Gender gap in math (and reading)
Science News and The Economist are reporting on a study published in Science by Guiso et al. that shows a correlation between ratings of a country’s gender equality and the size of the mathematics testing gap between male and female students. Countries with more gender equality exhibit a smaller gap. (Hat tips to Women in Science and to Skepchick for the links.)
My first reaction: well, duh. In places where it’s less likely for a girl to be laughed right out of the math classroom, girls tend to do better in their math classes. The only thing surprising about this is that someone thought it was worth doing a study to prove it.
On second glance, things get a bit more complicated. As with any social science study, interpreting the data is a tricky task. I’m going to start out by assuming for simplicity’s sake that the testing data and gender equality rankings are reliable, because the details of that are really more than I can cover here. I do want to talk about ways to interpret the data and implications for policy decisions.
Those of us who are equity-minded and sensitive to issues of political correctness want studies of this sort to show that girls and boys are equally capable of performing well in different types of tasks. We’re particularly aware of discrimination in STEM fields and eager to show that women are just as capable as men of succeeding there. Naturally, the likely way to approach this study is to note that in a country with a higher Gender Gap Index (GGI) the difference between girls’ and boys’ math scores basically tends to zero, while the difference is large in favor of boys in a country with a low GGI. Therefore women and men are equal, QED, let’s go home.
Wait, let’s not — because the reading scores matter too. This is the aspect of the study that stands out most to me, but it gets little attention even in the original paper: girls consistently score better than boys in reading, and their performance improves in both reading and math when gender equality improves. The gap in math closes, but the gap in reading widens. There are several ways we could choose to interpret this. Perhaps girls are naturally better at reading than boys, and are equally good at math. Or maybe girls are better at reading and boys are better at math, but because girls spend more time on homework (see the supplement) the math gap gets closed. That supposes that the effect of girls’ studiousness is of equal magnitude to the boys’ math advantage. There are plenty more possible interpretations, mostly variations on those two. Any way you slice it, though, the data show a difference between the genders — in math ability, reading ability, intelligence in general, dedication to schoolwork, or some combination thereof.
These differences could be innate or learned. Obviously the PC answer is that differences like these are learned, that is, based upon socially imposed discriminatory messages, but after a lot of thought I have come to the conclusion that there are few if any plausible explanations for this data that don’t rely at all on inherent differences between the genders. We might say that prejudices don’t simply discourage girls from learning, but rather encourage gender roles, causing girls to be more focused on reading and boys more on math — but that would mean that as the math gap disappeared, the reading gap should too. In order to explain the rise in girls’ reading scores, there has to be some other societal factor differentiating between the genders that remains even in more equal societies. For example, the gender role effects could coexist with some prejudice that works against boys overall, dissuading them from their studies. In order to explain the data, the gender role effect must exist in countries with all different levels of gender equality, while the anti-boy (or equivalently pro-girl) effect must be negligible in countries that are less gender-equal and grow in strength alongside equality. However, this seems like a silly explanation, and I can’t come up with any that would be more believable. If you have one, please do suggest it in the comments.
Let’s suppose, then, for the sake of argument (seeing no compelling counterargument at least at the time of posting), there is some inherent difference between girls and boys that causes this gap in testing performance. It looks like girls always outperform boys at reading, so it might be the case that girls have a natural aptitude for it, an absolute advantage. Even if the math gap disappears when gender discrimination lessens, boys still have a comparative advantage in math over girls on average, while girls’ comparative advantage is in reading. What, if anything, should we do about it?
Before you answer that question, remember this: the testing gender gap referred to in the study is the difference between the average of girls’ scores and the average of boys’ scores. If we aggregate the scores of each gender, there will be some boys who score high on reading and some girls who score low. If you pick a girl and a boy and random, it is quite possible that the girl will have earned a lower score than the boy — the exact probability of this depends on the gap size and the shape of each score distribution. However, when people know about a statistical effect like this gap in mean scores, they tend to generalize and misapply their knowledge. For example: it is a well-known fact that men are, on average, taller than women. Studies have shown that when asked to evaluate the heights of men and women in photographs, people tend to rate the man taller than the woman even when the photo does not reflect that difference, and even when informed that both people in the photo were the same height. If you were asked to guess the height relationship between a man and a woman you knew nothing else about besides their genders, it would be strategically correct to guess that the man was taller than the woman because that is statistically more likely. On the other hand, if you have some evidence about the particular heights of the individuals in question, that information clearly ought to weigh more heavily in your judgment than statistical generalizations about the population as a whole… yet people still ignore it.
This sort of problem could easily happen in evaluations of boys’ and girls’ academic performance. It could mean that girls get disproportionately tracked into advanced-level English classes regardless of their individual strengths and weaknesses, while boys who would do well in those classes based on past performance are told to focus on math instead. Scholarship committees might subconsciously award grants to study math to more male applicants, denying them to better-qualified female applicants, or award more grants to study English to female students, denying them to better-qualified males. A manager looking to hire a computer programmer could be more likely to toss out female applicants’ résumés thinking them not suited for mathematically-minded work, or an editor might be more likely to reject a manuscript written by a man believing it to be too technical-sounding and not sufficiently articulate. In all these cases, plenty of evidence about the individuals exists, but that evidence may be ignored in favor of the statistical generalization.
The data might also guide us to make some policy decisions that are rational but still go against our equity-minded intuitions. If women have the advantage in reading skills and men have the advantage in math skills, we ought to allocate our resources accordingly in order to maximize society’s productivity. Maybe women’s colleges should be distribute their funds disproportionately toward the humanities, neglecting their math and science departments. Maybe technical or liberal arts universities with student populations that are overwhelmingly male or female shouldn’t bother implementing affirmative action in their admissions policies to correct for the difference — and nor should our scholarship committee in the last paragraph. The study noted that boys’ biggest advantage was in geometry and smallest advantage was in algebra, so if a math department finds a boy and a girl both performing equally well in an algebra class this year and is trying to decide which of them to promote to a higher-track geometry class next year (and there’s only room for one), it makes sense for them to bet on the boy.
Given the information that this study found, should we be taking policy action in cases like these where it does seem rational? Is unequal treatment based on gender wrong even if it’s grounded in fact rather than irrational prejudices? Clearly in some cases, like decisions about medical treatment, such information would be relevant to include — but where should we draw the line? It’s a very tricky call.
I suspect that the public reception of these studies differs greatly depending on exactly what is found. We would probably ask for a second opinion if Guiso and his colleagues found a persistent gender gap even when controlling for societal pressures. There hasn’t been any outcry over the fact that women consistently score better in reading, but then, there’s no taboo against saying that. All I’ve said about gender is at least as true for race. Remember James Watson’s comments about the relative intelligence of people of European and African descent? Imagine three possibilities for what a scientific inquiry might show: Watson was right and white people are on average smarter than black people; Watson had it backwards and black people are smarter than white people; or, both black and white people’s intelligence levels map to the same distributions with equal average values. Assuming equal scientific rigor in all three hypothetical studies, I think we’d be more likely as a society to embrace either of the latter two than the first. But are we really doing science if we have a correct answer in mind before we even look at the data? It seems like we’re not… but maybe it is okay. Maybe we should have a higher standard for findings that reinforce existent biases than for findings that contradict them, since the potential downsides from an erroneous result are that much worse.
At this point you might be wondering why anyone would do these studies in the first place. After all I’ve said it might sound like they create more problems than they could possibly resolve. However, I do think questions about inherent differences between people are worth asking and investigating. I believe this first of all because knowledge is valuable to pursue for its own sake, and if we find knowledge that makes us unhappy we should reevaluate our happiness rather than shut out that knowledge. I also think these types of studies are useful for understanding the underlying biochemistry of the brain, and for understanding what sorts of traits are linked to what genes, and in turn to what other traits. They could help us decipher exactly what it means to think or to understand a concept, and might help with finding causes of real learning disabilities.
Of course, we do need to remain conscious of all those problems. While scientists continue to investigate questions about our differences, we need to have open public dialog about which types of discrimination are bad and which types if any are sensible. We also need to educate people about statistics, about distributions and averages, so that findings are less likely to be misused. We should be aware of how easy it is to misuse statistical generalizations even when we’re well-informed about what those generalizations mean, and we need to plan accordingly.
Comments
One Response to “Gender gap in math (and reading)”
Leave a Reply
-
Recent Comments
sandeep on Thoughts on religious tolerance
bruce on Alternative medicine: worth a try?
Mike on Think before you speak
Zychan on What is Acts 29?
Emily on Grad students, 1…
Interesting analysis. Clearly there are biological and physiological differences between male and female brains. But how much (if any) do these differences translate into differences in academic aptitude and performance?
The evidence (as I interpret it) seems to increasingly suggest that the effect is very small.
Thanks for the post.