Introduction
Realist evaluation, a methodology for understanding complex health interventions, is a form of theory-driven evaluation which acknowledges that interventions and their outcomes are subject to contextual influences. As such, a realist evaluator’s duty is to understand ‘how, why, for whom and under which conditions’ interventions work.1 To do this, realist evaluators identify Context-Mechanism-Outcome Configurations (CMOCs).1 These configurations describe how specific contextual factors (C) work to trigger particular mechanisms (M), and how this combination generates outcomes (O), thus introducing the concept of generative causality (the idea that mechanisms operate in specific contexts to generate outcomes). By exploring these configurations of change, realist evaluations aim to understand how a programme is expected to work within specific contexts and what conditions may hinder successful outcomes,1 2 in order to produce policy relevant findings that can be transferred across settings and contexts.3 4
The process of conducting a realist evaluation follows a cycle, as shown in figure 1. The evaluation will usually start and end with a theory or hypothesis about how the programme works. The study design, data collection, data analysis and data synthesis all contribute to testing and refining that theory. A multimethod evidence base and inclusion of various stakeholder groups as research participants is usually recommended,5 6 though not always necessary. Regardless of the data source, it is important that data collection and analysis work to refine the programme theory or theories by identifying generative causality. Data analysis should be retroductive, which refers to ‘the identification of hidden causal forces that lie behind identified patterns or changes in those patterns’.7 Retroduction includes the researcher’s insights and can use both deductive and inductive reasoning to identify generative causation.8 By applying principles of generative causation and retroduction, ‘engaged realist’ researchers6 can elicit CMOCs and refine contextually relevant programme theories that explain how, why and for whom, interventions work (or do not work).9
The Realist And Meta-narrative Evidence Synthesis: Evolving Standards (RAMESES) II Project has produced important guidance to support researchers in conducting and reporting realist evaluations.10 11 Their development was informed by published realist evaluation studies, the large majority of which occurred in high-income countries, and a Delphi study with 33 participants, all of whom were affiliated with institutions in high-income countries.10 Considerations for the use of this methodology within a wide variety of contexts is an important endeavour moving forward.
While realist evaluation offers an alternative approach to study complex health interventions, they are not without challenges, some of which can be attributed to a lack of procedures and precedent regarding its practice.12 A concern most frequently highlighted by realist researchers is the difficulty of defining ‘mechanisms’, and distinguishing them from ‘context’, both of which are tightly intertwined.3 5 12 13 It is also possible to find multiple combinations of mechanism and context which can bring about a variety of outcomes.14 There is limited guidance on how to code for and identify CMOCs and theories,8 which requires a substantial amount of time,15 researcher reflection and creativity.16 Striking a balance between theory and pragmatism, while adhering to realist ontological underpinnings of generative causation and retroduction, is no easy task and often requires researchers to fall back on their own experience.