Causality, Impartiality and Evidence-Based Policy - Mechanism and Causality in Biology and Economics - page 212

participated exchanged the pills between them (at the cost of halving the dose) or
took them to independent laboratories to verify the active principle. They
completely undermined the trial protocol.
Drug trials in developing countries illustrate how access to experimental
treatments becomes a politically contentious issue within the country (Macklin
2004 ; Petryna 2009 ). We can probably expect the same from RFEs in economics: if
they address interventions about which the potential participants have preferences,
randomisation may elicit a different type of self-selection. Participants may behave
differently depending on their taste for a treatment, over-complying if they want it
to succeed or the opposite if they want to see it fail. Randomisation will only
succeed in breaking any correlation between the participants' preferences and the
trial outcome if these former remain ignorant about the comparative nature of the
experiment. But if they have strong preferences about the treatments, how far can
we go in deceiving them about the comparison?
In order to control for such post-randomisation effects, Duflo et al. ( 2007 )
suggest two additional strategies. The first is continue collecting data after the
experiment is terminated in order to verify whether the interaction with the experi-
menter was making any difference in the behaviour of the participants (e.g. Duflo
and Hanna 2006 ). One way or another, we need participants to remain ignorant
about the controls: they should not know they are still being observed. And we need
to test this ignorance, just as in medical trials with blinding; we just cannot take it
for granted.
To sum up, in RFEs, randomisation may generate a self-selection bias; we can
only avoid with a partial or total masking of the allocation procedure. We have
argued that this is a viable solution only insofar as the trial participants do not have
strong preferences about the trial outcome. If they do, we cannot assume that
blinded randomisation will be a control for their preferences unless we test for its
success. We will only be able to claim that the trial has been impartial regarding the
participants' preferences if we have a positive proof of them being ignorant of the
comparative nature of the experiment. Hence, in RFEs, randomisation is not a
strong warrant of impartiality per se: we need to prove in addition that it has been
masked successfully. 1
6 Can Field Trials Ground an Evidence-Based Policy?
In order to use RCTs as regulatory tools, it is necessary to provide some warrant of
their external validity and impartiality. If we could have perfect causal knowledge
of the effects of an intervention, impartiality would be warranted by default. But if
there is uncertainty about it, RCTs should incorporate some warrants of impartiality.
1 For a further discussion of the possibility of dispensing with randomisation in field experiments,
see Deaton ( 2010 ) and Imbens ( 2010 ).