Reply to: Should we require balance t-tests of baseline observables in randomized experiments?
Sorry to revive such an old post, but there's an additional dimension here. When doing experiments with very large samples (think program administrative data) and with many treatment groups even very slight differences between groups will be statistically significant...
Reply to: What’s the latest in development economics research? A round-up of 140+ papers from NEUDC 2017
Skipping the macro papers speaks volumes about the current dismal status of development economics
Reply to: Should we worry about home bias in development research?
I agree with Alexis when he says that working on your own country has a comparative advantage in terms of knowledge and experiences. But it is also important to be aware of the development taking place in other developed and developing countries so that the pitfalls faced by there countries could be avoided and achieve research objectives at much lesser time
Reply to: The Latest Evidence on Gender and Development
These are really an interesting summaries but not observed that child marriage, internal and cross country migrations are not considered as major challenges in in Africa,
Reply to: Spatial Jumps
Thanks! Berk discusses this in a past post on this blog on RDD more generally (including on manipulation) here. As that post points out, common solutions include density checks and to have a baseline before the policy is implemented in order to test for balance before selection can plausibly occur (particularly for balance on variables indicative of manipulation, e.g. transaction history in agriculture).
However, key to evaluating the impacts of such infrastructure developments on geography rather than individuals (plot vs household outcomes) is documenting the selection of individuals across plot types. This is likely an important mechanism through which the policy has impacts, as clearly land itself is not moving!
Florence & John