Women tend to underperform in introductory STEM (Science, Technology, Engineering and Mathematics) courses, but tweaking how courses are graded could help change that.
By Diana Crow
In many undergrad STEM courses, high-stakes exams — such as mid-terms and finals — determine as much as 60-70% of the student’s overall grade. However, this emphasis on tests may be inadvertently putting some students at a disadvantage.
A study published in PLOS ONE late last year found that in introductory biology classes where homework assignments and smaller quizzes made up a larger percentage of the grade (and exams only accounted for about 40%), the gender gap decreased or disappeared. In these courses, women tended to perform better not just in terms of overall grade but also on the exams themselves.
Aside from the way exam grades were weighted, the exams didn’t change, says study co-author Sehoya Cotner of University of Minnesota. The tests were the same length and the questions were just as difficult. “This isn’t about watering things down,” she says. “It’s about reducing anxiety.”
When Cotner presents this idea, some STEM instructors are reluctant to de-emphasize exams. Often, their argument is that working scientists are expected to perform in high-stakes situations, but Cotner counters that high-stress exams may not be the most accurate predictors of who does well later on.
“There’s a need to make sure that the scientists we graduate are competent,” she says. “But…I think there are a lot of ways we can reveal competence and nurture competence without giving developing scientists these really high-threat testing environments.”
She points out that de-emphasizing exam grades is a relatively low-effort intervention to try. In many undergrad STEM classes, weekly homework assignments, lab reports, and class participation are already part of the overall course grade; all the instructor has to do is make those non-exam assignments a larger chunk of the grade.
Cotner says it’s not entirely clear why women tend not to do as well on high-stakes exams. One hypothesis is that women may be experiencing stereotype threat, a psychological phenomenon where members of a group become preoccupied with a stereotype that ostensibly describes that group. Women who have repeatedly received the message that “women aren’t good at math and science” may be more anxious and distracted during high pressure tasks that remind them of that stereotype — like exams — even if they are highly proficient. Lowering the stakes on exams may reduce that anxiety and allow students to focus more easily.
Stereotype threat isn’t the only possibility. “The differences we see could be driven by women doing better when it’s lower-stakes, but it could also be driven by men doing worse when it’s lower stakes,” says Cotner. “It could be our male students who are just phoning it in when they don’t perceive [the small assignments] being that important.” Without further research, though, this remains speculative.
Psychologist Toni Schmader of the University of British Columbia, who was not involved in the study but who studies stereotype threat, says that since the study did not measure whether students felt self-conscious about stereotypes going intothe exam, this study by itself doesn’t show whether stereotype threat is the cause. However, the idea of women feeling heightened stereotype threat on higher stakes exams is plausible, given the existing literature, she says.
One of the most pernicious stereotypes about women and minorities in STEM is the belief that success requires innate talent or natural genius. Fields where belief in “fixed ability” is more prevalent, such as maths, physics, and philosophy, tend to have wider achievement gaps for women and minorities. While reading Cotner’s paper, Schmader wondered whether high stakes exams might inadvertently send the message that the instructors are looking for “raw talent” rather than effort and growth. “If that’s the case, then perhaps high stakes testing is bringing to mind those kinds of beliefs,” Cotter says. In other words, students who see exams as rewarding raw talent over growth might be primed to suffer from stereotype threat. Schmader is quick to point out that “the evidence isn’t there to connect all those dots.” But, she says, “that would certainly be an interesting avenue for further research.”
Cotner is following up the study with further work through aNational Science Foundation-funded collaborative network that will test out and collect data on strategies designed to lower barriers for women and minorities in STEM classes. In one experiment, they’ll see whether bringing in therapy dogs for students reduces stress and improves exam performance.
In another, they’ll see whether allowing students to choose between grading schemes at the beginning of the semester (before exams) helps improve performance. In theory, students who tend to lose track of small assignments could choose to have their exams weighted more heavily, while students who tend to psych themselves out on high-stakes exams could choose the opposite. Whether this strategy works would depend on whether students know themselves well enough to accurately choose the grading scheme that is best for them.
Cotner’s long term goal is to identify strategies that make a quantitative difference for students. “If it doesn’t close the gap, I’m not going to recommend it,” she says.
However, re-weighting grades isn’t the only strategy for boosting engagement and performance in STEM class that Cotner recommends. Here are a few other tweaks that instructors can make without changing exam content:
- Don’t “spread out” women and minorities between groups. Some instructors try to set up small groups such that women and minorities are distributed across all the groups (rather than all sitting together), but being the only woman and/or only person of color in the group can make students feel tokenized and self-conscious.
- Give students time to discuss material in small groups. Fear of asking a stupid question in front of a large class can hit students from under-represented groups especially hard. (Even among professional scientists, men are more likely to ask questions at conference seminars.) Cotner finds that letting students bounce ideas off each other in small groups before starting the full-class discussion helps students feel more confident in their questions and comments.
Diana Crow is a freelance science writer in Providence, Rhode Island.