"The aim of this paper is to provide guidance on the evaluation of complex interventions in international development. Our contribution to the literature is threefold. First, unlike other reviews on the same subject, our focus is exclusively directed to evaluations of development interventions. Second, we identify methods that are new or little used in the evaluation literature. Third, we map methods to project types with the aim of identifying methods that are useful to the evaluation of particular projects.
We start by making a distinction between complex interventions and complex systems. While much recent research has been devoted to understanding how interventions can be implemented and understood in complex systems, our interest rests primarily with the goal of understanding and evaluating complex interventions. We therefore abstract from complexities arising from the interaction between interventions and complex systems. In our framework, complex interventions can be implemented in simple as well as in complex systems, but the latter are not the primary focus of our paper. Removing the consideration of interactions between interventions and complex systems reduces the scope of our review, but helps focus attention on a manageable number of issues and methods.
We define complex interventions as interventions characterised by multiple components, multiple stakeholders, or multiple target populations, or interventions incorporating multiple processes of behavioural change. These interventions cannot be represented by single-intervention or single-outcome models, and present several challenges to evaluation. Occasionally, or purposely, these interventions give rise to emergent outcomes such as non-linear effects, tipping points, and multiple equilibria. The presence of emergent outcomes of this type adds additional difficulties to evaluation because existing methods are ill-equipped to detect and estimate non-linear impacts of interventions." (Inroduction)