The scheme depicted in Figures 3.1 and 3.2 is really quite general, and applies not only to the evaluation of selection criteria, but to the assessment of policy more generally. A fundamental difficulty with effective policy evaluation is that we rarely get to observe what would have happened if the policy had not been put into effect. Policies are not implemented as controlled experiments, but as concerted actions.
Randomised controlled trials have been suggested to the British Government by Ben Goldacre and colleagues.
