Like incentivised laboratory experiments Randomized Control Trials (RCTs) are all the rage in economics. RCTs are commonplace in the health sector starting with Pasteur’s first controlled trials 200 years ago. While application of RCTs to social sciences is relatively recent.
However, by their very nature social sciences involve researching social groups and networks where information is distributed and co-ordinated with relative ease and frequency.
This creates a unique problem for RCTs in social research because it is very difficult to construct experiments that are able to completely seal information within evaluated units. Importantly, the closer the social networks of individuals the more likely there will be information contamination and that individuals in the ‘control’ condition will act on this information.
Individuals in the ‘control’ group on learning about the conditions of the ‘intervention’ group may seek their own alternative solutions or act adversely due to the perception of being excluded. Heckman’s discussion of how social networks in groups make RCTs difficult in social sciences is particularly relevant to policy interventions in the education space. Education by its very nature is a learning space within which social groups actively share information.
‘The Problem With Evidence-Based Policies’
A good review of RCT in economics and policy interventions.
‘James Heckman on Facts, Evidence, and the State of Econometrics’
Heckman raises the problem of trying to do RCTs in a social environment when social networks of groups are closely linked (as in schools & social groups). He discusses this impact on HIV RCT clinical trials and education/work based interventions.
‘Experimental Conversations: Nobel Winner Angus Deaton’
Deaton is in the longitudinal study camp.
For those interested the problem of endogeneity in economic behaviour due to social interactions and beliefs:
‘Striving for balance in economics: towards a theory of the social determination of behaviour’.
Karla Hoff & Joseph Stiglitz