It's that time of year again, when eager undergraduates start thinking about their futures, including the possibility of graduate school. This inevitably leads to emails of the form "Hi, Professor, could you write recommendations for me for these nine schools? And by the way, they're due Friday. Thanks!"
Happily, Sean Carroll comes to the rescue of those of us in need of a way to put off writing recommendation letters, by offering unsolicited advice on getting into graduate school in physics. The advice he gives is mostly good, and comes from the perspective of someone who has read applications at a top graduate school, and thus has some actual knowledge of what goes on. There are a couple of comments and additions I would make from the perspective of a faculty member at an undergraduate institution.
Sean writes, in Q&A format:
Are GRE scores important? Yes. At least, in the following sense: while bad GRE's won't kill your chances, good GRE's make it much easier to admit you. (We're speaking of the Physics GRE, of course; the general tests are completely irrelevant.) It stands to reason: given two applicants from similar schools with similar grades and interests, there's no reason for a department to choose the student with lower GRE scores. At the same time, you can certainly overcome sub-par GRE's by being outstanding in other areas; this is particularly true for students who want to do experiment.
This is true, and can be a problem for undergraduates from small departments or small schools. The Physics GRE is a very difficult and comprehensive test, and students at small schools are in many ways at a disadvantage when it comes to preparing for it.
For one thing, small schools often can't offer the range of courses that larger places do. My department, for example, doesn't have anyone on staff who can really teach a course on solid-state physics. We get the occasional visitor who can teach it, but most years, we don't cover it at all beyond a week or so in our sophomore-level modern physics class. My undergrad alma mater was even more limited-- they didn't even offer classes on particle or nuclear physics while I was there. They did an excellent job teaching quantum mechanics and optics, but there wasn't anybody who did research in particle or nuclear physics.
Another problem, in a perverse sort of way, is that we teach small classes, even at the intro level. If the largest class you teach is fifteen or twenty students, you can feel free to assign long and complex problems, even on the exams, and emphasize problem-solving over memorization. If you're teaching classes of eighty or a hundred, you're more likely to lean on multiple-choice and short-answer questions, just for reasons of logistics. The GRE is a bubble-sheet multiple-choice test, and to score well, you need to have a large array of formulae memorized, and know how to get answers quickly. You just don't pick that up as easily from the sort of small classes we teach (even though those classes are pedagogically superior in other ways).
Still, the fact is that the Physics GRE does matter for graduate admissions. If you're a student at a small school, you need to take it seriously, and start studying for it well in advance, as it's going to be very different from what you're used to.
How do I get good letters of recommendation?
Go to your faculty advisor's house and disable his cable and Internet access, so he can't do anything other than write your letter. Or:
It's more important to have letters from people who know you well than from people who are well-known themselves. One of the best side benefits of doing research is that you can get your supervisor (who hopefully has interacted with you quite a bit) to write letters for you. It's really hard to write a good letter for a student who you only know because they took one class from you a year or two ago. Over the course of your undergrad career, you should find some way to strike up a personal relationship with one or more faculty members, if only to sit in their office now and then and ask some physics questions. Then they can write a much more personal and effective letter.
This is excellent advice, and I can't stress that enough. I've had to write letters for students I've only had in one or two classes, and it's excruciating. Writing for students who have done research in my lab, on the other hand, is much easier, because I have something to say about them.
We've recently hired a Fellowship and Internship Coordinator at the college, and she says that she tells students who are being groomed for prestigious awards (Watson, Fullbright, Goldwater, Rhodes, etc.) that they need to get to know one of their professors well every term. That way, when it comes time to get supporting letters, they've got people to ask. The advice is just as good for students who aren't necessarily Rhodes Scholarship caliber-- try to make sure that at least a few faculty members know you as something other than "that kid in the hat, who always sat in the back of the room."
Do I need to know exactly what I will specialize in? Not really, although in certain circumstances it can help. Professors like to know that someone is interested in their own area of research, and might push a little harder to accept someone whose interest overlaps with their work; on the other hand, most people understand that you don't know everything after three and a half years of being an undergraduate, and it can take time to choose a specialization.
I would actually put a little more emphasis on this than Sean does, for strategic reasons. But here, I'm really talking to the "B" students in the audience.
If you've got the grades and the test scores to get into one of the very top schools-- Harvard, MIT, Caltech, Berkeley, Chicago (more for theory than experiment), Princeton, a few others-- it doesn't really matter whether you know what you want to do or not, as they're almost guaranteed to have an excellent research group in any subfield you decide to pursue. Below those schools, though, there's a tier of schools that are not as broadly excellent as the very top programs, but that have specific areas in which they are every bit as good as the big-name places. In AMO physics, these would be schools like Colorado, Texas, UConn, Rice, and these days Penn State and Maryland (among others). Those schools don't have the cachet of a Harvard or a Caltech, so they're a little easier to get into, but if you go there to study the right subfield, you'll be working with people who are every bit as good as anyone at the big-name places. And within the field, everybody knows that, so the "lesser" degree will still get you a good post-doc, and carry some weight when it comes time to look for a job.
You can believe me when I say this, because that's how I got where I am today. I ran a B/B+ average in college, and had middle-of-the-road GRE scores (quite literally), so I wouldn't've stood a chance of getting into Harvard or MIT. I knew what I was interested in, though, and despite going to a second-tier graduate school, I ended up in a world-class research group. And things have gone pretty well since then.
Which brings us to:
Is my life over if I don't get into my top grad school? Yes. Well, only if you let it be. The truth is, how you do in grad school and beyond (including how you do on the postdoc and faculty job market) depends much more on you than it does on where you go to school.
What he said.
Sean promises a second installment talking about how to choose a graduate school, so I'll save my comments on those issues for that. Which I guess means I need to go write that letter...