Empirical validity of the evaluation of public policies: models of evaluation and quality of evidence

There is a wide range of evaluation methods. On the basis of which criteria should one method be chosen? On what scientific foundations are the results of numerous evaluations based? How can reliability and empirical validity be tested? The relevance of such questions is heightened in new fields of action such as agri-environmental policy. This paper aims to demonstrate that theoretical advances on level and types of evidence (existence, causality, effectiveness) can help resolve these issues. The main evaluation methods are classified into three main categories, according to their main goal (to learn, measure, understand) and linked to the debate on types of evidence. The analysis is illustrated by comparing evaluation methods in the field of agroenvironmental policies and farm advisory services. Attention is drawn to the shortcomings of each method with respect to corroborating facts as well as existing complementarities and trade-offs between methods in terms of empirical validity.