My view is not that there should be an impact evaluation of each and every project which an aid agency supports. (I also don’t believe that a doctor should conduct a randomised control trial every time she prescribes a medicine. What I want to know is that every medicine that is generally prescribed has been previously proven to be effective and safe by a rigorous study.)
That is Owen Barder talking about different motivations for supporting the results agenda in Aid. The rest of the paragraph contains sensible qualifications about when such evaluations should not be required. It is possibly not the extract that Owen is looking for people to take away from that post, and neither should it be taken as a blanket excuse for not assessing project impact. But nonetheless a useful reminder amidst the chorus of calls to evaluate almost everything, and that includes all those conservation biologists who seem more interested in their fine grain biodiversity indices than big picture statements like: how much coral reef do we have?