Are realist randomised controlled trials possible? A reflection on the INCLUSIVE evaluation of a whole-school, bullying-prevention intervention.

We previously proposed that realist randomised controlled trials could be used to evaluate how, for whom and under what conditions complex interventions can be used to activate mechanisms to improve health. While this idea was accepted by some, it was also met with resistance, particularly from some realist evaluators who believe that trials are inextricably positivist and dependent on constant conjunctions to understand causation, and that realist trials are unfeasible because participants and contexts will be insufficiently diverse to enable the testing of context-mechanism-outcome configurations. In this paper, we reflect on analyses of qualitative and quantitative data from the Initiating Change Locally in Bullying and Aggression through the School Environment (INCLSUIVE) trial, and whether these are useful and aligned with realism. We summarise the concerns expressed by realists and reflect on the philosophical and practical challenges that we encountered and whether or not they are related to the trial's design. Finally, we reflect on the trial's weaknesses and highlight areas that future researchers might consider when running realist trials. We conclude that realist randomised controlled trials are philosophically coherent, practically feasible, and can produce nuanced findings.