Exponential Random Graph Models Research Articles

Typical structure of sparse exponential random graph models

Typical structure of sparse exponential random graph models

Formation mechanisms of intra-organizational membership overlap: a longitudinal network analysis of membership data from the International Communication Association

Abstract Social entities are linked and interdependent through shared members, a phenomenon described as niche overlap. Using archival data on yearly affiliations of International Communication Association (ICA) members with divisions and interest groups over a 9-year period, this study conducts a longitudinal network analysis to examine the formation mechanisms of intra-organizational niche overlap among organizational subunits. The results of the temporal exponential random graph model suggest the absence of dominant ICA subunits whose topic areas are broad enough to integrate different specializations through membership overlap. Triadic closure facilitates the sharing of members among three subunits, leading to a cohesive internal structure and limited brokerage opportunities. Subunits with smaller membership sizes are more likely to have membership overlap, particularly with larger subunits. Positive affinity relations between subunits, characterized by the sharing of scarce resources within the organization, contribute to the formation of membership overlap.

Evolution, resilience and causes of global petroleum gas trade networks: 1995–2020

Based on the HS 4-digit code trade data in UNCOMTRADE from 1995 to 2020, this paper analyzes the characteristics of the evolution of the global PG trade network using the complex network approach and analyzes the changes in its resilience at the overall and country levels, respectively. The results illustrated that: (1) The scale of the global PG trade network tends to expand, and the connection is gradually tightened, experiencing a change from a “supply-oriented” to a “supply-and-demand” pattern, in which the U.S., Russia, Qatar, and Australia have gradually replaced Canada, Japan, and Russia to become the core trade status, while OPEC countries such as Qatar, Algeria, and Kuwait mainly rely on PG exports to occupy the core of the global supply, and the trade status of other countries has been dynamically alternating and evolving. (2) The resilience of the global PG trade network is lower than that of the random network and decreases non-linearly with more disrupted countries. Moreover, the impact of the U.S. is more significant than the rest of countries. Simulations using the exponential random graph model (ERGM) model revealed that national GDP, institutional quality, common border and RTA network are the determinants of PG trade network formation, and the positive impact of the four factors not only varies significantly across regions and stages, but also increases with national network status.

Structural Evolution and Its Determinants of Domestic Value-Added Network of Digital Service Exports Based on Temporal Exponential Random Graph Model

ABSTRACT Digital service trade has become an important engine of world economic growth, and its competitive landscape is also evolving. We construct domestic value-added networks of digital service exports (DVAN-DSEs) from 1995 to 2018 based on the world input-output table, and explores their structural evolution and determinants using Temporal Exponential Random Graph Model. We find that (a) the DVAN-DSEs present the features of agglomeration, heterogeneity, reciprocity and unevenness. (b) The United States and European countries are the core in DVAN-DSEs. (c) Preference attachment, structural embeddedness, delayed reciprocity, stability, variability, Matthew effect in GDP, population and Internet coverage, geographical proximity, cultural similarities, and regional trade agreements play important roles in the evolution of DVAN-DSEs. (d) Structural embeddedness and geographic proximity only facilitate new relationship construction, while other factors influence both relationship establishment and maintenance. Countries should build network facilities, expand trading partners and promote trade agreement negotiations to facilitate the development of digital service trade.

Analysis of the Driving Mechanism of Urban Carbon Emission Correlation Network in Shandong Province Based on TERGM

Analyzing the driving factors and mechanisms of urban carbon emission correlation networks can provide effective carbon reduction decision-making support for Shandong Province and other regions with similar industrial characteristics. Based on industrial carbon emission data from various cities in Shandong Province from 2013 to 2021, the spatial correlation network of carbon emission was established by using a modified gravity model. The characteristics of the network were explored by using the Social Network Analysis (SNA) method, and significant factors affecting the network were identified through Quadratic Assignment Procedure (QAP) correlation analysis and motif analysis. The driving mechanism of the carbon emission correlation network was analyzed by using Temporal Exponential Random Graph Models (TERGMs). The results show that: (1) The spatial correlation network of urban carbon emission in Shandong Province exhibits multi-threaded complex network correlations with a relatively stable structure, overcoming geographical distance limitations. (2) Qingdao, Jinan, and Rizhao have high degree centrality, betweenness centrality, and closeness centrality in the network, with Qingdao and Jinan being relatively central. (3) Shandong Province can be spatially clustered into four regions, each with distinct roles, displaying a certain “neighboring clustering” phenomenon. (4) Endogenous network structures such as Mutual, Ctriple, and Gwesp significantly impact the formation and evolution of the network, while Twopath does not show the expected impact; FDI can promote the generation of carbon emission reception relationships in the spatial correlation network; IR can promote the generation of carbon emission spillover relationships in the spatial correlation network; GS, differences in GDP, differences in EI, and similarities of IR can promote the generation of organic correlations within the network; on the temporal level, the spatial correlation network of urban carbon emission in Shandong Province has shown significant stability during the study period.

Research on scientific knowledge evolution patterns based on ego-centered fine-granularity citation network

A deep comprehension of the development of scientific knowledge is necessary to properly appreciate technological advancement and the process of knowledge innovation. However, fine-grained citation networks are less frequently utilized to characterize the evolution of scientific knowledge. To bridge this gap, we first constructed the fine-grained citation networks based on the PubMed and Web of Science (WOS) databases, then extracted the ego-centered networks of scientific knowledge, and finally employed Exponential Random Graph Models (ERGMs) to analyze the factors influencing the formation of Ego-centered Fine-granularity Citation Networks (EFCNs), taking into account the endogenous network structure and exogenous knowledge attribute variables. The results reveal that both types of variables play pivotal roles in the evolution of scientific knowledge. Furthermore, we found that (1) the in-degree and the out-degree centrality have a positive effect on knowledge evolution, respectively, in the 98.5 % and 99.9 % sample networks, while the clustering coefficient only has a positive effect on the edge formation of the 6.8 % sample network at the 0.05 significance level. (2) The citation behavior and domain impact of authors positively influence the scientific knowledge evolution, respectively, in the 63.2 % and 78.3 % sample networks. (3) There is a tendency to form citation relationships between scientific knowledge units of similar age in 67.1 % of the sample networks. (4) There is a greater possibility of developing a citation relationship between scientific knowledge with the same journal impact rank and knowledge type. Our findings indicate that the evolution of scientific knowledge is influenced not only by the process of scholarly communication but also by self-organizing mechanisms at the fine granularity level.

Formation of professional social networks via physicians’ online engagement: Evidence from Sina Weibo and Sina Health

The professional social networks of physicians are an essential source of health-related user-generated content and online social capital. Yet limited research has examined how social networks of physicians emerge at the level of relationships between physicians. To help fill this gap, this study investigates the formation of followship between physicians (i.e., when one physician follows another) from the perspectives of network structures, actor attributes, and external contextual factors. Using exponential random graph models and strategic network formation models, this study analyzes the effects of reciprocity and transitive closure, physicians’ social media and online healthcare community (OHC) engagement, and physicians’ interaction networks. The results reveal that physicians’ professional social networks have a reciprocated and open structure. Physicians’ social media connectivity and content, online service performance, professional capital, and social identity homophily facilitate the formation of professional social networks. In addition, the interaction networks have an entrainment effect on professional social networks. This study reveals the mechanism of social processes, including reciprocity, preferential attachment, and homophily, in the formation of social networks. The results also extend social cognitive theory and signaling theory into the context of physicians’ professional social networks.

Approximate inference for longitudinal mechanistic HIV contact network

Network models are increasingly used to study infectious disease spread. Exponential Random Graph models have a history in this area, with scalable inference methods now available. An alternative approach uses mechanistic network models. Mechanistic network models directly capture individual behaviors, making them suitable for studying sexually transmitted diseases. Combining mechanistic models with Approximate Bayesian Computation allows flexible modeling using domain-specific interaction rules among agents, avoiding network model oversimplifications. These models are ideal for longitudinal settings as they explicitly incorporate network evolution over time. We implemented a discrete-time version of a previously published continuous-time model of evolving contact networks for men who have sex with men and proposed an ABC-based approximate inference scheme for it. As expected, we found that a two-wave longitudinal study design improves the accuracy of inference compared to a cross-sectional design. However, the gains in precision in collecting data twice, up to 18%, depend on the spacing of the two waves and are sensitive to the choice of summary statistics. In addition to methodological developments, our results inform the design of future longitudinal network studies in sexually transmitted diseases, specifically in terms of what data to collect from participants and when to do so.

Invisible Iterations: How Formal and Informal Organization Shape Knowledge Networks for Coordination

AbstractThis study takes a network approach to investigate coordination among knowledge workers as grounded in both formal and informal organization. We first derive hypotheses regarding patterns of knowledge‐sharing relationships by which workers pass on and exchange tacit and codified knowledge within and across organizational hierarchies to address the challenges that underpin contemporary knowledge work. We use survey data and apply exponential random graph models to test our hypotheses. We then extend the quantitative network analysis with insights from qualitative interviews and demonstrate that the identified knowledge‐sharing patterns are the micro‐foundational traces of collective coordination resulting from two underlying coordination mechanisms which we label ‘invisible iterations’ and ‘bringing in the big guns’. These mechanisms and, by extension, the associated knowledge‐sharing patterns enable knowledge workers to perform in a setting that is characterized by complexity, uncertainty and ambiguity. Our research contributes to theory on the interplay between formal and informal organization for coordination by showing how self‐directed, informal action is supported by the formal organizational hierarchy. In doing so, it also extends understanding of the role that hierarchy plays for knowledge‐intensive work. Finally, it establishes the collective need to coordinate work as a previously overlooked driver of knowledge network relationships and network patterns.

Hierarchical Bayesian adaptive lasso methods on exponential random graph models

The analysis of network data has become an increasingly prominent and demanding field across multiple research fields including data science, health, and social sciences, requiring the development of robust models and efficient computational methods. One well-established and widely employed modeling approach for network data is the Exponential Random Graph Model (ERGM). Despite its popularity, there is a recognized necessity for further advancements to enhance its flexibility and variable selection capabilities. To address this need, we propose a novel hierarchical Bayesian adaptive lasso model (BALERGM), which builds upon the foundations of the ERGM. The BALERGM leverages the strengths of the ERGM and incorporates the flexible adaptive lasso technique, thereby facilitating effective variable selection and tackling the inherent challenges posed by high-dimensional network data. The model improvements have been assessed through the analysis of simulated data, as well as two authentic datasets. These datasets encompassed friendship networks and a respondent-driven sampling dataset on active and healthy lifestyle awareness programs.

Policy influence and influencers online and off

AbstractSocial media is an important arena for policy contestation. Although online social media debates can yield notable power over political processes offline, little research has examined the relationship between policy actors' influence in offline policymaking and social media policy arenas. We explore the relationship between four types of influence: reputational and formal‐institutional influence offline, and broadcasting and boosting influence online. We ask (1) are influential policy actors better able than others to broadcast their own messages on social media? and (2) are they better able than others to boost the broadcasting influence of other policy actors in social media? Using exponential random graph models on survey and Twitter data from the Finnish climate policy domain, we find that actors with high reputational influence in offline policymaking are also influential online, when measuring influence as the ability to broadcast one's own message. The pattern does not hold for those with formal‐institutional influence offline. Additionally, offline influence does not translate to the ability to further shape online influence by making boosting other actors' visibility. Our results suggest that although online influence partially corresponds to influence in policymaking, influence varies across arenas of policy contestation.

Transboundary cooperation in Arctic climate change governance under geopolitical tensions

Political conflicts or geopolitical tensions can create uncertainty in addressing climate change and environmental management in the Arctic. Dissecting how actors interact with each other and form networks is important for understanding ecological and environmental management challenges during geopolitical tensions, as well as promoting better governance. We construct transboundary networks for Arctic climate change governance (ACCG) from 2013 to 2021 based on the Global Database of Events, Language, and Tone (GDELT). Further, we used network descriptive statistical analysis and Temporal Exponential Random Graph Models (TERGM) to explore the structure of ACCG networks and the key factors influencing cooperation formation. The findings suggest that the overall cooperation density of the ACCG is low, and the dominant position of core actors is continuously strengthening. Non-state actors are less likely to be seen as partners and their participation depends largely on cooperation with states. The results also show that actors with similar stances and problem exposure are more likely to cooperate, but those exposed to high latitudes often choose not to cooperate; first-comers are more likely to perceive as cooperating yet they are inclined to establish internal cooperation. Additionally, two geographically proximate actors are more likely to cooperate. This indicates that under geopolitical tensions, the ACCG faces challenges not only due to the limited capacity of non-state actors to perform transboundary functions but also because the cooperation mechanisms are influenced by regional political logic. Accordingly, we further suggest policy recommendations from developing binding international frameworks to guide transboundary cooperation, enhancing cooperation among non-state actors, and ensuring the representativeness and fairness of non-Arctic actors' participation. This research provides insights into transboundary environmental management under political tensions, while also offering new pathways for analysing large-scale environmental governance structures.

Political relations and trade cooperation of the Belt and Road initiative: research on trade network based on strategic emerging industries

PurposeThe purpose of this study is to examine the impact of political relations on trade in strategic emerging industries (SEIs) in the Belt and Road initiative (BRI) associated countries. This investigation encompasses not only from the perspective of bilateral political relations but also the political intervention of third parties.Design/methodology/approachThe study employs the temporal exponential random graphmodel to analyze the dynamic structure and influencing factor of SEIs trade network among 150 BRI-associated countries from 2015 to 2020.FindingsThe results indicate that the trade of SEIs in the BRI-associated countries exhibits a pattern of concentrated exporters and decentralized importers. Amicable bilateral political relations foster trade cooperations in SEIs, while political pressure from the United States has the opposite effect. Furthermore, compared with the influence of third parties, the BRI has created a more robust trade environment characterized by political mutual trust.Practical implicationsBRI-associated countries should strengthen their political communication, and endeavor to transform political consensus and shared vision into concrete collaborative projects, while mitigating geopolitical uncertainties through a sound risk evaluation system. Moreover, they should establish a more transparent and consistent consultation mechanism and leverage the BRI trade network to foster balanced and mutually beneficial partnerships that minimize rivalry and dependence on a single market.Originality/valueThis study goes beyond observed trade cost and incorporates the political factor into the determinants of the BRI trade, thereby expanding the theoretical boundaries of existing BRI research. Also, this study employs bilateral trade data to construct SEIs trade networks (SEITNs) along the BRI route. It provides a comprehensive understanding of the dynamic determinates of the SEITNs will provide valuable practical guidance for enhancing and expanding trade and cooperation among BRI-associated countries.

Analysis of the Spatial Correlation Network and Driving Mechanism of China’s Transportation Carbon Emission Intensity

From 2008 to 2021, this study analyzed the spatial correlation characteristics between provincial transportation carbon emission intensity and explored ways to reduce transportation carbon emissions. This study used the modified gravity model, social network analysis (SNA) method, and temporal exponential random graph model (TERGM) to analyze the spatial correlation network evolution characteristics and driving mechanism of China’s transportation carbon emission intensity. This study found that China’s transportation carbon emission intensity and spatial correlation network have unbalanced characteristics. The spatial correlation network of transportation carbon emission intensity revealed that Shanghai, Beijing, Tianjin, Guangdong, Fujian, and other provinces were at the center of the network, with significant intermediary effects. The spatial correlation of transportation carbon emission intensity was divided into four functional plates: “two-way spillover”, “net benefit”, “broker”, and “net spillover”. The “net benefit” plate was mainly located in developed regions, and the “net spillover” plate was primarily located in underdeveloped regions. Endogenous structural and exogenous mechanism variables were the main factors affecting the evolution of the spatial correlation network of provincial transportation carbon emission intensity.

Social network analysis to predict social behavior in dairy cattle

Dairy cattle are frequently housed in free stalls with limited space, impacting social interactions between individuals. Social behavior in dairy cattle is gaining recognition as a valuable tool for identifying sick animals, but its application is hampered by the complexities of analyzing social interactions in intensive housing systems. In this context, precision livestock technologies present the opportunity to continuously monitor dyadic spatial associations on dairy farms. The aim of this study is to evaluate the accuracy of predicting social behavior of dairy cows using social network analysis. Daily social networks were built using the position data from 149 cows over 14 consecutive days of the study period. We applied the separable temporal exponential random graph models to estimate the likelihood of formation and persistence of social contacts between dairy cows individually and to predict the social network on the subsequent day. The correlation between the individual degree centrality values, the number of established social contacts per individual, between the predicted and observed networks ranged from 0.22 to 0.49 when the structural information from network triangles was included in the model. This study presents a novel approach for predicting animal social behavior in intensive housing systems using spatial association information obtained from a real-time location system. The results indicate the potential of this approach as a crucial step toward the larger goal of identifying disruptions in dairy cows expected social behavior.

Life-Space Mobility, Transportation, and the Companionship Network of Members of a Hispanic Senior Center.

Companions (i.e., friends who spend time together) are important for the well-being of older adults. Senior centers in the United States are places for older adults to participate in group activities and form and maintain companionships. However, differences in mobility and transportation may affect the ability of older adults to leverage senior center activities into actual companionships. This social network analysis was conducted to characterize the companionship network among members of a senior center in relation to their life-space mobility and transportation resources. An exponential random graph model was estimated to identify mobility- and transportation-related correlates of the likelihood of a companionship tie among senior center members (N = 42). Members had an average of 2 companionships with one another (M = 2.2, SD = 2.7). Companionships were more likely for members with greater life-space mobility (p = .009), who attended the senior center more frequently (p = .004), with automobile ownership in their households (p = .034), and who were not transportation cost-burdened (i.e., spent less than 15% of their income on transportation, p = .005). Demographic characteristics, limitations on instrumental activities of daily living, and being at risk for depression were not significantly associated with the likelihood of companionships. These findings extend previous knowledge of the role of life-space mobility and transportation in supporting general social participation for older adults to include the importance of transportation and mobility for having companions within a senior center.

Organised crime movement across local communities: A network approach

This paper explores the structure of organised crime movement across local communities and the drivers underpinning such movement. Firstly, it builds on network analysis to offer a novel methodological approach to empirically and quantitatively study the movement of organised crime offenders across geographical areas. The paper then applies this approach to evidence from Cambridgeshire in the United Kingdom. It reconstructs the movement of organised crime members across local areas based on a large-scale police dataset that includes 41 months of recorded crime events. It identifies organised crime “turf” and “target” areas and then explores the drivers of movement from the former to the latter using Exponential Random Graph Models. Findings confirm that geographical distance matters; however, socio-demographic, urban, economic and crime-related characteristics of communities play a key role. Organised crime group members target urban communities with higher than average illegal market opportunities (proxied by drug-related activity). The work also finds the effect of socio-demographic homophily between turf and target communities, suggesting that organised crime group members might target territories that are similar to their own. While a high level of deprivation makes a community more likely to send organised crime members, its impact on a community’s probability of being a receiver is less clear. Finally, the paper offers a way to identify communities (local areas) at risk of being targeted by criminal organisations, thus providing practitioners with a tool for early interventions.

Connectivity at the human-wildlife interface: starling movements relate to carriage of E. coli

Synanthropic bird species in human, poultry or livestock environments can increase the spread of pathogens and antibiotic-resistant bacteria between wild and domestic animals. We present the first telemetry-based spatial networks for a small songbird. We quantified landscape connectivity exerted by spotless starling movements, and aimed to determine if connectivity patterns were related to carriage of potential pathogens. We captured 28 starlings on a partridge farm in 2020 and tested them for Avian influenza virus, West Nile virus WNV, Avian orthoavulavirus 1, Coronavirus, Salmonella spp. and Escherichia coli. We did not detect any viruses or Salmonella, but one individual had antibodies against WNV or cross-reacting Flaviviruses. We found E. coli in 61 % (17 of 28) of starlings, 76 % (13 of 17) of which were resistant to gentamicin, 12 % (2 of 17) to cefotaxime/enrofloxacin and 6 % (1 of 17) were phenotypic extended spectrum beta-lactamase (ESBL) carriers. We GPS-tracked 17 starlings and constructed spatial networks showing how their movements (i.e. links) connect different farms with nearby urban and natural habitats (i.e. nodes with different attributes). Using E. coli carriage as a proxy for acquisition/dispersal of bacteria, we found differences across spatial networks constructed for E. coli positive (n = 7) and E. coli negative (n = 9) starlings. We used Exponential Random Graph Models to reveal significant differences between networks. In particular, an urban roost was more connected to other sites by movements of E. coli positive than by movements of E. coli negative starlings. Furthermore, an open pine forest used mainly for roosting was more connected to other sites by movements of E. coli negative than by movements of E. coli positive starlings. Using E. coli as a proxy for a potential pathogen carried by starlings, we reveal the pathways of spread that starlings could provide between farms, urban and natural habitats.

Does Formal Interorganizational Relationship Shape Interpersonal Knowledge Exchange in Multiteam Construction Projects? A Multi-Level Network Analysis

This study investigates the differences in structural patterns of intra-and inter-team knowledge exchange networks in multiteam construction projects and explores the cross-levelinteraction between knowledge exchange networks and interorganizational formal networks, as represented by networks of contractual and task interdependence relationships. An empirical study is developed by using multiple exponential random graph models. The findings show that a discrepancy between intra- and inter-team knowledge exchange structural patterns arises with respect to reciprocity. Interorganizational task interdependencies are found to align with cross-team knowledge exchange. Significant homophily in work experience is found in cross-team knowledge exchange. Interorganizational task interdependence promotes cross-team knowledge exchange among project members with heterogeneous expertise. This study expands the research on project-based multiteam system knowledge exchange networks from a multi-level network analysis perspective.

Participatory approaches and Social Network Analysis to analyse the emergence of collective action for rural development: a case study in the Spanish Pyrenees

This paper explores the social capital of an emergent beef quality brand in two valleys (Ansó and Hecho) in the Spanish Pyrenees through a combination of participatory approaches (focus groups) and Social Network Analysis. We organised three focus groups with stakeholders from the administration, tourism and commerce, and farming sectors to evaluate the interest for the initiative. In the Social Network Analysis, we surveyed 32 local stakeholders to construct an advisory network, including the surveyed actors and non-surveyed actors that they mentioned, and a trust network, including only the surveyed actors and their trust level in each other. The trust network was analysed using the Louvain’s algorithm to identify potential internal communities and the exponential random graph model (ERGM) to infer the shaping effects of actors’ attributes on the network’s structure. Our results revealed a general interest of the stakeholders in the beef quality brand and a sparse network characterised by a propensity to mutual and open interactions with four clusters based on employment sectors/educational levels and connected by two prominent actors. Therefore, the social capital of the initiative showed both risks because a loss of these few prominent actors could inhibit the network, and opportunities, because their presence, along with the mutual and open interactions, would ensure efficient information exchange. The investigation revealed also two possible limiting factors: the absence of a coordinator and the actors’ unwillingness to assume this role. The results confirmed the efficacy of the methodology used to investigate the social capital of emergent initiatives, but further research should be carried out including stakeholders’ concerns on its economic implications. HIGHLIGHTS Focus groups and Social Network Analysis allowed to explore the social capital of a beef production quality brand initiative. A sparse network with mutual interactions but few central actors emerged. The absence of a coordinator and unwillingness to cover the role are possible limiting factors.