New Evidence in Favor of a Gender Math Gap
From a new working paper by Steven Levitt (U Chicago economist, author of "Freakonomics") and Roland Fryer (Harvard economist), "An Empirical Analysis of the Gender Gap in Mathematics":
In this paper, we utilize the Early Childhood Longitudinal Study Kindergarten Cohort (ECLS-K) to shed new light on the gender gap in mathematics. ECLS-K is a data set administered by the Department of Education. The survey covers a sample of more than 20,000 children from roughly 1,000 schools entering kindergarten in the fall of 1998. An enormous amount of information is gathered for each individual, including family background, school and neighborhood characteristics, teacher and parent assessments and expectations, and test scores, which allows us to test several important theories for gender differences within a unified framework. The original sample of students has been subsequently re-interviewed in the spring of kindergarten, first grade, third grade, and fifth grade.
Consistent with the prior literature, when children enter kindergarten, girls and boys are observationally equivalent in both math and reading. By the end of fifth grade, however, girls have fallen more than 0.2 standard deviations behind their male counterparts in math. The math gap is equivalent to 2.5 months of schooling.
Girls are losing ground in math in every region of the country, every racial group, all levels of the socio-economic distribution, every family structure, and in both public and private schools. By the end of the sample, girls do significantly worse than boys on every math skill tested. Underperformance by girls is evident not just in mean test scores, but also in the upper tail of the math distribution. On entry to kindergarten, girls make up 45% of the top five-percentiles in math test scores; by the end of fifth grade just 28% of the top five percent are female. Girls are underrepresented in the bottom tail of the math distribution in kindergarten, but overrepresented in the bottom tail by fifth grade.
We explore a wide range of possible explanations in the U.S. data, including less investment by girls in math, low parental expectations, and biased tests, but find little support for any of these theories. Our evidence suggests that the gender math gap is especially large among children who attend private schools, have highly-educated mothers, and have mothers working in math-related occupations -- all factors that one might think under some theories would be conducive to girls’ success in math.
Although highly speculative, in the cross-country data we identify one factor that correlates strongly with a small gender math gap: gender-segregated schools. This is an area ripe for experimentation.
MP: Does this mean that Lawrence Summers gets his job back as president of Harvard University?