Key points

•  The evaluation of an intervention requires a framework within which the evaluation can be designed, data analysed and results interpreted. This framework will generally be based on the intervention's logic model and decisions made about the evaluation objectives.

•  Developing the logic model enables the assumptions, processes, impacts and outcomes (both intended and unintended) of the intervention to be identified and articulated, which in turn helps to identify the evidence required to answer the evaluation questions.

•  Reviewing existing evidence relating to the broad evaluation questions is important for enabling the objectives of any new evaluation research to be identified and refined. Systematic review, rapid evidence assessment and meta-evaluation are approaches to assessing existing evidence.

•  Many evaluations of complex interventions or impact pathways will require a theory-based evaluation framework which seeks to triangulate evidence from multiple sources to test and refine the assumptions made in the logic model. Within this framework the evaluation could draw on evidence gathered through process evaluations and counterfactual impact evaluations as well as using analytical techniques, such as simulation modelling.

•  Simulation models can be used to combine existing and new evidence to answer the evaluation questions, but can be subject to some uncertainty due to the need to make assumptions about how the different pieces of evidence are related.