Messier 82

Messier 82
Beautiful Hubble shot of a starburst galaxy, M82

Sunday, November 30, 2008

Surprise! There's Still No Magic Bullet.

As a female student of a physical science, I have a certain amount of interest in the subject of female students in the physical sciences. It turns out we're pretty rare - the gap is practically gone in the biological sciences, closing rapidly in chemistry, and closing slightly more slowly in mathematics, but it's remained stubbornly large in physics, engineering, and computer science. It also turns out we underperform compared to our male colleagues both in introductory classes and on standardized tests. The biggest and most influential gap I'm aware of is on the Physics GRE, but we also apparently underperform on the physics SAT II, math SAT, quantitative GRE, and the standardized pre/post-test used in many introductory physics classes called the "Force Concept Inventory."

Every few years, one study or another comes up with great data showing that girls or women do a lot better in math or science classes when certain changes are made. These changes almost invariably correspond to the techniques that just happen to be generally fashionable in the educational community at the time. When the "new math" was in vogue, there were studies showing that it closed the gender gap (it may have, to some extent, by crippling boys and girls equally); "child-led learning" was supposed to do the same; then there were charter schools, smaller class sizes, and recently a proposed return to gender-segregated schools championed by the sorts of people who said that girls would learn math better by counting flower petals instead of solving equations. It's almost enough to make one cynical.

The method in vogue today is something alternately called "active," "interactive," or "collaborative" learning. It's implemented at the grade school level as "think-pair-share"; for adults in college classrooms, a similar technique is used, but the silly terminology is (thankfully!) left out in favor of more age-appropriate phrases like "peer instruction." The general idea is that a significantly larger portion of class time than usual is devoted to various structured collaborative small-group activities, with or without ensuing full-class discussions.

Now, there are all sorts of ad-hoc rationalizations about why this is supposed to be better for female students in general - we're supposedly more collaborative, less competitive, more timid about answering questions in class, and better at coming up with answers if we can verbalize them in small groups first. The fact that male students almost universally also benefit from changes in teaching techniques designed to benefit girls is always ignored. The observation that nearly all students in serious college-level for-majors math and science courses are atypical in some way, and female students are more often than not gender-atypical, is never taken seriously.

Doing my best to lay aside my own reaction to these teaching techniques (I hate it! I hate talking to people when I haven't had a chance to work out a problem on my own! I hate feeling locked in to a solution because someone has already seen it and listened to me explain it! I hate feeling like I'm teaching my classmates when I'm completely and utterly unqualified to do so! I hate feeling responsible for other people's misconceptions!) Anyway...doing my best to lay all that aside and realize that the generalizations made in studies aren't necessarily intended to apply to me personally, I took a serious look at the Harvard study released a while back that showed that the gender gap could be significantly reduced with the introduction of interactive learning techniques. It seems fairly well-done, with a typical narrative for the sort of study that it is: women were worse off than men coming into a calculus-based physics course and the differences were magnified by the end of the course, new teaching technique was adopted, both women and men did better but women did so much better that the gap was erased.

As I sad, this is typical for the sort of study that it is. You could easily substitute any of the myriad of other educational fashions in for "interactive learning techniques" and find a study that produces basically the same results in some field or other in some age group. To the extent that these studies demonstrate anything, it's that when you take decent instructors, give them a new tool in their teaching toolbox, tell them how to use it, and force them to pay more attention to their teaching (because they're using the new tool), their students do better. This may indirectly help to close the gender gap in some cases by lifting all students up to a similar level of understanding - students, including girls, who come in unprepared are more reliant on being "taught" - but the evidence that any gender-equalization effect is really linked to the techniques' catering to sex-stereotyped learning styles is in my opinion weak at best.

So, as you may guess, I was completely unsurprised to discover that a new study, this one from the University of Colorado, failed to show any statistically-significant gender-gap reduction using the same techniques as the Harvard study. Again, all students did better with the new teaching style, but women didn't gain on men as they did at Harvard. In fact, men made greater gains than women.This ought to be shocking; a "feminine" sex-stereotyped program aimed at improving women's learning actually benefits men more than women. But it's barely worthy of mention, and the study authors make sure to appease the sex-stereotypers by noting that women in the classes did in fact perform in accordance with their stereotypes, doing better than men on "collaborative" homework and worse on "competitive and time constrained" exams, achieving overall grades that were on average equal to the men's. Altogether, the results were impressive as a demonstration of the effectiveness of a new teaching style implemented well, but failed to show any implications for gender equality whatsoever.

So there's still no silver bullet; students, including women, who come into a physics class with less preparation will usually leave with a weaker understanding of the material, the Harvard study notwithstanding. It's possible there may have been a confounding variable like class size, instructor availability, or simply the general higher preparation level of the Harvard students as compared to the Colorado students. But the use of some new teaching techniques can be of general benefit to most students. It doesn't make for good headlines - but responsible science, especially in the social sciences, generally doesn't.


  1. As many a journalist has lamented, the headline writers may not even have read the article. Much less the issues we have with journalists not hearing the science.

    The weaker background issue has an obvious and readily addressed solution -- provide stronger backgrounds. And, harder to implement, require stronger backgrounds. In taking the physics (et al.) classes, you are going to be in trouble if taking a derivative was something you passed in your calculus class only with a great deal of struggle, and which remains a struggle. In later classes, it's assumed that you can do all that material well.

    There are also serious issues with calculus being taught by people who have no consideration of their students ever trying to (gasp!) use the subject to solve problems. Hard to get the strong background in that case, and then success in physics et al. turns to more a matter of who thinks more in applied mode than taking the appropriate background classes.

    Maybe in some later posts you can put up some of your own ideas about how to teach science to women.

  2. I do agree on providing stronger backgrounds. Requiring stronger backgrounds, as you note, is tricky; in theory, women and men entering these classes have the same background in formal education. My guess is that women's background gaps are more informal; many don't spend as much time building and designing things, playing with rockets and projectiles, and generally doing things that provide an informal, intuitive understanding of physical concepts that can serve as a background on which to build the formal knowledge. This is of course not universal, but my guess is that it can account for a significant portion of the gender gap for students entering the introductory physics and engineering sequences.

    On calculus: the amount of calculus required for the level of physics the studies were investigating is really quite trivial. Waves and especially fields require significantly more, but the research focused on introductory mechanics courses, which can be handled with basic algebra and a minimal understanding of how to take a derivative of a simple polynomial and do an occasional integral of the same. However, an algebra weakness (which is in my admittedly anecdotal experience quite common) can be a killer. I don't think that's a problem that can be solved with a more "applied" teaching methodology, though; if anything, the weakness seems to lie in basic symbolic manipulation.

    On my own ideas - it depends on the science, and it depends on the women. As I noted, I think that most women who make it to college-level physical science and engineering courses have gender-atypical learning styles to begin with - the filtering starts in early high school or possibly even earlier.

    And I think that any question phrased as "how do we teach (insert subject) to (insert group)?" is founded on some pretty tenuous assumptions about the homogeneity of that group. Women taking calculus-based physics are different from women taking physics for flautists, and both are different from middle-school girls taking introductory earth science. Even worse, the variations within each group are probably enormous, especially in the middle-school group which hasn't yet self-selected into different science tracks.

    More on this later - I have calculus homework :)