Two new experiments seem to show that both college students and instructors carry biases when one rates the other. One study shows that better-looking female students have an advantage in the classroom, while another shows that male professors get better teaching evaluations than their female peers.
In a large study at Metropolitan State University of Denver released in January, researchers found that attractive female students received measurably better grades than their less attractive peers — but only in traditional classrooms. That apppearance boost disappeared in online courses.
The second study found that female instructors were consistently rated lower than male instructors. This held true in the U.S., where the study swapped names in online courses to change gender perception, and in France, where researchers compared actual student achievement levels to instructor ratings.
Both of these studies, experts say, fit into a growing body of research on “implicit bias,” or subconscious judgments that most of us make on gender, race and other obvious personal characteristics.
Surprising results
In the MSU Denver study, researchers compared over 168,000 course grades given to 6,500 students. Researchers used photo ID cards to measure attractiveness, asking respondents to rate students' appearance on the familiar 10 point scale.
The researchers then compared grade results between traditional classroom courses and online courses, where the students' appearance would be less evident.
"As the a female student's attractiveness rises, the difference between traditional course grades and online classes gets larger," said Rey Hernandez-Julian, the study’s lead author, an economics professor at MSU Denver.
The results did not hold for attractive male students, who did not get a boost in either class format.
While previous studies have shown that appearance and grades correlate, it was widely assumed that this effect was really the result of "unobserved productivity," which is the notion that productive people appear more confident and care for themselves better, which makes them register as more attractive.
In keeping with this theory, Hernandez-Julian says he originally expected to find that "attractiveness" either reflects or breeds confidence, or both. That is to say, someone may appear more attractive in a photo because they look more "with it," and that confidence enhances their appearance.
More successful students may "pay more attention to grooming, exercise and the clothes they wear,” and those behaviors may translate into perceived attractiveness, Hernandez-Julian notes.
Another possibility is that more attractive students may have received positive reinforcement over the course of their lives, leading to greater confidence and efficacy.
In fact, by showing that the looks boost for grades is lost in online classes, this new study suggests that discrimination does play a role.
There are still some key questions unanswered, Hernandez-Julian noted. "Do professors invest more time with more attractive students?" he asks. "Or do they reward appearance alone with higher grades?" He suspects the answer may be yes and yes.
Evaluating teachers
In another paper released this month, another team of researchers released results showing that teacher evaluations are biased against female professors, based on studies in both France and the United States.
The researchers in the U.S. used online courses and swapped names of the professors so that the students would not know whether the teacher was a man or a woman. Because the courses were online, students did not know the actual gender of the instructor, and the same instructor was given either a male or female name.
In France, students in a six large courses rated the instructors in their smaller sections within that course. These ratings were then compared to the results of the standard final exam. The French study demonstrated that female instructors received lower ratings, but that their students actually did better on the final exam.
Both studies showed bias against female instructors, but there is a puzzling oddity in that in France it was primarily male students who showed bias against female instructors, while in the U.S. it was primarily female students.
Stark said there are too many differing variables to really untangle this discrepancy.
The studies differed markedly in their methodology, he said, and there are likely also differences in culture that may be masked by the fact that French law will not allow ethnicity data to be collected for such studies. It is possible, Stark suggests, that a large cohort of nonwestern males in French universities could skew evaluations against female professors.
"It does suggest that these biases are incredibly complicated," Stark said. "And the bottom line is that these evaluations clearly don't measure teaching effectiveness."
Looking at a teacher evaluation to measure the effectiveness of instruction, Stark said, is like using a bathroom scale to measure someone's IQ. "It clearly doesn't measure teaching effectiveness," he said.
Stark pointed to other studies that suggest that even if male and female instructors were to get similar ratings, it may simply show that female professors work harder to get similar results.
Stark said some data suggests men are gauged more on whether they have "command" of the classroom, while women "have to work much harder on the underlying pedagogy." In short, students are more likely to respect the aura of authority carried by a male instructor, while women must clear a higher hurdle to get the same results.
Why does gender matter in teaching evaluations? Stark doesn't have firm answers, but he speculates that on a visceral level, a female instructor's expected role may differ from that of a man. After exams, he says, a female teacher often has to "replenish the Kleenex" more than a male. That is, students feel more free about unloading emotional burdens on female teachers.
Gender-biased teaching evaluations may be, Stark suggests, more of an archetypal problem than a socially constructed one.
Implicit bias
These findings also intersect with a growing area of psychology that explores "implicit bias," or "thoughts and feelings outside of conscious awareness and control," usually based on gender, race or other physically apparent traits.
One of the leaders in this field is Tony Greenwald, a psychology professor at the University of Washington and a co-founder of Project Implicit, a high-profile collaboration center for implicit bias research now based at Harvard University.
The attractiveness and grades study does not surprise Greenwald. This a form of implicit bias known as the "halo effect," he says, and it has been known for at least 50 or 70 years.
"If you know one thing favorable about someone," Greenwald said, defining the Halo effect, "you tend to judge other things favorable, and often the first thing you pick up is biased."
Greenwald has collaborated for years on these issues with Mahzarin R. Banaji, a Harvard sociologist. They published a 1995 article in Psychological Review and a 2013 book, "Blind Spot: The Hidden Biases of Good People."
Certain cues, such as age, race and ethnicity, are so obvious that they swamp all other data. "We like to think we can ignore them," Greenwald said, "but we can't."
Project Implicit offers an online self-diagnosis and research tool that asks viewers to respond to a blend of positive and negative verbal cues combined with pictures of black and white men.
The photo test is followed by a personal reflection section where test taker gives some personal background a more conscious response to related questions. This section does not factor into the test results, but is used to collect data.
Project Implicit has also found that many African Americans have an implicit white preference, Greenwald said, though they are only beginning to understand why.
One of the most intriguing early findings, he said, is that some blacks with automatic white preferences may have had positive relationships with highly admired white teachers in elementary school.
The potential for such a dynamic to work in the opposite direction is also there, he said, and there is some evidence that a highly admired black figure could similarly help counter implicit bias.
There is a widespread misconception that once you know about your biases, you are cured, Greenwald says, and some companies are marketing training about implicit bias and major corporations are using it. But there is no evidence that it produces beneficial effects, he said.
Email: eschulzke@deseretnews.com