How the gender gap in STEM might get its start in elementary school

On average, girls do as well as boys on elementary- and middle-school math tests. But by the time students enter the workforce, a big gender gap has emerged, with men earning nearly 80 percent of bachelor’s degrees in engineering and computer science.

A new study offers evidence that the disparity might be getting its start in elementary school classrooms.

The paper offers data from a single district, but the pattern it uncovers is striking. Girls are less likely to be nominated for, selected for, and continue in the district’s advanced math program. Despite their comparable math scores, the program loses girls at every step — a phenomenon that could contribute to fewer women entering math-focused fields later in life.

“It’s heartbreaking,” said Marisa Shuman, a math and computer science teacher at an all-girls public school in New York City and a Math For America master teacher. “Not only are we not taking them or asking them to be in it, whatever they’re doing in this program for some reason is not keeping [girls] and showing them that they have a place in this world of STEM.”

The study, which was recently published in a peer-reviewed academic journal, focuses on a math acceleration program in Wake County, North Carolina.

Students in the Raleigh-area elementary schools can be nominated to participate by their teachers or parents. Then, if they score well on a special qualifying test, they skip a grade in math. For most students, that means walking down the hall to the next grade’s class for math. Fifth-graders take an online class focused on sixth-grade content.

Researchers Steven Hemelt of the University of North Carolina and Matthew Lenard of Harvard found that the third-graders who scored particularly well on the state math test are about evenly split, gender wise — 53% male, 46% female. But there was a bigger disparity among those nominated for the accelerated math program. That group was 61% male, 39% female.

The gender gap grew when looking at students who passed the qualifying test and joined the program, who were 67% male, 33% female. And when the researchers looked at who remained in the program into middle school, the gap had extended to 69% male, 31% female.

There are other signs that this program is not effectively supporting girls. Hemelt and Lenard show that girls who do join the accelerated math program see their state test scores decline as a result, while boys don’t. (A complicating factor here is that students still take grade-level tests, so they’re not being directly assessed on the advanced content that they’re being taught.)

There is also some evidence that participating girls have lower educational ambitions because of the program. Those girls were less likely than other high-achieving girls who didn’t move ahead in math to say that they hope to continue their education after high school or that school is important to achieving their future goals.

“Acceleration may have a somewhat demoralizing effect on girls’ aspirations,” write Hemelt and Lenard. “All told, this collection of findings implies a diminished opportunity for girls to accumulate (and benefit from) advanced math skills as they move through middle and high school.”

Similar trends have been shown to play out in college, according to separate research. In high school, girls score as well as boys in math, and girls are more likely to graduate and enroll in college. But they’re far less likely to earn a college degree in a STEM field. There’s some evidence that much of that disparity traces back to the fact that girls report less interest in engineering as a career as early as ninth grade — interest that may be shaped by factors like experiences in math class and societal expectations.

It’s not clear what exactly explains the North Carolina results. Hemelt and Lenard hypothesize that “stereotype threat” might be at play. That’s the idea that individuals perform worse on tasks when they feel at risk of confirming a stereotype about their group, like girls being worse at math.

A 2013 study found evidence that the phenomenon reduced math scores for young girls, though a recent overview of research found that the impact of stereotype threat is generally quite small.

Another theory is bias working against girls in math, perhaps from educators and family members. One striking study, for instance, showed that sixth-grade girls in Israel did worse on math tests and were less likely to complete advanced math and science courses in high school as a result of being assigned to teachers who showed evidence of bias against girls in math. The paper found that this form of prejudice — estimated using scores on teacher-graded exams versus exams that were blind-graded — was fairly prevalent among teachers.

Shuman, the math teacher, says that girls may get the message that math isn’t for them. “I think it comes from a lot of things,” she said. “When you think of a mathematician, you think of an old man.”

She also suggested that girls in advanced math programs might feel isolated. “Is she the only girl in a group of boys? If so, then that is going to have an impact,” Shuman said. “If you view yourself as different, then you’re going to be less likely — often, not always — to share. You’re less likely to take a risk. You’re less likely to volunteer an answer. And all of those things are how you learn and how you get better.”