Saturday, May 22, 2010

Search committee math, or what does it mean to have 200 applicants for a position?

Hiring has been on my mind, inspired by the wide ranging discussion of spousal hires. Although I am an academic neophyte, seeing three searches in two departments this year (I was on two of the committees as well) has really opened my eyes as to how searches work. Here's how we did it this year:

210 applications. First of all, for all three searches, we had 200+ applications. Initial evaluation and grading is done by each member of the committee independently. Then, we come together to make a "long list" and then after more consideration, a "short list". We can throw away at least 1/3 of them right away due to lack of qualifications (no PhD, no postdoc*, wrong field, wrong sub-field*, or application incomplete). I should note that we do keep applications in the no postdoc or wrong sub-field if they are SUPER AMAZING, but that is not most of them. When we write the ad, we have some type of expertise in mind that we are looking for to fill out the department, move into a new area, or cover a needed area of teaching. Sometimes we will hand off applications to another (more suitable) department that has an opening. As for postdocs, we generally require at least one. In the past (before my time), the department has hired someone with a deferred start so they can complete a postdoc first, so it isn't a total waste of time to apply without one, if you have something good lined up. This is very, very rare though. A surprising number of applications are not complete. The most common missing thing is one or more reference letters (we ask for three). We will often over look one missing letter (we assume the writer flaked out), and contact the listed writer if we need more information. If more than one letter is missing, we assume it is the candidate.

140 applications. We score all the remaining applications in A, B, or C. Applications scored C are removed from consideration. Applications scored B are (briefly) reviewed a second time in case we missed something, especially if the decision was not unanimous. Usually, the A's can pretty clearly be separated from the B's and C's. This cut is made mostly on CV and letters of reference. We look primarily at publication record--we like to see first author papers from both PhD and postdoc. We also like to see some signs of collaborations (like non-first author papers). We look at where the candidate is publishing (Nature/Science papers are not required, but 17 publications in Joe's Journal of Microscopy won't cut it). In the searches I saw, I'd say around half of the remainder were C's.

70 applications. Now it gets tougher. We need to separate the A's from the B's. At this point, we consider the proposed research plan, as well as previous track record. We are looking for a colleague who can 1) be successful in our department and 2) bring something to improve the department. If the research plan is completely unrealistic ("I'll cure cancer in 5 years!" or "I need the entire contents of my massively successful advisor's lab to get started") or incomprehensible, the applicant is a B, no matter how great their CV is. If the applicant can't write a decent research plan, we have grave doubts they will be able to write a decent grant proposal. When I am scoring the applicants, I look a little at citations, especially for publications more than a year old, but only with a super-quick citation search (definitely not comprehensive in any way). This is just to get an idea of how relevant the prior research is, particularly for sub-fields I don't know that well. We don't generally care too much about where the applicant did their PhD/postdoc as long as the productivity is good. It helps to have at least one recommender who is known someone on the committee by reputation, if nothing else. Admittedly this is harder if someone is coming from a less well-known university, but that is where conference networking comes in. We don't consider teaching experience at all at this point. If there are any weirdnesses (no letter from advisor, unusually long postdoc, gap in CV, etc) we expect them to be addressed in the cover letter or by one of the reference letters. The ratio varies, but let's say the applications are split 50-50 between A's and B's (not too unusual these days with so many people chasing so few jobs).

35 applications. These are the people we are considering for invitations to interview (the "long list"). If any are missing letters, we contact the listed recommenders. At this point, we are considering everything we asked for in the application, but most important is research experience and research fit with the department. On the committees I was on, at the first meeting(s), we went down the whole list, and graded every candidate as A,B,C, or F. At the next meeting, we considered if we should promote any of the B's and worked on a rank order for the A's. We now make the "short list" after much debate. At this point, most/all of the candidates are well qualified, and the distinctions are based on research fit, and how much people like the research proposals. This is where the luck part comes in. We'd probably have a great pool if we invited any of these 35 candidates, but we can only invite 5. So being in the right place at the right time is important. If we already have a great person who studies widgets, we are not going to interview another person who does really similar research, no matter how great they are when we have so many other great choices available.

At this discussion, we will also think about whether the number of women (as determined by name, generally--we don't see the EE cards) on the list is representative of the pool to make sure we didn't accidentally overlook anyone. If we could identify other underrepresented minorities, we would do the same, but that is pretty much impossible from the applications. This is NOT AA. We are doing this to make sure we didn't overlook a file due to unconscious bias (as has been shown by many, many studies and anecdotes). In both searches where I was on the committee, this was not a problem.

We rank all of the A's and decide who to invite to campus. This was not too contentious, which I found pretty surprising. In both cases, we converged on a list of 10 or so, and then spend the most time whittling that down to 5. We generally picked the "best" person (on paper) doing a particular technique or studying a particular subject, and in general, only one person from a given research group is invited out. This does help prune a little bit.

5 applications. These are the people we bring to campus. At least one of them will bomb. We hope that one or more will dazzle us and make the choice difficult. The ranks start out fresh, and we let the candidates place themselves. I'll talk about interviews in a future post or two. We do not notify anyone else of their status, in case we need to revisit the pool.

17 comments:

  1. Thanks for the post! Very true.

    In my department (fairly large) there is the additional initial step of which subarea priority to hire a person. There's some powerful intradepartmental politics at that step, and when the decision to make an offer is finally made. Candidates are usually not aware how much this aspect shapes the outcome of the search.

    ReplyDelete
  2. I like the idea of doing the subarea priority FIRST, rather than last.

    I recently interviewed for a position where they went with someone who works on something that is almost the opposite of what I do. I didn't feel like that was a fair comparison and I kind of wish they had spared me from getting my hopes up when I apparently didn't really have a chance. I found out later from another source that the department has some internal political stuff going on, but I wish they hadn't wasted my time with it.

    ReplyDelete
  3. Yeah, that happened to me too, and it totally sucks. Wastes everyone's time.

    In our department, we prefer to argue about the department's needs in the abstract (rather than as embodied by candidates), so that is why we like to pick the target sub-field first.

    When I was the outside committee member for another department (which is a nice idea, I might add--people are on better behavior in front of outsiders), they places a very general ad, and looked for the "best" person. This worked OK because they definitely favored people who would bring new techniques/subjects to the department. It did make evaluating the candidate's merits somewhat harder, because different sub-fields have different publication standards.

    ReplyDelete
  4. This is so useful, PA, thanks for posting! Interesting to know you look for 2nd-author papers--I hadn't really thought about that!

    ReplyDelete
  5. A step you apparently don't do, but that we do routinely, is phone interviews. We do this at about the equivalent of the 35 applicant stage you describe here; that is, after going through CVs and the like.

    Phone interviews are often highly revealing, and can give a very good sense of things that you can't get from reading someone's CV.

    ReplyDelete
  6. If we could identify other underrepresented minorities, we would do the same, but that is pretty much impossible from the applications. This is NOT AA. We are doing this to make sure we didn't overlook a file due to unconscious bias


    And what, pray tell, do you think Affirmative Action is if not this?

    ReplyDelete
  7. DM, I suppose you are right. I was thinking about more direct AA--I know a couple of places that will provide extra money to bring out a 6th or 7th candidate to improve the diversity of the interview pool. What we actually did was do a quick estimate of the number of women in the pool, and found that our long list and short list both reflected that fraction pretty well.

    ReplyDelete
  8. I still find all nepotism, including this repugnant spousal hiring practice, to be repulsive and extremely unethical. You said nothing about the other candidates with much higher qualifications than the trailing spouse. The limiting reagent in life is opportunity.

    ReplyDelete
  9. Dr. ATD (and this is a serious question)--after someone is well over the minimum qualified bar to be a TT professor what does it mean to have higher qualifications? Once someone is qualified, they are qualified, and it is very difficult to determine who is "higher qualified" without context. Is it more publications (but what if their advisor favors least publishable units, while another advisor prefers long meaty papers?) Is it more citations (also advisor-linked)? Is it a better research fit?

    If departments use research fit as a guide (and they do), how is this different from either hiring someone AND their spouse to get the benefit of both? Or a different department taking a chance on someone who they don't pay full freight for and get a new line to accommodate? No department that I know of will accept someone unqualified for the TT as a TT spousal hire--they might offer a postdoc or research associate position, but no way a TT position UNLESS the spouse is qualified.

    I can say that I personally know a couple that chose a lower ranked university because they got a spousal hire there. That university got a great deal at minimal cost to themselves, since they attracted a much bigger fish than normal and got the well-qualified spouse as well. It will also be that much easier to retain them.

    At my University I know three people who are bi-coastal (they are on one coast, spouse is on the other). Although they are awesome scientists, great colleagues, and excellent teachers, my University does not receive the full benefit of employing them, because they are always flying back and forth to their families (or hosting their families locally). They try to get 3 or 4 day weekends as much as possible which they do not spend here at the University. They pack their service obligations into mid-week. They are hardly ever on campus during the summer. Is it worth it to tell people this is what it takes to be on the TT? I would say no.

    ReplyDelete
  10. I am sorry, but looking at a name to determine someone's gender (or ethnicity) is bias and AA! How do you know that Sam Thomas is not short for Samantha or Sequita? And what about tall people--do you have enough tall people? What about people with red hair or those with pimples?

    Under no circumstances would I want to be selected for a position based on the fact that I am a woman. In fact, I would refuse a job offered to me under those conditions because it is an insult.

    ReplyDelete
  11. Well said, Drug Monkey.

    ReplyDelete
  12. K. Tyson--where did I say that we selected people based on gender or ethnicity? Given all the data out there about problems in judging qualifications and gender/ethnicity, we do a quick check to make sure we didn't accidentally overlook someone because their name triggered an unconscious bias.

    It is a fact that the percentage of women and minorities in professional orchestras dramatically increased when they went to blind auditions, while the number of female conductors is stagnant (and blind auditions are impossible). We are in the same boat--nominally selecting by merit, but possibly blinded by unintentional bias, so we try to make sure we aren't as best as we can.

    True AA would try to compensate for lack of opportunity for women and underrepresented minorities, but we don't do that formally.

    ReplyDelete
  13. How does citizenship factor in? The fundamental basis for visa applications (and Congressional legislation) is that there is a shortage (that we know to be a myth) of skilled workers (researchers, faculty, engineers, etc.). Your blog has nicely detailed that there is no shortage, but instead a very large pool of highly qualified candidates.

    ReplyDelete
  14. We don't care about citizenship, as long as the person is eligible to get a visa to work. We want the best person we can find, and if that person is not a US citizen, we support their visa application. Prodigal U does NOT pay immigration fees, but does pay moving costs (even for overseas moves). This is highly university dependent (some pay both, some pay neither).

    ReplyDelete
  15. Thank you so much for this helpful blog!! I just sent out my first application. It was rather exciting to draft out my first research proposal. Anyhow, I just found out that I didn't make the short list (n=9) but they are keeping me on file in case they need to interview more candidates. In your experience, how often do committees look beyond their first short list?

    ReplyDelete
  16. At my school there is one other big factor that rises above all others in the hiring process - research grant funding. It is a de facto requirement that there be both 1) a record of past research funding and 2) the applicant will be bringing an active grant with them as a new hire. Yes, this is for beginning TT positions, with people coming directly from postdoc. A top 20 medical school.

    Grant funding is a must-have to get a job in biomedical research. We use that to cull the initial list down to a short list. It generally is true that all the top canidates, as judged by the other criteria of papers, school pedigree, etc, will also have funding. So get that K99, Burroughs, K08, etc, or be automatically lumped in the bottom of the applicant pool.

    ReplyDelete
  17. @Anon 10:17
    In my experience, it is unusual to return to the pool unless something unusual happens (like more than one of the short list candidates declines the interview). You never know, though. All kind of weird things happen in searches.

    @Anon 7:54
    I've heard that things are totally nuts in the life sciences. Scary that portable funding is a requirement for a TT position!

    ReplyDelete