Grad School Recommendation Letters

This week I wrote graduate school recommendation letters for two students whom I don’t know very well at all. One received an AB and another a B in the classes they took with me.

When they asked me for letters (they did so separately and are in no way connected),  I explained to each that I didn’t know them very well and that I usually didn’t write for students who hadn’t gotten an A in my class. The thing is, they each have one professor who knows them well, but need at least 3 letters for their applications. I make a point of remembering everyone’s’s face and learning their names  (it’s doable even with a 100+ people in the class, but requires some work), so I guess they came to me because they were at least sure that I knew who they were.

How many undergraduates really have 3+ professors who know them well? A couple of my former graduate researchers come to mind, but they were stellar (it helps with getting remembered if you do great in someone’s class), engaged with a research group for over a year, and having college-level extracurriculars. But most students aren’t superstars; even if they do research, that advisor is likely the only professor who knows them well enough to write something specific.

Why do undergraduates need 3+ letters to get into graduate school? Couldn’t we say, “At least 1, and make it substantive?” When I look at applicant’s letters, there are always lukewarm ones from the people who clearly don’t know the student. At best, these letters add nothing of substance to the case. At worst, they make the applicant seem weaker through the well-known “damning by faint praise” effect. Students from large universities cannot be expected to be well connected with multiple professors the way they are at some smaller schools.

I wrote the letters that are as specific and as positive as I could. The B student is switching majors, and is passionate about the topic he is about to pursue for his Master’s more than the one he’s received a Bachelor’s degree in; I knew as much and focused on his motivation and ambition for the new field. The AB student is actually very bright but I would say somewhat lethargic; I focused on his capability and how it manifested itself in my class.

I told both students that I would write the letters, but that they would be fairly generic. I said I would try to make the letters as positive as I could to the extent to which I knew the students, which isn’t much. They were fine with it.

I did what I promised and I wish them success with their applications.


  1. Given the competition levels in and after grad school, are undergrads who aren’t stellar really going to make it? I went to a undergrad university of over 40,000 students and I had no problem getting my three letters from professors who knew me well. But, I knew I had to do things like go to office hours, do research, etc. in addition to being stellar in class. Students need mentoring to know these things of course, but well…can you ma eit in grad school if you aren’t self-motivated?

  2. This is easier to an extent in my field, as all my classes are small and writing intensive. But yes, I run into this issue regularly. I spent about 2 hours yesterday revising and submitting a letter for a top student (as in she was a university nominee for Rhodes scholarship) and it was still difficult to do, as her honors thesis isn’t close to finished. And that’s one of the easy ones! What of those who do well but don’t distinguish themselves? And what of those who sabotage themselves? I had a student ask me a couple weeks ago if I’d write her a letter due Dec 1…and this is a student who has missed enough classes that she should technically fail. She’s a great writer, but that’s not a letter I could agree to do for many reasons. I do wonder if you’re right, that one letter should be enough for undergrads applying to grad programs…

  3. Maybe students who haven’t really distinguished themselves in the eyes of multiple faculty shouldn’t be going to grad school. The world won’t be hurt by one less grad school applicant, and the graduate can get an earlier start on their career and learning the industry.

  4. @postdoc: In many physical science fields, grad school without a path to becoming a PI is a still good use of the students’ time and effort because the job prospect are very good. These two students want to get an MA or an MS (one of them is contemplating MS/PhD) and then go and work, which they will be able to do. One of them (the B student) is very motivated by specific commercial applications and has an entrepreneurial spirit; he wants to learn more and get a Master’s degree in a specific field, so can do what he envisions better. I think that’s an excellent use of a Master’s degree.

    I don’t think we should discourage people from all forms of graduate education because they won’t make good professors or PIs. We should try to inform them honestly of the cost/benefit ratio and the different opportunities. These days, having a college degree is a baseline requirement, like what having a high school diploma used to be. It seems having a graduate degree of some kind is necessary for most well-paying jobs, and definitely for many in the physical sciences.

  5. thanks for answering. My perspective is a bit different coming from the social sciences. I tend to advise students that grad school is something they should only do if they love the field and can’t imagine doing anything else. Cause it will take 6-8 years and the job prospects are not good.

  6. Thanks for a thoughtful post on an important issue. I write letters for many, many different reasons, ranging from those for undergraduates in a 200 person class, where I usually can do little more than report a grade, to those for neighboring grad students and postdocs, to those for my own trainees.

    Having served on many a grad school admission committee, I know how difficult it is for students to find three professors who know them well at a largish school. I thus generally am happy to write for students in the grade range you’re discussing (As and Bs). For students who did more poorly in my class or lab (low B, C), I’d try declining to write if the letter would not be supportive. As a committee member, I generlaly only looked closely at the letter(s) from faculty with whom the student had done research–those letters have much more detailed information, often gathered over several years.

    I”d also note that in my field, there is now good data to show that many of our standard quantitative metrics (GPA, quantitative GRE scores) have no predictive value with respect to success in graduate school in biomedical science (e.g.,

  7. We make sure that our undergrads who wish to go to grad school can get at least one really meaningful letter and a few decent ones, by requiring a capstone research (or engineering design) experience that is 2-3 quarters long. In addition students take intensive upper-division courses with faculty who get a good view of student abilities. Students who should be going on to grad school have no trouble getting letters. Students whose work is barely passing have a hard time getting letters, but shouldn’t be contemplating grad school.

    On the grad admissions committee, I find that most academic letters are fairly informative (except those that just tell me that a student took a class and got a good grade—I can get that from the transcript), but letters from industry are generally so contentless that they tell me nothing about the person other than that they worked for the company for a while.

    To reduce the time it takes me to write a letter of recommendation, I require that students give me a draft of the letter of recommendation to edit. I usually end up rewriting most of the letter, but the draft gives me a detailed place to start from (so I don’t forget what the student did—these letters are often several years after the student has graduated), and the better students usually give me better drafts to work from, so there is a higher quality letter for the higher quality candidates. It also serves to discourage students from randomly picking me as a letter writer just because they passed a class from me.

  8. I’m surprised you can’t connect the dots on this. It’s exactly because those students who actually make a positive impression on multiple faculty are likely to be genuinely outstanding as grad students.

  9. The study above makes a classical mistake: they look at a highly selected group of students which cleared the admission hurdle. So sure, a low GRE had no correlation with success since the student had already proved to the committee through other outstanding abilities that they were worth having.

    The properly controlled experiment is to grab a bunch of students with similar qualities with the GRE scores being the only discriminator. We did this somewhat unwittingly a while back at my institution: we would ask for the GRE but not make it available to the admissions committee (for some weird legal reasons, don’t ask). This went on for a few years and eventually we went back and revisited the lowest scored accepted grad students. For our field GRE quantitative was a strong predictor among that group. Today we use the GRE scores exactly that way: as an extra input in borderline cases and ignore it in cases where students already have extensive and distinguished research experience.

    Google had similar results: if you use extensive selective procedures you will find low correlation with grades, meaning “among very bright students there are some that sucked at exams”, however this does not disprove the converse “percentage wise there are more bright students among the high GPA ones”.

    Indeed today the Google job interview is pretty similar to acing a final exam: they subject you to known questions which are available on the web and you can study beforehand. They find that if you do well in that phase, which you can do either by sheer talent or by studying the material, then with high correlation you are a bright top engineer.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s