We describe study recruitment and lessons learned for two case studies of frontline communities engaged in RCTs regarding extreme weather. Study 1 (recruitment December 2020 – July 2021) enrolled N=219 frontline decision makers who manage stormwater and local planning for 60 small- and medium-sized communities along the U.S. Gulf Coast. Study 2 (recruitment June 2022 – August 2023) enrolled N=303 frontline community members from low-income, mostly Spanish speaking households in four communities in the San Francisco Bay Area, California. Participants in each study trialed interventions to support decision making – for Study 1, local adaptation planning for storm surge and inland flooding, and for Study 2, individual protective actions for wildfire smoke and extreme heat.
We describe the resources required to meet research goals and support ethical engagement with frontline communities. In particular, we emphasize the human resources – multidisciplinary teams, relationships with communities, and dynamic team building – needed to engage frontline decision makers and community members. Results include comparisons of when multimodal remote recruitment (Study 1) works and when in-person recruitment (both studies) is needed, noting ethical concerns around community health and capacity for each. We also describe approaches to partnerships with boundary organizations (both) and community members via promotoras de salud models (Study 2). Additionally, we describe two modes of developing and iterating on protocols that improve our ability to recruit hard-to-reach populations and maintain relationships with communities: phased recruitment (Study 1) and piloting (Study 2).
Finally, we detail ways of measuring “recruitment failure.” Across both studies, research teams collected empirical data during recruitment, including outcomes from a database of 700+ phone calls (Study 1) and questionnaires from eligible decision makers (Study 1) and community members (Study 2) who did not enroll. Using this data along with geographic, Census, and historic hazard data, we demonstrate who may be “missed” in recruitment, including places where people may have both lower capacity and higher barriers to participate (e.g., recent hazard experience, immigration status). We demonstrate how adding research team capacity to efforts measuring “failure” can improve long-term project success. In particular, we argue this matters for applying study results to policy decisions that matter most for frontline communities.
Identifying effective interventions for supporting adaptation in frontline communities requires implementing studies in context now. However, few social science researchers can reach these communities, namely because they are often not members of them. Research with professional samples is often the status quo due to the reliability of using a set budget to achieve sufficient statistical power. Yet this greatly limits what we can learn about what interventions work where and for whom. Increasing the transparency and replicability of study designs that recruit samples in frontline communities can decrease the hidden costs of this work, providing blueprints for other researchers to follow.
Increasingly, agencies and funders are seeking to make impact in work focused on environmental justice. More attention is needed on the care and resources needed to make this work successful for research outcomes and, more importantly, for the people for whom interventions are intended to generate benefits.

