Key points

•  Impact evaluations have special requirements which benefit from being considered during the policy design stage, because of the need to understand what would have occurred in the absence of the policy (generally through examining a comparison group of unaffected individuals or areas).

•  Minor changes to policy design can dramatically improve evaluation options and quality. Conversely, failure to consider the evaluation early enough can limit those options and the reliability of the evidence obtained.

•  When thinking about an impact evaluation technique such as randomised controlled trials and piloting should be considered. Where this is not feasible, alternative ways of implementing the policy, such as phased introduction and allocation by scoring, can strengthen evaluation significantly.

•  These types of adjustments need not introduce delays or complications to policy implementation. However, if policy makers intend to by-pass these considerations due to other factors which are seen as over-riding, they should do so only after a full examination of the implementation options and the pros and cons entailed by each.