Wednesday, April 15, 2015

New Study Demonstrates Shocking Truth About Faculty Hiring

Screen grab from CNN for the OpEd by the authors of the PNAS study. 
(*) In a paper just published in PNAS, Cornell professors W. M. Williams and S. J. Ceci have demonstrated conclusively that the process that all university departments use to hire new faculty is completely unrelated to the actual process they modeled in their study of fictitious faculty searches.

When the Harvard University Department of Astronomy undertook a recent faculty search, the Harvard faculty asked applicants to submit a CV, a list of publications, statements of research and teaching interests, and to arrange for confidential letters of recommendation. The department reviewed these materials, selecting a half-dozen applicants for interviews. Each individual visited for two days, during which time they delivered a colloquium, and met with faculty and students, including several dinner meetings. The faculty then convened for several hours to decide on whom should receive the offer.

The complex interplay of these various elements in a job search was noted by some of the Harvard faculty: "Indeed, I can't imagine anyone actually hiring a new professor without reading their CV and even some of their papers, attending their colloquia, and undertaking a critical examination of their ideas and plans," said David Charbonneau, Professor of Astronomy at Harvard.

Study authors Williams and Ceci noted the critical importance of these materials in the faculty search, writing in the introduction of their PNAS article that "Hiring tenure-track faculty... entails selecting among highly accomplished candidates, all of whom have completed Ph.D.s and amassed publications and strong letters of support."

However, by the methods section, the PNAS authors had dispensed with these requirements.

Instead, the authors proposed a process by which voting faculty consider only brief, written narratives of only three candidates: They should not be sent CVs, publications, research and teaching statements. Moreover, since confidential (and often critical) letters of reference are often given enormous weight in deliberations, the study authors did not include these.

"At first, I didn't find the study persuasive," wrote Prof. Charbonneau. "Maybe their method was a reasonable proxy for our faculty searches? But then I read in the methods that the authors hadn't included anything akin to publications, reference letters, job talks, or dinner conversations. I was convinced: The PNAS study proves conclusively that the faculty hiring simulation in the PNAS study is totally unrelated to any faculty search that I have ever been part of."

As a final step to distinguish their proposed process from actual faculty searches, the study authors ensured that survey respondents knew that it was a fictitious survey. Potential respondents were asked to distinguish between otherwise identically qualified individuals who differed in gender. Only 1/3 of survey recipients responded.

"For me, this last part was the nail in the coffin, " commented Charbonneau. "In our searches, all faculty members are required to participate and vote. But in the PNAS study, it was likely that only those survey recipients with a stake in the survey responded. Surely this subset of the nation's top faculty wouldn't discern the deeper purpose of two brief narrative summaries of knowingly fictitious individuals who differed only in gender?"


(*) Yes, Dear Reader, this is my attempt at satire. I actually do have a nuanced view of the PNAS article. I certainly don't want anyone to delude themselves into thinking I was being sincere here. Doing so would be akin to conducting a survey along the lines described above and then writing a prominent CNN Op-Ed piece concluding that "The only sexism [women] face in the hiring process is bias in their favor."

1 comment :

Anonymous said...

I have seen many criticisms of the study that center around your final paragraph ("surely the respondents knew it was about gender!"), and was initially quite skeptical about this myself. However, in their supplementary materials (here, page 6), the authors state "when a subset of 30 respondents was
asked to guess the hypothesis of the study, none suspected it was related to applicant gender." None! That fairly strongly indicates that their disguising techniques worked.

The rest of the study has problems, and the authors' generalizations and op-ed on CNN are atrocious. Still, lets be honest about the rigors of the paper itself.