Dynamic Network Actor Models (DyNAMs) assume that an observed sequence of relational events is the outcome of an actor-oriented decision process consisting of two decision levels. The first level represents the time until an actor initiates the next relational event, modeled by an exponential distribution with an actor-specific activity rate. The second level describes the choice of the receiver of the event, modeled by a conditional multinomial logit model. The DyNAM assumes that the parameters are constant over the actors and the context. This homogeneity assumption, albeit statistically and computationally convenient, is difficult to justify, e.g., in the presence of unobserved differences between actors or contexts. In this paper, we extend DyNAMs by including random-effects parameters that vary across actors or contexts and allow controlling for unknown sources of heterogeneity. We illustrate the model by analyzing relational events among the users of an online community of aspiring and professional digital and graphic designers.

This article considers the link removal problem in a strongly connected directed network with the goal of minimizing the dominant eigenvalue of the network’s adjacency matrix while maintaining its strong connectivity. Due to the complexity of the problem, this article focuses on computing a suboptimal solution. Furthermore, it is assumed that the knowledge of the overall network topology is not available. This calls for distributed algorithms which rely solely on the local information available to each individual node and information exchange between each node and its neighbors. Two different strategies based on matrix perturbation analysis are presented, namely simultaneous and iterative link removal strategies. Key ingredients in implementing both strategies include novel distributed algorithms for estimating the dominant eigenvectors of an adjacency matrix and for verifying strong connectivity of a directed network under link removal. It is shown via numerical simulations on different type of networks that in general the iterative link removal strategy yields a better suboptimal solution. However, it comes at a price of higher communication cost in comparison to the simultaneous link removal strategy.

Is it always beneficial to create a new relationship (have a new follower/friend) in a social network? This question can be formally stated as a property of the centrality measure that defines the importance of the actors of the network. Score monotonicity means that adding an arc increases the centrality score of the target of the arc; rank monotonicity means that adding an arc improves the importance of the target of the arc relatively to the remaining nodes. It is known that most centralities are both score and rank monotone on directed, strongly connected graphs. In this paper, we study the problem of score and rank monotonicity for classical centrality measures in the case of undirected networks: in this case, we require that score, or relative importance, improves at both endpoints of the new edge. We show that, surprisingly, the situation in the undirected case is very different, and in particular that closeness, harmonic centrality, betweenness, eigenvector centrality, Seeley’s index, Katz’s index, and PageRank are not rank monotone; betweenness and PageRank are not even score monotone. In other words, while it is always a good thing to get a new follower, it is not always beneficial to get a new friend.

The order of the authors in the published article is incorrect. The authors should appear as follows:

J. Schneider, B. Cornwell, A. Jonas, R. Behler, N. Lancki, B. Skaathun, L. E. Young, E. Morgan, S. Michaels, R. Duvoisin, A. S. Khanna, S. Friedman, P. Schumm, E. Laumann, for the uConnect Study Team

The authors regret the error.

We study the incentives that agents have to invest in costly protection against cascading failures in networked systems. Applications include vaccination, computer security, and airport security. Agents are connected through a network and can fail either intrinsically or as a result of the failure of a subset of their neighbors. We characterize the equilibrium based on an agent's failure probability and derive conditions under which equilibrium strategies are monotone in degree (i.e. in how connected an agent is on the network). We show that different kinds of applications (e.g. vaccination, malware, airport/EU security) lead to very different equilibrium patterns of investments in protection, with important welfare and risk implications. Our equilibrium concept is flexible enough to allow for comparative statics in terms of network properties, and we show that it is also robust to the introduction of global externalities (e.g. price feedback, congestion).

Isolation is a concept originally conceived in the context of clique enumeration in static networks, mostly used to model communities that do not have much contact to the outside world. Herein, a clique is considered isolated if it has few edges connecting it to the rest of the graph. Motivated by recent work on enumerating cliques in temporal networks, we transform the isolation concept to the temporal setting. We discover that the addition of the time dimension leads to six distinct natural isolation concepts. Our main contribution is the development of parameterized enumeration algorithms for five of these six isolation types for clique enumeration, employing the parameter “degree of isolation.” In a nutshell, this means that the more isolated these cliques are, the faster we can find them. On the empirical side, we implemented and tested these algorithms on (temporal) social network data, obtaining encouraging results.

In this paper, we study how the topology of the International Trade Network (ITN) changes in geographical space, and along time. We employ geographical distance between countries in the world to filter the links in the ITN, building a sequence of subnetworks, each one featuring trade links occurring at similar distance. We then test if the assortativity and clustering of ITN subnetworks changes as distance increases, and we find that this is indeed the case: distance strongly impacts, in a non-linear way, the topology of the ITN. We show that the ITN is disassortative at long distances, while it is assortative at short ones. Similarly, the main determinant of the overall high-ITN clustering level are triangular trade triples between geographically close countries. This means that trade partnership choices and trade patterns are highly differentiated over different distance ranges, even after controlling for the economic size and income per capita of trading partners, and it is persistent over time. This evidence has relevant implications for the non-linear evolution of globalization.

Social networks, particularly those defined by friendships, influence many childhood and adolescent health behaviors such as the use of alcohol, tobacco, and other drugs, as well as diet and physical activity. Few, if any, studies have examined the concordance between friendship networks and sun exposure/safety behaviors. This study examines the friendship networks and sun safety behaviors for a group of fourth and fifth grade students taking part in a larger sun safety intervention, “SunSmart” (n = 128). Intra-class correlation, homophily hypothesis testing, and exponential random graph models were used to test friendship homophily based on sun safety behaviors. Peer Leaders were identified through social network popularity, and sun safety change scores were compared between Peer Leaders and non-leaders. Results show that students cluster based on shared demographic characteristics and some sun safety behaviors, and that there was a trend for Peer Leaders to respond better to the SunSmart intervention than non-leaders. Implications for future sun safety interventions using Peer Leaders as champions for sun safety behavior change are discussed.

Despite the importance of diverse expertise in helping solve difficult interdisciplinary problems, measuring it is challenging and often relies on proxy measures and presumptive correlates of actual knowledge and experience. To address this challenge, we propose a text-based measure that uses researcher’s prior work to estimate their substantive expertise. These expertise estimates are then used to measure team-level expertise diversity by determining similarity or dissimilarity in members’ prior knowledge and skills. Using this measure on 2.8 million team invented patents granted by the US Patent Office, we show evidence of trends in expertise diversity over time and across team sizes, as well as its relationship with the quality and impact of a team’s innovation output.

While network research often focuses on social integration as a predictor of health, a less-explored idea is that connections to dissimilar others may benefit well-being. As such, this study investigates whether network diversity is associated with changes in four health outcomes over a 3-year period of time in the U.S.A. Specifically, we focus on how an underexplored measure of network diversity—educational attainment assortativity—is associated with common self-reported outcomes: propensity to exercise, body-mass index, mental health, and physical health. We extend prior research by conducting multilevel analyses using this measure of diversity while adjusting for a range of socio-demographic and network confounders. Data are drawn from a longitudinal probability sample of U.S. adults (n=10.679) in which respondents reported information about themselves and eight possible alters during three yearly surveys (2013–2015). We find, first, that higher educational attainment is associated with more educationally insular networks, while less-educated adults have more educationally diverse networks. Results further suggest that having educationally similar networks is associated with higher body-mass index among the less educated. Further exploration of the relationship between ego network diversity, tie strength, and health is warranted.

We consider a network of interacting individuals, whose actions or transitions are determined by the states (behavior) of their neighbors as well as their own personal decisions. Specifically, we develop a model according to two simple decision-making rules that can describe the growth of the user population of a newly launched product or service. We analyze 22 sets of real-world historical growth data of a variety of products and services, and show that they all follow the growth equation. The numerical procedure for finding the model parameters allows the market size, and the relative effectiveness of customer service and promotional efforts to be estimated from the available historical growth data. We study the growth profiles of products and find that for a product or service to reach a mature stage within a reasonably short time in its user growth profile, the user growth rate corresponding to influenced transitions must exceed a certain threshold. Furthermore, results show that individuals in the group of celebrities having numerous friends become users of a new product or service at a much faster rate than those connected to ordinary individuals having fewer friends.

We consider the problem of generating uniformly random graphs from a constrained distribution. A graph is valid if it obeys certain constraints such as a given number of nodes, edges, k-stars or degree sequence, and each graph must occur with equal probability. A typical application is to confirm the correctness of a model by repeated sampling and comparing statistical properties against empirical data. Markov Chain Monte Carlo (MCMC) algorithms are often used, but have certain difficulties such as the inability to search the space of all possible valid graphs. We propose an improved algorithm which overcomes these difficulties. Although each individual iteration of the MCMC algorithm takes longer, we obtain better coverage of the search space in the same amount of time. This leads to better estimates of various quantities such as the expected number of transitive triads given the constraints. The algorithm should be of general interest with many possible applications, including the world wide web, biological, and social networks.

We consider a dynamic network cascade process developed by Duncan Watts applied to a class of random networks, developed independently by Newman and Miller, which allows a specified amount of clustering (short loops). We adapt existing methods for locally tree-like networks to formulate an appropriate two-type branching process to describe the spread of a cascade started with a single active node and obtain a fixed-point equation to implicitly express the extinction probability of such a cascade. In so doing, we also recover a formula that has appeared in various forms in works by Hackett et al. and Miller which provides a threshold condition for certain extinction of the cascade. We find that clustering impedes cascade propagation for networks of low average degree by reducing the connectivity of the network, but for networks with high average degree, the presence of small cycles makes cascades more likely.

Does the human brain have a central connective core, and, if so, how costly is it?

Noninvasive imaging data allow the construction of network maps of the human brain, recording its structural and functional connectivity. A number of studies have reported on various characteristic network attributes, such as a tendency toward local clustering, high global efficiency, the prevalence of specific network motifs, and a pronounced community structure with several anatomically and functionally defined modules and interconnecting hub regions (Bullmore & Sporns, 2009; van den Heuvel & Hulshoff Pol, 2010; Sporns, 2011). Hubs are of particular interest in studies of the brain since they may play crucial roles in integrative processes and global brain communication, thought to be essential for many aspects of higher brain function. Indeed, hubs have been shown to correspond to brain regions that exhibit complex physiological responses and maintain widespread and diverse connection profiles with other parts of the brain. We asked if, in addition to being highly connected, brain hubs would also exhibit a strong tendency to be mutually interconnected, forming what has been called a “rich club” (Colizza et al., 2006). Rich club organization is present in a network if sets of high-degree nodes exhibit denser mutual connections than predicted on the basis of the degree sequence alone. We investigated rich club organization in the human brain in datasets that recorded weighted projections among different anatomical regions of the cerebral cortex, recorded from several cohorts of healthy human volunteers (van den Heuvel & Sporns, 2011; van den Heuvel et al., 2012).

This study considered the role of adult children in the core networks of U.S. older adults with varying levels of functional health. Taking a multidimensional perspective of the ego network system, we considered (a) presence of child(ren) in the network, (b) contact with children network members, and (c) embeddedness of children within the network. We observed older parents from three waves of the National Social Life, Health, and Aging Project (NSHAP). The common ‘important matters’ name generator was used to construct egocentric network variables, while self-reported difficulty with activities of daily life was used to measure disablement transitions. Parameters were estimated with Generalized Estimating Equations (GEE). Though child turnover was common in parents’ core networks, there was no evidence linking disablement transitions to systematic forms of child reshuffling. Children that remained in parents’ networks, however, showed increased contact with parents and with other members of the network when the parent underwent disability progression. Disability onset was not significantly linked to either outcome. There was limited evidence of gender variation in these patterns. Overall, results strengthen the view that children are distinctive members of older adults’ core networks. Further, the role of adult children shifts most noticeably at advanced stages of the disablement process.

Hyperbolic random graphs (HRGs) and geometric inhomogeneous random graphs (GIRGs) are two similar generative network models that were designed to resemble complex real-world networks. In particular, they have a power-law degree distribution with controllable exponent $\beta$ and high clustering that can be controlled via the temperature $T$ .

We present the first implementation of an efficient GIRG generator running in expected linear time. Besides varying temperatures, it also supports underlying geometries of higher dimensions. It is capable of generating graphs with ten million edges in under a second on commodity hardware. The algorithm can be adapted to HRGs. Our resulting implementation is the fastest sequential HRG generator, despite the fact that we support non-zero temperatures. Though non-zero temperatures are crucial for many applications, most existing generators are restricted to $T = 0$ . We also support parallelization, although this is not the focus of this paper. Moreover, we note that our generators draw from the correct probability distribution, that is, they involve no approximation.

Besides the generators themselves, we also provide an efficient algorithm to determine the non-trivial dependency between the average degree of the resulting graph and the input parameters of the GIRG model. This makes it possible to specify the desired expected average degree as input.

Moreover, we investigate the differences between HRGs and GIRGs, shedding new light on the nature of the relation between the two models. Although HRGs represent, in a certain sense, a special case of the GIRG model, we find that a straightforward inclusion does not hold in practice. However, the difference is negligible for most use cases.

Genetic risk is particularly salient for families and testing for genetic conditions is necessarily a family-level process. Thus, risk for genetic disease represents a collective stressor shared by family members. According to communal coping theory, families may adapt to such risk vis-a-vis interpersonal exchange of support resources. We propose that communal coping is operationalized through the pattern of supportive relationships observed between family members. In this study, we take a social network perspective to map communal coping mechanisms to their underlying social interactions and include those who declined testing or were not at risk for Lynch Syndrome. Specifically, we examine the exchange of emotional support resources in families at risk of Lynch Syndrome, a dominantly inherited cancer susceptibility syndrome. Our results show that emotional support resources depend on the testing-status of individual family members and are not limited to the bounds of the family. Network members from within and outside the family system are an important coping resource in this patient population. This work illustrates how social network approaches can be used to test structural hypotheses related to communal coping within a broader system and identifies structural features that characterize coping processes in families affected by Lynch Syndrome.

Inspired by socio-political scenarios, like dictatorships, in which a minority of people exercise control over a majority of weakly interconnected individuals, we propose vulnerability and power measures defined on groups of actors of networks. We establish an unexpected connection between network vulnerability and graph regularizability. We use the Shapley value of coalition games to introduce fresh notions of vulnerability and power at node level defined in terms of the corresponding measures at group level. We investigate the computational complexity of computing the defined measures, both at group and node levels, and provide effective methods to quantify them. Finally we test vulnerability and power on both artificial and real networks.

We describe here a notion of diffusion similarity, a method for defining similarity between vertices in a given graph using the properties of random walks on the graph to model the relationships between vertices. Using the approach of graph vertex embedding, we characterize a vertex vi by considering two types of diffusion patterns: the ways in which random walks emanate from the vertex vi to the remaining graph and how they converge to the vertex vi from the graph. We define the similarity of two vertices vi and vj as the average of the cosine similarity of the vectors characterizing vi and vj. We obtain these vectors by modifying the solution to a differential equation describing a type of continuous time random walk.

This method can be applied to any dataset that can be assigned a graph structure that is weighted or unweighted, directed or undirected. It can be used to represent similarity of vertices within community structures of a network while at the same time representing similarity of vertices within layered substructures (e.g., bipartite subgraphs) of the network. To validate the performance of our method, we apply it to synthetic data as well as the neural connectome of the C. elegans worm and a connectome of neurons in the mouse retina. A tool developed to characterize the accuracy of the similarity values in detecting community structures, the uncertainty index, is introduced in this paper as a measure of the quality of similarity methods.

Social networks, wherein the edges represent nonbehavioral relations such as friendship, power, and influence, can be difficult to measure and model. A powerful tool to address this is cognitive social structures (Krackhardt, D. (1987). Cognitive social structures. Social Networks, 9(2), 109–134.), where the perception of the entire network is elicited from each actor. We provide a formal statistical framework to analyze informants’ perceptions of the network, implementing a latent space network model that can estimate, e.g., homophilic effects while accounting for informant error. Our model allows researchers to better understand why respondents’ perceptions differ. We also describe how to construct a meaningful single aggregated network that ameliorates potential respondent error. The proposed method provides a visualization method, an estimate of the informants’ biases and variances, and we describe a method for sidestepping forced-choice designs.