Observing Leadership: Using Non-participant Observation to Support Programme Theory Development in Realist Evaluation

Non-participant observation is a recognised method in qualitative research but remains underutilised in realist evaluation, with most published examples focused on frontline service delivery settings. This paper offers a methodological reflection on the specific contribution of non-participant observation to programme theory development within a realist evaluation of leadership and governance in an Integrated Care System (ICS) in England. The article argues that governance settings, where leadership is enacted through discourse, negotiation, and formal decision-making, present distinctive methodological challenges that make observation particularly valuable.

Drawing on eleven observations of governance meetings, conducted alongside interviews and documentary analysis, this paper demonstrates how non-participant observation contributed to the iterative development and refinement of context-mechanism-outcome (CMO) configurations. Observation provided access to interactional and relational processes that participant accounts did not fully capture, including non-verbal cues, patterns of participation, and divergences between espoused and enacted practice. Where interview accounts described collaborative decision-making, observations of the same forums sometimes revealed more hierarchical dynamics, with these divergences becoming productive starting points for retroductive reasoning. Fieldnotes also informed the design of subsequent interviews, enabling specific observed incidents to be used as prompts to explore participant reasoning, consistent with realist interviewing principles.

Limitations of observation in governance settings, including restricted access to informal interactions, interpretive uncertainty, and the constraints imposed by the formality of meeting structures, are presented. The paper concludes that non-participant observation can function as an active component of early theory development in realist designs involving complex organisational settings, and calls for greater methodological transparency in how observation is used and reported within realist evaluation.