Thread by @realscientists

The use of randomized controlled trials (RCTs) to study the impact of specific interventions, has over the last decade become a dominant methodology in development microeconomics

However, some argue that socioeconomic RCTs do not test hypothesis rooted in theory and ignore mechanisms of causality

For example,

"In 2006, approximately 1,300 men and women were tested for HIV. They were then offered financial incentives of random amounts ranging from zero to values worth approximately four month’s wages if they maintained their HIV status for approximately one year..."

"Throughout the year, respondents were asked about their sexual behavior three times, through interviewer-administered sexual diaries. Respondents were then tested for HIV, and financial incentives were awarded based on whether they had maintained their HIV status..."

"After the second round of testing, the incentives program stopped."

Taken from the article 'Conditional Cash Transfers and HIV/AIDSPrevention: Unconditionally Promising?'

After the study provided no significant effects on the cash transfer on reported sexual behavior, the researchers hypothesize that the monetary reward was too far in the future for the participants

And for a reduction in risky sexual behavior, the participants would need compensation in the present

The World Bank and others have looked to medical, particularly pharmaceutical, research as a model and as a means of seeming legitimate

But, the use of RCTs in development explicitly seeks to remove or downplay the importance of social, political, and cultural contexts

And humans are less controllable than bodily functions

The pursuit of causality comes at the expense of generalizability which is crucial to expanding programming into different contexts

Complex socioeconomic interventions combine multiple interacting components, which interact in a way that their sum is greater than the effects of the individual parts

Socioeconomic RCTs differ from medical RCTs because participants in the latter usually do not know how the treatment will affect them, whereas, in the former, interventions often require individuals to understand effects well enough to evaluate benefits

Double-blinding is common in medical RCTs but fairly impossible in socioeconomic RCTs

Complex interventions interact with socioeconomic and environmental conditions, organizational readiness, policy context, and target population

The socioeconomic RCTs can also create a treatment sample that differs from the general population that may skew results