Abstract
This study investigates the overlooked yet pivotal role of local sports communities (LSCs) as data agents in the collaborative governance of urban public spaces. Rather than framing citizens’ leisure sports solely as intervention, we develop a triadic framework, comprising Governance Structure, Data Production, and Decision Mediation, to analyze how LSCs generate, manage, and translate data for urban governance. Three integration patterns are identified through the theory-informed case studies (n = 12). The findings show that LSCs function with multifaceted roles, providing trust and legitimacy, generating grounded data, and mobilizing narratives to shape governance pathways. Ultimately, we conclude with a Quadrant Model to map LSCs’ embeddedness along the axes of institutional legitimacy and grassroots agency, offering a diagnostic tool for tracing the evolution from data silos to active data agency. The study contributes to debates on data-driven urbanism, collaborative governance, and leisure sports.
Introduction
In many cities, everyday public life is increasingly mediated by local sports communities (LSCs). Running clubs chart safe and engaging urban routes; swimming groups monitor the quality of rivers and coastlines; neighborhood fitness collectives assess the accessibility of parks and open spaces. Beyond promoting health and sociability, these grassroots practices generate situated data and knowledge on how urban spaces are used and experienced (Liu et al., 2016; Wang and Vermeulen, 2021). Such data, whether spatial concentrations (Balletto et al., 2021), movement trajectories (Liu and Lai, 2024), or feedback on facilities (Wu et al., 2018), captures dimensions of urban life often absent from official datasets and commercial platforms (Tian, 2020).
Although the data generated by the LSCs has the potential for planning and governance, it remains on the fringes of the institutional framework (Glebova and Desbordes, 2022; Leszczynski, 2016). Large-scale acquisition and application by technical experts, relying on public sensor networks or commercial platforms, is the mainstream approach at present (e.g. Chen et al., 2018; Robinson et al., 2024). Self-organized communities emerge from shared interests, social ties, or geographic proximity (Misener and Doherty, 2014), representing an underestimated but promising source of self-generated datasets that are embedded in everyday urban practices (Sileryte et al., 2016).
In addition to providing localized data, LSCs also mediate between individual actions and collective outcomes, shaping identities, and social ties (Couldry and Hepp, 2018). They bring together athletes, organizers, and brands, motivated by fitness, sociability, identity, and market positioning (Eime et al., 2022). Their organizational trajectories often evolve from informal networks to structured, sometimes commercially oriented entities (Hindley, 2022a; Robinson et al., 2024). This positions LSCs as mediating units in polycentric governance. Inspired by the increasing mobilization of such data by sports brands (Popp and Woratschek, 2016), we propose that LSCs can act as legitimate data proxies for collaborative decision-making in urban governance.
A systematic understanding of the various data types, processing methods, and application mechanisms generated by LSCs can raise awareness of the value and utility of these data in the urban domain. Therefore, this study explores how LSCs can be embedded as data agents within urban public space governance frameworks, aiming to mitigate data silos and advance collaborative governance of public spaces.
In the following sections, we developed a triadic theoretical framework, integrating three layers: the governance structure layer (GS), the data production layer (DP), and the decision mediation layer (DM). This approach enables a systematic examination of how rules (GS), data practices (DP), and actions (DM) interact to activate LSCs as data agents that facilitate urban governance. Through theory-informed case studies (n = 12; George and Bennett, 2005), we identify three integration patterns and synthesize a Quadrant Model. The model maps LSCs’ embeddedness along axes of institutional legitimacy and grassroots agency, and offers a diagnostic tool for tracing their evolution from data silos to active data agency. The study provides practical insights for public space governance, while contributing to broader debates on healthy cities, collaborative governance, and the institutionalization of citizen-generated data.
Literature review
Civic agency, volunteered geographic information (VGI), and data agency
Integrating the LSC-generated data into urban governance structurally represents the entry of civic agency into the digital collaborative governance system. The concept of civic agency dates to Jacobs’s (1961)“Eyes on the Street,” which highlighted how a diverse mix of storefronts, residences, and pedestrians collectively cultivates informal surveillance and a micro-scale spatial governance. These civic actors now extend beyond traditional residential communities and street-based units to groups including LSCs (Bui, 2025; Palmer, 2024). By activating parks, waterfronts, streets, and embedding their norms and routines in urban life, grassroots sports organizations increasingly intersect with governance processes (Middle et al., 2017; Robinson et al., 2024).
The core of civic agency usually lies in voice or social mobilization (Grzymala-Kazlowska and O’Farrell, 2023), not data progress or infrastructures. As data is regarded as the lifeblood of decision-making and the raw material for accountability (UN, 2014), data agency has emerged as a more evidential and traceable form analytically distinct from civic agency. It encompasses the ability to intentionally or unintentionally generate, interpret, and strategically mediate data that may affect governance decisions (Kitchin, 2014; Liu and Lai, 2024), reconfiguring civic agency’s traction in digital governance. LSCs exemplify such data agency through digital traceability, evidence-based claims, and platform-driven advocacy, distinguishing it from broader forms of knowledge brokerage or civic participation.
Advancements in sensors, mobile devices, and remote sensing are transforming cities into “perceptive systems” (Van Ameijde, 2023). Earlier VGI and “citizens as sensors” traditions focused on data production and accuracy (Goodchild, 2007). Yet scholarship has called for moving beyond this accuracy discourse to governance, mediation, and institutional arrangements (Elwood et al., 2012). Our study answers this call by investigating how LSCs interpret, mediate, and translate fragmented data into governance-relevant forms, what we term data agency. The collection and use of citizens’ sport-related data (e.g. location, movement trajectory, activity intensity) largely relies on digital infrastructures and platforms. However, these developments risk exacerbating tensions among privacy delineation power, data ownership, and public interests (Couldry and Mejias, 2019), while also deepening platform capitalism and digital exclusion (Barns, 2019, 2021).
How multiple actors cooperate and negotiate has been widely examined in modern governance research, including polycentric (Ostrom, 2010) and collaborative governance theories (Ansell and Gash, 2008; Emerson et al., 2012). These theories help understand how LSCs are embedded in institutional environments, but insights into LSCs’ data practices remain limited. Moreover, existing scholarship habitually defines LSCs as public health or leisure, neglecting their potential as positive data agents. Although some studies have noted that citizen sports data inform urban renewal (Bui, 2025; Middle et al., 2017), these initiatives are still dominated by technocrats and government actors.
Historic timing of LSCs’ governance roles
With growing public health awareness, low-threshold and community-oriented sports groups are becoming important. They have flexible structures, rapid responsiveness, and sensitivity to urban design and policy outcomes (Doherty and Cuskelly, 2020; Middle et al., 2017; Yu et al., 2023). They align with what this study terms Local Sports Communities (LSCs), defined as self-organized groups tied by shared sports interests and operating in public spaces, embedded across community, space, and data.
Multiple intersections facilitated the transformation of LSCs from traditional public health service units to data agents within urban governance (Garamvölgyi et al., 2022). Since the approval of civilian GPS use in the late 1990s (Appiah et al., 2025), the popularity of GPS devices and fitness platforms has enabled LSCs to digitally capture collective spatial practices and high-precision geocoding data (Robinson et al., 2024). Additionally, social events and public crises (e.g. SARS, COVID-19 pandemic) have accelerated the recognition of public spaces’ social value and have increased institutional demand for real-time, fine-grained data (Cinnamon, 2020). Policymakers increasingly emphasize the use of vivid community-generated data to complement general statistics, highlighting data-driven governance (Ansell and Gash, 2008). These recent changes have made previously invisible and informal LSCs’ practices traceable and actionable, creating institutional interfaces through which the LSC-generated data can be integrated into urban policies and planning.
Most LSCs are grounded in recreational and social proximity; acting as data agents is not an explicit self-identification. When their activity spaces, facilities, or opportunities are disrupted, such capacity might be activated (Wu et al., 2018). This lack of intentionality does not weaken their analytical relevance. Governance-related data arise from their situated practices in response to their environments, rather than from deliberate data production. These locally rooted practices constitute the uniqueness of citizen-generated data, recording and interpreting data endow it with initiative (Kitchin, 2014; Milan and Treré, 2019).
Theoretical framework and research questions
To explore how LSCs act as data agents embedded in collaborative governance processes, this study synthesizes existing theories into an integrated triadic theoretical framework:
1. Governance Structure Layer (GS) captures the institutional environments that shape LSCs’ data practices. Ostrom’s polycentric governance (McGinnis and Ostrom, 2012; Ostrom, 2010), self-governance (Ostrom, 1990), and collaborative governance (Ansell and Gash, 2008; Emerson et al., 2012) together emphasize institutional diversity and negotiated co-production. Within such environments, LSCs position themselves as collaborative actors embedded in governance networks (Healey, 2020).
2. Data Production Layer (DP) focuses on the attributes of data and its technical processing. The citizen-generated data (Heeks and Renken, 2018) characterizes the data produced by LSCs as social agents, illustrating how grassroots communities assert interpretive and narrative power through data production (Kitchin, 2014). This layer centers on the ontological qualities of LSCs’ data and its role in everyday collaborative practices.
3. Decision Mediation Layer (DM) demonstrates how grassroots actors can reframe data as governance-relevant knowledge and gain traction in governance processes (Gray et al., 2016), which is consistent with the concept of civic data intermediation. LSCs’ intermediary roles extend to organization, persuasion, and institutional alignment, creating strong couplings with GS and DP.
Within this framework (GS-DP-DM), LSCs are conceptualized as data agents capable of collecting, processing, and transforming data into decision-relevant information (Kitchin, 2014). Building on this, the study addresses three sub-questions:
Methods
This study adopts a theory-informed multiple case study approach, designed to advance a dialogic exchange between theoretical constructs and empirical evidence (George and Bennett, 2005).
Case selection and comparability protocol
LSCs samples that engage in public space governance or act as collective actors in governance-oriented initiatives were selected, employing theoretical and purposive sampling (Patton, 2014). We established the following recursive three-stage selection to ensure the representativeness and explanatory power of the final samples.
The inquiry began with two seed cases, namely London Parkrun and the Shanghai Escape Plan Running Club (EPRC). The researchers explored these two cases as heuristic anchors through archival data and participatory observation and operationalized the theoretical concepts in conjunction with the literature. Although EPRC was finally excluded due to limited governance integration, it facilitated refining the GS-DP-DM framework.
To test the framework’s explanatory boundaries and capture mechanistic variations, we employed two theoretical sampling steps, yielding an initial repository of 28 potential cases: (1) Snowball sampling traced network and literature connections from the seed cases. (2) Maximum variation sampling deliberately targeted LSCs across diverse geographical and organizational contexts.
The screening and verification followed four operational criteria: (a) public spaces as the primary activity site; (b) evidence of traceable data practices; (c) documented linkage to governance processes; and (d) verification through multiple independent sources (triangulation). This screening ceased upon reaching theoretical saturation and yielded 12 final cases (Patton, 2014).
Data from 12 cases were extracted and triangulated to construct a traceable analytical archive (see Table 1). The archive integrates policy evidence, platform/digital records, and secondary interpretative sources (academic and media). Each case is supported by at least two evidence types to ensure evidentiary robustness for analysis.
Case selection process and data triangulation sources.
We acknowledge that the selected cases are mainly publicly documented, which might exclude the less visible ones. However, to ensure analytical comparability and enhance the internal consistency of the results, such trade-offs were necessary. In addition, cases involving commercial actions might yield higher-quality platform records, whereas marginal cases relied more on policy or academic documentation. To reduce bias, we adopted pattern matching (George and Bennett, 2005) instead of statistical comparison, ensuring insights were drawn from consistent data connections within the GS-DP-DM framework. Therefore, the analysis comparability was constructed at governance mechanisms and data-mediated pathway levels.
Theory-informed analysis
Based on the triadic framework, the analysis proceeded in three steps, supported by the continuous writing of analytical memos (Lempert, 2007) to document conceptual development, theoretical insights, and comparative insights.
Step 1: Transformation of sensitive concepts and initial encoding categories. The GS layer focused on participant roles (GS1) and participation stages (GS2). The DP layer examined data types (DP1), recording & storage (DP2), and data ownership & privacy (DP3). The DM layer captured translation instruments (DM1) and intermediaries (DM2). Analytical memos documented code rationale insights, retaining exploratory codes to capture emergent themes.
Step 2: Coding and refining categories, combining theory-driven and inductive approaches. Inter-coder reliability was enhanced through consensus-based procedures, reconciliation meetings, and systematic memo use (Guest et al., 2011). For instance, discrepancies in coding the primary governance actor (GS) for Case 12 were resolved by returning to the original case archives and triangulating source materials to reassess the evidence and consolidate interpretive coherence.
Step 3: Synthesizing findings to construct theoretical propositions. (1) descriptive analysis clarified LSCs’ roles across the three layers (SQ1); (2) pattern matching traced data flows, identifying three integration patterns (SQ2); (3) theoretical refinement linked empirical patterns with theories, providing request conditions and actionable guidance for governance (SQ3).
Findings and analysis
Performing roles: LSCs as multi-layered data agents
The multi-case analysis indicates that LSCs play a set of interrelated data roles within urban governance. They unfold the three theoretical layers and produce nine sub-dimensions in total (see Online Supplemental Appendix 1), providing trust and legitimacy, generating grounded data, and mobilizing narratives to shape governance pathways. The findings directly address SQ1.
The Governance Structure (GS) layer: Navigating legitimacy
In the GS layer, LSCs act as governance actors who initiate, co-create, and sustain networks. They provide legitimacy and trust that external bodies cannot replicate.
Certain LSCs exhibit proactive initiating roles. For instance, Case 5-The Bluetits Chill Swimmers advocated and initiated water-quality monitoring to enable swimming enthusiasts on the Thames to understand the water quality conditions (Edwards, 2024). This LSC participated in the consultation on proposals to designate 27 new sites as bathing waters under the Bathing Water Regulations 2013 (S.I. 2013/1675) held by the Department for Environment, Food & Rural Affairs (DEFRA). The swimming associations offered detailed and engaging feedback, conveying that the blue spaces are commons for people to use and enjoy (DEFRA, 2024). Here, legitimacy is derived from lived experience, promoting the grassroots priorities and their practical experiences to enter the municipal agenda.
Other LSCs play a cooperative role in institutionalized cooperation. Case 1-Parkrun UK maintains a structured partnership with the Royal College of General Practitioners (RCGP) and local authorities. This cooperative initiative (Public Health England, 2020) promotes social prescription sports activities through participation in 5 km parkrun events. Clinic staff can socially prescribe physical activities by recommending such activities to patients (over 16% of practices have been registered as parkrun practices), demonstrating their legitimacy and social trust. Such collaborative data agency bridges the gap between informal community activity and formal public health interventions, blending grassroots vitality with expert authority to co-constitute urban policy (Hindley, 2022b).
The Data Production (DP) layer: From practice to evidence
The Data Production layer is delineated by data types, generation, and management, and is primarily centered around two agent mechanisms: making data Useful or Usable. Among the various data processed by LSC, the ones with spatial governance significance mainly consist of three types: Participant, Behavioral, and Feedback data (see Table 2).
Detailed data agent mechanisms of the 12 selected cases.
LSCs often act as stewards of sensitive and identity-based participant data, recording personal narratives. In Case 4-Brooklyn Banks Skate Park, the LSC preserved the cultural memory of the skateboarders’ collective in Lower Manhattan (Alleyne-Davis, 2025). When urban renewal and gentrification in this area eliminated their activity space, these non-anonymous, qualitative, and narrative data were desensitized and became usable for community claims (NYC, 2025). As lead advocate, Steve Rodriguez noted: It felt like something we could own, something that belonged to us. No one was kicking us out… Just skateboarders doing their thing.
For behavioral data, LSCs’ role depends on technical actionability. In sports with low data-recording thresholds, like running, activity initiators work within predefined data frameworks. LSCs aggregate scattered data from platforms such as Strava, whose Metro data stream informs municipal cycling and pedestrian infrastructure planning (Garrity, 2024; Robinson et al., 2024). As a repository hub, LSCs compile fragmented data into usable formats. For complex environmental monitoring (e.g. water quality), LSCs face distinct aggregation challenges. Case 6-Flussbad Berlin, a long-running civic initiative advocating for a swimmable Spree, has translated community monitoring into an institutionalized planning process with public funding and formal partners (Kraemer, 2021). Volunteers compile water quality readings from citizen samples, public sensors, and real-time flow data (Flussbad Berlin e.V, 2025). These inputs feed into the SWIM: AI Forecasting System, a model developed by the Kompetenzzentrum Wasser Berlin (2025; Seis et al., 2024), which continuously compares forecasts against actual measurements. Sensor measurements are validated twice weekly through laboratory testing (Flussbad Berlin e.V, 2025).
Feedback data is an asset that is cooperated by LSCs and institutional partners. Such cooperation can enhance rigor and institutional legitimacy but may also recalibrate data proxies. For instance, in Case 2-Filipino Migrant Sport Communities, academic partners facilitated systematic analysis and professional intermediary outputs that were difficult for grassroots communities to achieve, enabling community experience to serve as reference knowledge for policy-making (Aquino et al., 2022). The potential concern is that the archiving and dissemination rights of feedback data are often influenced by institutional priorities. Therefore, LSCs exhibit a conditional data proxy, enhanced in visibility and credibility, but restricted in narrative control.
The Decision Mediation (DM) layer: Translation and advocacy
The last layer demonstrates that LSCs go beyond the realm of data generation and play the role of an explanatory mediator, shaping the framework, communication, and operability of evidence in the governance process.
The mediation capabilities of some LSCs are manifested in alliance relationships, conveying common initiatives to the general public. Case 9-Oslo Fjord Sauna mobilizes facility usage and access data to claim the social value of waterfront public spaces, sparking debate on coastal accessibility and protection (Gurholt and Kronsted Lund, 2026). Similarly, Case 7-Swimmable Birrarung Alliance renders complex environmental data in a readable and resonant way to non-expert audiences and policymakers, thereby bridging specialized knowledge and public deliberation (Bellato et al., 2024).
Some other LSCs play a more exemplary role in governance implementation. In the hybrid governance model of Case 11-Minnesota Cross Country Skiers Community, LSCs and regional networks generate and publish grooming logs and trail updates on park webpages (e.g. Duluth Cross-Country Ski Club, Cities of Duluth Minnesota, n.d). These data interpreted by LSCs inform the public and administrators and facilitate the park in securing the Cross-Country-Ski Trail Grant Statutes (Minnesota Statutes, 2025). This exemplary effect has also become a policy implementation tool. Case 3-Village Super League demonstrates that local engagement, live-stream visibility, and integrated cultural programming prompt local authorities and stakeholders to reevaluate the socio-economic and spatial value of community sports, modeling rural revitalization for other regions (Yan and Xu, 2024).
If the role of these grassroots organizations stops at being a civic agency or data provider, the extensive collaboration and implementation may degenerate into perfunctory initiatives or idealistic utopian practices. Collaborative governance can truly be promoted when structured, interpretive data agent roles are identified and empowered. This is conducive to better leveraging communities and promoting collaborative governance.
Building connections: Integration across the three dimensions
Collaborative governance often faces institutional barriers or technical obstacles (Kitchin and Moore-Cherry, 2021). Our findings indicate that LSCs bridge these gaps by synthesizing Governance Structure, Data Production, and Decision Mediation into a cycle anchored in community values, ensuring that data remains contextually grounded.
Three patterns of LSCs’ integration across the three layers
The efficacy of LSCs as Data Agents stems from their multi-roles across the three layers. They are simultaneously governance actors, data stewards, and narrative translators. Such structural overlap allows LSCs to internalize the translation process, converting fragmented metrics into coherent governance intelligence. Based on the principal forms of coordination and leadership roles of LSCs, three integrative patterns were identified (see Figure 1), addressing SQ2.
A) LSC-Led Pattern (Cases 4, 5, 6, 9): LSCs in this pattern proactively project their community narratives to claim a seat at the governance table. They initiate governance, produce data, and advocate for decisions, demonstrating strong grassroots agency. External actors (e.g. academics, media, government) engage later, mainly to amplify or legitimize the community’s actions. Legitimacy stems from deep community trust, cultural identity, and shared memory. The action objectives of such LSCs usually focus on the requirements for public spaces and the environment (e.g. skate parks demolished in urban renewal, rivers prohibited for swimming or polluted), bridging data silos at the governance structure level.
B) Coalition-Based Pattern (Cases 1, 7, 8, 11): This pattern reveals a negotiated order where LSCs and institutional actors (governments, NGOs, professionals) align around a common vision, thereby mitigating data isolation. LSCs contribute community trust, local knowledge, and data; partners provide expert authority, technical resources, and institutional support. Case 8-Surfrider Foundation Japan (SFJ) demonstrates this pattern and differs from the institutionalized means of confrontational public environmental claims (litigation or policy advocacy) of its international headquarter Surfrider Foundation (2026). SFJ shows coalition-based actions that each entity (surfing enthusiast communities and clubs, government departments, nearby residents, etc.) made efforts to safeguard their shared marine lifestyle and promote the prosperity of their surfing communities (Cristina and Clément, 2024).
C) Institutional-Cooptation Pattern (Cases 2, 3, 10, 12): When the potential (commercial value or social value) of LSCs is recognized, external institutions will extend “olive branches” to them or recruit from them, providing platforms, resources, and legitimacy, formalizing and scaling LSCs’ activities and data. In Case 10-Grand Rapids Skating, the Destination Marketing Organization (DMO) has gathered various local entertainment resources and the community’s power of ice and snow sports to enhance local tourism. These decentralized groups, including LSCs, have all received support and benefits to varying degrees. Although this pattern expands the impact and scale, it often triggers a tension between efficiency and the erosion of community spontaneity.
Three patterns of LSCs’ integration across the three layers.
The paradox of legitimacy: Operational dynamics and critical tensions
The three patterns identified above provide a descriptive account of how LSCs connect across governance layers. Building on this, their operational mechanisms and enabling conditions can be distilled.
The LSC-led pattern typically emerges within institutional vacuums or governance failure, where political opportunity structures allow grassroots initiatives to thrive. Yet its grassroots nature also constitutes vulnerability, such as resource precarity and difficulties in scaling outcomes. Sustained operation often hinges on a critical mode transition toward the coalition-based model. In Case 4-Brooklyn Banks Skate Park, local advocates secured the return of a skateboarding space through lobbying municipal authorities and external organizations (Alleyne-Davis, 2025). While LSCs retained autonomy throughout, the process bore the imprint of a coalition-based pattern. Strategic alliances with professional or governmental bodies provide the institutional scaffolding.
The coalition-based pattern approximates an idealized standard of collaborative governance. However, its high transaction costs should not be ignored. For instance, Case 7-Swimmable Birrarung Alliance demonstrates a shared vision and complementary resources that foster a diverse coalition (Bellato et al., 2024). The alliance has gradually taken shape through continuous convening since 2020, and its momentum was boosted by the political timing of the Seine River clean-up during the 2024 Paris Olympics (Benton, 2025). It reveals that coalition-based patterns usually require unforeseeably long periods of negotiation and may make its momentum dependent on contingent policy windows (Kingdon, 1995). The burden here also lies in the power asymmetry. Resource-rich institutional partners may inadvertently dominate the agenda and undermine data institutions’ autonomy.
The institutional-cooptation mode harbors an inherent tension between power and agency, revealing a key paradox: gaining institutional legitimacy often requires constraining grassroots agency. Although the legitimacy of LSCs is rooted in grassroots representativeness and local knowledge, ensuring institutional legitimacy (e.g. state recognition, funding) often requires domesticating radical institutions. Legitimacy in urban governance is not a static resource but constructed, contested, and instrumentalized (Docherty et al., 2001; Legacy, 2017). Case 3-Village Super League has been heralded as a model pilot project imitated by other cities and rural cultural-tourism initiatives in China (Yan and Xu, 2024). Such standardization improves efficiency and scale, but at the cost of reducing community radicalism and subordinating social goals to economic logic. The core tension lies between incorporation and empowerment, whether LSCs can retain autonomy across data narratives (DM layer) and community identity even after ceding operational control (GS layer).
These patterns indicate that integration into governance structures is a negotiated and often contested process, neither linear nor explicitly beneficial. As LSCs seek legitimacy, data privacy ethics are being reconfigured. Data disclosure is mainly regarded as a risk to be managed or a defense boundary (cf. Curran and Smart, 2021). In contrast, our cases suggest that data visibility serves as a performative claim for the collective (e.g. Cases 2, 3, 4, 7). With this logic, data becomes ontological proof of who the participants are and their community values (Alleyne-Davis, 2025). By voluntarily disclosing its data to support collective claims, LSCs point toward softening data silos. Here, data connections stem from social belonging and identity recognition rather than top-down technological integration, fostering a more relational form of collaborative governance.
The quadrant model of grassroots communities as data agency: Embedding LSCs in urban governance
The three empirical patterns describe data agent mechanisms of LSCs. However, considering these patterns in isolation may manifest static or context-specific configurations. To address this limitation, we synthesize the cross-case analysis and develop a Quadrant Model to conceptualize how LSCs can be embedded in urban governance through data agents (Figure 2), responding to SQ3.
The quadrant model of grassroots communities as data agency.
The model integrates the patterns into a dynamic relationship oriented toward governance practice. It situates LSCs along two intersecting dimensions: institutional legitimacy (X-axis), reflecting the degree of formal recognition and integration into governance arrangements, and grassroots agency (Y-axis), capturing communities’ capacities to mobilize, interpret, and mediate data in daily practices.
The Quadrant Model does not introduce additional types beyond the three patterns. It encompasses Quadrant IV Latent Potential (data silos), a site from which multiple governance trajectories may emerge, depending on intervention logics. Through the Quadrant Model’s diagnosis, diverse entities can plan appropriate governance pathways under different conditions.
The Community-led quadrant reflects autonomous community data practices that enable rapid grassroots innovation but remain institutionally fragile due to limited formal support. LSCs typically enter this quadrant when activity spaces are constrained or institutional backing is lacking, requiring communities to actively exert initiative to secure rights. The Coalition-based quadrant captures data practices organized through negotiated partnerships, while simultaneously increasing coordination burdens and negotiation costs. This quadrant is suitable for governance issues where multiple parties share a vision (e.g. environmental monitoring). Entry typically occurs when government and institutions guide LSCs toward collaborative arrangements. LSCs that demonstrate strong data agency from the Community-led quadrant can further integrate here. In the Institutional-Cooptation quadrant, community practices gain formal recognition, while subsequent standardization of data practices or bureaucratic assimilation may limit grassroots agency initiative. This quadrant applies when government intervention significantly outweighs LSCs’ agency, or when LSCs’ social or economic influence signals governance potential.
The model does not prescribe an ideal outcome. It enables practitioners to locate communities and consider which axis to intervene on first and through which pathway, providing transferable governance mechanisms (SQ3). It engages in dialog with Arnstein’s (1969) Ladder of Citizen Participation, introducing a second key dimension, community agency (Y-axis), beyond the degrees of power transfer emphasized by the ladder (X-axis). This model emphasizes that effective interventions must advance both axes simultaneously to promote capacity building (Chaskin, 2001).
Discussions
This study demonstrates that LSCs can function as data agents in urban public governance. Such capacity is not by default, but emerges through specific configurations of data practices, organizational capacities, and institutional relations.
Using the quadrant model as a governance diagnostic tool
The Quadrant Model can serve as a starting point for decision-making workflows, enabling practitioners to map community groups, assess their data capacities, and design targeted engagement strategies. To integrate the Quadrant Model into existing planning and data governance workflows, we propose a 3A three-step protocol for public agencies:
1) Accessing: Identify LSCs’ position by assessing two observable dimensions. Their degree of institutional mediation (X-axis): Do the LSCs have formal agreements, funding relationships, or data-sharing agreements with a public body? Their data production capacity (Y-axis): Do the LSCs produce structured, reusable data (e.g. spatial, sensor-based) and have the capacity to interpret it?
2) Analyzing: Diagnose typical gaps or bottlenecks in data flows. Common issues include a) data not in usable formats for agency workflows; b) no formal decision-making entry point for community inputs; or c) temporal mismatches between community data collection (e.g. weekly) and planning cycles (e.g. annual).
3) Acting: Determine context-appropriate engagement strategies. For data silos (Quadrant IV), prioritize legitimacy-building (e.g. data-sharing pilots). For Community-led LSCs, explore coalition models. The protocol enables context-sensitive strategy design without prescribing an ideal outcome.
Data agency in collaborative governance
Theoretically, this study reframes data agency as a relational and situated capacity. LSCs cannot replace institutional expertise to dismantle data silos uniformly, but under specific conditions, they can translate embodied practices into governance knowledge. This aligns with behavior–environment system design approaches that emphasize reciprocal adaptation (He et al., 2025), and also resonates with calls to treat infrastructure as both process and product (Glass and Addie, 2025).
Yet the pathways identified in our findings are not without significant tensions. As elaborated in the Paradox of Legitimacy Section, LSCs face risks, suggesting that expanding community-generated data is not an unqualified good but a contested terrain of data power (Barns, 2019; Couldry and Mejias, 2019). The findings also indicate that data silos should not be regarded as purely technical failures; governance structures and data collaboration modes should be considered. We take the GS layer as a strategic entry point, echoing Healey’s (2020) argument on the importance of institutional thickness for collective action, and emphasize that governance capacity should be enhanced instead of merely focusing on physical spaces (Janssen et al., 2024). In doing so, it responds to Milan and Treré ’s (2019) critique of techno-solutionism and frames data infrastructures as socio-technical systems valued for procedural justice and efficiency.
Finally, the findings also reveal a phenomenon worth delving into. The Coalition-Based Pattern is often problem-driven with action demands mostly related to environmental protection (e.g. Cases 7 and 8, river/ocean pollution). While the motivation of the Institutional-Cooptation Pattern is usually interest-driven, often associated with governance advantages (e.g. Cases 2 and 10, fiscal revenue and social stability). When mapped to the Quadrant Model, the question shifts from “how to break data silos” to “what drives the breaking of data silos,” inviting a promising research proposition on participatory governance.
Limitations and future research
Several limitations point to avenues for future work. First, although triangulation strengthened case validity, the sample of 12 cases remains limited for broad generalization. Comparative studies incorporating ethnographic or interview-based methods could further validate the Quadrant Model across diverse urban and cultural contexts. Second, the inherent spontaneity and instability of some LSCs challenge their capacity as reliable, long-term data providers. Moreover, the risks identified above (platform dependence, uneven representation, and instrumentalization) remain unresolved in our analysis. Future research should examine mitigation strategies such as community data cooperatives or governance safeguards against co-optation. Third, this study conceptualizes LSCs as data agents without developing a full technical or institutional governance framework. Subsequent research could engage with data governance theories (ownership, ethics, standardization) and explore their integration into LSC-led or coalition-based models, advancing both scholarly and policy debates on collaborative urban data governance.
Conclusion
This study explores how LSCs function as Data Agents within urban public space governance. Through the analysis of 12 global cases, we identified three core patterns of LSCs embedded in the governance framework and developed a Quadrant Model along the dual dimensions of institutional legitimacy and grassroots agency. This model provides a theoretically grounded and practical tool for understanding and enhancing the role of community data in cities, shifting focus from technical integration to socio-institutional recognition of community actors. Yet these pathways also carry risks. Platform dependence, uneven representation, and instrumentalization, documented in our cases, caution against an uncritical embrace of community-generated data. The visibility LSCs gain through data disclosure may come at the cost of autonomy. Overall, by recognizing LSCs not as data points but as strategic partners, we open new pathways for building more inclusive, adaptive, and responsive urban futures.
Supplemental Material
sj-docx-1-usj-10.1177_00420980261454086 – Supplemental material for From data silos to data agency: The role of local sports communities as data agents in urban governance
Supplemental material, sj-docx-1-usj-10.1177_00420980261454086 for From data silos to data agency: The role of local sports communities as data agents in urban governance by Yijia Jiang, Kin Wai Michael Siu, Jiayu Yang, Keren Zhang, Shiyu Liu and Huiyi Zhang in Urban Studies
Footnotes
Acknowledgements
The authors would like to acknowledge the research fund support of the Research Postgraduate Scholarship of The Hong Kong Polytechnic University and Eric C. Yim Endowed Professorship in Inclusive Design (8.73.09.847K). The corresponding author also thanks the Fulbright Scholarship and MIT’s support.
Acknowledgments of AI
This study employed artificial intelligence tools (ChatGPT and DeepSeek) for language editing purposes. These tools were used solely to enhance readability and clarity for English-language audiences, without altering the intellectual content or analytical framework of the manuscript.
Author contributions
Author 1: Conceptualization, Methodology, Data Curation, Visualization, and Writing – Original Draft. Author 2: Supervision, Resources, Fund Acquisition, Project Administration, Writing – Review, Editing & Final Proving. The following authors contributed equally to this work. Author 3: Investigation, Writing – Review & Editing. Author 4: Validation, Writing – Review & Editing. Author 5: Validation, Writing – Review & Editing. Author 6: Formal Analysis, Writing – Review & Editing.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Research Postgraduate Scholarship of The Hong Kong Polytechnic University. Eric C. Yim Endowed Professorship in Inclusive Design (8.73.09.847K) of The Hong Kong Polytechnic University.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data availability statement
All data analyzed during this study are included in this article and its Supplemental Information Files (Appendix Table 1).
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
