In case you're interested - I'm halfway through writing a proposal for the Mathematica Summer Fellowship program. I'm proposing doing an evaluation of the National Science Foundation's historically black college or university undergraduate program (HBCU-UP). It's a capacity building grant program to HBCUs to help promote science and engineering education. There's one evaluation of it published, but it's largely qualitative and I think the design could be more rigorous. I'm proposing using some institution level Department of Education data to create a more plausible counter-factual using propensity score matching. Since I have panel data on these schools, I can do a difference-in-differences analysis once I have matching weights from the propensity score analysis.
I think it's a strong design, I have good background in this sort of work - I've implemented very similar research designs evaluating some child welfare reforms and evaluating President Bush's High Growth Job Training Initiative. So I think it's a strong proposal - we'll see.