In the past decade economists have begun to carry out randomised field trials in social settings to examine effectiveness of developmental projects that ranged from micro-financing to vaccination in a school setting. There is strong recognition among economists that experimental design in a social setting cannot erode such problems as selective contamination and compliance. Estimation methods can be used to correct such problems. Although strongly related to the approaches long practiced in epidemiology, there are significant differences in how economists have corrected such problems as selection and heterogeneity of impact.
A central theme in the economics literature is that agents can choose to participate in an intervention even in the case of randomised assignment. This paper examines illustrative interventions in social settings aimed at improving health where programmes were implemented with a randomised design. The paper is not meant to be systematic in its review, nor does it focus on any particular policy issues; we also do not attempt to provide a critique of the literature on randomised trials. Instead the paper chooses to report on a few key studies to describe how particular econometric techniques can help answer policy relevant questions from randomised trials where there might be problems such as compliance, contamination and presence of varied plausible causal explanations as to why a change might have occurred.
