go to careers that are perceived to have less human interaction? Maybe because men are more insecure about their gender-identities, they are more likely to gravitate towards already male-dominated fields?
Seriously, any of those “things boys are bad at” could have just as much play in the final results as “things girls are bad at.”
I have some anecdotal evidence for this in the howls of outrage I got for a quiz I once gave some CS students that required them to read one side of a page’s worth of text and then answer some short questions about it. (I was the TA, the instructor got angry emails telling him that they didn’t go into CS in order to read essay questions. Mind you this quiz was worth less than 1% of their grade.)
However, I have further anecdotal evidence that suggests that good language skills are a prerequisite for *success* in CS, and many of the people who couldn’t read/write well were predictably the ones who didn’t come to me in office hours but should have (this is the worst-off set, behind the people who wanted me to do their assignments for them in the guise of “help”.)
It’s all so complicated. In general, any strong cognitive claim about anything should be taken with about a truckload of salt, including claims by the expertliest of experts.
And to continue on the theme of anecdotal evidence, I’ll suggest that many of the scientific papers I read do not demonstrate the sort of language skills that I would hope for my colleagues. My theory is that language skills help computer science students greatly, but many undergraduate programs test them so infrequently that a lot of people with poor language skills skip through the cracks even though learning more language skills could have made them better at their jobs, possibly even made them better programmers (as communication of what your code does is not a skill to be under-rated).
Indeed. I’m occasionally struck by the hostility that is sometimes expressed towards literature classes, by tech students. More than once, I’ve seen people claim that they can’t enjoy reading anymore, because the idea of reading fiction analytically ruins it for them.
When I was an English major, I took very seriously the idea that the virtue of fiction was the teaching of empathy with the inner lives of others. It strikes me that quite often, those others are women. It wasn’t until well into the 20th century that novels were considered an appropriate subject for scholarship, and there’s reason to believe that has a lot to do with the extent to which novels were written by women and read by women.
So, I suspect an aspect of the hostility towards reading and writing is that these things still seem “womanly” to some people — aspects of life that they would prefer to ignore.
I quite enjoy reading, just not reading analytically. I’m always waiting for the author to come back from the grave, bust into the classroom, and shout “It’s a STORY! JUST A STORY! Stop analyzing it to death and have a good laugh!” But hey, analysis takes the fun out of everything.
See, personally I love reading analytically. To me it’s not just a case of trying to figure out what was said, but how it was said. The reason is that because there are fewer things that tell us, as a collective and as individuals, what we privilege than by the stories that we tell and how we tell them.
I can understand that people don’t like to read things analitically, and that people can get a little too invested into the analysis, but sometimes you can get a bigger laugh when you look at it critically.
Though this may just be me defending the fact that I’ve spent over 60k and 6 years doing just that.
Hmm, I’d suggest that hostility towards literature classes comes from a number of sources. So, due to time constraints, I never took a lit class in university, but in high school I experienced different English class teaching styles, and some were fun and some were alienating, but I got the impression, perhaps incorrectly, that the style I found alienating was the orthodoxical one.
And then we have the whole pomo warz which is the first thing a nerd finds when s/he searches the interwebz for what the lit critters are doing.
Where I suspect gender comes into it is where “objectivity” is considered to be a masculine characteristic, while “subjectivity” wishy-washy feminine. We can respond to this in a couple of ways that may or may not be mutually exclusive. To be sure, where I found high school English classes most engaging was where we looked at the book in light of its historical context and the intentions of the author, not my own irrelevant/inconsequential reaction to it :)
Asad:
The thing that bothers me most about the analytical classes is that the teacher will ask “what do you think….” and then you answer, and then they tell you’re interpreting it wrong. There are two sides to every coin! And hey, this story might even be a D20!
Yeah, precisely. It was very hard for me to understand how my gr10 English exams were marked, because I’m quite convinced that I answered the essay questions honestly and in clear English; but because I missed the Grand Hidden Metaphor that didn’t seem to have anything to do with the story, I regularly lost points. Was more comfortable reading Saint Joan in gr11; I could hang my hat on history and religion relatively reliably. (Yeah yeah sour grapes…)
There seems to be a conspiracy of contemporary writers who write specifically to give English teachers something to teach.
Somehow I don’t think that this reaction is particularly gendered, but maybe that’s just me. I think it does bias tech students away from this sort of thing because it seems like the point is to figure out the right BS that will please teacher.
It’s not just a conspiracy. From what I’ve read and been told by working writers, there are two main tracks for professional fiction writers. One track is writing for mass market publishers; the other is basically an academic career, in which case publishing fiction takes the role that publishing research does for an academic researcher. The two tracks have surprisingly little contact between them.
There’s plenty to be criticized about the humanities side of things. I just wanted to point out what’s pretty much a commonplace about literature in English: that the novel was generally understood, until the early to middle twentieth century, to be a form of literature written for women, often by women, about women’s lives. So, most survey courses on literature in English will spend at least some time discussing relationships between men and women, and the personal lives and differing points of view of women and men.
My sense is that a fundamental aspect of sexism is that masculinity is defined as a discipline of ignoring anything feminine — thus, for people who wish to assert their masculinity, discussing the lives and priorities of women is an interruption of discipline. This is, I think, why some technically oriented men will boast, early and often, that they hated English classes.
Your suspicious would be borne out by a follow-up post and thread re SF from the anti-feminist blog The Spearhead (link here)—a followup, actually, to the post that inspired a quick hit here. In the thread, it is quite explicitly expressed and repeatedly that women’s entry into SF novelry presages a tapering of male interest in the genre, and in fact there will be only a few entertainment venues left for men, as it were. Precisely and explicitly because the presence of women prevents the formation and experience of masculinity as such.
In fact, if you go over the whole site, even the innocuous articles ultimately reveal a terror that there might eventually be very little about the world that is specifically male, and little that is specific to males, culturally or biologically. Whereas women have wombs.
(Aside: I just noticed that The Spearhead’s subtitle is “Piercing the Shield of Ignorance”. O. M. G. HAHAHAHAHAHA. They’re complaining about women’s takeover of SF while using a metaphor and style lifted directly from bad historical bodice-rippers…)
Good point. It might be more informative to look at the language skills of a given group rather than their math skills. As you move along the spectrum across IT, applications programming, and systems engineering, your practical skills are measured by your ability to talk to a machine rather than other people. I agree that there is a strong social bias against women in computer fields, but if this bias was the primary reason for the gap, the gap should narrow in all the computer fields at the same rate. However, it seems it remains the widest in fields where most of the work involves close interaction with a machine.
One of my favorite quotes from the Hacker’s Manifesto: “I made a discovery today. I found a computer. Wait a second, this is cool. It does what I want it to. If it makes a mistake, it’s because I screwed it up. Not because it doesn’t like me… Or feels threatened by me.. Or thinks I’m a smart ass..”
If a certain anonymous commenter is wondering why their comments don’t show up, it’s due to the 4th word in this comment.
Great slideshow, thanks. Now I just need it in brick form for hitting all those guys who told me girls can’t do maths, when I got into Cambridge and they didn’t.
Not to say whether or not grades are the best indicator of ability or not…. but if you did measure ability based on grades, I’m pretty sure the few women in my comp sci program would have had the men beat in terms of GPA.
Hi there,
Biology may not explain it, but sociology does! ;)
Actually, it doesn’t. There’s currently no consensus on what causes the gap, so scientifically speaking, no discipline explains it.
That said, sociology and psychology have some really interesting insights into the problem.
Sociology does explain it, but not in the general sense.
I’ve seen the phenomenon in the UK and in France, they occur for quite different reasons, because the two societies are quite different contexts for this situation.
There’s not one single reason or a single set of reasons anyway.
Thank you for this!
Of course, I know that if I post this to my friends, they will immediately say you misrepresented the math (and probably go right to the flatter curve thing).
I’m happy to have it, though.
And add me to the list of those who think it is a constellation of factors, many of them social, that result in a gradual magnification effect resulting in the dramatic skew one sees now.
Yeah, “the curve isn’t perfect” and “I want more math” are common complaints. Of course the curves aren’t perfect! I drew them in OpenOffice, for goodness sakes! It’s a simple slide presentation meant to give to morons on slashdot, not a graduate level lecture in ability differences! ;)
Sarcasm aside, someone on another list I read recommended this book, which seems to talk about the magnification effect:
From this washington post article.
I’m a female mathematician/statistician, and I think that your argument in the presentation would be strengthened by specifying exactly what those curves are that you are referencing. I tried to download the paper but couldn’t find an active link in my university online resources. Will keep searching, certainly.
The thing is, I’ve done some analysis of psychological measurement and I’ve noticed that psychologists have this tendency to treat everything as a measurement of normal means problem, up to and including making probability statements about them. And so my question is: Are those curves in your presentation population curves (ie, representing the spread and diversity of ability within the population), or are they margins of error resulting from the calculation of a sample average (a so-called “data point” that is averaged across your sample, with its associated error bars, normal as per the Central Limit Theorem when the sample size is large-ish)? It is one thing to compare the populations with your cutoff points (which are basically relative probability statements), but doing those comparisons on the confidence in average ability is much different, especially if the shapes of the curves differ. Means are influenced by outliers– they only match up with medians or modes if the population distribution is symmetric. Even if your confidence in the value of the mean is this bell-shaped distribution, depending on the shape of the population, there could be many more or fewer people extended out beyond those means. From my reading of the caption in the picture, it’s looking at an effect size, which generally is a comparison of means.
Not that I’m arguing for a biological distinction between men and women on mathematics skills, but universally, it seems like everyone except statisticians are terrible at stats.
I’ll say it again: This is not intended to be a full scientific proof; this is meant to a be a quick explanation understandable by someone with fairly basic math skills. Years of presenting before audiences has taught me that a compelling simplification can be more persuasive and useful in building a mental model than a detailed analysis, especially when your intended audience is not skilled in your field of study.
It’s like explaining atoms to kids: you give them this initial model of the electrons rotating around like planets, and even though it’s not entirely accurate as a model, it provides a basis for you to later explain the electron shell if they pursue further study in physics. This presentation is meant to be the basis for understanding that the difference between the means is not large enough to explain the gender gap by itself, since that basic concept seems to elude a lot of people.
You asked specifically about the graph: I guess maybe you can’t quite read the information below the graph in the slideshare embedded version (it’s quite legible in the original, which you can download), but it’s just a graph showing the approximate magnitude of the difference in the means, not showing the full distribution. The paper itself has some fascinating numbers about ability gaps over time, and I believe if you search for it in google scholar and click on a few of the pdf links, you can eventually find an online copy that doesn’t require subscription.
There *are* papers with more accurate graphs, but I didn’t have any handy and, given that I’m trying to build simplified mental models here, it would have just added to the mental task load with very little benefit. As I said earlier, the tails represent few individuals, and wouldn’t have much effect on the final stats. And on top of that, I’m relatively sure that genius-level math folk don’t often go into computer science, so one would likely need to chop off the tail entirely… all in all, it would have added several slides and complicated the explanation to very little benefit. The key to great presentations is often knowing what to leave out.
Should be easy enough to find better graphs if you to go a university library and ask a librarian, though. Alternatively, you could check the references from one of the many books on the subject of gender and ability. And finally, an organization like http://www.ncwit.org/ might have a reading list or provide some suggestions if you asked.
Just a comment to let you know I enjoyed the presentation. It was extremely clear, and I do indeed feel that I have learned something.
Thanks :)