Although realist evaluation (RE) requires multiple data collection methods, qualitative interviews are considered most valuable and are most frequently used. The guiding principles of RE may limit the emergence of new Context-Mechanism-Outcome (CMO) configurations by evoking particular underlying mechanisms. This paper proposes a new method for conducting semi-structured interviews in the RE context by drawing on the literature and examining the ability of vignettes to explore perceptions about specific situations. Vignettes are developed based on researchers' knowledge of the setting and program theory and are updated through an iterative process throughout data collection. Interviews focus on situations illustrated in the vignette to capture variations in interviewees' perceptions. This method constrains interviewees to using retroduction to identify the hidden underlying mechanisms that link contextual elements to outcomes based on their experiences. This method allows researchers to focus on CMO configurations without evoking mechanisms, which contributes to the rigor of the method.
Keywords: interview; realist evaluation; vignette.