Ex Ante Evaluation as a Precondition for Evaluating Impact

Abstract

Since the dawn of evaluation, the challenge of attributing impacts to a particular intervention has been subject to theoretical and practical research. Numerous books, journal articles and grey papers have proposed evaluation designs, discussed the strengths and weaknesses of methods and instruments as well as the prerequisites for reliable assignment of observable changes to a specific measure or activity. White (2013, 2010), Duflo et al. (2005, 2006) and Bamberger et al. (2004), just to name a few, refer in that regard to experimental or quasi-experimental designs as so-called ‘gold standards’,¹ which require (at least) a before and an after treatment (as in intervention) data collection at the target group and a comparison group. Furthermore, at each point in time, data has to be collected from the target group, that is, from those who are directly affected by the intervention, and from a comparison group (individuals who are not affected). The advantage of these designs is that they allow for controlling external confounding factors by measuring not only the development of relevant (societal, political, economic or ecologic) parameters affected by the intervention but also by comparing this development with what happened in the meantime outside its scope (as in so-called double-difference approach).