Free Research Idea: NSF GRFP Evaluation

If anyone wants a cheap research idea – though one that may well have already been done – I’ve got one I have no plans on using. I’d be very curious to know about the long-term career trajectories of NSF Graduate Research Fellowship winners vs. honorable mentions. The differences in “ability” or “quality” should be tiny or non-existent – the differences in evaluations are quite small, and so are the differences in institutional mix (or you could control for that somehow). It’d be interesting to see if there is a huge, small, or non-existent effect, and if it varies by discipline. Does this already exist? Anyone interested in trying it? You could examine how much that difference predicts winning other future competitive and prestigious awards, placement inside academia, graduation rate, etc. I bet there would be interesting variation between disciplines. What I don’t know is how big the effect would be in fields I’m interested in – e.g. sociology, economics, anthropology. Any speculation or data?



  1. Max Besbris

     /  October 3, 2010

    Speculation: NSF GRFP recipients go on to receive a higher number of grants and fellowships than honorable mentions. Or maybe a higher dollar amount…

    • Elizabeth

       /  October 3, 2010

      Over what time frame? Presumably for the years the fellows are funded, they are less interested in applying for other grants than are those that received honorable mentions.

      For a field like sociology, where we generally take a long time to finish, this might not matter so much since GRFP winners (especially if they won it in their first year, and especially if they used the funded years early) will still need some sort of funding as dissertators. For economics, which I think is a much faster time to degree, I’m not so sure.

  2. Isaac

     /  October 3, 2010

    I actually have a friend who contacted the NSF to see if it was possible to use this variation. It isn’t easy, because you would have to hand collect the long-term outcomes, since the NSF doesn’t do it.

    • I feel like hand collecting some outcome measures – graduation within x years, placement – should be relatively easy. I know some folks who’ve done this for other reasons, and it’s a pain, but not an epic one, especially if you’re willing to lose people who go into the private sector (plausible for Soc, maybe less so for Econ or some of the hard sciences).

  3. Russ

     /  October 6, 2010

    A good starting point is this NBER working paper ( They even have qualitative data!