Last week, I got my first request of the season to submit a reference letter for a former student who’s applying to grad school (with a crazy early deadline in September, too…). This particular student has a sizeable list of places he’s applying to, so this is just the beginning of a process of creating 27 new logins for all the different schools using different services, and getting paranoid about whether this one has flipped the order of the radio-button options for the quantitative portion since the last one. It’s a bit of a grind, really, not one of my favorite parts of being faculty.
I basically never turn these down, though, because I know how important they are for the students. (The one case I can think of where I tried to say “No” didn’t actually work— I replied to the student asking me for a letter with something like “I think you would be better served by a letter from somebody other than me,” but he put my name in anyway and I got the automated request for a letter.) I had another case of a student who wasn’t going to get an objectively good recommendation from me, but had burned all his other bridges more comprehensively, so I told him “This is the kind of thing I can write for you, if you still want me to do this, I will.” (He took it, and I put the best face I could on what I had to say.)
I’ve never done graduate admissions (we don’t have grad programs at Union), so I don’t know how much weight these actually get in the process. I suspect it’s pretty substantial, though, because I know I find reference letters maybe the most useful part of the process I do take part in, which is faculty hiring.
This is a somewhat controversial position, I know. We’re at the start of the season where I get requests to provide references for students, which means we’re also at the start of the season of social-media posts denouncing the practice of asking for reference letters with faculty applications. The usual argument is that references should only be checked at the shortlist stage of the process, after the pool has been cut down to people who might get invited for interviews. I’ve gone a few rounds over this on Twitter through the years, but I’ve started basically ignoring them there, because it’s never a productive conversation. I’ll take a crack at a long-form version of it, though, just for posterity.
Just to set the stage a bit, Union is a small private liberal-arts college, which means we teach only undergrads. In physics, when we advertise an open position we typically get something like 100-150 applications, depending on the year and the nature of the position (tenure-track jobs get more applications than visiting positions, ads expressing a preference for a particular research specialty get fewer than open ads, etc.). Something like 30-50 of these will just be flatly inappropriate— chemical engineers in former Soviet republics claiming they can teach physics, and that sort of thing— leaving about 100 plausibly qualified candidates. The materials we generally ask for are a curriculum vitae, a statement of research interests, a statement on teaching, and 2-3 letters of reference. (In the last couple of years we’ve added a request for a statement on diversity, equity, and inclusion issues, but I haven’t been on the search committee since we’ve been getting those.)
Given that set of material, I find the letters absolutely essential for making the cut from the 100-odd applications that meet the formal requirements down to the oversized handful we need to seriously discuss inviting to campus for interviews (we normally get 3 interviews, sometimes 4). The CV is good for checking that the formal requirements have been met, but not much beyond that— you can do some paper-counting and journal-impact-factor tallying, but both of those are problematic. The statements are, for me, a lot like the college admissions essays I was complaining about last week: the format is very stilted and artificial, and as a result they basically just set a (fairly low) threshold to be cleared. There are sometimes disqualifying items in these— sentences like “Given three graduate students and two post-docs, the lab should be up and running in two years.” (We’re undergrad-only, and only rarely able to hire post-docs.) For the most part, though, anyone can put together research and teaching statements that satisfy the requirements (competently written, proposing a feasible and interesting research program, saying reasonable things about teaching), given that they have a long time to write them and access to colleagues who can offer editorial feedback.
(Aside to anyone who might be on the market: Get editorial feedback from colleagues, and listen to it. Ask people you know who have the kinds of jobs you’re applying for to take a look at your application materials (well in advance of the deadlines!) and tell you what they think. If you’re not doing this, you’re a fool.)
The formulaic nature of the statements is why I find the letters so helpful. Some of this is a function of our unusual setting— liberal arts colleges are a strange and intense environment, and many people at R1 institutions just don’t know what it’s like. Outside of a handful of candidates who have come through a similar system and can speak directly to that in their statements, this is one of the best places to find out whether they’re really interested in what we do, or just shotgun-applying to every ad in Physics Today. Positive letters from faculty who know something about the environment are gold, but even a letter that says something like “This person would be great at an R1, but they have this crazy insistence on wanting to teach undergrads no matter how much we try to talk them out of it…” is a really good sign.
But even outside the quirks of our format, I find these incredibly helpful, because they can offer commentary on the candidate that brings perspective we don’t necessarily have access to. A good reference can position the research proposed by the candidate within the broader context of the field and subfield in a way that we usually can’t (it’s exceedingly rare for us to be considering a candidate who works in the same subfield as somebody already on the faculty). People who have taught with the candidate can put that in context, too— there’s a big difference between using a pre-written syllabus and lecture notes and inventing new materials, and faculty can compare the candidate to other instructors at the same level. They can also give a sense of how the candidate is as a colleague that goes beyond lines on a CV— it’s easy to be on committees or projects without contributing anything significant, and a good reference will make clear which side of that the candidate is on.
The arguments against asking for letters up front also fail to convince me that this is a problem. The primary complaint is that it’s a lot of work for PIs to write letters for their students and post-docs who might not make it to the interview stage which, cry me a river— I have to read 300 of these. If your students are competitive for jobs, you’re going to need to write those reference letters sometime, so this would really only be putting it off until later, when it would be asked for on a much shorter time-scale.
The slightly more convincing argument is that it puts too much of the candidate’s fate in the hands of people they can’t control, which is true— they generally can’t force one of their references to send in a letter on time, and they may be subject to the whims and biases of their references. But both of those things also apply to the search committee at the other end. We don’t have any more power to compel the production of letters than the candidates do (again, at the shortlist stage this would be on a much shorter time scale than letters up front), and it’s not like the people on the hiring side are any more free of weird preoccupations and biases than the letter-writers would be.
(There’s a variant of the no-letters proposal that calls for never making the references put anything in writing, but doing phone or Zoom interviews with them later. This strikes me as absolutely insane in a number of ways— if you want a process afflicted by wildly problematic biases, that seems like the best way to get it. The usual response to this is “People will tell you things on the phone that they wouldn’t put in writing,” but my feeling is that the vast majority of the things people aren’t willing to put in writing are things they shouldn’t be telling me on the phone, either.)
And look, this process sucks— I know for a fact that I lost out on any chance for a couple of positions I applied for (lo these many years ago) because one of my references sent their letters really late (and then only after some badgering). But the fundamental problem here isn’t the references or any of the other requirements, it’s that 95% of the applicants for any individual position are necessarily going to be rejected before the interview stage.
Any process with those odds is going to suck for the candidate. Doing it with letters is like the Churchill line about democracy: it’s the worst possible system except for any of the others. If you’re just going on CVs and statements, you’re going to end up with even more of the stuff that’s problematic now— “Well, this one’s Ph.D. is from an Ivy League school, where that one went to Flyover State…” and “This one’s published in higher-impact journals than that one…” and “This one’s got seven papers, that one only three…” I would honestly rather pick a shortlist by rolling dice (after weeding out the chemical engineers…) than resort to that sort of thing.
All things considered, my first preference is… well, to not be on the search committee for a new faculty hire at all. Assuming we have to do it, though, I think it’s important to have the references at the start of the process, as all the other tools available are much worse.
So, there you have it; I’m sure this will be non-controversial and regarded as the definitive Take on the question, in which case you’ll want to click this button to get more of these in your inbox:
And this one to send it to all your friends:
Should you somehow find fault with this, though, and want to yell at me about it, I guess you can do that in the comments.
I'm pretty much in the same place. I don't want letters to be about Oh yOu kNOw THat GUy or otherwise be exercises in pedigree comparisons. I don't want letters to be mindless superlatives OR really gross kinds of gatekeeping and backstabbing. But sometimes a letter tells you something there is no other way to know that materially affects how you read a candidate--often for the better. A project's context gets explained in new ways, an intellectual biography gets narrated in ways a candidate can't do for themselves, and so on. It's why we cite people, in some sense--we form intertextual, interreferential, interoperative intellectual communities that help explain why a project exists, what an intellectual is hoping to do. So we have to have something like letters, and yet the ways in which we write letters are pretty dysfunctional.