Monday, April 20, 2015

Some Professors are More Biased Than Others

Motivated by a New York Times article, I looked up a paper in the Journal of Psychology about a study done to assess bias in university professors.  The results were based on a large blind audit of professors in various fields and were remarkable.  In all fields except the humanities, professors systematically replied to correspondence from professors than students with white male names differently than students with female names or names suggesting non-white race.

The study was performed by Katherine L. Milkman (U. Penn), Modupe Akinola (Columbia) and Dolly Chugh (NYU) and published on-line this year.  E-mails were sent to a large selection of professors in 89 disciplines and 289 institutions  The e-mails were from fictitious students expressing interest in a professor's research and asking for a short meeting during the student's visit to campus.  The student names were chosen to suggest gender and race, such as Meredith Roberts and Raj Singh.  A total of 6548 e-mails were sent to randomly selected professors.

The result were that professors on average were more likely to reply to students with white male names than any other group.  (E-mails were immediately sent to cancel any appointments with the fictitious students.)  All e-mails were identical except for the names.  With the students sight unseen, the bias in the thinking of these academics was revealed by value association based on names.  The results are shown in the figure (discrimination in grey and reverse discrimination in black) using percentages.  The number of samples for each disciplinary category ranged from fewer than 200 to more than a thousand.  There were 850 samples for the Natural, Physical Sciences and Math.

The results are most striking in the disciplines of business and education at the levels of 10's of percent response differences and lower in the science disciplines at the levels of few percent.  In fine arts, the bias was strongly reversed with professors replying systematically more to female and non-Caucasian students.  The only field where there appears to be little of no bias is in the Humanities.



The survey was designed to include a number of additional variables, including the demographics of the faculty who were sent the e-mails.  While the sampling numbers for those fine disciplines are smaller, I was surprised and disappointed to note that the bias appears to persist even with female and non-white professors.

It was interesting to learn the terminology in a study like this.  The experiment methodology was termed an "audit" since it did not survey the faculty, but determined information about their thinking through indirect e-mails.  The study concerned "pathways" to success and not the more usual "gateways".  Examples of gateways are entrance exams and college acceptance processes.  The pathway in this study was the informal step students take to contact a professor to learn about their research and study opportunities ... a pathway to the gate.

The result that bias occurs even in female and non-white professors is disturbing because it implies that their increasing representation will not immediately fix bias problems.  The only solution to this undercurrent of discrimination, probably mostly unconscious bias, is to raise awareness of the problem.  This study is one way to educate academics that bias exits so they and their administrations can re-double efforts to eliminate it.