A Review of a Book That Should Be Read Much More Widely Than It Will Be
Here are two facts:
- This may be the most important book I've ever reviewed.
- Most of you will never read it. In fact, as soon as I tell you what it's about, most of you will move on to the next article, because you're tired of hearing about the topic it discusses, and you figure you're not part of the problem.
The book Why Aren’t More Women in Science?, is a collection of 15 articles by leading researchers in the field. It was put together in the wake of Lawrence Summers’ controversial musings in 2005 about why there are so many fewer women in high-profile positions in science than in law, medicine, and other professions. It is emphatically not a one-sided rant, any more than Summers’ original speech was. Instead, it gives experts on all sides of the debate an opportunity to present their evidence and make their case. In doing so, it provides fascinating insight into how difficult the “slippery sciences” are, and how easy it is to let your beliefs shape your understanding of facts.
Let’s start with some of those facts. In North America, men and women make up roughly equal proportions of high school math and science classes. Despite this, the gender ratio in the professoriate is five to one. Is this because men are intrinsically better at math, or because of subtle and not-so-subtle discriminatory forces? Here are some of the arguments:
- The spread of abilities among men is greater than the spread among women, i.e., the male bell curve is flatter and wider than the female bell curve. This is why the male:female ratio in students scoring 700 or more on the SAT-M (the math portion of the Standardized Achievement Test) is 13:1. It's therefore not surprising that men outnumber women in numerate disciplines.
- But wait---that 13:1 ratio comes from 1983. By 2005, the ratio was only 4:1; what's more, the differences in ability between American students and those in Europe and Japan are greater than those between the genders. Surely that's proof that socialization is a, if not the, major factor?
- Well, no. Women tend to outperform men on tests of verbal fluency, arithmetic calculation, perceptual speed, and memory for spatial locations. Men, on the other hand, tend to do better at verbal analogies, mathematical word problems, and mental rotation (i.e., the ability to look at several pictures of 3D objects, and figure out whether they represent the same thing from different angles or not). If you statistically subtract the influence of rotational ability from SAT-M scores, it eliminates the sex differences. Being able to see things in 3D must therefore be important to mathematical thinking, and men just happen to be better at it.
- Not so fast. Excelling in science, as in any career, requires single-minded dedication: according to E.O. Wilson, you need forty hours a week for teaching and administration, twenty on top of that for basic research, and another twenty to do really important research. What's more, you need to put in these hours when you're in your twenties and thirties, which also happens to be the time most people start families, or start caring for elderly parents. In our society, these burdens fall disproportionately on women. The end result is about as fair as telling athletes who are just about to enter the major leagues that they have to cut their training time in half.
- Hang on---the fact that more women than men devote themselves to family is a choice. There's lots of research showing that on average, men tend to prefer working with "things", while women prefer working with "people". What if society's expectations simply reflect people's preferences? You don't have to shift the mean of a bell curve very far to have a dramatic effect on numbers in the upper tail...
- All right, let's talk about shifting the mean. Research by Dweck and others has shown that if students---both male and female---believe that something is a "gift", i.e., that people are either born good at it or not, then they are less likely to do well at it, because the first time they hit a setback, they conclude that they "just don't have the gene". If, on the other hand, they are told that mastery of the ability has been proven to depend only on hard work, they will, on average, do better. Couple that with societal stereotypes, like Barbie dolls saying, "Math is hard," and it's easy to see why women are underrepresented at the upper levels of math, science, and engineering.
There’s a lot more of this in the book, with men and women arguing on both sides. Some studies are quoted and explained (or explained away) three different ways in as many chapters; refreshingly, there are very few rhetorical questions and no obvious sign of political dogmatism. These are scientists, wrestling with an emotive issue as objectively as they can. For that alone, it’s worth reading.
To meet their standards, I should make my own biases clear. Several years ago, Michelle Levesque and I looked at the gender balance in open source (see “Open Source, Cold Shoulder” in the November 2004 issue of Software Development). While the male:female ratio in the software industry is between 7:1 and 12:1, depending on how you measure it, the ratio in open source is at least 200:1, and probably worse. For a community that talks so loudly about freedom and rights, I think that’s shameful; I think it’s even more shameful that so many people in that community choose not to notice, or say (rather defensively), “Well, it’s not my fault.” I think some social refactoring is long overdue; I think that programs like the one Margolis and Fisher led at Carnegie-Mellon, and described in their book Unlocking the Clubhouse, matter a lot more than copyright reform or the fight against software patents. Sadly, though, our profession is self-selected for people who don’t agree, and that, I think, is the greatest shame of all.
Stephen J. Ceci and Wendy M. Williams (eds): Why Aren’t More Women in Science? American Psychological Association, 2006, 159147485X, 254 pages.Jane Margolis and Allan Fisher: Unlocking the Clubhouse: Women in Computing. MIT Press, 2003, 0262632691, 182 pages.