This morning’s thought experiment is inspired by a recent conversation with a few colleagues about grad school. These particular colleagues went to a certain snooty uptown Manhattan institution; I went someplace a bit more downtown (but, frankly, still pretty snooty).

Here’s the problem: Whenever I give advice to young professionals on career paths and grad school, my evidence base is limited to my own experience and these sorts of conversations with friends or colleagues. Basically it’s only observational, anecdotal, and potentially idiosyncratic. But today’s aspiring development practitioners demand evidence-based educational decisions.

Therefore, I would like to propose a randomized control trial for development education: take a cohort of incoming graduate students, randomly assign them to a set of MPA programs, and then track their career progress, debt levels, and general life satisfaction in the years following. I wonder if Harvard’s Kennedy School, Columbia’s SIPA, NYU’s Wagner, and Princeton’s Woodrow Wilson School would go for that? I wonder if any students would?

Even more fun: Let’s randomize the randomistas. Take doctoral students who are utilizing RCT methods in their dissertation research, and randomly assign them to post-doc positions or junior faculty posts.

  1. Moctar Aboubacar January 11, 2014 at 1:00 pm

    Or, you know, just start making grads of those programs take surveys at regular intervals. I realize this is a humorous post, you don’t need an RCT to get good information on this for prospective students- you already have people who go to these schools, why assign other people? You’re also not adding any external variable to any group, you’re just seeing how different groups in different situations fare.
    Survey on acceptance, survey on entry, survey after year 1,2 graduation, post graduation years 1-10. I don’t understand what more you’d need.


    1. Well there’s an easy answer to that: selection bias.


  2. But… this experiment would only measure the impact of that education *in a situation where all admission is randomized.* If part of the benefit (or drawback) of certain schools is being around the other folks who got into and chose that school (and didn’t choose others) then you wouldn’t be measuring that at all!


    1. Yeah. There’s a more technical way to put that point, in terms of network/group outcomes, effect size, and external validity.


  3. I think they might very well go for that, if the study paid for the costs of tuition. Otherwise what incentive is there for the participants to participate?


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: