Did any author change or modify their view about RCTs in some way by participating in this book, given that they would have read one another’s work?
Lant: I personally learned from three of the chapters. One, the chapter on the rhetoric of Poor Eonomics is brilliant and helped me understand a book that is so obviously wrong about how to address poverty was so popular. Two, I learned from the chapter on health (a field that is not a full time research interesst of mine) a great example of distinguishing LATE (local average treatment effects) from universal impact as the chapter points out that many of the most effective and widely prescribed drugs have effects in only about 30 percent of patients. This justifies their prescription as cost effective, but doens’t mean they work on everyone. Again, good example. Third, I learned a lot from Jonathan Morduch and Timothy Ogden of how people in the middle of this debate understand the critics and the evolution of the field.
Editors: This is a good question and it should be put to all authors. From the point of view of the editors, we learned the following lessons: We had already worked on the critique of RCTs, and the book only confirms and supports arguments developed in previous publications.
Sector analyses are very useful because they really show what RCTs have contributed to a particular field of knowledge, and the result is: not much, or even counterproductive (strong crowding out effects in the health field, wrong questions asked in the field of sanitation and microcredit, and wrong conclusions (microcredit).
The book also has the merit of tackling head-on the ethical question, which is largely underestimated, even by critics (Abramowitz & Szafarz’s chapter).
Giving a voice to decision-makers is also extremely useful, and gives very concrete examples of the inability of randomistas to address the real problems of development and the perverse effects this has. Dialogue with “open” users of RCTs is also very useful: the added value of RCTs is not in the field of evaluation, but rather in testing behaviors (Morduch’s chapter).