Let's Can Recommendation Letters

If each applicant must submit three letters, and each letter takes three hours to write and submit, how many faculty work hours are consumed when 104 applicants apply for a single position?

Last night I checked my email before going to bed and saw a request for a recommendation letter from a student. It is the sixth request that I have received in the past week. I know many more will come.

Every year I teach more than 200 students. Some go on to doctoral programs, and others seek jobs following graduation. I receive, on average, about 20 reference requests per year -- including requests from former students.

To write a good letter takes me about three hours, not including the administrative time that could require other correspondence, and sometimes meeting with the student.

I must think about the candidate, review the CV, pull out old assignments and grades, and review their scholarly work. I must also consider the specific program or job to which they have applied and then think how this candidate would be good for this specific site. What qualities make this a good match?

It is true that some medical schools and graduate schools are opting for online platforms such as Interfolio and the SOPHAS systems that facilitate submission of a common letter to all programs to which the student is applying. However, not all programs have joined in. Some others have adopted the common platform but still require that each person be rated on a scale and other essay questions be answered. Many of these essay questions overlap in content with the letter.

So to sum up, for each student who requests a letter, I must spend approximately three hours to write the letter itself and between one and three hours on administrative details such as uploading, checking the questions on the rating scale for each site or program, and filling in my contact information on forms.

Multiply this by multiple applications per cycle (i.e., applying to several jobs and academic programs) and the result is that I burn several hours each time I write for a given student.

And here’s what I could have been doing with that time: scholarly work, treating patients, writing a lecture, working with a student on a thesis or publishing research that might help the world in some small way (or -- gasp -- hanging out with my family, since I end up doing most of this work on weekends).

I also know such letters are vital to my students. For most academic programs and jobs, applications cannot be submitted without three letters of recommendation. So why do I, and my colleagues, consider this to be a drain on our time?

Because I know that, in many cases, the letter of recommendation will not be reviewed. In my experience on admissions and selections committees, the academic CV, relevant experiences and quantifiable data are what put the candidate under consideration. It is only after a candidate has cleared the initial review that recommendation letters are read. That means in the majority of cases, the applicant’s letters will not be reviewed.

Moreover, the purpose of these required letters is somewhat suspicious. Presumably, the point of the letter is to indicate whether the candidate is a good worker -- the letter must be trusted to provide some information that cannot be ascertained from a CV. But research indicates that the letter of recommendation is a poor predictor of job and academic performance. The letter also introduces the potential for bias, calling into question its usefulness and appropriateness altogether. Finally, although it is seldom mentioned publicly, some professors adopt a “You write the first draft and I’ll sign it” policy, or they have a teaching assistant do the writing. Considering that these letters are sometimes an elaborate exercise in rubber-stamping, how necessary are they?

If we are seeking to have an unbiased system of student and employee selection, unencumbered by nepotism and personal favors, we should consider alternatives to the recommendation letter. In academics, I would like to think that ours is not a system of who you know but what you do. In which case, the CV and transcripts should be the first line of data.

In our lab at Yale School of Medicine, my mentor implemented a triage system for considering applicants. Applicants first submit their CV and personal statement/letter of interest. Applications are reviewed and letters requested only from those who are under serious consideration. Given the hundreds of applications that we receive every year, we have likely saved -- at minimum -- three work hours per applicant. That potentially translates to hundreds of hours of letter-writer time saved.

The next time I recruit for a postdoc, I intend to replace the letter with a request for recommenders’ names and contact information. From informal discussions and from research itself, I have learned that many people are unwilling to write negative things in letters, hoping that the review committees will read between the lines. So from what I can tell, it comes down to this: for a good applicant, the reference letter provides very little information beyond what can be ascertained from a CV.

Because when it comes to selecting graduate students, interns, postdocs or employees, the most predictive information comes from the quantifiable data. But if we are seeking subjective information (i.e., whether we would like to work with this person), wouldn't it be more efficient to place a phone call to a former supervisor?

Students also stand to benefit from a stepped process. Instead of investing their time and energies pursuing positions for which they are not competitive, they can instead focus on the programs or jobs for which they are viable candidates.

Can we consider -- for the sake of our own productivity and that of our students -- revamping the letter-writing process? Either skip it altogether in favor of a phone call or perhaps only request letters from those applicants who are truly in the running.

We could spare us all the sales pitch of the letter writer -- and a lot of wasted effort.