Deville et al. studied the performance of physics professors as they moved from one job to another.

They found that moving to a lesser school lowered the impact of one’s research papers on average a tiny bit (by a fraction of a standard deviation)…

However, moving to a better school had no measurable impact at all, on average. You can go from the worst to the best without improving the impact of your papers. Because the result is true on average, it implies that moving to a better institution sometimes help, sometimes hurt. The authors find the result surprising and appear to be unable to explain it.

However, the true story is that the productivity of scientists is not affected very much by their move… This is consistent with a lot of earlier work showing “it is not where you work, but who you work with” that matters. Moving physically closer to geniuses is pointless:

I find no evidence for peer effects at the local level. Even very high-quality scientists do not affect the productivity of their local peers. (Waldinger, 2012)

We are easily convinced that people who worked in the same building as the best minds must somehow have been improved by the experience. Or that they have access to better books and better managers. But the evidence is clear: brilliance is not a contagious disease you catch by the air ducts.

Source: The Economist.

10 Comments

  1. That’s a really interesting result, and somewhat surprising! I would generally expect better institutions to attract better students, which can make a big difference in doing good work. If that effect isn’t significant, then that suggests that good researchers can attract equally good students at any location.

    Comment by Daniel Lowd — 12/5/2014 @ 16:44

  2. I didn’t read their methods in detail, but it could be that moving to a better university does improve your productivity, it is just that there is an opposite regression-to-the-mean effect.

    In order to move to a better university, a scientist has to do well. Therefore, scientists who just moved to a better university are more likely to do so after having a productive period (relative to themselves), and you would therefore expect their productivity to go down after the move (towards their true mean ability)

    Comment by OZ — 13/5/2014 @ 2:00

  3. @OZ

    This appears to be more or less their argument… but I do not find it convincing.

    Comment by Daniel Lemire — 13/5/2014 @ 8:24

  4. Is “..being a better researcher…” quantified by the volume (i.e. frequency) of publications? (Not rhetorical – I’m actually asking). It seems a very simplistic measure (by itself) for what a researcher does.

    In “the business world” (as opposed to the “the research world”) the highest productivity comes from someone who is happy — plain and simple. If a person is happy with where they are and who they work with, they will have greater output of… whatever. If they are unhappy, productivity suffers (including both quantity and quality).

    For a researcher, I would think the same to hold true. Papers are up when they are happy and down when they are not. So, if I have a researcher who I’m moving from an environment where he’s happy to a state-of-the-art building that has nothing but the best geniuses and Bill Gates is personally funding the facility — but the researcher hates it there, then the measure would show that moving a researcher to a “better” facility is less productive.

    I would think you’d need to control for a researcher’s satisfaction — or choose a different measure for the productivity of a researcher.

    Comment by David Allyn — 13/5/2014 @ 12:14

  5. @David

    I believe that they use the number of citations per article as a quality measure.

    Comment by Daniel Lemire — 13/5/2014 @ 14:54

  6. Could it be that a better university has a higher rate of managerial staff (very lame translation of “encadrement”) so the burden of admin work is lighter, allowing the researcher to devote more time to his research?

    Comment by Djamé — 14/5/2014 @ 13:57

  7. @Djamé

    I am sure larger and richer schools have more “managerial staff” but whether it translates into more research time… I do not know.

    Comment by Daniel Lemire — 14/5/2014 @ 14:24

  8. I had watched an piece of interview to Richard Feynman in which he talk about this problem with students.

    His point was more or less this: “If you take a brilliant guy from a modest class and you put him in a class of brilliant guys, it is very likely he won’t sparkle any more. This is because among the brilliant guys, there is still space for only 3 of them to be super brilliant. The rest will feel sad and will go on feeling even worse than it is.”

    This makes somehow sense for me. To what extent it can be true for researchers too, I don’t know. What do you think?

    Comment by Michele — 15/5/2014 @ 12:55

  9. @Michele

    I do not know.

    However, I would say that if you are stuck in the mediocre class, you should not conclude that you are done for. I started primary school in a class for students having learning disabilities. Today, I have been a federally funded researcher in computer science for almost 15 years.

    I was lucky, however, to work with brilliant people who raised my level.

    Comment by Daniel Lemire — 15/5/2014 @ 13:15

  10. I saw his answer more as a scepticism against all the forms of elitism in academy, rather than some thing else.

    Comment by Michele — 15/5/2014 @ 13:18

Sorry, the comment form is closed at this time.

« Blog's main page

Powered by WordPress