Physics Grading Biased Against Women

By Emily Conover

A complex array of factors seems to be holding women back from the field of physics, and now it appears that grading biases may play a part. A recent study performed in several German-speaking countries indicates that physics teachers give better grades to male students than female students — even when their answers are identical. However, this bias disappears as teachers become more experienced. In Switzerland and Austria, secondary school teachers of both genders gave lower grades to female students, while in Germany, female teachers displayed the same bias. Surprisingly, male German teachers were an exception to the pattern, grading students equivalently regardless of gender.

The study focuses specifically on physics teachers, but such biases have already been shown to be present in science education in general. "The fact that teachers unknowingly evaluate students differently as a function of stereotypes about who is supposed to be smart and good at a given subject is pretty well established," says Nilanjana Dasgupta, a professor of psychology at the University of Massachusetts at Amherst, who was not involved with the study.

"What’s new in this is the effect of teacher experience, which I thought was really interesting," Dasgupta says. "It makes sense that as teachers become more experienced … that’s when their gender bias in student evaluation goes away."

The researchers asked 780 physics teachers to grade a simple exam question, answered by a fictional student. The question was a conceptual Newtonian mechanics problem, which asked students to explain in words what would happen to two skateboarders, each holding one end of a rope, when one skateboarder pulled on the rope.

The choice of such an open-ended question was intentional, says Sarah Hofer of ETH Zürich, the author of the paper, which was published in the International Journal of Science Education. "Such [questions] are more prone to biases or to stereotypes because stereotypes particularly take effect in situations that are ambiguous."

Each teacher was presented with the same student answer, which showed some understanding of physics concepts, but was not fully correct. The student's gender was randomly listed as either male or female. The student's academic focus — languages or sciences — was also randomly assigned.

For newbie teachers at the beginning of their careers, the fictional boy who answered the question had a statistically significant advantage over the fictional girl, even though their responses were identical. For example, in Switzerland, where grades range from 1 to 6, the mean grade difference was 0.7, meaning that a girl might get a 4.8, while a boy got a 5.5 for the same answer. This bias gradually decreased with teaching experience, disappearing for teachers who had taught for at least ten years. The student's academic focus had no influence on the grade the student received.

The only exception to this pattern was male German teachers, who graded boys and girls equally. The explanation for this difference is unclear. "It was really surprising because all other teachers really behaved the same; only this group was different," says Hofer. She speculates that the difference could be due to programs in Germany that attempt to sensitize teachers to gender issues in physics. "Perhaps this was effective, but especially for the male German teachers," Hofer says.

Because the finding was unexpected, in order to conclude there's really a difference in male German teachers, "The thing to do would be to see if you can replicate this effect," says Dasgupta. "If you get it more than once, then you begin to believe it’s a real thing, and not just something fluky."

The importance of experience for reducing stereotyping can be understood in the context of previous research on biases, Hofer says. "Stereotypes may influence the judgment process, particularly in situations that are cognitively demanding and provide the weakest information," she says. "With increasing experience, teachers can be expected to develop all the cognitive resources that are necessary to avoid the influence of biases on the judgment process."

Alternatively, the effect may be due to the repeated exposure to female students who are good at physics, which contradicts the teachers' implicit biases, Hofer says. "So the stereotype itself disappears."

Luckily, the problem has potential solutions, researchers say. For example, grading can be done using student ID numbers instead of names, so that the student's gender is unknown to the grader. "I do that routinely and I find it an interesting experience," says Dasgupta. "Often I’m surprised that my initial impression of a student in class doesn’t match the grade I give them on an assignment."