David Mckenzie’s post on Development Impact: An underappreciated benefit of experiments: convincing politicians when their pet projects don’t work is worth reading.
It agues that impact evaluations can help to stop projects that do not work. However, this raises a question often not answered: can we test everything?
Are there any ‘experiments’ that we should simply avoid? Think tanks can make use of new methods and approaches to strengthen the evidence base in their arguments but they should be careful not to remove all signs of values and principles from them.
In any case, the reason this post caught my attention was that impact evaluations (if we can do them on the cheap) could encourage policymakers and politicians to experiment without having to stick by their ideas (even if they end up being outrageous). If they were able to explain the concept of pilots to the public then they may be able to promote an acceptable culture of innovation and reduce the fear of loosing face that many politicians and bureaucrats have.