ABSTRACT
The increased use of randomised control trials in economics, and the evidence from such studies to inform policy decisions, has been premised on positive claims about causal identification and an associated ‘credibility revolution’. Normative considerations are only recognised as entering into this evidence-based policy process at the point of decision-making. This chapter provides two critiques of such claims. From a pluralist perspective, the choice of methodological approach eliminates a large class of policy options from consideration that pertain to structural factors which are not amenable to producing short-term findings through randomised experiments. A second criticism comes from within the neoclassical framework: researchers’ prior beliefs determine the hypotheses they choose to test. The resultant bias in policy advice is demonstrated with reference to education experiments that disregard structural constraints and adopt deficit models of teachers in developing countries.
