When I was an assistant professor, I was very anxious about every new hire, because so much rode on every group member—money was scarce and the pressure to produce NOW was immense.

I did have to let go several students in the first few years, of which in hindsight only one might have been a mistake. Everyone else was well justified and I would do it again.

I don’t think there is a foolproof way to determine if someone will work out or not. There are ways to get better at it, but I think the first step is making peace with the fact that sometimes things won’t work out. Either the student won’t be what you thought they would be, or you won’t be what they expected, or grad school itself wouldn’t be what they wanted. So it’s a good thing to try and internalize that hiring people into the group is an imperfect process and will sometimes fail without anyone being a monster.

When it comes to hiring international students sight unseen, if I have anyone in the group from that country and that university, I definitely solicit their input on the application. Often they will be able to interpret the grades and even sometimes the comments in the letters of recommendation and put the university rankings in appropriate context. For instance, there are schools where a grade of 14 or 15 out of 20 is a top grade in the class, and a student with 17-point-something GPA is considered absolutely awesome and the top of their class,  whereas the equivalent 4.0-scale GPA would be 3.4 and I wouldn’t consider it impressive from a US institution. I also solicit opinion of my colleagues from the same country if I don’t have students from there.

Also, I hand-pick students and have so far always selected from among those who have contacted me to express interest. I never go rummaging through the application pile to see who’s available; I always find suitable candidates via email first, then go look at their applications. This has served me well. That initial contact email conveys a lot of information about interest. Usually, the people who find me because they understand what I do and know that’s what they want end up working really well. So if there’s one bit of advice I would give, it would be to wait to be approached and then select from among those whose interest rings true.

There are also three other things that really helped as I progressed through the career.

a) What is the worst thing that could happen if you got to work with a bad student? You would have to let them go. Figure out what the procedure is, if there is any, to let students go. Your university should have something like a Dean of Students who should be able to help with this. Also, senior colleagues help. I found out that it is expected that the student would be told what they are failing at, in writing, given directions for improvement and a deadline, in writing, and do that perhaps twice, at which point you are OK to terminate.

Understanding how I am to act and that there is a way to be both humane and practical definitely took a lot of the pressure off, and helped me realize that it’s OK to give a few chances and sometimes people do shape up. But sometimes they don’t, and then your conscience is clear.

b) I have also become much better at advising, in that there is a much broader range of workable situations than I would have initially envisioned. With appropriate selection of topic, frequency of oversight, and generally time and mentoring invested, the range of student abilities, interests, and backgrounds that would yield a productive student who progresses well over time has expanded dramatically. So I am now not really afraid of bad apples because I know that most of them aren’t really bad (although a few are); maybe they won’t be eaten raw for lunch but could become caramel apples, or apple sauce, or apple cider…

c) I have also become better at having difficult conversations. Most difficult conversations are really of the kind where I lay out what I think I see, I say why that doesn’t bode well for a future in the group, and ask the student to tell me what they think about it. Usually I talk along the lines that if they are not happy and productive in my group, they should think about continuing elsewhere, no hard feelings. Often the student will take a few days to think about it and then we talk again.

I don’t think anyone should be miserable in my group, and that includes me. At some point, if either of us dreads meeting, we should definitely talk. I am much better now at initiating these conversations ASAP.

One interesting observation is that every student I have ever had to let go has been negligent when it came to attending group meetings. (We schedule group meetings anew every semester, after everyone’s got their schedules, so the time slot is something everyone can nominally make.) In my experience, blowing off group meetings is highly correlated with not working out in the long run.

I realize that this whole post, so far, has not actually been about predicting if someone will be productive. So here goes:

1. Look at the GPA, test results, letters of recommendation. Put those into culture-specific context (experience helps here, too). I personally look for people with high grades in math and physics and a high quantitative portion on the GRE (and a high score on the GRE subject, if applicable).

2. Prioritize consideration of students who took the time to do research on your group, understand what you do, and have contacted you with what what appears to be well-founded, genuine interest.

3. Try to get a feeling for what the person is like through a 15-to-30-min phone or Skype interview (to the extent possible). What you will likely learn is how good their English is, and only if it’s decent and not a barrier to the conversation will you be able to learn other things, such as if the person is energetic or not, naive or not, interested or not. I always ask if they have questions for me. They need to have questions. Not having questions means a lack of interest. But there are also cultural barriers here to which you have to be sensitive.

Basically, do all the expected things. If a person checks out and there are no red flags, boldly offer to bring them in.

4. Be aware that things still might not work out. Be prepared for it, and be confident that you will be able to deal with it if it arises. Arm yourself with the knowledge of the procedure, the belief that your gut is not wrong, and that if your gut says things are not working out, then you should have one or more difficult conversations. The more you wait, the worse things get. Do not wait for the problem to go away—it will not. If you think there’s a problem, then there is a problem, get on it ASAP. Try to fix it or part ways, but sooner is always better than later. With time, you will get better at fixing things, but parting ways is not the end of the world if done with the understanding that the student is usually not more thrilled about the match than you are.

5. Sometimes, to ensure a humane exit for the student, you might waste some money (e.g., fund student till end of semester or until they start with a new advisor). I have done that a few times. I consider that the cost of doing business. It comes out a wash, or, for me, I think I am in the black, as I have taken in and successfully brought to graduation more students who were first with someone else than there are those who were first with me and then went elsewhere.

For assistant professors out there, I wish I could give you a magic formula that would ensure you don’t have any unproductive students who increase your already high stress level, waste your startup funds, and leave you with nothing to show for it all. But there isn’t one. If you are at a top place, the chance of a random admitted student being great is higher than at a non-elite R1, but there are excellent students everywhere. Some need a little boost to become confident, and can get fired up and work out really well despite a wobbly start.

To me, one of the biggest lessons in teaching is that you get the students you get, and you need to find a way to teach those students, as opposed to some idealized super-motivated students who may or may not be a fantasy clone of yourself. The same holds for advising—with experience, we get better working productively with those we do hire, however they are, so we no longer have to strive to fulfill the pipe dream of flawless, prescient hiring.