Upstream: The Quest to Solve Problems Before They Happen

by Dan Heath

No donor would support an intervention that was proven not to work. On the other hand, many funders will support an untested intervention, based on the strength of anecdotal feedback.