Collective Action in Response to a Health Documentary Screening: Experimental Evidence from Artisanal Gold Mining Communities in Ghana

Policy Context

Artisanal and small-scale gold mining, known locally as galamsey, provides livelihoods in southwestern Ghana but also generates severe environmental damage and long-run health risks. Rivers and farmland are contaminated, and exposure to mining byproducts threatens community well-being. Governments and civil society increasingly rely on information campaigns to mobilize environmental action, yet there is limited causal evidence on whether credible health information translates into collective regulation, or how traditional leaders shape these responses. Collective action is central because formal state enforcement is often weak in these rural settings, and local governance must fill the gap. In communities such as Jema, local leaders and residents adopted a community mining regulation and sustained engagement through regular community meetings, helping the community resist new mining entry over time.

This project examines whether public access to credible information can spur collective action and how leadership incentives influence whether information produces consensus or polarization. The study focuses on mining-affected communities in Ghana’s Western and Western North Regions, where customary chiefs play a central role in governance and information flows, and where mining rents can create conflicts of interest. Here, conflicts of interest refer to cases where chiefs have private financial incentives from mining (for example, payments to allow mining) that may conflict with community environmental and health interests.

The project was developed in partnership with Ghanaian research institutions and civil society organizations, as well as with Erastus Asare Donkor, who won Ghana’s Journalist of the Year award in 2022 for his reporting on galamsey. He generously permitted the use of his documentary as a credible and powerful source of information in this study.

Study Design

The study implemented a clustered randomized controlled trial in 99 mining-affected communities. Household panel surveys were conducted at baseline and endline, alongside community town halls that typically drew 40 to 50 participants.

Communities were randomly assigned to one of two information-delivery approaches: seeding and broadcasting. In seeding communities, the documentary on mining-related health impacts was screened privately for chiefs only. In broadcasting communities, the same documentary was screened publicly for both chiefs and community members. This contrast is theoretically ambiguous and policy-relevant: seeding could leverage trusted leaders to translate complex information, but it may fail if information is filtered or strategically suppressed; broadcasting can ensure direct exposure, but may also trigger resistance or conflict when leadership incentives are misaligned.

Collective action was measured through structured town halls that elicited revealed preferences over the stringency of proposed community mining regulations and recorded real commitments to follow-up engagement. This design follows the local institution of durbars, where communities routinely deliberate, voice preferences, and coordinate collective decisions. Endline surveys, conducted about three weeks later, captured learning, discussions about mining, and beliefs about collective action. Leadership conflicts of interest were measured using community reports on whether chiefs would accept payments to allow mining, enabling comparisons between communities with a conflicted chief and communities with a non-conflicted chief.

Results and Policy Lessons

Learning and Engagement

Broadcasting substantially increased community understanding of long-run health risks and boosted follow-up engagement after town halls, as measured by concrete post-town-hall actions including sign-ups for follow-up community assemblies and reported discussion of galamsey during the endline period. Seeding did not produce comparable improvements in learning or shifts in follow-up engagement, even when non-conflicted chiefs reported actively sharing the information. Direct exposure to the video, rather than seeding through leaders, was the key driver of learning. The policy takeaway is that message complexity should be matched to communication technology: for complex health information, broad direct communication is more reliable than leader-mediated diffusion.

Leadership Incentives and Collective Outcomes

While we observed increased engagement in broadcasting communities regardless of whether the chief had a conflict of interest, whether public information translated into consensus depended on leadership incentives. Under non-conflicted chiefs, broadcasting increased support for stricter community mining regulations. Under conflicted leadership, broadcasting did not generate consensus. Instead, regulatory preferences polarized: community support moved toward both extremes, with more support for either no regulatory change or a full ban on all mining activities.

Accountability Dynamics

Public information also altered local political dynamics. Community members reported a greater willingness to question their chiefs’ actions regarding galamsey. Conflicted chiefs were less likely to attend town halls organized on the same day as the intervention, consistent with avoiding very-short-run accountability pressures. However, in the endline survey, they reported that community pressure had become a more important driver of their mobilization efforts. In these communities, external coordination channels also became more salient, reflected in increased sign-up for an NGO-moderated WhatsApp group when community members perceive chiefs to be not acting.

Overall, the findings show that information about environmental harms can spur collective action, but its effects depend on dissemination channels (seeding vs. broadcasting), leadership incentives (conflicted vs. non-conflicted chiefs), and the dimension of action (local policy preference vs. sustaining engagement).