Using realist evaluation to understand process outcomes in a covid-19-impacted yoga intervention trial: A worked example

Abstract

Realist evaluation offers a valuable way to understand how interventions function and thus how they can be improved and locally adapted. Consequently, realist evaluation is increasingly conducted in parallel with intervention trials. It comprises a clear philosophical foundation and view of causality, pragmatic mixed data collection methods, and a theory-driven approach in which hypothesised program theories are tested and refined. However, detailed methods for data analysis are seldom well-described in realist studies and no clear method for analysing and presenting realist evaluation data has yet emerged. In this methodological paper we use the worked example of our realist process evaluation of the SAGE yoga trial to illustrate an applied process of data analysis and presentation of findings. We show how we drew on other realist studies for ideas, provide examples of six key tasks involved in conducting a realist process evaluation (including coding data and structuring results) and describe strategies that did not work and our rationale for rejecting them. This detailed account of the decisions and methods that worked for us is intended to provide a practical and informed point of departure for researchers conducting a realist evaluation.