Hasan Bakhshi, of the U.K.'s National Endowment for Science Technology and the Arts (NESTA), posts about randomized control trials for assessing the outcomes of innovation policy. I have long struggled with concerns over selection bias in evaluations of policy and education programs. This is a very important discussion. It is an especially salient thought when considering evaluation of entrepreneurship training programs or policies designed to increase the number of/success of entrepreneurs, as there are huge self-selection issues associated with entrepreneurial intent.
The Fairlie et al. GATE study comes to mind as an example of a recent randomized control trial on entrepreneurship training programs. Do you know of any others? Let us know in the comments.