Key points are not available for this paper at this time.
Randomised control trials (RCTs) analyse what difference a programme makes through comparing those in the programme to a control group who do not receive it. Random assignment to the project and control groups overcomes selection bias which will otherwise occur from programme placement or self-selection. Conducting an RCT requires decisions regarding the unit of assignment, the number of 'treatment arms' and what, if anything, will be provided to the control group and when. A variety of RCT designs are available, including encouragement designs, raised threshold designs, randomising across the pipeline, and factorial designs, which are decided upon according to the intervention and the evaluation question. This article also addresses criticisms of RCTs, which are mostly argued to rest on misunderstandings of the approach.
Howard White (Fri,) studied this question.