12

Is there any good resource to figure out the track record of academic success of research lab’s alumni (e.g. how many of postdocs became a PI)?

I'm a grad student in the US who will finish PhD next spring, and I've just started to find a postdoc position. I realized it's crucial to find an postdoc advisor who has a good track record of alumni's getting job in academia, since I want to be a PI in the future.

It looks like many famous labs with big papers actually have too many postdocs and grad students so that not many of them have successful career. Productivity per person is diluted too much in those labs. I also doubt one's publication record is the only factor for success in getting a PI position as far as I hear from people around me. (Of course, I know that if you have a big paper in a big lab, that is the best way to get a position)

I am open to go to any country for work if the working environment is good. My field is biophysics with some flavor of materials science.

  • 2
    To my knowledge, there isn't a centralized resource which ranks PIs according postdoc success. Many PIs list lab alumni on their sites and what they are doing now. I suspect you'll just have to just go one-by-one and determine whether a given lab's postdocs have been successful based on looking each of them up. – tim peterson Aug 31 '14 at 4:12
  • I'd upvoted this question earlier, since I'm also looking forward to an answer, i.e. would like to know if there is any such database/search site for other branches as well. I mean, this is possible to some extent in High-Energy Physics, using SPIRES/INSPIRE, but obviously that doesn't cover biophysics. I hope this question doesn't go unanswered. :) – 299792458 Aug 31 '14 at 14:49
3

One approach is to simply ask the potential post doc supervisors. That said, if I received a random email from someone asking how many of my post docs have gone on to get academic positions, I would probably ignore it. If however, that email came from someone I had previously talked to about doing a post doc with me, I would be happy to tell them about my past post docs and where they went. You should then talk to former members of the lab and see if the PI's view of the world matches their views. While a single descriptive statistic like percentage of students who went on to get academic positions might seem like it is easier to interpret, it likely only provides a portion of the story.

2

This is not an answer to your question, so feel free to downvote me. But I'm going to answer the question you should have asked rather than the one you did ask. The implicit question here is

Should I use a lab's track record of getting academic positions for its postdocs as a surrogate for what my own chances will be (if I take a postdoc in that lab)?

In other words, the implied assumption is that if 50% of the postdocs from a given lab go on to get faculty positions, and if you go to that lab, then you will have a 50% chance of getting a faculty position. I believe this is poor statistical reasoning:

  1. Averages are not very useful in predicting the outcome of a single realization.
  2. Some postdocs may not desire an academic position. Therefore, their failure to get one has no bearing on what you are trying to assess -- but it will skew the measure you are using.
  3. Labs may not have had enough postdocs to provide a statistically significant sample. For every PI, there was a first postdoc who got an academic position. Unless she was the first postdoc that PI ever had, she went in with 0% odds by your measure.

In deciding where to go, I think talking to the PI and to current and former postdocs will be much more useful to you than computing this statistic.

  • 1
    Perhaps you should ask that question in a new post, then answer it there instead. I, for one, disagree with the assumption that the OP must be using the lab's track record uncritically as a surrogate for his own probability of landing a PI position. – ff524 Sep 1 '14 at 5:45
  • 1
    I'm responding to the OP's statement "I realized it's crucial to find an postdoc advisor who has a good track record of alumni's getting job in academia, since I want to be a PI in the future." – David Ketcheson Sep 1 '14 at 6:02
  • 1
    I don't think that necessarily implies he's only using that statistic, or that he wouldn't realize that statistic is not meaningful in many cases. I think a question asking how one should/shouldn't use such a statistic would be a good question, and a better home for this answer. – ff524 Sep 1 '14 at 6:04
  • Regarding point 3: If that post-doc was the PIs first post-doc ever, the odds would have been undefined. – silvado Sep 1 '14 at 11:17
  • @silvado Good point! I've corrected my answer. – David Ketcheson Sep 1 '14 at 11:57
2

I have a suggestion regarding the assessment (the actual question), not saying if this is a good or bad idea, along the lines of what was done here.

Write some code that:

  1. From a certain paper database, say, pubmed - gets all the papers in which your prospective PI is the last author. This is a pretty good estimate of all the papers from his lab (he might have joint work with other PIs in which he's not last, but let's ignore these).
  2. Gets all first-third authors as prospective graduate students from his lab. We can't really tell (automatically) if they were postdocs, and let's assume that papers in which you weren't on the first three authors don't mean much to your chances of becoming a PI. You also want to limit those papers to be a few years old (so we can assume that they finished their time in said lab).
  3. Look for papers in which those prospective graduate students are last authors. This suggests that they've become PIs.

Now you can get an estimate of the number of PIs out of number of students that published papers for each lab.

There are APIs to pubmed in many programming languages, and the researchers from the first link I posted have their code in github

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.