Friday, March 26, 2010

Choices Choices Choices

A reader asks how departments choose which advisers or research groups will get new graduate students. The reader's observation is that factors other than the apparent qualifications of the applicants seem to be involved in some of these decisions.

That's correct, although the allocation system need not be as corrupt as the one described in the e-mail by this reader. In that case, it seems that members of the graduate admissions committee assign themselves students in a preferential way. I would think that other faculty would object to that quite strenuously.

In departments with which I have been associated, decisions about who can accept new students, and how many, involve the following factors:

- quality of the applicants, as evaluated from the applications and/or from interviews;

- how many students an adviser already has;

- how many more students an adviser wants to admit and/or can reasonably take on (explained more below).

- what distribution of students among the various potential advisers will result in a good balance of students in the major subdisciplines.

Regarding the last point, consider that this year a certain research group in my department was described as the primary interest by many applicants, including most of the top applicants in the entire pool. If admissions decisions were based only on merit of the applicants, >50% of those admitted would have been interested in working with <10% of professors in the department. That would not be a good situation for anyone.

I am most familiar with departments that guarantee support for their students. In financially good times, there have been enough TA positions so that even faculty without major grants could advise a student or two and be sure that those students would be supported, at least for their salary, tuition, and benefits. In economic good times, any professor who wants to advise a qualified student can do so.

In times of economic crisis, faculty must demonstrate that they have the financial resources to provide an RA for at least some, if not all, of a student's support. For international students, advisers in some departments have to guarantee 12 months of RA support in the first year. Grad students are therefore preferentially allocated to faculty with resources, and the admissions committee makes decisions, in consultation with these faculty, about which applicants have the best qualifications for admission.

This year, during the admissions process, I had to provide detailed information about my existing funding, pending proposals, and planned near-future proposals. Based on my own estimation of what I could cover in the way of RA support, I proposed a certain number of students to be admitted.

Of course the number admitted does not correspond to the number who accept their offers, so there is a bit of guessing involved in deciding how many to admit.

Admissions committees may also consider factors related to a potential adviser's career stage and/or previous history advising students. For example, some departments want tenure-track faculty to advise graduate students, but the number of students advised by a new professor might be monitored so that it neither zero nor a huge number.

My correspondent described a senior professor who claimed that a certain applicant would be his "last student", although a current student was also supposedly his last student. This made me laugh because I have a colleague who has done this as well. He is either on his 2nd or 3rd "last student". The "last student" argument is compelling for making a case to advise one more student, but eventually an admissions committee should affirm that "last means last" and stop using hypothetical lastness as a factor in admissions decisions.

Furthermore, an adviser with a history of what I will vaguely term "difficult" interactions with graduate students might not be given input into admissions decisions and might not be allocated students.

These allocation decisions are made to provide balance among the disciplines, maximize the amount of RA support that admitted students can be guaranteed, and optimize the chances that a student will have a productive and successful graduate program. For these and other reasons, the decisions have to involve more than academic merit of the applicants. That doesn't mean that unqualified applicants are admitted (although I might not always agree with the some admissions decisions), just that other factors are used to decide among a pool of qualified applicants whose number always exceeds the available admission slots.


EliRabett said...

Students don't pick their advisers? Of course the adviser can decline. If Eli were a student again he would never go to your place

John V said...

Your system is much more top down than the ones I worked within.

We also have a central committee of 6 or so who represent the 30 or 40 who might advise students, and funnel TA and scholarship nominations, but the system in the end is driven by the applicants and advisors.

The primary determinant of which advisor has first shot at an applicant is who the applicant requests, or even more decisive, who has already been contacted.

The secondary determinant is whether one of the contacted advisors has inclination, time and/or energy to take on this particular student.

Few students have not already specified some preferred faculty, and those few generally are choosing us as a back-up, and unlikely to come if they end up having their choice of programs.

Nick said...

I am slightly confused by the "admission" and "allocation" of students to your program as described. Am I reading correctly that the graduate admissions committee has the final say in which professors advise which students? While such a procedure is neither inconceivable nor inherently bad, programs in my own field (chemistry) were much more flexible. After accepting admission to my preferred school and starting classes, I spoke to each of the eight professors in the section of the department to which I had been admitted and came to an agreement with one of them. Is this atypical?

Female Genetics Professor said...

Different fields - and different departments - have different ways of dealing with this issue. In my world, admissions committees have virtually nothing to say about which students are advised by which faculty members. First-year grad students "rotate" through three different labs, based on their interests and the willingness of the faculty to take them on for a few months. Everything gets sorted out at the end of the first year, when students and faculty match themselves up for the long term. It is understood that faculty who commit to taking students will support them from grants, although departments may have to pick up stipends later on if the prof's grant isn't renewed. This process does tend to result in an imbalance, with some labs having many students and others having few or none, but students usually ending up working on projects that excite them.

Kevin said...

Our department is small and our applicant pool is large, so the "admissions committee" consists of the entire faculty.

We don't allocate students to faculty on admissions---they have to do 3 rotations to find a lab that matches. We do make sure that the student has expressed interest in fields we have strength in, but do not attempt to balance the input.
It is rare that one faculty member wants someone that the rest don't (hasn't happened yet).

There is some natural sorting into wet-lab/dry-lab students, but quite often the students surprise us---more often by computational people deciding they want to do wet-lab work than wet-lab people turning out to be good computational researchers.

We've found that students generally want the big-name prof when they apply, but often end up preferring someone else's lab.

One exception is foreign students, who cost a lot more---a faculty member has to be willing to pay their first year (while they do at least 2 rotations with other faculty) for us to admit them. Generally only super-star foreign students are admitted to our program. This is very different from, say, the EE department, where almost all the students are foreign.
The difference is that we have a strong pool of domestic applicants, while for EE the domestic applicants tend to be rather weak.

female Science Professor said...

I didn't mean to give the impression that the admissions committee was quite so powerful, though the committee can make decisions about applicant qualifications and they are the ones who oversee the "balance" issue. For example, one professor couldn't decide to advise 15 new students, however excellent these students might be. The major decision/choice is between adviser and student, though. Applicants express a preference and the adviser can decide whether they want to advise that student or not.

Anonymous said...

Thank God for the lab rotation system in most biomedical sciences! I cannot imagine either: 1) picking a lab as a student without the benefit of more than a short visit, or 2) picking a student without some equivalent time to evaluate the match.

The "allocation" issue is an interesting one even in Departments or universities with a lab rotation system. In an earlier time, we accepted grad students as a Department (about 35 MCDB faculty). We generally were looking for the "best students" without a strong worry about their stated interest area, though that could lead to trouble as in a given year some labs are "in" and others "out". We now have switched to one portal for all 300 biomedical science faculty in ten Departments. We do think about things in very broad interest areas (in part to pacify Departments who feel they are not "getting enough students"). But at the end of the day, its now a free market and a faculty member needs to learn how to sell what she or he does, both when students are choosing rotations and afterward. Attention to webpages and active involvement in recruitment and teaching help at the front end and a happy lab and a good training record help at the back end.

Mark P

Anonymous said...

yeah, i'm in geosciences and kind of jealous of the whole lab-rotation setup... i never even knew that happened in some disciplines until i started reading blogs!

Anonymous said...

When I first started out as a prof, I had a hard time attracting any grad applicants. One of my colleagues suggested that I get myself on the admissions committee so I could get dibs on some prospective grad students. Thank goodness I didn't do it - that committee involves a fuckload of work. But, it is true that occasionally a seemingly well qualified people apply without any idea of who they would work with, or on what questions. And there is one person who seems to have a habit of staking a claim to these prospectives before anyone else even sees the files. Sometimes they turn out to be a good students...but not very often.

CSE prof said...

My department -- and a number of other engineering-ish departments I know -- use a very different model:

1) students are admitted to work with a particular professor, who commits to supporting them for the duration of their studies (technically, we only commit for one year, but the accepted standard is that you either continue supporting a student or stop advising them).

2) TAs are allocated to professors, who decide which students to support. In our extremly nonsocialist system, TA slots are allocated *proportionally to research funds (actually, to RA's supported in the recent past). In other words, you can't have students at all unless you're continually bringing in grant money and supporting RAs.

I don't think the system is optimal (especially part 2). But it is the system for now in my department (which lives on the border of Science and Engineering).

Anonymous said...

Wow, this is a very complicated system. Is it common in the US?

At my university, students are hired directly by professors/advisors, who pay 80% of their salary for five years. The university guarantees the other 20% for any student that an advisor is willing to support. So yes, one person could advise 15 (or more) PhD students if they have a lot of funds. It really never happens that someone has no students at all (unless they don't want any) - if someone doesn't have enough grant money for even one PhD student they aren't likely to be a successful scientist anyway.

I know of American colleagues in my field who advise 20+ PhD students, while others at their universities may have very few, so apparently this is not unheard of over there either.

Interesting anyway, to hear how others do it.

Thinkerbell said...

In spite of now having had some good experiences with rotation students, I still prefer the (European?) system where a PhD student is hired directly by a lab, like any other employee hired for job X, in this case the job of doing the PhD research. The student knows what the project is, interviews with the lab and off you go. That said, after finishing their MSc, most students (in my home country) have done multiple lab rotations (6+ months each), so they have a pretty good idea of what they want and often end up being hired by the lab they rotated in. I find the downside of the US rotation system that it is really seen as feeling if there's a good fit between the student and the lab. That's very useful, but from a rotation-period-output point of view, it's not always worth the investment on the hand of the supervising PhD student or postdoc. Some first year students are also so swamped with classes that they are barely able to spend any time in the lab to begin with.
That being said, the rotation system at least leaves the student in control: I think it should always be up to the student to ultimately decide where they want to spend the next years of their lives in a longer than average workweek. The advisor has to be willing to accept the student, but it's your career, so you pick the PhD spot!

female Science Professor said...

Students in my field don't have to work with a particular adviser if they don't want to; no one is ever forced to work with any one adviser.

Anonymous said...

We admit students with guaranteed funding for three (of five+) years. They are required to find a supervisor during that time and the supervisor must guarantee a portion of their funding for the remaining years. No one wants to supervise you? Well, then it gets complicated ...

The original admissions are made primarily by committee but any faculty member can say "Admit that one, I'll pay it all!" and they are in.

There can be some competition for students between profs once they are here but that seems quite reasonable.

My previous department was much more rancourous -- advisors emailing potential students subtly and indirectly slagging off colleagues, stealing them out from each other with money, promises, etc ... bah!

female Science Professor said...

I have found it difficult to co-advise students based in engineering departments, in large part owing to differences in how/when students select advisers and also differences in grad funding systems (e.g., an engineering student might TA a class even while being paid an RA from my grant). Owing to these differences, I decided not to co-advise any more engineering-based students, but I would be happy to co-advise a student with an engineering colleague if the student is admitted through my department.

Anonymous said...

See...this is where mathematics is soo much better. The success of faculty members and the success of their grad students hardly have any correlation.

Kevin said...


Is it legal for a student to get both an RA and a TA position simultaneously at your institution? It certainly isn't here!

Is the problem with engineering students, or with one who cheated on the system, figuring he wouldn't get caught because the science and engineering bureaucrats don't talk to each other?

Anonymous said...

Fascinating. At my institution, I pick them, I pay for them, and the admissions committee rubber stamps them, unless they fail some numerical criteria for admission, which can be accommodated with a probationary period. We don't have a lot of TA support in my department.

Anonymous said...


At my institution, there is actually a shortage of students to TA because there's a lot of money in biological sciences so ~85% of grad students are paid RA's. To fulfill this shortage, students actually TA for "free" as part of a requirement to graduate. Their salary is paid by either their adviser directly (RA) or from a pool all the PI's pay into if the student is in their first year and doesn't have a permanent adviser yet. So regardless the student makes the same amount even if they are technically both doing an RA and a TA. Maybe something similar is occurring at the engineering dept at FSP's school.

Kevin said...

"To fulfill this shortage, students actually TA for "free" as part of a requirement to graduate."

Can't do that here. TAs are unionized---can't work them for free. GSR salaries come from grants, and work done needs to be at least plausibly related to the grant. Some grad students have chosen to work with their adviser on a course without being TAs, but we certainly could not require it.

female Science Professor said...

I don't know if it was a temporary policy or standard operating procedure in that engineering department, but it seems to me that the RA-pay for a TA scenario would only work if the adviser had large amounts of guaranteed funding (perhaps from various sources) for the duration of a student's graduate program.