Impact Evaluation when N=1

Abstract

A common presumption holds that when there is only one unit of observation, such as in the case of a national-level policy or a small scale intervention, causality cannot be established and impact evaluation methods do not apply. Yet many development interventions have single communities or organisations as their target, just as in other cases we are interested in the impact of a programme in a particular community, not for the average community. The experimental observation of single subjects or groups is normally practiced in sciences such as psychology and biology. Social sciences do not enjoy the controlled environments that allow conducting single-case experiments. However, this CDI Practice Paper by Edoardo Masset shows that, under some circumstances, methods to assess impact when there is only one unit of observation are possible, and that we should try to create the right conditions for these ‘experiments’ to take place rather than neglect them.