Introduction
Gold mining worldwide has generated significant environmental and public health concerns, particularly through contamination of water and soil systems (Hilson 2002). Empirical studies across regions consistently document elevated concentrations of hazardous heavy metals associated with gold mining. Research from Kenya reports mercury, lead, and arsenic levels exceeding permissible limits (Ogola et al. 2002), while mining effluents in South Africa have degraded rivers, wetlands, and groundwater systems (Durand 2012). Similarly, extensive gold mining in Ecuador’s Amazon region has resulted in widespread contamination of river systems and soils (Mestanza-Ramón et al. 2022).
These environmental impacts are closely linked to adverse health outcomes among miners and nearby communities. Neurological symptoms consistent with chronic mercury exposure have been documented among small-scale miners in Myanmar (Kyaw et al. 2020), while communities near gold mining areas in Kenya and Ghana have experienced increased respiratory disorders and arsenic levels in blood exceeding World Health Organization (WHO) guideline values (Ogola et al. 2002).
Despite the scale of these risks, conventional environmental monitoring systems have often failed to prevent, mitigate, or effectively communicate mining-related impacts. Monitoring is typically conducted to satisfy Environmental Impact Assessment (EIA) requirements, yet access to data is frequently restricted, limiting transparency and public scrutiny (Condon 2017). Regulatory enforcement is further constrained by limited institutional capacity, weak regulatory frameworks, and financial constraints, as observed in Ghana and Ecuador (Adu-Baffour et al. 2021; Miserendino et al. 2013; Veiga and Fadina 2020). Although public participation is commonly mandated during EIA studies, stakeholder engagement is often absent during post-approval monitoring, where long-term risks emerge (Hasan et al. 2018).
In response, environmental governance has increasingly shifted from state-centered or corporate accountability toward models of shared responsibility involving citizens, civil society, and state institutions (Li 2009). Advances in digital technologies have enabled communities to independently collect environmental and health data, enhancing transparency and supporting regulatory oversight (Condon 2017). Within this context, citizen science (CS) has gained global prominence. In the United States, volunteer monitoring identified mining-related pollution that prompted state investigation (Jalbert and Kinchy 2016), while community-led soil sampling in Peru revealed elevated heavy metal concentrations near a mining complex (Molloy et al. 2020). Prior research suggests that CS can enhance public understanding of science and facilitate collaboration and trust among stakeholders (Kieslinger et al. 2018).
In Thailand, CS initiatives are emerging but remain largely top-down, with professional scientists leading projects and citizens primarily contributing data (Otwong 2022). Existing efforts focus mainly on biodiversity and marine monitoring (Magson et al. 2022), while bottom-up CS initiatives that enable communities to define research questions and leverage evidence for policy engagement remain limited.
Moreover, empirical research on CS in Thailand has largely emphasized project outcomes rather than design processes. This study addresses this gap through a formative, process-based evaluation of a community-based CS program for monitoring the health impacts of gold mining at Gold Mine Z in Thailand. Drawing on the evaluation framework adapted from (Kieslinger et al. 2018), the study examines program design across three analytical dimensions: scientific quality and governance, participant engagement and capacity, and societal impact and communication. By analyzing eight interrelated design activities, the article investigates how CS can be designed to address challenges of trust, legitimacy, and accountability in a contested governance context. In doing so, it conceptualizes CS not only as a mode of data production, but as an embedded accountability infrastructure within politically sensitive environmental governance environments.
Theoretical Approach
Citizen science (CS)
CS is broadly conceptualized as a means of reconnecting science and society by fostering scientific citizenship and enabling public participation in decision-making processes, particularly in environmental and sustainability contexts (Irwin 1995). Methodologically, CS commonly refers to the active involvement of non-professional scientists in scientific research activities, such as data collection and analysis (Vohland et al. 2021).
Over time, the concept of CS has expanded and diversified, shaped by varying objectives, disciplinary traditions, institutional contexts, and impact assessment frameworks (Haklay et al. 2021; Wehn et al. 2021). Phillips et al. (2019) frame CS as a form of informal science education and public engagement, while Dickinson and Bonney (2012) highlight its role as an innovative mechanism for public participation in scientific research. Other scholars define CS primarily as a process through which members of the public contribute to various stages of the scientific endeavor. Hecker et al. (2018), for example, highlight CS as a mode of knowledge production that enhances public participation, facilitates dialogue between professional scientists and non-experts, and promotes reflection on the societal implications of scientific outcomes. Similarly, Otwong (2022) interprets the expansion of CS as evidence of the democratization of science, marking a shift from expert-dominated knowledge production toward more inclusive practices. Robinson et al. (2018) even articulated ten widely cited principles of CS to guide the ethical and effective implementation of CS initiatives.
Scholars have proposed various typologies to classify CS initiatives, reflecting differences in governance arrangements and degrees of public involvement. Bonney et al. (2016) distinguish CS projects based on participant activities, including data collection, data processing, curriculum-based initiatives, and community science projects. Conrad and Hilchey (2011) classify CS according to governance structures, identifying consultative, collaborative, and transformative approaches. In consultative models, government agencies typically define objectives while communities collect data; collaborative models involve shared governance among stakeholders; and transformative models are community-led, with local actors defining problems, designing monitoring activities, and mobilizing evidence to influence policy processes. Haklay (2012) further conceptualizes CS along a continuum of participation intensity, ranging from limited forms of contribution to participatory and extreme CS, the latter involving citizens in collaborative problem definition, data collection, and analysis.
More recent developments in CS have increasingly emphasized community science and community-based monitoring approaches, both of which adopt a bottom-up orientation and require deeper citizen engagement. Community science focuses on collaborative efforts among community members, scientists, and other stakeholders to generate data that inform local decision-making and policy processes (Bonney et al. 2016). Such projects are often initiated by communities seeking scientific support, with local actors determining the roles of experts within the research process (Otwong 2022). Community-based monitoring, sometimes referred to as community-based CS, places even greater emphasis on community empowerment, positioning citizens at the center of all stages of the scientific process, including problem identification, research design, data collection, analysis, and interpretation (Haklay 2012). Frequently described as a “science-by-the-people” model (Otwong 2022), this approach promotes collaboration among communities, government agencies, civil society organizations, businesses, and academic institutions to jointly monitor, evaluate, and address pressing socio-environmental challenges (Conrad and Hilchey 2011).
Citizen science process evaluation framework
This study adopts the CS evaluation framework developed by Kieslinger et al. (2018) to assess the design of a CS program in Thailand. A process-based evaluation approach was selected because the CS initiative examined in this case study has been designed and piloted but has not yet been fully implemented. Process-based evaluation is particularly suited to identifying operational strengths, weaknesses, and areas for improvement during the early stages of program development. The original framework evaluates CS initiatives across three interrelated dimensions: scientific, participant, and socio-ecological and economic (Kieslinger et al. 2018). These dimensions encompass both the internal functioning of CS projects and their broader societal implications.
Table 1 presents the process-based evaluation framework developed and applied in this study. Rather than treating CS as a linear data-collection activity, the framework conceptualizes CS design as the interaction of three interrelated dimensions: (1) scientific quality and governance, (2) participant engagement and capacity, and (3) societal impact and communication. Together, these dimensions foreground the institutional, social, and communicative conditions through which citizen-generated knowledge becomes credible, usable, and impactful.
Table 1
Process evaluation framework for citizen science programs, adapted from Kieslinger et al. (2018).
| CORE CRITERIA | GUIDING FOCUS |
|---|---|
| Dimension 1: Scientific quality and governance | |
| Purpose and knowledge orientation | Are the scientific objectives clear, credible, and appropriate for a CS approach, and do they address socially relevant problems through joint knowledge creation? |
| Data integrity and governance | Are data validation, quality assurance, ethics, and data governance (ownership, access, consent) clearly defined, transparent, and understood by participants? |
| Openness and system integration | Does the project ensure appropriate openness, interoperability, and long-term data stewardship to enable reuse and institutional uptake? |
| Evaluation and adaptative management | Does the project include reflexive evaluation and adaptive management mechanisms to respond to scientific, social, and political risks over time? |
| Collaboration and interdisciplinarity | Does the project mobilize relevant interdisciplinary expertise and partnerships to enhance learning, credibility, and robustness? |
| Dimension 2: Participant engagement and capacity | |
| Inclusivity and role design | Are participation opportunities aligned with the capacities, motivations, and constraints of different participant groups, with diversified roles and pathways? |
| Depth and equity of participation | Are participants able to engage meaningfully across project phases, and are citizens and scientists positioned as mutually respected partners? |
| Capacity building and communication | Are training, facilitation, feedback, and communication practices sufficient to support learning, confidence, and sustained engagement? |
| Dimension 3: Societal impact and communication | |
| Outreach and dialogue | Does the project employ targeted, accessible, and two-way communication strategies to engage affected communities and broader audiences? |
| Amplification and uptake | Does the project connect with media, civil society, and policy-relevant actors to enhance trust, visibility, and the potential for social or policy impact? |
The scientific quality and governance dimension emphasizes that scientific credibility in CS depends not only on methodological rigor but also on transparency, accountability, and interdisciplinary collaboration—the governance conditions under which knowledge is produced and validated. It includes the clarity and societal relevance of objectives, data integrity and governance arrangements, openness and system integration, and mechanisms for evaluation and adaptive management.
The participant engagement and capacity dimension focuses on how participation is structured and supported. It addresses the inclusivity and alignment of participant roles with diverse capacities, the depth and equity of engagement across project phases, and the provision of training, facilitation, and feedback. Emphasis on capacity building highlights learning, empowerment, and mutual respect between scientists and citizen participants, which are especially critical in bottom-up and community-based CS initiatives.
The societal impact and communication dimension examines how CS initiatives extend beyond project boundaries to influence social, policy, and governance arenas. It focuses on outreach, two-way communication with affected communities, and collaboration with media and civil society actors to amplify outcomes. By treating communication as a core design variable, this dimension clarifies how CS can achieve societal relevance and policy traction in politically complex contexts.
Methodology
Case study
Gold Mine Z is a large-scale gold mining operation in Thailand that has been active since the mid-1990s and spans multiple provinces (Tran et al. 2022). Operated by a domestically registered subsidiary of an Australian multinational corporation, the mine comprises an open-pit extraction site and a processing plant, with gold extraction relying on cyanide-based processing (Sutthirat and Changul 2013).
Since operations began, numerous studies have documented environmental contamination and related health risks. Lead (Pb) and manganese (Mn) have been identified as critical contaminants due to their elevated concentrations and leaching potential (Sutthirat and Changul 2013). Significant risks of cyanide contamination from tailings storage facilities have also been reported, with potential exposure affecting residents in 3–23 surrounding villages (Tran et al. 2022). Additional investigations have raised concerns about hazardous substance leakage, with local residents reporting unsafe water sources and symptoms consistent with central nervous system toxicity, including weakness, headaches, and altered taste and smell (Phenrat et al. 2021).
Escalating environmental and health concerns prompted sustained campaigns by media and non-governmental organizations calling for the suspension of mining activities (Laurence 2021). In 2016, the Thai government, acting under Section 44 through the National Council for Peace and Order (NCPO), ordered the closure of Gold Mine Z prior to the expiration of its operating license, citing environmental protection, public health, and national security concerns. The mine operator subsequently initiated arbitration proceedings against the Thai government under the Investor–State Dispute Settlement (ISDS) mechanism of the Thailand–Australia Free Trade Agreement, seeking US$1.27 billion in damages (Khoman et al. 2024). Following negotiations, the dispute was settled, and in 2021 the mine was granted a renewed concession and an extended processing permit, allowing operations to continue until 2027.
Data collection and key informants
This study adopted a qualitative research design. In-depth interviews were conducted with 26 key informants (KIs) between July and October 2025 (Table 2). KIs included CS program designers, members of the program oversight committee, program partners, citizen scientists involved in the pilot phase, representatives from the mining company, and relevant stakeholders. KIs were selected using purposive sampling, supplemented by snowball sampling. Sampling continued until data saturation was reached, as no new themes or substantive insights emerged from additional interviews. Interview data were complemented by a field visit to the mining site in November 2024.
Data analysis
All interviews were audio-recorded, transcribed verbatim, and anonymized. Data were analyzed using thematic analysis guided by pre-defined themes aligned with the CS evaluation framework: (1) scientific quality and governance, (2) participant engagement and capacity, and (3) societal impact and communication (Table 1). The interview guide and detailed codebook are provided in the supplementary materials (Supplemental File 1: Appendix 1). To enhance analytical rigor, two researchers (KK and TU) collaborated to code the data and resolved discrepancies through discussion and consensus.
Researcher positionality
Researcher positionality was explicitly acknowledged given the authors’ involvement as members of the CS program design team. As insider researchers, both authors participated in all stages of program development and evaluation. To mitigate potential bias, the study employed a structured analytical framework, systematic coding procedures, and collaborative analysis, consistent with reflexive qualitative research practices (Holmes 2020).
Results
The results are presented in two sections. The first briefly explains the design of the CS program for gold mining impact monitoring. The second discusses the evaluation findings, structured according to the evaluation framework.
Design of the citizen science program
The pilot CS program was launched in July 2024 and initiated by a civil society organization with extensive experience working closely with communities, particularly in the area of community health impact assessment in Thailand. The program was designed to (1) identify key health and environmental indicators, (2) build community capacity to use monitoring tools, and (3) establish mechanisms for recording, interpreting, and responding to risk trends. Where elevated risks are detected, the program aims to support risk communication and trigger institutional responses. Based on these objectives, the program comprises eight interrelated activities.
Activity 1: establishing program governance
Given prolonged conflict between the mine operator and surrounding communities, the program adopted a governance-oriented design emphasizing plural expertise, institutional oversight, and procedural transparency (Ansell and Gash 2008; Bovens 2007). Rather than a conventional expert-driven model, the program was structured as a collaborative governance arrangement involving interdisciplinary researchers from environmental science, public health, and social sciences, alongside CS specialists familiar with the local context.
To strengthen vertical accountability and institutional linkage, public organizations were formally engaged as program partners, such as the Department of Health within the Ministry of Public Health, the National Health Commission Office (NHCO), the Office of the National Human Rights Commission of Thailand (NHRC), the Mining Industry Council, and the Thai Public Broadcasting Service (ThaiPBS). Their involvement was intended to facilitate institutional uptake of citizen-generated evidence and enhance the program’s accountability function.
An independent oversight committee was also established as a horizontal accountability mechanism. Comprising ten members from government, academia, media, and the private sector, the committee met bi-monthly to review program design and pilot activities and to provide strategic guidance, mediate competing claims, and reinforce procedural legitimacy. During the design and pilot phases, six committee meetings were convened, along with an initial site visit aimed at enhancing members’ understanding of the local context.
Activity 2: analysing pollution sources and developing a risk map
Program designers conducted a systematic analysis of pollution sources and exposure pathways using EIA and EHIA reports, government inspection data, prior research, and satellite imagery. These data were integrated into a GIS-based database on the C-Site platform.1 a public crowdsourcing infrastructure developed by ThaiPBS. Environmental surveillance data—including TSP, PM10, surface water, and groundwater quality—were used to visualize pollution pathways. Two stakeholder workshops were convened to support the analytical process. The first workshop involved technical experts, including representatives from the Department of Groundwater Resources, the Department of Health, and academic institutions, to identify the primary risks associated with arsenic contamination in groundwater. The second workshop engaged community members, local schools, local government agencies, and representatives of the gold mining company to validate the risk map and ensure its relevance to local conditions.
Activity 3: identifying indicators and selecting community monitoring tools
Health-related indicators were identified through document reviews and expert focus group discussions, with particular concern for household water safety and noise and vibration impacts. Monitoring tools for arsenic and cyanide detection were selected based on sensitivity, accuracy, cost, and feasibility. Tool validation followed six steps: selection, procurement, laboratory validation, quality control design, field testing, and documentation through standardized operating procedures. Final indicators included water quality (arsenic, pH, electrical conductivity, total dissolved solids), air quality (PM10 and PM2.5), and noise and vibration measured via a verified mobile application.
Activity 4: establishing a response system
A multi-stakeholder roundtable convened in July 2024 by the NHRC brought together government agencies, mining operators, the Mining Council, and academic experts to clarify roles and response pathways when monitoring results exceed regulatory thresholds. Relevant regulatory frameworks, including the National Minerals Act of 2017, were reviewed to identify potential institutional responses.
Activity 5: developing the citizen science curriculum
Curriculum development began with a review of existing CS curricula related to mining and environmental health. This was followed by consultations with local schools, communities, and the regional Office of the Basic Education Commission (OBEC) to assess their willingness to participate in the pilot program. Subsequently, a focus group discussion was convened with representatives from the regional OBEC and local schools across three provinces in the vicinity of the gold mine. The discussion aimed to collaboratively identify appropriate curriculum content, learning activities, and expected program outcomes.
Activity 6: designing a data recording system
Environmental indicators were integrated into data-entry modules on the C-Site platform. The system includes user manuals, GIS-based risk maps, citizen-generated survey data, and standardized digital forms. Data security, privacy, and quality assurance protocols were established, although the environmental dataset remained in a pilot, non-public phase at the time of evaluation.
Activity 7: piloting
A two-day pilot workshop was conducted in September 2025, involving 15 participants—including Village Health Volunteers and school representatives. It introduced CS principles, monitoring tools, risk maps, and data recording procedures. Participants conducted water testing and geotagging exercises. Feedback from the pilot informed revisions to both the monitoring system and training curriculum. These revisions included simplifying data entry procedures on the C-Site platform, incorporating instructional video clips demonstrating the use of test kits, and adding foundational content on key water quality parameters, such as pH and electrical conductivity (EC), to the curriculum.
Activity 8: evaluating the citizen science design process
A formative evaluation was embedded as a core design activity to assess feasibility, identify weaknesses, and refine governance arrangements prior to full-scale implementation. The findings reported below are derived in part from this evaluative process.
To sum, these eight activities constitute the empirical object of this study: the design architecture of the CS program. The evaluation herein examines how their interaction enables or constrains the program’s overall capacity to produce credible knowledge, sustain meaningful participation, and function as an accountability-oriented CS initiative in a contested mining context.
Evaluation results
The evaluation findings are grounded in the process-based analytical framework presented in Table 1 and are organized across three dimensions: scientific quality and governance, participant engagement and capacity, and societal impact and communication.
Dimension 1: scientific quality and governance
Purpose and knowledge orientation
The program exhibited a clear scientific orientation focused on standardized, community-based monitoring of environmental and health risks. Interviewees noted that prior government monitoring efforts were fragmented and episodic, whereas the CS approach enabled continuous data generation aligned with locally experienced risks.
“The program has a dust particulate matter (PM) measuring device. Every school here faces PM problems, especially during the winter… When the school installed the device, we saw, ‘Wow, this level is hazardous,’ and could warn students to wear masks. This is very beneficial.” (VS#2)
“Sometimes government monitoring is inconsistent due to budget limits. Community testing with kits can help, and any issues found can be reported for official rechecking.” (PP#4)
“The mine already monitors as required by law, but community monitoring is also useful. If an issue is detected, experts can verify the results and work together with the community to identify the cause, fostering collaboration between the mine and local residents.” (M#1)
In addition, local knowledge was integrated into the CS program. First, affected communities, local schools, and local government agencies were involved in validating the risk map. Second, local schools were closely consulted to design a curriculum that is appropriate for teachers and students and applicable in real-world contexts.
Data integrity and governance
Most interviewees expressed confidence in data integrity, citing expert oversight, laboratory validation, and structured protocols. Citizen scientists—particularly health volunteers and science teachers—were perceived as capable data collectors following training:
“They will be trained and mentored by experts, with local health centers providing advice and supervision… Most public health volunteers are selected for their knowledge and good skills.” (PP#3)
However, despite broad support for inclusive community-based monitoring, some interviewees continued to express concerns regarding the accuracy of data collected by non-specialists. While they acknowledged the value of community participation, they questioned whether individuals without formal scientific training possess sufficient knowledge and skills to ensure measurement accuracy. As one respondent commented:
“It’s good that communities collect samples and monitor by themselves. But I am not sure whether communities have sufficient knowledge and skills to ensure accuracy.” (PP#1)
Concerns were also raised regarding sustainability, perceived neutrality, and workload. Some feared that strong anti-mining sentiments could undermine objectivity, while others suggested modest compensation to sustain participation:
“If the funding includes compensation, they will feel independent… but they are also not working completely for free. There are expenses, like travel or internet use, that should be covered.” (M#1)
Although a Data Privacy and Security Plan aligned with the Personal Data Protection Act PDPA was developed, volunteers reported limited understanding of data protection. Moreover, no explicit agreements governed data ownership or sharing, generating tensions between advocates of restricted disclosure and those favoring transparency for accountability or legal action:
“We have to agree beforehand that it cannot be published unless filtered and approved… It should be a tripartite process among the mine, villagers, and academics.” (E#3)
“If there is high contamination, this will be a baseline data that villagers can use in their fight, whether in the form of a lawsuit or whatever.” (PD#5)
Openness and system integration
Citizen-generated data were designed to feed into the C-Site platform; however, the dataset remained non-public during the pilot phase. This somehow reflected tensions between openness, verification, and conflict management in a sensitive political context.
“If a substance exceeds the standard, it should be carefully investigated to understand the cause. If shared with the public too early, it may be misinterpreted due to incomplete data.” (M#1)
“We do not provide public access to raw data, as they are sensitive and may cause alarm or opposition if misinterpreted.” (PP#6)
“If an abnormal result is found, it will be reported to the local health office for verification. Lab confirmation is required before uploading to C-Site, and the data should not be made public at this stage.” (VS#3)
Evaluation and adaptive management
The inclusion of a formal evaluation process as an explicit program activity reflected an adaptive management orientation. Program designers viewed evaluation not as a post- hoc assessment, but as a mechanism for identifying design weaknesses, managing risks, and refining governance arrangements prior to full-scale implementation. This reflexive approach was widely perceived as enhancing procedural credibility, although its effectiveness was seen as contingent on continued institutional support and responsiveness.
Collaboration and interdisciplinary
Interdisciplinary collaboration, formal partnerships with public agencies, and the establishment of an independent oversight committee were widely perceived as strengthening scientific robustness and institutional legitimacy. Interviewees emphasized that these collaborative arrangements signaled accountability and reduced perceptions that the project was driven solely by activist or adversarial motives.
Dimension 2: participant engagement and capacity
Inclusivity and role design
The CS program sought to engage a broad range of stakeholders from the outset. The establishment of an oversight committee reflected an explicit effort to structure participation across institutional and societal boundaries. Interviewees widely regarded the involvement of recognized experts as effective, noting that their credibility facilitated access to difficult-to-engage stakeholders and helped build initial trust. Moreover, several informants emphasized that experts’ personal and professional networks helped frame the program as an academic inquiry rather than an adversarial intervention, thereby encouraging participation from industry actors.
Depth and equity of participation
Health agencies participated as research partners and reported co-learning benefits, with some expressing interest in future CS applications:
“The Department can apply this CS principle in the future to train monitors in this area or other similar areas, where we will see the benefits that we can utilize.” (PP#4)
However, gaps remained. Local government agencies with implementation authority were insufficiently engaged, and mining sector participation largely occurred at an operational rather than decision-making level:
“Local government agencies are still missing… For instance, for arsenic testing in urine, we need to involve the Regional Environmental Office; for testing in the body or food, we need to involve the Medical Science Center…. And integrate the Provincial Agriculture Office as well. This part may still be incomplete. It needs to involve more… just the Public Health Officers is not enough.” (PD#4)
“The program couldn’t reach the mine owners; it only reached the operational level. Therefore, the operational level is not the one that makes decisions. But they are the ones who come in and are informed that the work is being done, which is good—to inform them from the start and invite them to discuss from the start—but it must be in parentheses: they are only at the operational level, not the owners with the power to decide.” (PD#5)
Capacity building and communication
From interviews, training and piloting strengthened basic monitoring skills, yet deep-seated distrust persisted:
“Honestly, I am still suspicious about where this program will end up, because I fear it will be suppressed by other agencies in the end.” (VH#3)
Also, the involvement of a professional media partner enhanced communication capacity but also led to caution and self-censorship among participants:
“…once media is involved…It makes the communication guarded; the community is tended to self-censor, and so do the entrepreneurs. They cannot talk openly about anything.” (PD#3)
Dimension 3: societal impact and communication
Outreach and dialogue
The program incorporated a professional communicator to lead outreach targeting students, schools, and education policymakers. This is beneficial as school volunteers highlighted experiential learning benefits:
“If our students get to learn or practice being a basic scientist, it will be beneficial because it instills scientific thinking from a young age… it will be useful for learning, since most subjects today require active, hands-on approaches that make learning more lasting.” (VS#4)
Amplification and uptake
External partners reported increased understanding of CS and interest in scaling the approach.
“What I’ve learned is the [CS] curriculum. If it succeeds, we can apply it in the future for monitoring training in this or other areas, where we will see the benefits of using this knowledge.” (PP#3)
However, responsibilities for communicating sensitive monitoring results remained unclear. Volunteers were reluctant to assume this role, while others perceived excessive caution by the research team:
“The duty of public communication should not be ours. It should be the responsibility of the relevant agencies.” (VS#2)
“I think the study results are not being fully utilized. The research team might be handling the findings cautiously to avoid negative consequences for business operators or government authorities, both central and local.” (PD#5)
Accordingly, interviewees emphasized the need for clearer dissemination strategies and sustained public visibility to enhance policy attention.
In sum, the evaluation shows that in contested governance contexts, the credibility and policy relevance of CS depend on formal accountability structures, involve trade-offs between scientific rigor and inclusivity, and require strategic communication as a core design element rather than a peripheral activity.
Discussion
This study identifies several strengths in the design (and early implementation) of the CS program for monitoring health impacts of gold mining. First, the program was led by practitioners with extensive CS experience, enabling the adaptation of established methods, tools, and institutional networks to a complex and contested local context. Lessons from prior CS initiatives were successfully operationalized. Second, the integration of a professional journalist at the design stage strengthened strategic planning for dissemination and public communication—an element often underdeveloped in CS initiatives addressing politically sensitive environmental governance. Third, collaboration with external agencies as formal partners supported early co-learning and fostered institutional interest in applying CS approaches beyond the project.
More fundamentally, the program’s governance architecture is analytically significant. Rather than functioning primarily as a data-generation exercise, the CS initiative was deliberately designed as an embedded accountability infrastructure. By integrating collaborative governance, institutional oversight, and multi-actor participation, the program sought to transform citizen-generated data into a credible basis for dialogue, oversight, and potential corrective action in mining governance. Governance design thus enabled co-production understood as the integration of local knowledge and scientific expertise to generate policy-relevant evidence (Fung 2015; Irwin and Michael 2003; Stevenson and Dryzekll 2014).
This design choice is particularly salient in Thailand, where civil society actors often operate within constrained institutional spaces, and face limited formal access to decision-making arenas (Ungsuchaval 2022; Ungsuchaval 2025; Ungsuchaval and Ariyasirichot 2023). In this context, CS may serve as an alternative pathway through which civil society enters institutional governance processes. By producing evidence that is scientifically structured yet socially grounded, CS can enable a shift from protest-based or moral claims toward evidence-informed engagement with state institutions, as observed in prior Thai cases (Ungsuchaval et al. 2022). This role is especially pronounced in environmental and health governance, where technical expertise is frequently mobilized to gatekeep participation.
At the same time, the findings show that CS initiatives in contested governance settings require formal accountability structures to achieve both scientific and political credibility. Legitimacy was not derived from participation alone, but from institutionalized safeguards such as an independent oversight committee, documented tool-validation protocols, and response pathways linking citizen-generated data to authorities. Participants and external stakeholders consistently associated credibility with these mechanisms. This supports a testable proposition that formal accountability structures are positively associated with external acceptance of citizen-generated evidence in contested policy domains.
The evaluation also reveals challenges that constrain CS effectiveness and policy relevance in high-conflict contexts. The program operated amid entrenched distrust, asymmetric power relations, and long-standing conflict. Although diverse actors were engaged, gaps in continuous communication led some groups to perceive their involvement as insufficient, reinforcing scholarship that emphasizes deliberation, transparency, sustained engagement, and role clarity as prerequisites for trust-building in participatory governance (Kantamaturapoj et al. 2018; Rowe and Frewer 2000; Ungsuchaval et al. 2024).
Participants also differed in expectations regarding data credibility, ownership, and use. Ambiguity around validation, access, and downstream application limited the perceived legitimacy and utility of CS outputs, particularly where communities sought evidence for advocacy or legal action. This reflects broader tensions in Thai governance, where civil society is often expected to provide highly validated evidence to counter centralized state or corporate power, while state agencies retain discretion over whether such evidence is recognized or acted on (Ungsuchaval 2025; Ungsuchaval and Ariyasirichot 2023).
These dynamics point to a second key insight: Increasing scientific rigor in CS designs may raise participation thresholds and unintentionally constrain inclusivity. While validation and quality-control procedures strengthened scientific robustness, some participants perceived them as barriers to sustained engagement, particularly those with limited monitoring experience. This suggests a trade-off between credibility and accessibility that remains under-theorized in CS research. As CS initiatives become more formalized to meet accountability demands, participation may become more selective unless accompanied by compensatory capacity-building and support mechanisms.
A third insight concerns communication. The evaluation highlights strategic communication as a core design variable rather than a peripheral activity in accountability-oriented CS. Although participants trusted the technical aspects of data collection, uncertainty over how data might be interpreted, disseminated, or politically mobilized generated unease. This challenges assumptions that transparency alone builds trust. Instead, the findings suggest that CS initiatives lacking explicit communication strategies for managing external interpretation and political use of data are more likely to experience trust erosion—even when scientific protocols are robust.
Therefore, these findings support a conceptualization of CS in contested governance contexts as embedded accountability infrastructure rather than a standalone participatory research method. Credibility, participation, and political relevance are jointly produced through interrelated design choices concerning governance arrangements, scientific rigor, and communication practices. By foregrounding program design as an evaluative object, this study contributes to CS theory by identifying concrete, testable mechanisms through which citizen-generated data can function as trusted inputs into environmental health accountability processes.
Conclusions
This study contributes to CS scholarship by examining the design and formative evaluation of a community-based CS initiative implemented in a highly contested environmental governance context. Rather than focusing solely on outcomes, the analysis foregrounds the processes through which CS initiatives are designed, governed, and negotiated among diverse actors. In doing so, it responds to calls for greater attention to governance arrangements, data practices, and contextual conditions shaping participation and knowledge production.
The evaluation identifies three interrelated design considerations critical for future phases of the program and for CS initiatives in similar settings: (1) systematic stakeholder analysis and management, (2) explicit agreements on data ownership and access, and (3) intentional external communication strategies that support trust.
First, the findings underscore the importance of rigorous stakeholder analysis to understand power relations, expectations, and roles, and to design appropriate participation mechanisms (Brugha and Varvasovszky 2000; Bryson 2004; Kantamaturapoj et al. 2023; Luyet et al. 2012). Although the program encouraged broad participation, insufficient engagement of key decision-making actors limited institutional uptake and generated uneven perceptions of inclusion. Participation in CS should therefore be treated as a strategic design choice rather than an assumed outcome, with continuous communication playing a key role in sustaining engagement (Vann-Sander et al. 2016).
Second, the study highlights data governance as a central yet often under-specified dimension of CS practice. Divergent expectations regarding the use of citizen-generated data—ranging from internal learning to public disclosure and legal advocacy—created tensions among participants. While ethical safeguards were addressed, the absence of clear agreements on data ownership, access, and use constrained trust and collaboration. These findings reinforce that data governance in CS is a social and political process, not merely a technical one, particularly in conflict-affected contexts (Golumbic et al. 2020; Vann-Sander et al. 2016).
Third, although the program demonstrated strong scientific rigor through expert validation and quality assurance, these strengths were not consistently visible to external stakeholders. This supports prior research showing that scientific credibility in CS depends not only on methodological rigor, but also on transparent communication about how data are produced and validated (Haklay 2015). Communication should therefore be treated as an integral component of scientific practice rather than a post-hoc activity.
More broadly, this study advances CS theory by illustrating how CS can function as an accountability-oriented knowledge practice rather than solely as a participatory or educational intervention. In conflict-affected settings, CS initiatives are often expected to produce evidence capable of supporting claims, challenging authority, or enabling corrective action. This places heightened demands on governance design, data validation, and institutional linkages, suggesting that co-learning-oriented models may be insufficient without explicit accountability pathways.
This study has several limitations. It is based on a single case and focuses on the formative design stage prior to full implementation, limiting conclusions about long-term outcomes. In addition, the researchers’ insider positionality may introduce interpretive bias, although this was mitigated through a structured evaluation framework and systematic analysis. While not statistically generalizable, the findings offer analytical and practical transferability. The design principles, governance mechanisms, and tensions identified here are likely relevant to other CS initiatives operating in high-conflict or low-trust governance contexts where citizen-generated data are expected to inform accountability processes.
Data Accessibility Statement
For privacy and identification reasons, interviewees were assured that transcripts of recorded interviews would not be shared beyond the research team. Data in the form of a codebook has been made available.
Supplementary File
Supplemental File 1: Appendix 1
Interview guide and a detailed codebook used for the thematic analysis. DOI: https://doi.org/10.5334/cstp.973.s1
Notes
[1] The C-Site platform is publicly accessible online at https://csite.thaipbs.or.th/home. However, this specific CS program has not yet been made publicly available due to concerns about potential misinterpretation and the risk of conflict.
Ethics and Consent
Ethics approval was acquired from the Institute for the Development of Human Research Protection (COA No. IHRP2024132, dated September 16, 2024).
Acknowledgements
This study is conducted as an integral component of the larger project, “Community Health Impact Surveillance: [The gold mine Z]” The authors gratefully acknowledge Ms. Somporn Pengkam for facilitating and granting permission to conduct the evaluation phase of this project.
Author Contributions
TU and KK contributed equally to all aspects of the study, including the conceptualization, design, methodology, data acquisition, project administration, formal analysis, original draft preparation, and subsequent review and editing.