35

This is an attempt to rescue a strongly down-voted question.

Are there any studies investigating possible unconscious gender bias in evaluating recommendation letters? Specifically, is there any published evidence that recommendation letters with female authors are (or are not) less effective than recommendation letters with male authors? Studies considering letters for graduate admission, faculty hiring, or promotion and tenure are all relevant.

Let me emphasize that I am not asking about intentional sexism, which I assume is sufficiently rare to be insignificant, but rather unconscious bias. I am also not asking for anecdotes, but pointers to actual published literature.

Similar studies have revealed significant gender disparity in several related academic contexts, including recommendation letters for male vs. female applicants. Other examples include:

  • Are requests in lieu of literature searches really on topic? – keshlam Aug 17 '15 at 1:01
  • 11
    @keshlam that is basically what the reference-request tag is for, yes. – ff524 Aug 17 '15 at 1:18
  • 1
    It's hard to see how a test with recommendation letters from imaginary recommeders (varying the gender of the recommender) could be done. In decreasing order of influence the recommender could be known personally to the reader ("Jack and I coauthored a paper together 12 years ago", or by reputation ("I've read lots of Jane's papers"), or by association ("Prof X. is at MIT, so I'll give this letter some credit even though I've never heard of him.") A recommendation letter with non of these features wouldn't give a realistic test. – Brian Borchers Aug 17 '15 at 2:26
  • 2
    Are you asking about studies, for example, that use identical letters and swap the writter's name, or differences in how men and women write letters. While both are issues, they seem different, – StrongBad Aug 17 '15 at 2:41
  • @BrianBorchers Depending on the field, you could do the test with imaginary recommenders, but it's hard to see how you could mine the data from existing processes, or how you could run an end-to-end test on the whole admissions process. In a system where people bid on folders they want to review, and some fraction of folders have all unknown recommenders, one could imagine seeing how bids correlate across recommender names. Of course, most letters are PDF files, so not so easy to edit the recommender name. – user3188445 Aug 17 '15 at 7:33
11

I couldn't find any recent (written after 2000, that is) study that directly addresses the issue of how the recommender's gender influence how the letter is perceived. However, I did find one interesting survey study (doi: http://dx.doi.org/10.1111/j.1468-2389.2009.00453.x) that asked experts in personnel-related professions how they think about LORs. One item asked them to rate on a scale from 1 (strongly disagree) to 5 (strongly agree) the following statement:

I believe that applicant, referent, or reader gender may influence how one interprets the contents of LORs

The average response was 2.86 with an SD of 1.22. About 42% strongly agreed, and 43% strongly disagreed. Clearly there is some level of controversy when it comes to the role gender plays in LORs. The study did not, however, ask any question directly about the referent's gender, which makes it even harder to draw any specific conclusion for our purpose here.

  • Interesting catch. And RQ 14: I believe that applicant or writer gender may influence how one writes a LOR has interestingly 43.8% agree 39.8% disagree. – Zenon Aug 17 '15 at 4:07
  • 2
    ^ I would guess the part Zenon mentioned is the primary source of the controversy. As far as I know, it's fairly well established that there are significant gender differences in writing styles. Here's a paper on the topic and here's a website that guesses the author's gender of a given text based on the aforementioned work. – reirab Aug 17 '15 at 6:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.