Providence, RI -- "I'm too pretty to do math": This year, a T-shirt carrying that slogan was marketed to young girls. After outraged objections, the shirt was pulled from stores, but is still available for sale on the internet -- and its familiar message continues to echo: It's boys, not girls, who excel in math. Was the outrage over the shirt knee-jerk political correctness? Is it perhaps time just to accept the fact that boys are better at math than girls?
Not unless you ignore the data. A major new study appearing in the January 2012 issue of the Notices of the American Mathematical Society (http://www.ams.org/notices) marshals a plethora of evidence showing that many of the hypotheses put forth to account for the so-called "gender gap" in mathematics performance fail to hold up. The article, "Debunking Myths about Gender and Mathematics Performance" by Jonathan Kane and Janet Mertz, takes a scientific, fact-based look at a subject that too often is obscured by prejudice and simplistic explanations.
To start with, Kane and Mertz note that, by several measures, girls actually *do* perform as well as boys in mathematics. In many countries, there is no gender gap in mathematics performance at either the average or very high level. In other countries, notably the United States, the gap has greatly narrowed in recent decades. For example, some U.S. test score data show that girls have reached parity with boys in mathematics, even at the high school level, where a significant gap existed forty years ago. Another piece of evidence is found among U.S. students who are highly gifted in mathematics, namely, those who score 700 or higher on the quantitative section of the SAT prior to age 13. In the 1970s, the ratio of boys to girls in this group was 13:1; today it is 3:1. Likewise, the percentage of U.S. Ph.D.s in the mathematical sciences awarded to women has risen from 5% to 30% over the past half century. If biology were destiny and boys had a "math gene" that girls lack, such large differences would not be found over time or between countries.
Nevertheless, other measures continue to show a significant gender gap in mathematics performance. Various hypotheses have been advanced to explain why this gap occurs. Kane and Mertz analyzed international data on mathematics performance to test these hypotheses. One is the "greater male variability hypothesis", famously reiterated in 2005 by Lawrence Summers when he was president of Harvard University. This hypothesis proposes that variability in intellectual abilities is intrinsically greater among males---hence, in mathematics, boys predominate among those who excel, as well as among those who do poorly.
To test this hypothesis, Kane and Mertz calculated "variance ratios" for dozens of countries from throughout the world. These ratios compare variability in boys' math performance to variability in girls' math performance. For example, using test scores from the 2007 Trends in International Mathematics and Science Study (TIMSS), Kane and Mertz found that the variance ratio for Taiwanese eighth graders was 1.31, indicating that there was quite a bit more variability in math scores among boys than among girls. However, in Morocco, the ratio was 1.00, indicating the amount of variability observed in the two groups was identical. In Tunisia, this ratio was 0.91, indicating there was somewhat more variability in math scores among girls than among boys.
If the "greater male variability hypothesis" were true, boys' math scores should show greater variance than girls' math scores in all countries; one should also not see such big, reproducible differences from country to country. Therefore, Kane and Mertz conclude that this hypothesis does not hold up. Kane and Mertz suggest that there are sociocultural factors that differ among countries; some of these factors, such as different educational experiences and patterns of school attendance, lead to country-specific differences in boysÕ variances and girls' variances and, thus, their variance ratios.
Kane and Mertz took the same kind of data-driven approach to examine some additional hypotheses for explaining the gender gap, such as the "single-gender classroom hypothesis" and the "Muslim culture hypothesis", both of which have been proposed in recent years by various people, including Steven Levitt of "Freakonomics" fame. Again, Kane and Mertz found that the data do not support these hypotheses. Rather, they observed no consistent relationship between the gender gap and either co-educational schooling or most of the country's inhabitants being Muslim.
They also examined the "gap due to inequity hypothesis", which proposes that the gender gap in math performance is due to social and cultural inequities between males and females. To examine this hypothesis, they used an international gender gap index that compares the genders in terms of income, education, health, and political participation. Relating these indices to math scores, they concluded that math achievement for both boys and girls tends to be higher in countries where gender equity is better. In addition, in wealthier countries, women's participation and salary in the paid labor force was the main factor linked to higher math scores for students of both genders. "We found that boys as well as girls tend to do better in math when raised in countries where females have better equality, and that's both new and important," says Kane. "It makes sense that when women are well educated and earn a good income, the math scores of their children of both genders benefit."
Mertz adds, "Many folks believe gender equity is a win-lose zero-sum game: If females are given more, males end up with less. Our results indicate that, at least for math achievement, gender equity is a win-win situation."
The article by Kane and Mertz will appear on the Notices web site (http://www.ams.org/notices) on December 12, 2011.
Specific questions are best directed to the authors, whose email addresses are given in the article. General questions can be directed to:
Mike Breen and Annette Emerson
AMS Public Awareness Office
Founded in 1888 to further mathematical research and scholarship, today the more than 30,000 member American Mathematical Society fulfills its mission through programs and services that promote mathematical research and its uses, strengthen mathematical education, and foster awareness and appreciation of mathematics and its connections to other disciplines and to everyday life.
American Mathematical Society
201 Charles Street
Providence, RI 02904
Notices of the American Mathematical Society