Organizational Ecology of Hyperlocal Groups: How Does Category Fuzziness Relate to Participation in an Event-Based Social Network?

Abstract

Social categories serve as key anchors for organizations and groups, helping to establish their identities and positions within conceptual spaces such as market and cultural niches. Researchers have examined the social categorization of both online communities and offline organizations through tags, folksonomies, and news outlets. While these studies have provided valuable insights into the success of online communities and formal organizations, the relationship between social categories and the organizational processes of hyperlocal groups remains unclear. Hyperlocal groups typically form and interact online through event-based social networks (EBSNs) while their activities take place offline. As a result, the structures of competition and individuals’ participation patterns may differ from those of online communities and formal organizations. This paper examines how ecological features, such as legitimation, competition, and category fuzziness, shape group membership and participation on an EBSN. Based on a retrospective analysis of 2017 Meetup data, a widely used EBSN platform, this study applies and extends organizational ecology theory to offer new insights into the ecological dynamics of hyperlocal groups in major U.S. cities.

Plain Language Summary

How Local Meetup Groups Grow and Stay Active, and Why Their Categories on the Websites Matter?

Online platforms like Meetup make it easy for people to organize local gatherings around shared interests, from hiking and dancing to coding and language exchange. These platforms help form what we call “hyperlocal groups,” which are rooted in specific cities or neighborhoods and often meet in person. But not all groups are equally successful. Some grow quickly and host frequent events, while others struggle to attract members or quietly disappear. This study explores why some groups do better than others. Specifically, it looks at two important factors: how many other groups are in the same category (like “Art & Culture” or “Health & Wellness”) and how clearly that category is defined. We analyzed Meetup data from 14 large U.S. cities, covering over 2 million events. We wanted to know whether having many groups in the same category makes it harder to stand out, and whether unclear or overlapping categories (what we call “category fuzziness”) affect group growth and participation. We found that groups in vague or fuzzy categories tend to do worse. They attract fewer new groups, have higher dropout rates, and struggle to get people to RSVP for events. Surprisingly, we also found that competition from similar groups doesn’t seem to hurt performance much, likely because it’s easy and free to join multiple groups. On the other hand, people are more likely to join and participate in categories that are well-known and active. Our results suggest that platform design matters. Clearer category labels and more flexible tagging could help users find relevant groups more easily, support group organizers, and make local communities stronger.

Keywords

event-based social networks hyperlocal groups competition legitimation category fuzziness

Introduction

Event-Based Social Networking (EBSN) platforms such as Meetup and Eventbrite have reshaped how people organize and participate in social life. These platforms reduce coordination costs and enable users to form interest-based groups through online forums (K. Han et al., 2014; Liu et al., 2012; Tama et al., 2023). Meetup.com, in particular, exemplifies this model of hybrid social infrastructure. As of September 2025, it hosts over 52 million members, 300,000 groups, and 100,000 events per week across 193 countries (Meetup, 2025b), underscoring its widespread adoption and influence.

Groups formed on EBSNs can be understood as hyperlocal groups, as they are geographically embedded, community oriented, and centered on shared interests rather than economic goals (Metzgar et al., 2011). These groups typically operate through informal, low-commitment participation, allowing individuals to join multiple groups freely and engage intermittently. Because digital infrastructure requires minimal resources for users, many groups persist despite low levels of activity. These features make EBSNs ideal for fostering decentralized and flexible local networks. At the same time, they make it difficult to identify the factors that shape group success, member retention, and long-term sustainability. Accordingly, scholars have examined various social dynamics in hyperlocal groups, including leadership (Ricken et al., 2014; Seering et al., 2018), activity diversity (Lee & Butler, 2020), network structures (López et al., 2015), geographic characteristics (J. H. P. Hsu et al., 2023; Robert & You, 2018), and online posting and engagement behaviors (Tao & Aubin Le Quéré, 2024).

However, less attention has been paid to how platform-level and ecological features influence hyperlocal group dynamics. Research in organizational ecology has long emphasized the roles of legitimation and competition in shaping organizational success (Hannan & Freeman, 1977; Wang et al., 2013), while studies of online communities have examined how categorical identity, audience expectations, and niche positioning affect performance (E. Pontikes & Hannan, 2014; Zhu, Chen et al., 2014). Yet, how these ecological mechanisms operate in hyperlocal groups, which are less profit-driven, more geographically bounded, and often constrained by fixed platform categorizations, remains underexplored.

This study addresses this gap by examining how ecological features, including legitimation, competition, and category fuzziness, are associated with the success of hyperlocal groups on EBSNs. Drawing on Meetup data from 14 major U.S. cities in 2017, we investigate two core research questions: (R1) How do density-based ecological features affect hyperlocal group performance? and (R2) How does category fuzziness shape user participation, group growth, and group sustainability? By linking these research questions to organizational ecology and social categorization theory, this study extends these frameworks to informal, hyperlocal communities and demonstrates how platform architecture itself can function as an ecological force shaping group outcomes. In doing so, we contribute to research on digitally mediated local communities in the informal, interest-based contexts. We also offer practical implications for platform design, including clarifying category structures to reduce ambiguity and foster healthier group ecosystems.

Literature Review

While hyperlocal groups and online communities often share similar characteristics, as both are typically formed around common topics and interests, they differ in several ways. Unlike most online communities, which operate primarily in digital spaces, hyperlocal groups are geographically embedded within local contexts (Bingham-Hall & Law, 2015) and often facilitate offline activities. EBSNs, for example, serve as key platforms that enable local engagement while maintaining an online presence for coordination. As a result, scholars have argued that EBSNs represent a distinct form of “hybrid community,” in which people connect online to organize offline gatherings (J. Chen et al., 2019).

Success Factors in Hyperlocal Groups

The growing prominence of EBSN platforms such as Meetup has sparked a rich body of research examining the social dynamics that shape hyperlocal group formation and sustainability. This research spans both individual and group levels. At the individual level, scholars have explored how information and communication technologies (ICTs) foster people’s civic engagement and local awareness (X. Chen et al., 2025; López & Farzan, 2015), enhance social cohesion (Üblacker et al., 2024), and support the infrastructuring of civic life (Daly et al., 2015). Research on group organizers has also highlighted challenges such as leadership burden (Ricken et al., 2014; Seering et al., 2018) and perceived success (Ricken et al., 2017), emphasizing the human labor and affective investments required to sustain hyperlocal groups.

At the group level, studies have investigated factors contributing to the success or failure of hyperlocal groups, including the size of initial event participation (Ricken et al., 2017), the structure of participant networks and their embeddedness (López et al., 2015), and the spatial dispersion of members (Robert & You, 2018). Gender has also emerged as an important dimension, with evidence that the gender of organizers can influence the diversity of group members and event participants, as observed in Meetup groups focused on software development (Sharma et al., 2022). Additionally, recent research has examined the mechanisms of long-term engagement (Atrey et al., 2024), highlighting the importance of topic alignment and collective identity in sustaining group momentum. These studies underscore that hyperlocal group dynamics are shaped not only by individual behavior but also by group-level activities and the social networks that emerge among members.

Performance in Broader Online Groups and Communities

Group performance in social computing spans two broad domains: group work performance and online community success, both of which are directly related to how “success” is characterized in hyperlocal groups. Group work refers to tasks performed by individuals collaborating either online or offline and is defined by how effectively members complete shared tasks. These tasks include, but are not limited to, content tagging (Thornton & McDonald, 2012), knowledge curation (Lam et al., 2010), and learning (Ounnas et al., 2007). These outcomes are often shaped by collaboration patterns (Quinn & Bederson, 2011), crowdsourcing structures (Franzoni & Sauermann, 2014; Kittur et al., 2013; Spicher et al., 2024), the diversity of interests (Ren et al., 2016), diversification within social networks (Silva et al., 2019), gender dynamics (Schiller et al., 2024), technology adaptation (Ounnas et al., 2007), and coordination goals (Suffill et al., 2019).

Meanwhile, studies on online communities focus less on task outcomes and more on resource-based indicators such as membership growth, retention rates, and content generation, all of which directly affect a community’s sustainability and influence (Butler, 2001; Kraut & Resnick, 2012). Studies have found that online community success depends on social factors such as founders’ social capital and prior experience (Kraut & Fiore, 2014), member heterogeneity and engagement patterns (Raban et al., 2010), strong social bonds or shared identity (Chiu et al., 2006; Ridings & Gefen, 2004), and the ways in which technology is used by group organizers. Conversely, the absence of formal leadership, rules, norms, and procedures has been found to negatively affect community continuity (Torres, 2020).

When it comes to hyperlocal groups, their performance may need to be reassessed. Measures such as membership growth and retention remain important. For example, López et al. (2017) identified attraction (the ability to draw new members), retention (the ability to keep existing members), and the ability to generate content and impact as key indicators of the sustainability of hyperlocal information systems. These measures align with performance indicators commonly used in online community research and organizational theory (Butler, 2001; Hannan & Freeman, 1977). In addition, greater emphasis may need to be placed on offline engagement, particularly the frequency, scale, and participation levels of in-person events. Unlike online communities that prioritize content generation and digital interaction, the success of hyperlocal groups is reflected in the level of participation in offline gatherings (J. Han et al., 2012), which has been shown to sustain ongoing contributions and engagement within the community (Schwitter, 2024). This includes not only metrics such as the number of offline events and attendance rates, but also the relational characteristics between group organizers and participants that shape long-term sustainability (Chen et al., 2019; J. Lu et al., 2017).

Building on prior research (Cunha et al., 2019; Hannan & Freeman, 1989; López et al., 2017; TeBlunthuis, 2025; Wang et al., 2013), this study focuses on four key measures to assess the performance of hyperlocal groups on EBSNs: (1) entrance rate, the rate at which new groups are created; (2) exit rate, the rate at which groups become idle or disappear; (3) the number of events organized per group; and (4) the number of RSVPs per group. These measures operationalize group sustainability, membership dynamics, and participant engagement, particularly offline engagement, which is central to the success of EBSNs. The first three indicators assess group performance from the organizer perspective, while the fourth reflects performance from the participants’ perspective. Hyperlocal groups may thrive when both organizer- and participant-side dimensions of performance are in good standing. We note that these measures are intended to capture category-level (or niche-level) performance, as this study focuses on the ecological dynamics of hyperlocal groups rather than the outcomes of individual groups.

Technology Adoption and Categories

The success of hyperlocal groups is closely tied to individuals’ adoption of EBSN platforms, as participation in local gatherings through these platforms implies technology adoption. From that perspective, technology adoption research provides valuable insights into the role of platform-driven categories. Research on the Technology Acceptance Model (TAM) and its variants highlights two key factors influencing people’s technology adoption: perceived usefulness and perceived ease of use (Davis, 1989; Venkatesh & Davis, 2000). Studies applying the Technology Acceptance Model (TAM) to knowledge management systems (Money & Turner, 2004), mobile applications (Mohamed et al., 2011; Rahaman et al., 2021), and sustainability-related technologies (Ajina et al., 2024; Wei et al., 2025) consistently find that positive perceptions of system utility and usability increase both initial uptake and long-term engagement. This line of research has been extended to demonstrate how communities of practice and their associated technologies influence organizational sustainability (Abdulmuhsin et al., 2026), how the integration of social support mechanisms fosters the sustainable use of digital learning technologies (Alshammari & Alkhwaldi, 2025), and how Industry 4.0 technologies contribute to advancing knowledge management practices in emerging economies (Al-Sartawi et al., 2023).

Similarly, research on the diffusion of innovations (Rogers et al., 2014) and the social construction of technology (Pinch & Bijker, 1984) highlight institutional factors that shape technology adoption within broader social, cultural, and institutional contexts. Users often do not adopt technologies in isolation; rather, perceptions of legitimacy, peer practices, and local norms play significant roles in shaping whether digital platforms are embraced or abandoned. For instance, studies of emerging peer-to-peer and sharing platforms show that trust, system reputation, and local embeddedness strongly influence adoption trajectories (Belk, 2014; Botsman & Foley, 2010). These insights underscore that technology adoption is not merely a matter of individual choice but also reflects collective meaning-making within broader social and cultural institutions.

Local group categories provided by EBSNs are among the key features that determine system utility and usability. Categories shape system utility because they are heavily used by filtering and search algorithms when groups and events are displayed in response to users’ search queries or topic preferences selected during sign-up. They are also closely related to the platform’s usability, as users’ perceptions of how established, legitimate, and popular these categories are for particular types of groups or events directly affect the ease of using the system when searching for or navigating local gatherings. Thus, studying the ecological impact of categories is also crucial from a technology adoption perspective, as it directly affects platform users’ engagement and retention.

Ecological Factors in Hyperlocal Groups

Ecological features have received less attention in social computing research compared to organizational factors. Understanding their impact is crucial for hyperlocal system designers, as it can inform the design of group categories on EBSNs. Without a strategic approach to designing topics and categories, hyperlocal systems risk unintended declines in the number of events and participation—a dynamic well documented in organizational ecology theory (Hannan & Freeman, 1977). In the following, we review several key ecological characteristics studied in organizational studies that lead to our research questions.

Niche Density

In organizational ecology, a niche refers to a segment of the environment defined by the resource requirements and productive capacities of organizations within a population (Baum & Singh, 1994). Niche density, often measured by the number of organizations occupying a given niche, is a critical ecological feature that influences organizational survival through two processes: legitimation and competition (Hannan & Freeman, 1977). Legitimation refers to the recognition of an organizational form as appropriate and valuable, which increases access to resources and enhances survival chances. Competition arises when organizations within the same niche vie for limited resources, reducing their likelihood of survival as density grows (Lai, 2014; Nickel & Fuentes, 2004; Wang et al., 2013).

This density-based dynamic has been observed in online communities, where they compete for users’ attention and participation. Users’ time and contributions are finite; participation in one community often comes at the expense of another, affecting group resilience (Ransbotham & Kane, 2011; Zhu, Kraut, & Kittur, 2014). Also, membership overlap between two or more communities can increase competitive pressures and hinder growth (J. H. P. Hsu et al., 2023; Wang et al., 2013). Others report that moderate levels of overlap can enhance community survival, particularly in younger groups (Zhu, Kraut, & Kittur, 2014; Zhu, Chen et al., 2014). Recent work suggests that online communities function as “nested niches,” where individuals engage with multiple groups to curate their experiences, shaping patterns of retention and satisfaction (Hwang & Foote, 2021).

When it comes to hyperlocal groups, these processes may operate differently than in traditional organizations. On the participant side, participation typically requires relatively low commitment and cost (often free), allowing users to join multiple groups without significant resource investment. As a result, legitimation is likely to be less tied to resource scarcity and more influenced by perceived relevance, social endorsement, and engagement patterns (J. Chen et al., 2019; Manikandan et al., 2024; Ransbotham & Kane, 2011). Additionally, competition may be shaped more by individuals’ attention to and participation in events, both online and offline (Hwang & Foote, 2021; Zhu, Kraut, & Kittur, 2014).

When it comes to membership overlap, group organizers must select a single predefined category, creating non-overlapping label spaces (E. Pontikes & Hannan, 2014). In other words, while platform users can sign up for multiple groups at the organizational level, EBSN groups cannot span multiple niches at the categorical level. This unique structure makes it difficult to understand how competitive dynamics unfold and how they are shaped by rigid categorical boundaries and localized interests. These characteristics underscore the need to adapt the original theory to reflect the hybrid, double-sided nature of hyperlocal communities. Thus, RQ1 examines how density-based features, shaped by platform-defined categories, influence hyperlocal group performance. Specifically, we investigate whether niche density promotes legitimation effects or intensifies competition, and how these mechanisms are associated with groups’ entry into, exit from, and activity on EBSN platforms.

RQ1: How are density-based ecological features associated with hyperlocal group performance on EBSNs?

Category Fuzziness

Categories are cognitive tools that help individuals make sense of complex realities, guiding everyday decisions such as choosing between French or Indian restaurants, selecting wines (e.g., Merlot vs. Pinot Noir), or identifying social identities (e.g., hipster vs. normcore). Yet, these boundaries are often fluid and overlapping, with entities displaying varying degrees of affiliation across multiple categories. This inherent ambiguity (i.e., category fuzziness) complicates classification, influences individual and collective decision-making, and shapes the social and cognitive dynamics that underpin behavior. From an organizational ecology perspective, categories define organizational membership by grouping social entities or actors with similar resource requirements, productive capacities, and audience expectations (Baum & Singh, 1994; Freeman & Hannan, 1983; Mervis & Rosch, 1981). In reality, many organizations operate within fuzzy categories, where membership boundaries are ambiguous and overlap across multiple domains (Hannan et al., 2007; Navis & Glynn, 2010; Romanelli & Khessina, 2005).

Category fuzziness affects market participants such as producers. Producers that span multiple categories (or market niches) may be perceived as less legitimate, lacking clear specialization (Zuckerman, 1999). They also differ in their niche strategies: specialists fulfill a narrow set of resource needs and align closely with audience expectations, whereas generalists address a broader range of needs to appeal to a wider audience but often fit less precisely within any single category (Hannan & Freeman, 1977). Fuzzier positioning, therefore, offers greater flexibility but makes it more difficult to maintain a coherent identity and clear branding.

On the flip side, audience perception plays a crucial role in mediating the effects of fuzziness. Studies have found that category ambiguity can deter stakeholder support across various domains. For example, Wry and Lounsbury (2013) show that firms filing patents across multiple technology classes are less likely to receive venture capital funding, suggesting that blurry category boundaries reduce perceived value. Similarly, films and auctions that span multiple categories experience lower success rates (G. Hsu et al., 2009); restaurants with ambiguous category labels receive poorer reviews (Kovacs & Hannan, 2010); and wine producers with weak categorical contrast attract less favorable audience attention (Negro et al., 2010). Firms with ambiguous identities are also more likely to receive lower credit ratings (Ruef & Patterson, 2009). Together, these findings underscore a consistent pattern: categorical clarity supports legitimacy and success across diverse organizational contexts.

However, not all category ambiguity leads to negative outcomes. When organizations strategically bridge related categories or introduce new category labels to clarify their identity, they may regain legitimacy and improve their chances of survival (Chliova et al., 2020; Durand & Paolella, 2013; E. G. Pontikes, 2022). Such “category innovations” allow firms to navigate ambiguity by reclassifying themselves in ways that are more intelligible and appealing to stakeholders (Durand & Paolella, 2013; E. G. Pontikes, 2022).

In the context of hyperlocal groups, the characteristics of niches and their formation may differ significantly from those in other organizational settings. Instead of emerging organically through member interactions, niche structures are largely imposed by platform owners. In EBSNs, group organizers must select a single, predefined category, which shapes niche fuzziness through three factors: (1) the intrinsic breadth of the topic (e.g., “sports” being fuzzier than “photography”); (2) platform owners’ strategic or ad hoc category design choices; and (3) group organizers’ selection behavior within these fixed options.

Unlike in traditional niche theory, where specialist and generalist organizations naturally differentiate along a resource continuum, hyperlocal groups are constrained by externally defined categorical boundaries. As a result, niche fuzziness reflects not only group identity and activities but also the structural constraints of the platform itself. This creates a distinct form of category fuzziness–one shaped more by platform architecture than by strategic positioning. By linking theoretical insights on category fuzziness and legitimacy from organizational ecology to hyperlocal group dynamics, we examine how these mechanisms influence observable outcomes such as participation, retention, and the prevalence of event activity. This leads to our second research question:

RQ2: How is category fuzziness associated with hyperlocal group performance on EBSNs?

Hypotheses

Building on theories of niche density, legitimation, competition, and category fuzziness, we propose hypotheses concerning how ecological features are associated with hyperlocal group performance on EBSNs.

RQ1: Density Effects: Legitimation and Competition

Prior research in organizational ecology shows that increasing niche density triggers two opposing processes: legitimation, which enhances the attractiveness and perceived legitimacy of a niche, and competition, which intensifies struggles for limited resources and reduces the survival rate of entities (Hannan & Freeman, 1977; Wang et al., 2013). In the context of hyperlocal groups, although competitive structures may differ from those in formal organizations, groups still compete for a finite pool of local participants. Greater legitimation may encourage new group formation by signaling that a category is credible, active, and well established. However, as niche density increases, intensified competition within a crowded category may discourage new entrants by reducing the perceived opportunity for success (Nickel & Fuentes, 2004). We therefore hypothesize:

H1-1: Higher legitimation within a category is associated with an increase in the number of new groups, whereas higher competition is associated with a decrease in new entrants.

At the same time, the literature shows a general pattern in which legitimation of a niche improves member retention, leading to a reduced exit rate from the market (Cattani et al., 2008). Consequently, higher density may decrease group exit rates. Conversely, intensified competition for limited local resources heightens the challenges of attracting and retaining members, thereby increasing the likelihood of group exits. Thus, we hypothesize:

H1-2: Higher legitimation within a category is associated with a decrease in group exit rates, whereas higher competition within a category is associated with an increase in exit rates.

Because the primary purpose of hyperlocal groups on EBSNs is to facilitate offline gatherings, group performance is better measured by the number of in-person events than by the amount of online content or virtual interaction (J. Han et al., 2012). Greater legitimation may encourage more frequent offline events by signaling category vitality, while heightened competition may constrain groups’ ability to sustain such activities since a crowded category divides the finite pool of local participants and organizer attention, making it harder for any single group to fill and justify frequent events. Thus, we hypothesize:

H1-3: Higher legitimation within a category is associated with an increase in the average number of events per group, whereas higher competition within a category is associated with a decrease in event frequency.

From the participant’s perspective, direct competition among groups within a category is less visible. Instead, participants respond to the perceived vibrancy and relevance of the category, both of which are shaped by legitimation processes. Higher legitimation may signal a more active and credible category, potentially leading to greater user participation in events. Conversely, greater competition may undermine event participation by dispersing attention and reducing perceived value. Thus, we hypothesize:

H1-4: Higher legitimation within a category is associated with an increase in the number of RSVPs per event, whereas higher competition within a category is associated with a decrease.

RQ2: Fuzziness Effects: Identity and Retention

Category fuzziness, defined as the ambiguity or breadth of a category’s identity, can shape both organizer decisions and participant engagement in hyperlocal groups. From the organizer’s perspective, fuzzy categories may weaken a group’s perceived distinctiveness and make it more difficult to communicate its value proposition. This ambiguity can deter the formation of new groups by making it hard for organizers to position themselves within the niche (Freeman & Hannan, 1983; E. G. Pontikes, 2022; Wry & Lounsbury, 2013) and may also increase group exits. Moreover, unclear category identities may reduce organizers’ motivation to maintain active programming, leading to a decline in event frequency. In the ecology of hyperlocal groups, category fuzziness reflects varying degrees of identity clarity, which may either attract or discourage group organizers’ entry and persistence. We therefore hypothesize the following:

H2-1: Higher category fuzziness is associated with a lower number of new group entrants.

H2-2: Higher category fuzziness is associated with an increase in group exit rates.

H2-3: Higher category fuzziness is associated with a decrease in the average number of events organized per group.

From the participant’s perspective, the crispness of a category’s identity may enhance its attractiveness, whereas category fuzziness may reduce the clarity and appeal of group identities. When categories are too broad or ill-defined, individuals may find it difficult to assess whether a group aligns with their interests, which can lower engagement in events. Therefore, we hypothesize:

H2-4: Higher category fuzziness is associated with a decrease in the average number of in-person participation (RSVPs) per event.

Methodology

This section provides the details of our empirical study using Meetup data. We first provide a brief overview of the data, followed by a description of our feature extraction process. This process generates the variables that are used in our regression study using Linear Mixed-Effects Models. A detailed list of the notations that we use is provided in Table 1.

Table 1.

Notations.

Name	Notation	Description
Group	g	Online community for organizing local events
Category	ci	i^th of 33 meetup categories
Tag	tj	j^th of n user-defined tags
City	s	One of 14 large US cities
Month	m	One of 12 months of 2017
Co-occurrence matrix	W m,s	Matrix whose elements wij contain the number of groups that belong in category ci and were tagged with tag tj, in city s during month m
HHI	HHI(ci)	Herfindahl-Hirschman Index of the ith category, given by Equation 1.
Legitimation	Di,s,m	Number of groups in category ci at city s during month m.
Competition	D2,i,s,m	Square of number of groups in ci at city s during month m.
Fuzziness	Fi,s,m	Metric of fuzziness of ci at city s during month m. We use the Gini-Simpson index: 1 − HHI(ci).
Avg events	Vi,s,m	Average number of events per group in ci at city s during month m
Avg RSVPs	Ri,s,m	Average positive responses per event in category ci at city s during month m
Active groups	Gi,s,m	Total number of active groups (i.e., groups that have not stopped creating events) within category ci, at city s during month m.
New groups	G+i,s,m	Number of new groups created within category ci, at city s during month m.
Idle groups	G−i,s,m	Number of groups within category ci, at city s, that stopped creating events (i.e., became idle) at month m.
Entry rate	Ei,s,m	Fraction of new groups created within ci at city s during month m, over the number of active groups in ci the previous month.
Exit rate	Xi,s,m	Fraction of groups that became idle in ci, at city s during month m, over the number of active groups in ci the previous month.

Data

We use data from Meetup.com, an online EBSN platform that allows people with shared interests to form groups and organize offline meetings (Vaughn, 2015). The data cover activities from December 2016 through August 2018 across 14 major U.S. cities. Because these data predate more recent global developments, including the SARS-CoV-2 pandemic, they should be interpreted as reflecting historical patterns of EBSN engagement during the late 2010s and should not be assumed to generalize to periods marked by major disruptions to offline activity. These cities were selected based on their high levels of tech start-up activity. We assume that a strong presence of technology-driven culture in these cities increases residents’ likelihood of adopting and participating in EBSNs such as Meetup.com, so to ensure the credibility of EBSN use among their populations. The selected cities are: Austin (TX), Baltimore (MD), Boston (MA), Denver (CO), Durham (NC), Los Angeles (CA), New York City (NY), Philadelphia (PA), Pittsburgh (PA), Raleigh (NC), San Diego (CA), San Francisco (CA), San Jose (CA), and Washington, D.C. This means that the study samples are not representative of all U.S. cities; rather, they are purposive samples selected based on technology-focused cultures to ensure sufficient volume and diversity of Meetup-based activities within each city.

The study focuses on data from January 1 to December 31, 2017, during which approximately 2.3 million events were recorded across the selected cities—an average of 450 events per city per day. Data from December 2016 were used to identify groups that were active just prior to 2017, helping to capture newly formed groups. Data in 2018 were used to determine whether groups became idle: groups that held no events during this 8-month follow-up period (from January 1 to August 31, 2018) were considered exited. This approach mirrors prior work on online community life cycles, such as Zhu, Kraut, and Kittur (2014), who used a shorter 3-month window to identify group death on Wikia.

The data were obtained in JSON format using Meetup.com’s public API and include detailed records of groups and their associated events in each city. Attribution is given to Bending Spoons, the current owner of Meetup.com as of 2024, as the data proprietor. All data used in this study were derived from publicly accessible information on the platform, and their use complies with Meetup.com’s Terms of Service. Additionally, the first author received verbal approval from Meetup to use the data for research purposes. No personal or private user information was accessed.

For each group, we retrieve its unique id, name, city, creation date, member count, description (provided by the organizer), assigned category, and a list of user-generated tags that describe the group. We track membership changes for each group every month. For events, information on their group id, category, date, location, description, and RSVPs are retrieved. Each group is associated with 1 out of 33 predefined categories, such as “fitness” or “technology.” These categories are predefined by Meetup.com. We note that Meetup.com has updated its category system since the time of our data collection. Among the 33 categories, those with the highest number of events were “career/business” (273,050 events), “socializing” (217,065 events), “technology” (199,638 events), and “sports/recreation” (160,619 events). Let C = {c₁, c₂, . . ., c₃₃} be the set of meetup categories.

Each group is also associated with one or more user-defined “tags.” These tags can be added by group organizers, and they are not predefined. For instance, as of 2025, the categories “dancing” and “music” each include more than 50 user-generated tags and share overlapping terms such as “party people” and “creators.” Similarly, the categories “hobbies & passions” and “art & culture” include over 110 and 40 tags respectively, with shared tags like “book lovers” and “art-preneurs.” Other categories, such as “travel & outdoor” and “identity & language,” share common tags such as “travels” and “hikers.” Let T = {t₁, t₁,…, t_n} be the full set of n unique tags that users have added to groups.

Variables

To extract organizational and ecological features from the data, which represent the fuzziness of group categories in each city over time, we use metrics based on the co-occurrence of tags and categories within each group. Specifically, we compute the temporal sequence of category-tag co-occurrence matrices for each city, on a monthly basis. Let W_m,s = [w_ij] be a 33×n matrix, where each element w_ij represents the number of co-occurrences of the meetup category c_i with tag t_j, during month m and at city s. That is, w_ij is equal to the number of groups that existed in a particular month and city, and were associated with category c_i, and also tagged by users with the tag t_i. Each such local matrix was used to calculate the Herfindahl-Hirschman Index (HHI; Hirschman, 1980) for each group category, in each city over time as a measure of a category’s fuzziness. Let $W_{i} = (w_{i 1}, w_{i 2}, \dots, w_{in})$ be the $i^{th}$ row of matrix $W_{m, s}$ , and $sum (W_{i}) = \sum_{j = 1}^{n} w_{ij}$ be the sum of its elements. Let $p_{ij} = w_{ij} / W_{i}$ be the fraction of co-occurrences that category $c_{i}$ has with tag $t_{j}$ . The HHI score of a category $c_{i}$ is given by:

\begin{matrix} HHI (c_{i}) = \sum_{j = 1}^{n} p_{ij}^{2} \end{matrix}

(1)

The higher the value of HHI (i.e., range 1/n to 1) the less ’fuzzy’ is the category. For our regression models, we use the Gini-Simpson index, which is given by 1 − HHI(c_i), where HHI is given by Equation 1. This measure reflects category fuzziness at the city-month level and, when aggregated, enables analysis of variation over time and across cities.

The ecological features used in this study to measure the effects of group success are as follows:

Legitimation (D_i,s,m), the number of groups in category c_i and city s during month m,

Competition (D_i,s,m² ), the square of the number of groups in category c_i and city s during m, and

Fuzziness (F_i,s,m) of a category c_i in city s during month m, measured by the Gini-Simpson index.

We examine how the aforementioned organizational and ecological features influence group success indicators. The success indicators are measured for each city s and at each time period (month) m. The following success indicators used:

R_i,s,m, the average number of RSVPs per event in category c_i,

V_i,s,m, the average number of events per group in category c_i,

E_i,s,m, the monthly entry rate of new groups in category c_i, and

X_i,s,m, the exit rate of groups in category c_i.

It is important to note that among these four success measures, the first (RSVPs) relates to audience success as it reflects users’ responsiveness, while the other three measures are related to producers success, referring to the creators of the groups.

For the entry rate, we consider the ratio of new groups created in a city $i, s, m$ during a specific month (G⁺ ) to the number of active groups that existed in that category for the previous month ( $G_{i, s, (m - 1)}$ ):

E_{i, s, m} = \frac{G_{i, s, m}^{+}}{G_{i, s, (m - 1)}}

Since groups are never deleted from the platform, for exit rates we consider the ratio of groups that became idle in a month (i.e., they stopped organizing events until the end of the collection period) to the number of active groups that existed in that category for the previous month ( $G_{i, s, m}^{-}$ ):

X_{i, s, m} = \frac{G_{i, s, m}^{-}}{G_{i, s, (m - 1)}}

All variables were scaled for analysis. The means and standard deviations of raw variables are presented in Table 2. This table shows that the Avg RSVPs variable has a mean of 66.102 and a standard deviation of 107.331, indicating that some events receive significantly more RSVPs than others, with a relatively high degree of variability. The Avg events variable shows an average of 5.584 events per group, with a standard deviation of 4.409, meaning the number of events organized by groups varies substantially. The Entry rate has a mean value of 0.044 and a standard deviation of 0.079, suggesting that, on average, only 4.4% of groups are newly created each month, but this rate can vary greatly from month to month. For the Exit rate, the average is 0.020 and the standard deviation is 0.034, indicating that, on average, 2.0% of groups exit each month, but the exit rate varies across months. The Legitimation variable has a mean of 78.954 and a large standard deviation of 121.058, suggesting significant variation in how legitimized the categories are across different cities and times. The Competition variable has an average of 20,888.752, but the standard deviation is very large at 88,609.108, reflecting substantial variation in the level of competition within different categories. Finally, the Fuzziness variable has a mean value of 0.977 with a low standard deviation of 0.024, indicating that, overall, categories tend to have high fuzziness on average, with minimal variation across time and cities.

Table 2.

Description of Variables.

Name	Mean	S.D.
Avg RSVPs	66.102	107.331
Avg events	5.584	4.409
Entry rate	0.044	0.079
Exit rate	0.020	0.034
Legitimation	78.954	121.058
Competition	20,888.752	88,609.108
Fuzziness	0.977	0.024

Lagged Values and Smoothing

It is sometimes the case that a dependent target variable is not only affected by the current values of its predictors, but also their past values (Hyndman & Athanasopoulos, 2018). To account for this effect, we incorporate lagged values of the predictors in our models. These lagged value are multiplied by their corresponding coefficients and treated as separate predictor variables in the model. In our experiments, we examine the effects of the predictors (i.e., legitimation, competition, and fuzziness) not only in the current month but also in the preceding months. In Table 3, “Lag 0” refers to a model that was trained using only the current values of the predictors. In other words, the success metrics (i.e., entry rate, exit rate, average events, and average RSVPs) in month m depend only on the legitimation and competition (and fuzziness for the fuzziness-based model) measures observed in month m. On the other hand, “Lag 1” models include predictors for the current values of legitimation, competition, and fuzziness at month m, as well as the values of these variables at month m− 1. Similarly, “Lag 2” models incorporate predictors for legitimation, competition, and fuzziness at months m, m−1, and m−2, while “Lag 3” models further include predictors for 3 months prior. This lagged threshold was partly influenced by prior work by Trinh et al. (2020), which showed that the average activity between successive events on Meetup.com typically spans 3 months.

Table 3.

Models With Lagged Predictors.

	RQ1: Baseline Model								RQ2: Fuzziness-based model
	Lag 0		Lag 1		Lag 2		Lag 3		Lag: 0		Lag: 1		Lag: 2		Lag:3
Entry Rate	Est	p	Est	p	Est	p	Est	p	Est.	p	Est.	p	Est.	p	Est.	p
Fuzziness									0.010	0.609	−0.068	0.001***	−0.099	0.000***	−0.023	0.136
Legitimacy	−0.001	0.969	−0.057	0.128	−0.030	0.365	−0.023	0.520	−0.003	0.930	−0.048	0.209	−0.019	0.574	−0.016	0.658
Competition	0.002	0.951	0.044	0.181	0.023	0.417	0.018	0.561	0.003	0.922	0.039	0.247	0.017	0.567	0.014	0.663
Exit Rate	Est	p	Est	p	Est	p	Est	p	Est.	p	Est.	p	Est.	p	Est.	p
Fuzziness									0.016	0.420	0.033	0.110	0.041	0.056^†	0.060	0.006**
Legitimacy	−0.075	0.070^†	−0.036	0.416	−0.024	0.612	−0.020	0.692	−0.079	0.059^†	−0.045	0.316	−0.036	0.452	−0.041	0.427
Competition	0.042	0.221	0.014	0.706	0.004	0.919	0.000	0.999	0.044	0.199	0.020	0.602	0.012	0.772	0.014	0.758
Avg Events	Est	p	Est	p	Est	p	Est	p	Est.	p	Est.	p	Est.	p	Est.	p
Fuzziness									−0.058	0.001***	−0.011	0.538	0.000	0.982	0.036	0.051^†
Legitimacy	0.081^†	0.081^†	0.093	0.062^†	0.073	0.148	0.062	0.247	0.082	0.077^†	0.094	0.061^†	0.073	0.148	0.060	0.258
Competition	−0.026	0.428	−0.030	0.385	−0.020	0.579	−0.013	0.740	-0.026	0.413	−0.031	0.382	−0.020	0.580	−0.012	0.758
Avg RSVPs	Est	p	Est	p	Est	p	Est	p	Est.	p	Est.	p	Est.	p	Est.	p
Fuzziness									−0.108	0.000***	−0.055	0.008**	−0.033	0.102	0.046	0.028*
Legitimacy	0.166	0.001***	0.178	0.001***	0.160	0.003**	0.157	0.007**	0.169	0.001***	0.179	0.001***	0.161	0.003**	0.155	0.007**
Competition	−0.062	0.082^†	−0.069	0.074^†	−0.059	0.135	−0.058	0.177	−0.064	0.071^†	−0.070	0.070^†	−0.059	0.130	−0.056	0.185

†

p < .1. *p < .05. **p < .01. ***p < .001.

A moving average (MA) is a smoothing technique traditionally used as part of time series decomposition (Hyndman & Athanasopoulos, 2018). It helps estimate the overall trend-cycle of a time series by reducing the effects of short-term fluctuations. MA is calculated as the average of values within a moving window of a time series. For example, the 2-MA of legitimation would be:

\begin{matrix} {\frac{1}{2} (D_{i, s, 1} + D_{i, s, 0}), \frac{1}{2} (D_{i, s, 2} + D_{i, s, 1}), \\ \dots, \frac{1}{2} (D_{i, s, 12} + D_{i, s, 11})} \end{matrix}

In our experiments, we apply moving average smoothing to the predictor variables to study the combined effect of the most recent months of a predictor on the target variable. We include only current and past values (not future ones) in the predictors’ moving averages. In Table 4, “Lag 0” refers to the model trained using the current values of the predictors (legitimation, competition, and fuzziness). “Lag 1” uses the 2-MA of these predictors, which is the average value of each predictor at month m and month m−1. Similarly, “Lag 2” and “Lag 3” in Table 4 correspond to the use of 3-MA and 4-MA predictor variables, respectively.

Table 4.

Models With Smoothing.

	RQ1: Baseline model								RQ2: Fuzziness-based model
	Lag: 0		Lag: 1		Lag: 2		Lag: 3		Lag: 0		Lag: 1		Lag: 2		Lag: 3
Entry rate	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p
Fuzziness									0.010	0.609	−0.024	0.262	−0.072	0.000***	−0.078	0.001***
Legitimation	−0.001	0.969	−0.026	0.489	−0.018	0.577	−0.014	0.693	−0.003	0.930	−0.021	0.569	−0.008	0.810	0.001	0.972
Competition	0.002	0.951	0.020	0.528	0.014	0.603	0.011	0.713	0.003	0.922	0.018	0.584	0.008	0.767	0.002	0.953
Exit rate	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p
Fuzziness									0.016	0.420	0.027	0.244	0.050	0.057^†	0.072	0.019*
Legitimation	−0.075	0.070^†	−0.058	0.190	−0.040	0.395	−0.029	0.555	−0.079	0.059^†	−0.065	0.147	−0.054	0.255	−0.052	0.303
Competition	0.042	0.221	0.029	0.424	0.015	0.695	0.007	0.870	0.044	0.199	0.034	0.364	0.024	0.539	0.022	0.610
Avg events	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p
Fuzziness									−0.058	0.001***	−0.041	0.049*	−0.068	0.006**	−0.050	0.106
Legitimation	0.081	0.081^†	0.089	0.073^†	0.073	0.144	0.058	0.265	0.082	0.077^†	0.090	0.070^†	0.075	0.133	0.060	0.250
Competition	−0.026	0.428	−0.028	0.416	−0.020	0.561	−0.012	0.750	−0.026	0.413	−0.029	0.405	−0.021	0.535	−0.013	0.726
Avg RSVPs	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p	Est.	p
Fuzziness									−0.108	0.000***	−0.107	0.000***	−0.163	0.000***	−0.169	0.000***
Legitimation	0.166	0.001***	0.174	0.001***	0.156	0.004**	0.149	0.008 **	0.169	0.001***	0.177	0.001***	0.163	0.002**	0.159	0.004**
Competition	−0.062	0.082^†	−0.067	0.080^†	−0.056	0.138	−0.053	0.190	−0.064	0.071^†	−0.069	0.069^†	−0.061	0.104	−0.059	0.14

†

p < .1.*p < .05. **p < .01. ***p < .001.

Linear Mixed-Effects Models

To assess the relationship between group performance measures (i.e., average RSVPs, average events, entry rate, and exit rate) and the independent variables (i.e., legitimation, competition, fuzziness), we fit various linear mixed-effects models (LMMs) to our data. LMMs are an extension of traditional linear models (LMs) and allow for the incorporation of both fixed and random effects. A fixed effect is a parameter that remains constant across different groups. For example, assume there is a true regression line in the population that defines the linear relationship between two continuous variables, with β as its slope. We can estimate the slope, β^, from the data and assume that this estimate is representative for all instances of the model. In contrast, a random effect is a parameter that varies randomly. For instance, we could assume that β follows a normal distribution with mean µ and standard deviation σ. In traditional linear regression, the data are treated as random variables, while the parameters of the model are considered fixed effects. In mixed-effects models, we assume that both the data and the parameters are random variables at lower levels, but fixed at the highest level (e.g., assuming some overall population mean).

LMMs are particularly useful when the assumption of independence of observations is inappropriate (Faraway, 2006). Such situations arise in our study, where our data are grouped by fully crossing two factors: category and city. Each group in our data is sampled from one of 33 categories. The variability in a group performance measure can be interpreted as occurring either within a category or between categories. Therefore, observations are not independent because groups within the same category share more common characteristics than those in different categories. As a result, the effects of category were treated as random effects. Similarly, the effects of city were also treated as random, as the groups represent samples of many other hyperlocal groups organized within those cities. Modeling the effects of category and city as random effects allows us to draw inferences about the wider population of groups formed on EBSNs. For our work, the general framework of the LMMs is as follows:

y_{ij} = b_{0} + \sum_{k = 1}^{p} b_{k} x_{ijk} + v_{i 0} + \sum_{k = 1}^{q} v_{ik} z_{ijk} + e_{ij}

(2)

where,

y_ij∈ℝ is the response for the j-th measurement of the i-th observation.

b₀∈ℝ is the fixed intercept for the regression model.

b_k∈ℝ is the fixed slope for the k-th predictor.

x_ijk∈ℝ is the j-th measurement of the k-th fixed predictor for the i-th observation.

v_i0∼ N(0, σ²₀) is the random intercept for the i-th observation.

v_i0∼ N(0, σ²_k) is the random slope for the k-th predictor of the i-th observation.

z_ij_k∈ℝ is j-th measurement of k-th random predictor for the i-th observation.

e_ij∼ N(0, σ²_e) is the Gaussian error term.

We fit two sets of linear mixed-effects models (LMMs) to our data: the baseline model and the fuzziness-based model. For both sets of LMMs, the response variable for each of the four models is a group performance measure. The independent variables for the baseline models include legitimation, competition, and month (fixed effects), along with category and city (random effects). The fuzziness-based models include an additional independent variable, fuzziness, which is derived from the Gini-Simpson index. The dataset contains 5,544 observations across categories and cities. The LMMs are fit using the lmer function in the lme4 package in R (Bates et al., 2015).

Results

Table 3 presents the regression results for lagged predictors, while Table 4 reports those for moving average smoothing. Significance levels are color-coded in the tables, with blue indicating a positive relationship and red indicating a negative one.

RQ1: Effects of Legitimation and Competition

H1-1: Entry Rate

The results for hypothesis H1-1, which tests whether legitimation and competition are associated with the entry rate of new groups, were not statistically significant in either model, with or without smoothing. These findings do not reject the null hypothesis of H1-1.

H1-2: Exit Rate

The results for H1-2 are not statistically significant at the 95% confidence level. When the time lag is zero, there is a marginally negative relationship between legitimation and the exit rate (p = .070) in both models, with and without smoothing. The significance of the model disappears once time lags are introduced. While the direction of the estimated coefficients supports the hypothesized relationship (i.e., legitimation is negatively associated with exit rates), the overall results, including the estimated effect of competition on exit rate, are either marginal or statistically insignificant. Taken together, these findings do not reject the null hypothesis of H1-2.

H1-3: Average Number of Events Per Group

There are marginally positive associations between legitimation and the average number of events per group when no time lag is applied (p = .081) and at a 1-month lag (p = .062) in both models, with and without smoothing. While the direction of the relationship supports H1-3, the effect is not statistically significant and weakens further with longer time lags. Regarding the competition component of H1-3, the estimated effect of competition on event frequency is not statistically significant in any of the models. Overall, these results do not reject the null hypothesis of H1-3.

H1-4: Average Number of RSVPs Per Event

The results indicate that legitimation is positively associated with the average number of RSVPs per event across all lag lengths (p≤ .008), providing support for the first part of H1-4 and rejecting the null hypothesis. The estimated coefficients in both models are similar. The effect size for legitimation decreases slightly as the time lag increases (while not linear), with the largest effect observed at a 1-month lag. In contrast, competition is only marginally associated with RSVPs at lag lengths of 0 and 1 month (p = .082 and p = .074, respectively), and therefore does not reject the null hypothesis.

Overall, density-based ecological features (i.e., legitimation and competition within categories) exhibited limited and inconsistent associations with hyperlocal group performance, except for the legitimation effects in the RSVP-based model (H1-4). While neither factor significantly predicted group entry, exit, or event frequency (i.e., producer-side performance), legitimation consistently and significantly increased participant engagement, as measured by the number of RSVPs per event. This suggests that hyperlocal groups operate differently from traditional organizations in the ecology, due to the influence of technological features (e.g., category-based effects) and the two-sided nature of the platform (i.e., online coordination vs. offline gatherings). Also, it indicates that an established category identity may play a more salient role in shaping EBSN users’ participation than in influencing producer-side dynamics, a theme we explore further in the next section.

RQ2: Effects of Category Fuzziness

H2-1: Entry Rate

As shown in the right panel of Tables 3 and 4, category fuzziness is negatively associated with the entry rate of new groups when controlling for density-based features such as legitimation and competition. This effect is statistically significant at 1- to 2-month time lags in the models without smoothing and at 2- to 3-month intervals in the moving average model (p≤ .001). These findings suggest that the effect of fuzziness on group formation is not immediate but may emerge with some delay, possibly reflecting the time it takes for organizers to perceive category ambiguity and adjust their behavior accordingly. Overall, the results provide partial support for H2-1 and reject the null hypothesis.

H2-2: Exit Rate

Controlling for density-based features, category fuzziness is positively associated with group exit rates. This relationship becomes notably stronger as time lags increase, with the effect peaking at a 3-month lag in both models (β = .060; p = .006 and β = .072; p = .019, respectively). Specifically, the estimated coefficients increase and the p-values decrease with longer lags. This pattern suggests a delayed effect, wherein groups operating in fuzzier categories are more likely to become inactive after several months, possibly due to difficulties in maintaining a coherent identity or attracting sustained engagement. The consistent trend across models provides partial support for H2-2 and rejects the null hypothesis.

H2-3: Average Number of Events Per Group

When controlling for density-based features, category fuzziness is negatively associated with the average number of events per group, although the strength of this association depends on the model specification. In the model without smoothing (Table 3), this effect is statistically significant only at lag 0 (β = −.058, p = .001), but it weakens at lag 1 and 2 and turns marginally positive by lag 3 (β = .036, p = .051), suggesting a non-linear effect of time lag.

By contrast, the moving-average model (Table 4) shows a robust negative association that is significant from lag 0 (p = .001) through lag 2 (p = .006) and remains negative, though non-significant, at lag 3 (p = .106). Because the positive reversal at lag 3 appears only in the non-smoothed specification and disappears once predictors are averaged over time, it should be interpreted cautiously: it may reflect adaptation or selection among groups that persist in fuzzier categories, but it may also stem from the instability of individual lag coefficients when highly correlated monthly values are entered jointly. Taken together, the results provide partial support for H2-3 and reject the null hypothesis.

H2-4: Average Number of RSVPs Per Event

When controlling for density-based features, category fuzziness is negatively associated with the average number of RSVPs per event in both the lagged and smoothed models, though the pattern of effects differs across model specifications. In the model without smoothing (Table 3), the effect is strongest at lag 0 (β = −.108, p < .001) and remains significant at lag 1 (β = −.055, p = .008), but weakens at lag 2 (β = −.033, p = .102) and reverses direction at lag 3 (β = .046, p = .028). This suggests that in the short term, fuzzier categories reduce participant engagement, but for those events that continue over time, participation may improve, potentially due to adaptation or selection effects. As with the average-events model, however, this reversal appears only in the non-smoothed specification. The smoothed model remains negative and strengthens across lags. The model with moving average smoothing shows a robust and sustained negative association between fuzziness and event participation across all time lags. The coefficients remain strongly negative and statistically significant from lag 0 through lag 3 (ranging from β = −.108–β = −.169, all p < .001), indicating that when participation is averaged over time, category fuzziness consistently reduces turnout. Moreover, the effect size increases as the time lag lengthens, suggesting a potential cumulative effect of category fuzziness.

These findings offer strong support for H2-4, rejecting the null hypothesis. While the model without smoothing suggests short-term suppression followed by potential recovery, the smoothed model indicates that, on average, fuzzier categories are consistently associated with lower levels of participant engagement. This suggests that it likely reflect the instability of individual lag coefficients rather than a genuine temporal turnaround.

Overall, category fuzziness has a measurable impact on hyperlocal group performance, with its effects most evident in participation-related outcomes. Specifically, fuzzier categories appear to negatively affect producer-side performance (i.e., decreasing new group formation, increasing group exit, and reducing the number of events organized per group). They also negatively affect audience-side performance (i.e., lowering average event participation). These patterns suggest that ambiguity in category identity may discourage engagement from both organizers and participants.

Discussion

This study investigated how ecological features (specifically, density-based dynamics and category fuzziness) shape the performance of hyperlocal groups on an event-based social network. By drawing on organizational ecology and categorization theory, we tested how legitimation, competition, and category fuzziness on EBSNs influence both group-level outcomes (formation, exit, and organizing activity) and participant engagement. Below, we offer implications for theory and platform design.

Legitimation as a Driver for Participation?

In organizational ecology, legitimation has been studied as a factor that enhances organizational survival by attracting new members and reinforcing organizational identity (Hannan & Freeman, 1977). Our findings challenge this assumption in the context of EBSNs. While legitimation did not lead to new group formation, survival, or new event creation, it strongly predicted participant engagement, as measured by the average number of RSVPs per event. This suggests that legitimation functions more as a perceptual signal to participants than as a structural driver for organizers.

This insight extends prior work in social computing and online community research, which has shown that visibility and perceived vibrancy influence member participation. On EBSNs, participant engagement appears to be more responsive to social signals than to resource constraints, emphasizing the behavioral and perceptual nature of legitimacy in digitally mediated, low-barrier environments.

An alternative explanation is that the legitimation of categories on EBSNs may be closely related to search engine performance. In other words, rather than representing a niche within the conceptual space of hyperlocal groups, clearly defined category identities may play an important role in improving search results in response to users’ queries, thus increasing participation. In this case, the crispness of category boundaries may be compounded with search engine performance. Further research is needed to better understand the ecological effects of categories on EBSNs, particularly by comparing them with search engine performance across categories.

Practically, these findings underscore the importance of designing EBSN platforms in ways that enhance the legitimation of categories, either by dynamically adapting them or by allowing group organizers to define them organically. From the perspective of group organizers, when overlapping categories exist on the platform, they may benefit from selecting those that provide a clearer identity to attract participants to their events.

Is Competition on EBSNs Low or Unnecessary?

Contrary to classic ecological models that emphasize resource competition (Hannan & Freeman, 1977), our analysis finds little evidence that competition among groups within a category affects group formation, survival, activity, or participation. Two cautions temper this null. First, competition is operationalized as the squared density term (D²), which is by construction collinear with the linear legitimation term (D); non-significant competition coefficients may therefore partly reflect this collinearity rather than a true absence of competitive pressure. Second, in density-dependence terms, finding a legitimation effect but not a competition effect is equivalent to a density relationship that is approximately linear rather than curvilinear over the observed range, which is consistent with most categories sitting well below any saturation point. We therefore read these results as evidence that crowding has not yet become costly on EBSNs, not that competition is structurally irrelevant. In other words, on EBSNs, the cost of group entry is low (unlike in traditional organizations). Hyperlocal groups on EBSNs may not need to, or may not yet have reached the stage where they need to, compete for resources (i.e., participants), as the majority of these groups are small and interest-based. This suggests that the degree of organizers’ motivation to sustain their groups matters in organizational ecology and that the cost of entry into the market may moderate the effect of competition on their survival.

From the audience side, users can easily join multiple groups on EBSNs without incurring additional costs (Meetup, 2025a). Participation is typically low-commitment, non-exclusive, and cost-free, allowing members to engage with multiple groups simultaneously without significant trade-offs in time or money. In this context, user attention may be abundant enough to sustain overlapping groups, thus reducing competitive pressures. Furthermore, the platform’s architecture, such as fixed categorical boundaries and recommendation mechanisms, shapes how groups are discovered and joined, potentially redistributing participation in ways that weaken direct competition. Taken together, these structural and behavioral features suggest that competition on EBSNs might be mediated more by platform design and underlying algorithms than by resource constraints. This also implies that platform managers may need to focus less on filtering redundant groups and more on supporting diverse expressions of interest, as coexistence of groups does not appear to diminish performance.

Category Fuzziness and Group Identity

Consistent with categorization theory and market sociology (Wry & Lounsbury, 2013; Zuckerman, 1999), our findings indicate that category fuzziness significantly undermines hyperlocal group performance on both the producer and audience sides, though the effects are lag-dependent and uneven across outcomes. Fuzzier categories are associated with lower entry rates, higher exit rates, and fewer events on the producer side (effects that generally emerge with a one- to three-month delay rather than contemporaneously), and with reduced participant engagement on the audience side, which is the most robust and consistent of the four. Together, these findings indicate that unclear group identities deter both organizers and members, even in non-commercial, community-based settings, while this influence tends to unfold over time rather than immediately. Prior research has largely focused on formal organizations or profit-oriented platforms. Our study extends the concept of category fuzziness to informal groups and digital communities with minimal economic stakes, demonstrating that identity clarity remains critical for both organizer commitment and member engagement.

Another noteworthy finding is that, in the models without smoothing, the effect size of category fuzziness weakens and even reverses the direction at longer lags. Because this reversal does not appear once predictors are averaged over time, and because fuzziness varies little across categories (SD = 0.024), it should be read with caution. This could possibly reflect adaptation by persisting groups, but also possibly an artifact of collinearity among the lagged terms. This also suggests that, as noted in the legitimation-based discussion, temporality may moderate the effects of category fuzziness by improving search engine performance as additional data accumulate. Either way, it points to temporality as a factor worth examining directly in future work.

Taken together, our findings highlight the critical role of platform architecture in shaping category fuzziness. On EBSNs such as Meetup, group organizers are constrained by fixed and often overly broad taxonomies, which can hinder the accurate identification of their groups. This structural rigidity in categorization may generate identity ambiguity and influence group viability, potentially through the mediating effects of search engine algorithms that rely on categorical structures.

This suggests that classification systems on platforms operate as ecological forces in their own right, while also being adjusted by actors over time. To mitigate the effects of fuzziness, platform designers could adopt more flexible categorization systems, such as nested labels, multi-category tagging, or adaptive category schemes. Future research should also examine how design features, such as search algorithms and recommendation systems, interact with categorical structures to shape visibility, participation, and group sustainability. Our findings extend ecological theory into digitally mediated, hyperlocal contexts and foreground platform architecture as a novel determinant of group dynamics. Overall, our results inform both theory and design in the study of digitally mediated hyperlocal groups.

Limitations and Future Research

This study offers several strengths. It leverages a large-scale dataset of over 2 million events across 14 technologically vibrant, high-population U.S. cities, providing detailed insight into hyperlocal group dynamics on EBSNs. The combination of organizational ecology and categorization theory provides a novel theoretical lens through which to examine group formation, survival, and participation. By capturing multiple dimensions of group activity and category structure, the study sheds light on mechanisms that are often overlooked in online community research.

The study also develops methodological approaches to measure category fuzziness, in addition to other variables. By computationally modeling the fuzziness based on a large dataset, this study provides useful and precise measurements of category fuzziness, which can be applied to other domains that focus on social categories, social media data, and market niches.

As with other empirical studies, this work has several limitations. The data were collected exclusively from Meetup.com between 2016 and 2018, prior to major global events such as the COVID-19 pandemic, which may have substantially altered patterns of online and offline social engagement since then. As a result, the findings should be interpreted as reflective of historical dynamics before the major health crisis rather than current trends. Our study provides a baseline for understanding the performance of hyperlocal groups, rather than a direct characterization of current ecological patterns.

Moreover, our findings may not be generalizable to other platforms and could be specific to Meetup’s technological configurations, which likely influence group formation, participation, and ecological dynamics. Additionally, digital platforms evolve rapidly, as evidenced by the introduction of a new set of categories on Meetup as of 2025. Accordingly, user engagement and ecological interactions observed during our data collection period may not fully represent the current landscape of hyperlocal communities.

Finally, this study focuses on a set of U.S. cities with highly technology-driven cultures, which may limit the generalizability of the findings to other geographic or cultural contexts. While the dataset captures a rich array of events and group characteristics, unobserved factors, such as changes in platform design, social norms, or local regulations, could also introduce bias into the estimation of group outcomes.

These limitations point to several directions for future research. Studies incorporating multiple EBSNs and more recent datasets would help assess whether the observed patterns persist across platforms and over time. Comparative analyses across geographic and cultural contexts could further illuminate how local factors shape hyperlocal group dynamics. Finally, integrating platform design changes, user demographics, and broader societal shifts could clarify how these elements interact with ecological factors in digitally mediated communities.

Conclusion

This study explored how ecological features, specifically density-based features (i.e., legitimation and competition) and category fuzziness, shape the performance of hyperlocal groups on EBSNs, using a large dataset from Meetup. Our findings provide new perspectives on ecological processes in digitally mediated hyperlocal groups.

We find limited evidence that legitimation and competition influence group formation or exit, diverging from traditional density-dependence models and reflecting key differences between EBSNs and formal organizations. Hyperlocal groups on EBSNs such as Meetup operate with low costs, minimal risk, and few barriers to entry or exit. Unlike firms, these groups can persist even with minimal participation, and their organizers are rarely resource-constrained in the same way. While legitimation does not appear to drive new group creation or reduce group exit, it consistently increases participation on the audience side, underscoring its importance within EBSNs.

In contrast, category fuzziness exerts a more pervasive influence on group outcomes, though one that is often delayed and uneven across measures. Groups in fuzzier categories tend to experience lower entry rates, higher exit rates, reduced event activity on the producer side, and, most robustly, diminished participant engagement on the audience side. These effects are consistent with prior findings in formal organizations, where ambiguous identities undermine stakeholder support and market success. Our study extends these insights to informal, interest-based groups, highlighting that identity clarity remains critical even in digital environments not defined by profit-driven competition. As there is a possibility of compounding effects related to search engine performance, further research is needed to incorporate and examine these mechanisms.

Footnotes

ORCID iDs

Myeong Lee

Ron Mahabir

Jieshu Wang

Tokunbo Fadahunsi

Ethical Considerations

George Mason University’s Institutional Review Board has confirmed that this study is not human subject research (IRBNet #1534672).

Consent to Participate

This study does not require consent, as we have not collected human-subject or privacy-sensitive data.

Author Contributions

Myeong Lee: Writing, Project administration, Investigation, Visualization, Resources, Formal analysis, Conceptualization, Supervision, Methodology, Data curation. Olga Gkountouna: Writing, Investigation, Resources, Formal analysis, Conceptualization, Methodology. Ron Mahabir: Writing, Investigation, Formal analysis, Conceptualization, Methodology. Jieshu Wang: Writing, Editing, Investigation. Tokunbo Fadahunsi: Editing, Investigation.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project was supported by resources provided by the Office of Research Computing at George Mason University (URL: , NSF #2018631) and funded in part by grants from the National Science Foundation (Award #2217706).

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

The data cannot be shared directly by the authors, but the raw data are available through the Meetup APIs.

References

Abdulmuhsin

A. A.

Alkhwaldi

A. F.

Dbesan

A. H.

Tarhini

(2026). Empirical insights on organisational sustainability: Exploring the influence of communities of practice and associated technologies. Sustainability Accounting Management and Policy Journal, 17, 204–232.

Ajina

A. S.

Islam

D. M. Z.

Zamil

A. M. A.

Khan

(2024). Understanding green it adoption: Tam and dual-lens of innovation resistance. Cogent Business & Management, 11(1), 2403646.

Al-Sartawi

A. M.

Abd Wahab

Hussainey

(2023). Global economic revolutions: Big data governance and business analytics for sustainability [Conference session]. Second international conference, ICGER (pp. 27–28). Springer.

Alshammari

S. H.

Alkhwaldi

A. F.

(2025). An integrated approach using social support theory and technology acceptance model to investigate the sustainable use of digital learning technologies. Scientific Reports, 15(1), 342.

Atrey

Zakaria

Balan

Shenoy

(2024). W4-groups: Modeling the who, what, when and where of group behavior via mobility sensing. Proceedings of the ACM on Human-Computer Interaction, 8(CSCW1), 1–29.

Bates

Mächler

Bolker

Walker

(2015). Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software, 67(1), 1–48. https://doi.org/10.18637/jss.v067.i01

Baum

J. A. C.

Singh

J. V.

(1994). Organizational niches and the dynamics of organizational founding. Organization Science, 5(4), 483–501. https://doi.org/10.1287/orsc.5.4.483

Belk

(2014). You are what you can access: Sharing and collaborative consumption online. Journal of Business Research, 67(8), 1595–1600.

Bingham-Hall

Law

(2015). Connected or informed?: Local twitter networking in a london neighbourhood. Big Data & Society, 2(2), 2053951715597457.

10.

Botsman

Foley

(2010). What’s mine is yours. Tantor Media, Incorporated.

11.

Butler

B. S.

(2001). Membership size, communication activity, and sustainability: A resource-based model of online social structures. Information Systems Research, 12(4), 346–362.

12.

Cattani

Ferriani

Negro

Perretti

(2008). The structure of consensus: Network ties, legitimation, and exit rates of U.S. feature film producer organizations. Administrative Science Quarterly, 53(1), 145–182.

13.

Chen

Wang

Beck

Carroll

J. M.

(2019). Beyond leaders and followers: Understanding participation dynamics in event-based social networks. International Journal of Human–Computer Interaction, 35(20), 1892–1905.

14.

Chen

Cheng

Zhang

Guo

(2025). How do information and communication technology platforms shape rural e-governance: The case of Zhao-lou Village on the WeCounty platform. Information Systems Journal, 35(2), 545–576.

15.

Chiu

C. M.

Hsu

M. H.

Wang

E. T. G.

(2006). Understanding knowledge sharing in virtual communities: An integration of social capital and social cognitive theories. Decision Support Systems, 42(3), 1872–1888.

16.

Chliova

Mair

Vernis

(2020). Persistent category ambiguity: The case of social entrepreneurship. Organization Studies, 41, 1019–1042.

17.

Cunha

Jurgens

Tan

Romero

(2019). Are all successful communities alike? characterizing and predicting the success of online communities [Conference session]. The World Wide Web Conference (pp. 318–328).

18.

Daly

Erete

Farzan

Hsieh

Lampe

Lopez

Monroy-Hernandez

Quercia

Schwartz

Voida

(2015). Supporting cities, neighborhoods, and local communities with information and communication technologies [Conference session]. Proceedings of the 18th ACM Conference Companion on Computer Supported Cooperative Work & Social Computing (pp. 277–281).

19.

Davis

F. D.

(1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13, 319–340.

20.

Durand

Paolella

(2013). Category stretching: Reorienting research on categories in strategy, entrepreneurship, and organization theory. Journal of Management Studies, 50(6), 1100–1123.

21.

Faraway

(2006). Extending the linear model with R. Chapman and Hall/CRC.

22.

Franzoni

Sauermann

(2014). Crowd science: The organization of scientific research in open collaborative projects. Research Policy, 43(1), 1–20.

23.

Freeman

Hannan

M. T.

(1983). Niche width and the dynamics of organizational populations. American Journal of Sociology, 88(6), 1116–1145.

24.

Han

Niu

Chin

Wang

Tong

Wang

(2012). How online social network affects offline events: A case study on Douban [Conference session]. 2012 9th International Conference on Ubiquitous Intelligence and Computing and 9th International Conference on Autonomic and Trusted Computing (pp. 752–757). https://doi.org/10.1109/UIC-ATC.2012.149

25.

Han

Shih

P. C.

Rosson

M. B.

Carroll

J. M.

(2014). Enhancing community awareness of and participation in local heritage with a mobile application [Conference session]. Proceedings of the 17th ACM Conference on Computer Supported Cooperative Work & Social Computing (pp. 1144–1155).

26.

Hannan

M. T.

Freeman

(1977). The population ecology of organizations. American Journal of Sociology, 82(5), 929–964.

27.

Hannan

M. T.

Freeman

(1989). Organizational ecology. Harvard University Press.

28.

Hannan

M. T.

Hannan

M. T.

Polos

Carroll

G. R.

(2007). Logics of organization theory: Audiences, codes, and ecologies. Princeton University Press.

29.

Hirschman

A. O.

(1980). National power and the structure of foreign trade (Vol. 105). University of California Press.

30.

Hsu

Hannan

M. T.

Koçak

Ö.

(2009). Multiple category memberships in markets: An integrative theory and two empirical tests. American Sociological Review, 74(1), 150–169.

31.

Hsu

J. H. P.

Shin

Park

Lee

(2023). Two-sided cultural niches: Topic overlap, geospatial correlation, and local group activities on event-based social networks [Conference session]. Proceedings of the 11th International Conference on Communities and Technologies (pp. 54–63).

32.

Hwang

Foote

J. D.

(2021). Why do people participate in small online communities? Proceedings of the ACM on Human-Computer Interaction, 5(CSCW2), 1–25.

33.

Hyndman

R. J.

Athanasopoulos

(2018). Forecasting: Principles and practice. OTexts.

34.

Kittur

Nickerson

J. V.

Bernstein

Gerber

Shaw

Zimmerman

Lease

Horton

(2013). The future of crowd work [Conference session]. Proceedings of the 2013 Conference on Computer Supported Cooperative Work (pp. 1301–1318).

35.

Kovacs

Hannan

M. T.

(2010). The consequences of category spanning depend on contrast. Research in the Sociology of Organizations, 31, 175–201.

36.

Kraut

R. E.

Fiore

A. T.

(2014). The role of founders in building online groups [Conference session]. Proceedings of the 17th ACM Conference on Computer Supported Cooperative Work & Social Computing (pp. 722–732).

37.

Kraut

R. E.

Resnick

(2012). Building successful online communities: Evidence-based social design. MIT Press.

38.

Lai

C. H.

(2014). Can our group survive? An investigation of the evolution of mixed-mode groups. Journal of Computer-Mediated Communication, 19(4), 839–854.

39.

Lam

S. K.

Karim

Riedl

(2010). The effects of group composition on decision quality in a social production community [Conference session]. Proceedings of the 16th ACM international conference on Supporting group work (pp. 55–64).

40.

Lee

Butler

B. S.

(2020). Cultural activity diversity and community characteristics: An exploratory study [Conference session]. International Conference on Information (pp. 32–49). Springer.

41.

Liu

Tian

Lee

W. C.

McPherson

Han

(2012). Event-based social networks: linking the online and offline social worlds [Conference session]. Proceedings of the 18th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 1032–1040).

42.

López

Farzan

(2015). From community networks to hyper-local social media [Conference session]. Proceedings of the 18th ACM Conference Companion on Computer Supported Cooperative Work & Social Computing (pp. 239–242).

43.

López

Farzan

Lin

Y. R.

(2017). Behind the myths of citizen participation: Identifying sustainability factors of hyper-local information systems. ACM Transactions on Internet Technology, 18(1), 1–28.

44.

López

Lin

Y. R.

Farzan

(2015). What makes hyper-local online discussion forums sustainable? [Conference session]. 2015 48th Hawaii International Conference on System Sciences (pp. 2445–2454). IEEE.

45.

Farzan

Lopez

(2017). To go or not to go! what influences newcomers of hybrid communities to participate offline [Conference session]. Proceedings of the 8th International Conference on Communities and Technologies (pp. 159–168).

46.

Manikandan

Sam

R. S.

Gilbert

S. F.

Srikanth

(2024). Event-based social networking system with recommendation engine. International Journal of Intelligent Information Technologies, 20(1), 1–16.

47.

Meetup. (2025a). Who pays for a Meetup group?https://help.meetup.com/hc/en-us/articles/39428334646541-Who-pays-for-a-Meetup-group

48.

Meetup. (2025b). Media. https://www.meetup.com/media.

49.

Mervis

C. B.

Rosch

(1981). Categorization of natural objects. Annual Review of Psychology, 32(1), 89–115.

50.

Metzgar

E. T.

Kurpius

D. D.

Rowley

K. M.

(2011). Defining hyperlocal media: Proposing a framework for discussion. New Media & Society, 13(5), 772–787.

51.

Mohamed

A. H. H.

Tawfik

Al-Jumeily

Norton

(2011). Mohtam: A technology acceptance model for mobile health applications [Conference session]. 2011 Developments in E-systems Engineering (pp. 13–18). IEEE.

52.

Money

Turner

(2004). Application of the technology acceptance model to a knowledge management system [Conference session]. 37th Annual Hawaii International Conference on System Sciences (p. 9). IEEE.

53.

Navis

Glynn

M. A.

(2010). How new market categories emerge: Temporal dynamics of legitimacy, identity, and entrepreneurship in satellite radio, 1990–2005. Administrative Science Quarterly, 55(3), 439–471.

54.

Negro

Hannan

M. T.

Rao

(2010). Categorical contrast and audience appeal: Niche width and critical success in winemaking. Industrial and Corporate Change, 19(5), 1397–1425.

55.

Nickel

M. N.

Fuentes

J. M.

(2004). Relationship between legitimation, competition and organizational death: current state of the art. International Journal of Management Reviews, 5-6(1), 43–62.

56.

Ounnas

Millard

D. E.

Davis

H. C.

(2007). A metrics framework for evaluating group formation [Conference session]. Proceedings of the 2007 international ACM conference on Supporting group work (pp. 221–224).

57.

Pinch

T. J.

Bijker

W. E.

(1984). The social construction of facts and artefacts: Or how the sociology of science and the sociology of technology might benefit each other. Social Studies of Science, 14(3), 399–441.

58.

Pontikes

E. G.

Hannan

(2014). An ecology of social categories. Sociological Science, 1, 311–343.

59.

Pontikes

E. G.

(2022). Category innovation in the software industry: 1990–2002. Strategic Management Journal, 43(9), 1697–1727.

60.

Quinn

A. J.

Bederson

B. B.

(2011). Human computation: a survey and taxonomy of a growing field [Conference session]. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp. 1403–1412).

61.

Raban

D. R.

Moldovan

Jones

(2010). An empirical study of critical mass and online community survival [Conference session]. Proceedings of the 2010 ACM Conference on Computer Supported Cooperative Work (pp. 71–80).

62.

Rahaman

S. U.

Abdul

M. J.

Patchipulusu

(2021). Exploring (technology acceptance model) tam in mobile banking: A qualitative analysis using the technology acceptance model. International Journal of Management, 12(8), 51–59.

63.

Ransbotham

Kane

G. C.

(2011). Membership turnover and collaboration success in online communities: Explaining rises and falls from grace in Wikipedia1. MIS Quarterly, 35, 613–628.

64.

Ren

Chen

Riedl

(2016). The impact and evolution of group diversity in online open collaboration. Management Science, 62(6), 1668–1686.

65.

Ricken

Barkhuus

Jones

(2017). Going online to meet offline: Organizational practices of social activities through meetup [Conference session]. Proceedings of the 8th International Conference on Communities and Technologies (pp. 139–148).

66.

Ricken

Grandhi

Zytko

Hiltz

S. R.

Jones

(2014). Anyone for bowling? coalescing for shared activities [Conference session]. Proceedings of the 18th International Conference on Supporting Group Work (pp. 122–130).

67.

Ridings

C. M.

Gefen

(2004). Virtual community attraction: Why people hang out online. Journal of Computer-Mediated Communication, 10(1), JCMC10110.

68.

Robert

L. P.

Jr You

(2018). Disaggregating the impacts of virtuality on team identification [Conference session]. Proceedings of the 2018 ACM Conference on Supporting Groupwork (pp. 309–321).

69.

Rogers

E. M.

Singhal

Quinlan

M. M.

(2014). Diffusion of innovations. In Stacks

D. W.

Salwen

M. B.

(Eds.), An integrated approach to communication theory and research (pp. 432–448). Routledge.

70.

Romanelli

Khessina

O. M.

(2005). Regional industrial identity: Cluster configurations and economic development. Organization Science, 16(4), 344–358.

71.

Ruef

Patterson

(2009). Credit and classification: The impact of industry boundaries in nineteenth-century America. Administrative Science Quarterly, 54(3), 486–520.

72.

Schiller

Nah

F. F. H.

Luse

Siau

(2024). Men are from mars and women are from venus: dyadic collaboration in the metaverse. Internet Research, 34(1), 149–173.

73.

Schwitter

(2024). How offline meetings affect online activities: The case of wikipedia. EPJ Data Science, 13(1), 67.

74.

Seering

Yao

Kaufman

(2018). Applications of social identity theory to research and design in computer-supported cooperative work. Proceedings of the ACM on Human-Computer Interaction, 2(CSCW), 1–34.

75.

Sharma

Prana

G. A. A.

Sawhney

Nagappan

(2022). Analyzing offline social engagements: An empirical study of meetup events related to software development [Conference session]. 2022 IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER) (pp. 1122–1133). IEEE.

76.

Silva

M. O.

Rocha

L. M.

Moro

M. M.

(2019). Collaboration profiles and their impact on musical success [Conference session]. Proceedings of the 34th ACM/SIGAPP Symposium on Applied Computing (pp. 2070–2077).

77.

Spicher

Wesemeyer

Deserno

T. M.

(2024). Crowdsourcing image segmentation for deep learning: Integrated platform for citizen science, paid microtask, and gamification. Biomedical Engineering/Biomedizinische Technik, 69(3), 293–305.

78.

Suffill

Branigan

Pickering

(2019). Novel labels increase category coherence, but only when people have the goal to coordinate. Cognitive Science, 43(11), e12796.

79.

Tama

B. A.

Kim

Park

Manovich

Cha

(2023). Quantitative analysis of cultural dynamics seen from an event-based social network. arXiv preprint arXiv:2306.06176.

80.

Tao

Aubin Le Quéré

(2024). The neighborhood prism: How localized social networks reflect offline demographics [Conference session]. Companion Publication of the 2024 Conference on Computer-Supported Cooperative Work and Social Computing (pp. 323–329).

81.

TeBlunthuis

(2025). Niche dynamics in complex online community ecosystems [Conference session]. Proceedings of the International AAAI Conference on Web and Social Media (vol. 19, pp. 1880–1892).

82.

Thornton

McDonald

D. W.

(2012). Tagging Wikipedia: Collaboratively creating a category system [Conference session]. Proceedings of the 17th ACM International Conference on Supporting Group Work (pp. 219–228).

83.

Torres

E. N.

(2020). Online-to-offline interactions and online community life cycles: A longitudinal study of shared leisure activities. Leisure Sciences, 42(1), 32–50.

84.

Trinh

Huang

J. Z.

Azhar

(2020). Activeness and loyalty analysis in event-based social networks. Entropy, 22(1), 119.

85.

Üblacker

Liebig

Hamad

(2024). Catalysts of connection. The role of digital information and communication technology in fostering neighbourhood social cohesion: A systematic review of empirical findings. Urban Studies, 61(16), 3167–3186.

86.

Vaughn

D. N.

(2015). MeetUp and social capital: Building community in the digital age [PhD thesis, University of Denver]. https://www.proquest.com/openview/efb9bc71e395715d0446130e18f5b6ce/1?cbl=18750&pq-origsite=gscholar

87.

Venkatesh

Davis

F. D.

(2000). A theoretical extension of the technology acceptance model: Four longitudinal field studies. Management Science, 46(2), 186–204.

88.

Wang

Butler

B. S.

Ren

(2013). The impact of membership overlap on growth: An ecological competition view of online groups. Organization Science, 24(2), 414–431.

89.

Wei

Prasetyo

Y. T.

Belmonte

Z. J. A.

Cahigas

M. M. L.

Nadlifatin

Gumasing

M. J. J.

(2025). Applying the technology acceptance model – Theory of planned behavior (TAM-TPB) model to study the acceptance of building information modeling (BIM) in green building in China. Acta Psychologica, 254, 104790.

90.

Wry

Lounsbury

(2013). Contextualizing the categorical imperative: Category linkages, technology focus, and resource acquisition in nanotechnology entrepreneurship. Journal of Business Venturing, 28(1), 117–133.

91.

Zhu

Chen

Matthews

Pal

Badenes

Kraut

R. E.

(2014). Selecting an effective niche: an ecological view of the success of online communities [Conference session]. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp. 301–310).

92.

Zhu

Kraut

R. E.

Kittur

(2014). The impact of membership overlap on the survival of online communities [Conference session]. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp. 281–290).

93.

Zuckerman

E. W.

(1999). The categorical imperative: Securities analysts and the illegitimacy discount. American Journal of Sociology, 104(5), 1398–1438.