Abstract The formal engagement of diverse stakeholder groups in policy design and implementation has become a mainstay governance strategy. While much has been learned about collaborative governance arrangements in terms of their structure, processes, and participant dynamics, one particularly salient dynamic has been relatively underexplored: the factors contributing to sustained participation in collaborative venues. Public administration and policy have developed a variety of conceptual theories which draw attention to interpersonal interactions and the ability of stakeholders to garner political wins as important contributing factors to sustained participation. In this article, we test four theoretically rooted hypotheses to investigate stakeholder attendance in the collaborative governance case of an environmental justice council. We analyze council meeting minute data using computational text analysis tools and a Stochastic Actor-Oriented Model to assess: (1) conflict and concord, measured in terms of repeated (dis)agreement with others and (2) interest advancement, or the extent to which a stakeholder makes positive, neutral, and negative comments toward topics which become the council’s annual objectives in relation to an individual’s attendance. Our interpersonal interaction results show that increased positive relationships are associated with increased attendance, and increased negative relationships are associated with decreased attendance, but, for both cases, only when interactions are repeated and not one-off. Our interest advancement results show: (1) an increase in supported annual objectives is associated with decreases in attendance for an individual stakeholder, but (2) increased oppositional and neutral annual topics are associated with a greater likelihood of attendance. This article contributes to existing scholarship on collaborative governance by offering additional insights into individual factors associated with sustained engagement in the collaborative venue. Furthermore, the article offers a generalizable methodological approach for analyzing these dynamics drawing on computational text analysis of meeting minute data.