SEMINAR SERIES

DNAC hosts seminars regularly throughout the academic year. These events highlight cutting edge research from nationally and internationally renowned scholars and serves as a forum for interdisciplinary exchange.

Series Schedule:
Mondays   |  Lunch 12:15 – 12:45pm   |  Talk 12:45 – 2:00pm  |  230E Gross Hall

Date/timeLocationSpeaker(s)AffiliationEvent NameAbstract 
2020-02-24, 12:45–2:00 PMGross Hall, Rm 230EKossi Edoh & Elijah MacCarthyNorth Carolina A&T UniversityDNAC SeminarAbstractContact network models are recent alternatives to equation-based models in epidemiology. The spread of disease on contact networks is modeled using bond percolation and the results are compared to equation-based compartmental models such as Susceptible-Exposed-Infected-Recovered and Susceptible-Infected-Recovered. The weight of the edges in the contact graphs is assume as a function of several variables in which case the weight is the product of the probabilities of independent events involving each of the variables. We simulate equation-based assumptions on Watts-Strogatz networks with different node degrees to determine when the results are similar to that of equation-based models. We examine the impact of network structures on the spread of infection by considering various networks, including Poisson, Erdos Renyi, Scale-free, and Watts-Strogatz small-world networks, and discuss how control measures can make use of the network structures. As a case study, we use data from the 1918 Spanish flu pandemic and that from measles outbreak to validate our results. In the last experiment, the edge-weight is computed from a single variable involving the number of passengers on flights between two cities within the United States. In addition, we explore the dynamics and adaptive nature of contact networks. For example, when a node is infected, human tendency dictates that this node’s contacts decrease.
2020-02-17, 12:45–2:00 PMGross Hall, Rm 230ETodd BenDorUNC City and Regional PlanningDNAC Seminar
2020-02-10, 12:45–2:00 PMGross Hall, Rm 230ENeha GondalBoston UniversityDNAC Seminar
2020-02-03, 12:45–2:00 PMGross Hall, Rm 230EKeith WarrenOhio StateDNAC Seminar
2020-01-27, 12:30–2:00 PMGross Hall, Rm 230ESelin AviyenteMichigan State EngineeringDNAC Seminar
2019-03-25, 12:45–2:00 PMGross Hall, Rm 230EAshton VerderyDNAC Seminar
2019-03-18, 12:45–2:00 PMGross Hall, Rm 230EJon MorganDNAC Seminar
2019-03-04, 12:45–2:00 PMGross Hall, Rm 230EPeter MuchaUNC Applied MathematicsDNAC Seminar
2019-02-25, 12:45–2:00 PMGross Hall, Rm 230EDavid SiegelDuke Political ScienceDNAC SeminarAbstractLeadership is often perceived as a single trait relevant to the context at hand, but it is not: it is comprised of multiple aspects. The most commonly discussed aspect is the influence leaders have over others. While important, it is not unique in its downstream relevance. Also important are the motivations of leaders to lead and the context in which leaders are situated. The latter is a relational concept, indicating that network structure may matter in understanding the effect of leadership. In this talk, I explore two relational aspects of leadership: How does the structure of the network in which the leader is embedded affect her ability to employ influence? and How do differences in types of leadership—charismatic vs bureaucratic—alter the effectiveness of leadership?
2018-12-14 (probably 12:45–2:00 PM)Soc/Psych Bldg, Rm 329 (McKinney Rm)John Levi MartinSociology Jensen Seminar
2018-12-10 (probably 12:45–2:00 PM)Gross Hall, Rm 230EJeff LienertNIHDNAC Seminar
2018-11-19, 12:45–2:00 PMGross Hall, Rm 230EJoshua BeckKelloggCollected vs. Collective Intelligence in the Wisdom of Crowds
2018-11-12, 12:45–2:00 PMGross Hall, Rm 230ESandra Gonzalez-BailonUniversity of Pennsylvania Annenberg School for CommunicationDNAC Seminar
2018-11-09 (probably 12:45–2:00 PM)Soc/Psych Bldg, Rm 329 (McKinney Rm)Brian UzziSociology Jensen Seminar
2018-10-29, 12:45–2:00 PMGross Hall, Rm 230ERaquel AsencioDNAC Seminar
2018-10-26 (probably 12:45–2:00 PM)Soc/Psych Bldg, Rm 329 (McKinney Rm)Omar LizardoSociology Jensen Seminar
2018-10-22, 12:45–2:00 PMGross Hall, Rm 230ERichard BentonUIUC School of Labor and Employment RelationsDNAC Seminar
2018-10-15, 12:45–2:00 PMGross Hall, Rm 230ECharles NunnDuke UniversityDNAC Seminar
2018-04-09, 12:45–2:00 PMGross Hall, Rm 230EDaniel B. LarremoreUniversity of Colorado Boulder, BioFrontiers InstituteA Physical Model for Efficient Ranking in NetworksAbstractIn systems of many individual entities, interactions and their outcomes are often defined by hierarchies or rankings. While in most cases these rankings are hidden from us, their presence is nevertheless revealed in the asymmetric patterns of interactions that we observe. For example, social groups of birds, primates, and elephants are organized according to dominance hierarchies in which more powerful animals assert themselves over those less powerful. Social positions are not directly visible to researchers, but we can infer each animal’s position in social space by observing a sufficient number of interactions. Similar latent hierarchies exist in systems of endorsement in which status is due to prestige or reputation. For example, in academia, universities are more likely to hire faculty candidates from equally or more prestigious universities. I will present a principled model and algorithm called SpringRank to infer such latent hierarchies directly from data. By mapping a network of directed interactions to a physical system, this ranking process becomes as fast as your favorite sparse linear solver. Unlike other methods such as minimum violation ranking, SpringRank assigns real-valued scores to nodes rather than simply ordinal ranks, and it formalizes the assumption that interactions are more likely to occur between individuals with similar ranks. It provides a natural framework for a statistical significance test for distinguishing when the inferred hierarchy is due to the network topology or is instead due to random chance, and it can be used to perform inference tasks such as predicting the existence or direction of edges. I’ll illustrate these findings by analyzing real and synthetic data and show that our method outperforms others, in both speed and accuracy, in recovering the underlying ranks and predicting edge directions. This work is a collaboration with Caterina De Bacco and Cris Moore.
2018-04-02, 12:45–2:00 PMGross Hall, Rm 230EJaemin LeeDepartment of SociologyA Consolidation Model of Political Polarization
2018-03-26, 12:45–2:00 PMGross Hall, Rm 230EJennifer Lutz & David EagleNCSU, Sociology & Duke, SociologyReexamining Gender and Depression: Social Ties and Mental Health among an Occupational GroupAbstractThis study extends social-psychological research on social networks and mental health by assessing cross-gender differences in social integration and the relationship between social integration and depression among United Methodist Clergy in North Carolina. The relationship between social network characteristics and mental health is expected to differ across gender, as men and women experience depression in different ways, and depression is a gender appropriate response mechanism for women in the face of unfavorable life events. Using data from the fifth wave of the Clergy Health Initiative, a longitudinal analysis of North Carolina United Methodist Clergy, this study clarifies gender differences in depression using measures of social integration within a closed occupational network of clergy. Using negative binomial regression techniques, this study assesses the relationship between indegree and outdegree on depression (PHQ-9) among clergy. A split sample analysis of women (N= 433) and men (N= 887) reveal gendered differences in the association between of social integration and depression. Specifically, outdegree shows a statistically significant negative relationship with depression for men, but not women. Indegree was not associated with depression for men or women. However, too many social ties may be detrimental to mental health for both men and women. This study adds important insight into research on social-psychology research, social networks, and organizations by expanding knowledge of how gender shapes the relationship between social integration and depression, specifically, in a male-dominated workplace.
2018-03-05, 12:45–2:00 PMGross Hall, Rm 230ESrijan SenguptaVirginia Tech, Department of StatisticsA blockmodel for node popularity in networks with community structureAbstractNetwork data analysis is a fast growing research field with diverse applications spanning several scientific disciplines. The community structure observed in empirical networks has been of particular interest in the statistics literature, with a strong emphasis on the study of blockmodels. In this paper we study an important network feature called node popularity, which is closely associated with community structure. Neither the classical stochastic blockmodel nor its degree-corrected extension can satisfactorily capture the dynamics of node popularity as observed in empirical networks. We propose a popularity-adjusted blockmodel for flexible and realistic modeling of node popularity. We establish consistency of likelihood modularity for community detection as well as estimation of node popularities and model parameters, and demonstrate the advantages of the new modularity over the degree-corrected blockmodel modularity in simulations. By analyzing the political blogs network, the British MP network, and the DBLP bibliographical network, we illustrate that improved empirical insights can be gained through this methodology. If time permits, I will also briefly outline some related ongoing work on networks, that might be of interest to the NDAC community.
2018-02-26, 12:45–2:00 PMGross Hall, Rm 230EShai PilosofUniversity of Chicago, Department of Ecology & EvolutionStructure and Dynamics of Highly-Dimensional Ecological SystemsAbstractEcological systems are inherently multi-dimensional. For example, they vary in space or time or interconnect to form networks of species networks. However, description of ecological networks has been typically limited to one dimension, in part due to methodological challenges in analyzing multi-dimensionality. In this talk, I will present a framework to analyze networks with more than one dimension, namely Ecological Multilayer Networks. I will use this framework to explore two examples of temporal multilayer networks: (1) the structure of a community of host and parasite species; and (2) the population-genetic structure of the human malaria parasite. While these two examples differ in their underlying theory, types of data and questions, they unite to show that incorporating higher dimensions using ecological multilayer networks can profoundly change our understanding of ecological systems.
2018-02-19, 12:45–2:00 PMGross Hall, Rm 230ETeague HenryUniversity of North Carolina at Chapel Hill, Psychology and NeuroscienceRegistered Networks: Localization and HeterogeneityAbstractRegistered networks, networks where the identities of the nodes are known and relevant, are common in certain areas of social science. Brain connectivity networks are a prime example, as are psychometric networks, formed from the estimated relations between psychological constructs. In this talk, we present two related methods for analyzing samples of registered networks. The first, the network based statistic jackknife, allows for the localization of network statistic based inferences. This alleviates the issue of inference equifinality, where vastly differing network configurations can lead to the same conclusions. The second, the network homogeneity test, evaluates if a sample of registered networks comes from the same generating distribution. This test is based in graph limit theory and does not require the assumption of a specific generating distribution (e.g. ER, small world, etc.). Application areas of these tests are discussed, and several simulations are presented.
2018-02-12, 12:45–2:00 PMGross Hall, Rm 230EDerek Owens-OasDepartment of Statistical ScienceA Bayesian Model for Clustering Networked DocumentsAbstractWe introduce a novel Bayesian statistical model for simultaneously discovering topics and clustering documents which have a network structure. In much of existing literature for network topic models, links occur at a document-to-document level or a node-to-node level. Here, we model links at a document-to-node level, as they occur this way in our data. Specifically, we verify the model on political blog posts from 2012. Inference uses Gibbs sampling to sample from posterior distributions for topic assignments and block memberships. Top words from selected topics are displayed, discovered communities are discussed, and results are compared with a strong baselines from the joint network topic modeling literature.
2018-02-05, 12:45–2:00 PMGross Hall, Rm 230EEmily EriksonYale University, Department of SociologyCultural Holes and the Authorship of Early Economic Texts
2018-01-29, 01:00–2:00 PMGross Hall, Rm 230EPeter MuchaUniversity of North Carolina-Chapel Hill, Mathematics and Applied Physical SciencesPost-Processing Partitions to Identify Domains of Modularity OptimizationAbstractWe introduce the Convex Hull of Admissible Modularity Partitions (CHAMP) algorithm to prune and prioritize different network community structures identified across multiple runs of possibly various computational heuristics. Given a set of partitions, CHAMP identifies the domain of modularity optimization for each partition—i.e., the parameter-space domain where it has the largest modularity relative to the input set—discarding partitions with empty domains to obtain the subset of partitions that are “admissible” candidate community structures that remain potentially optimal over indicated parameter domains. Importantly, CHAMP can be used for multi-dimensional parameter spaces, such as those for multilayer networks where one includes a resolution parameter and interlayer coupling. Using the results from CHAMP, a user can more appropriately select robust community structures by observing the sizes of domains of optimization and the pairwise comparisons between partitions in the admissible subset. We demonstrate the utility of CHAMP with several example networks. In these examples, CHAMP focuses attention onto pruned subsets of admissible partitions that are 20-to-1785 times smaller than the sets of unique partitions obtained by community detection heuristics that were input into CHAMP. This work, coauthored with William Weir, Scott Emmons, Ryan Gibson and Dane Taylor, has appeared as Algorithms 10, 93 (2017), doi:10.3390/a10030093.
2018-01-22, 12:45–2:00 PMGross Hall, Rm 270Sharique HasanFuqua School of BusinessSocial Interaction and Performance in the Sharing Economy: Evidence from UberAbstractThe “sharing” or “platform” economy is rising in importance as more individuals work on technology-enabled platforms. Much of the existing research on these technologies examines the economic efficiency of the transactions they broker. However, this research has overlooked the “social” nature of the interactions they mediate. In this article, we use transactional data from a cohort of UberX drivers and examine how the drivers’ performance varies with the quality of the passengers they serve. We test three theories that make competing predictions about this relationship. Contrary to research in traditional organizational settings, we find evidence that drivers assigned to highly rated passengers subsequently decrease how much they drive by one hour and how much they earn. Our findings suggest that, at the margin, UBER drivers are willing to trade money for positive interactions with passengers.
2017-12-18, 12:45–2:00 PMGross Hall, Rm 230EChristopher A. BailDepartment of SociologyOngoing Work
2017-12-11, 12:45–2:00 PMGross Hall, Rm 230EShankar BhamidiUniversity of North Carolina, Statistics & Operations researchTheory for statistical models for networks: A tale of five topics
2017-12-04, 12:45–2:00 PMGross Hall, Rm 230EKathleen GatesUniversity of North Carolina, Quantitative PsychologyDo communities even exist? Evaluating the robustness of results obtained from a community detection algorithmAbstractThis talk describes a method for evaluating the robustness of a final community detection solution. The approach builds from a commonly used heuristic first introduced in physics literature and currently popular in functional MRI applications. The motivation for adapting it for use with sparse, weighted networks came from work on the unsupervised classification of individuals based on their brain processes (as opposed to diagnostic category, age, etc.). Having arrived at a technique that successfully classified individuals into expected communities, there remained a need for an objective evaluation of the solution robustness in an absolute sense. The unsupervised classification technique is briefly introduced with the primary focus of the talk being the evaluation of the resulting community solutions. Examples from functional MRI, daily diary studies, diffusion tensor imaging, and social networks will be presented as well as results from simulated data.
2017-11-20, 12:45–2:00 PMGross Hall, Rm 230EBrian SouthwellRTI InternationalOn Curbing the Diffusion of Misinformation
2017-11-13, 12:45–2:00 PMGross Hall, Rm 230EBen GolubHarvard EconomicsTargeting Interventions in NetworksAbstractIndividuals interact strategically with their network neighbors, as in effort investment with spillovers among peers, or production decisions among firms connected by a supply chain. A planner can shape their incentives in pursuit of some goal — for instance, maximizing utilitarian welfare or minimizing the volatility of aggregate activity. We offer an approach to solving such intervention problems that exploits the singular value decomposition of network interaction matrices. The approach works by (i) describing the game in new coordinates given by the principal components of the network on which the game is played; and (ii) using that to deduce which components, and hence which individuals, a given type of intervention will focus on. Some of the principal components turn out to be standard measures of centrality and polarization, while other vectors in this canonical decomposition seem economically interesting but have not been studied before. Across a variety of intervention problems, simple orderings of the principal components characterize the planner’s priorities.
2017-11-06, 12:45–2:00 PMGross Hall, Rm 230ENoam ZerubavelColumbia UniversitySocially Connected Brains: Neural Foundations of Liking, Reciprocity, and Popularity in Human Groups
2017-10-30, 12:45–2:00 PMGross Hall, Rm 230EAlex VolfovskyDepartment of Statistical ScienceCausal inference on networks: from experiments to observational studies
2017-10-23, 12:45–2:00 PMGross Hall, Rm 230EJonah BergerUniversity of Pennsylvania – the Wharton SchoolAtypicality, Emotional Variability, and Cultural SuccessAbstractWhy do some cultural items (e.g., songs and movies) succeed while others fail? While some have argued that success is random, we suggest that fit with individual-level psychological processes plays an important role. Two projects use natural language processing to test this possibility. The first project tests whether the similarity between cultural items shapes their success. Natural language processing of thousands of songs examines the relationship between lyrical differentiation (i.e., atypicality) and popularity. Results indicate that the more different a song’s lyrics are from its genre, the more popular it becomes. This relationship is weaker in genres where lyrics matter less (i.e., dance) or where differentiation matters less (i.e., pop) and occurs for lyrical topics but not style. The second project examines emotional variability. We use textual analysis to plot the emotional trajectories of thousands of movies, and examine the link between emotional variability (i.e., short-term shifts in emotion) and success. Results indicate that more emotionally variable movies receive higher ratings, and this relationship is stronger in genres where uncertainty and surprise should be more desirable (e.g., thrillers and mysteries). Taken together, these two projects shed light on cultural dynamics, why things become popular, and the psychological foundations of culture more broadly.
2017-04-17, 12:45–2:00 PMGross Hall, Rm 230EAaron ChatterjiDuke, FUQUALearning to Manage: A field experiment in the Indian startup ecosystemAbstractManagement styles and practices are important determinants of firm performance. Yet, substantial variation exists across organizations with regards to management, suggesting frictions in the broader diffusion of management knowledge. We argue that peer networks may allow for the diffusion of productive management across firms. Using a randomized field experiment with 100 high-growth technology firms, we show that founders who received advice from other founders with more “hands-on” management styles, were more likely to reorient their own management activity, and subsequently experience lower employee attrition and higher rates of firm survival 8-months after the intervention. For founders who already had a more hands-on management style themselves, these interactions also increase top-line employee growth via an increase in hiring rates. Our study demonstrates that management can indeed diffuse across young firms via networks, though the process might be uneven and slow in practice.
2017-03-27, 12:45–2:00 PMGross Hall, Rm 230EAndrew SlaughterSenior Research Psychologist at the Foundational Science Research Unit of the Army Research InstituteSocial and network science with the Army: Research directions, funding opportunities, and examplesAbstractThe DoD (including the Army) funds a broad variety of basic research, including research in the social sciences. This talk will discuss a few of the broad research topics of interest to the Army, including topics related to network science; funding opportunities for research, with advice on how to pursue those opportunities; and — time permitting — discuss some current examples of Army network science research.
2017-03-06, 12:45–2:00 PMGross Hall, Rm 230EWalter PowellStandfordLump, split, or elevate? How a classification created a château tradition in 19th century Bordeaux, FranceAbstractHow and why does something temporary become resilient? Social classifications, industrial categories, and technical standards can be more or less malleable; but they each have the effect that the more they are used, the more they are reinforced. Such classifications both enable coordination and create boundaries. They can also evolve into powerful symbolic representations. We examine how a classification that was temporarily created in the mid-19th century persisted in the face of numerous challenges (e.g. wars, depression, and pestilence) and survives intact through today. Drawing on a wealth of data sources — rival rankings, international exhibitions, architectural buildings, naming conventions — we show how a temporary classification became venerated. We propose the concept of transcendence, that is, how a classification gains a superior, external quality, detached from its birth, and widely accepted as a social fact. Empirically, we show how the Bordeaux wine classification evolved into an invented tradition robust to numerous rival claims. More abstractly, our study illuminates how the messy and contingent origins of an institution are often forgotten and ennobling accounts are constructed.
2017-02-20, 12:45–2:00 PMGross Hall, Rm 230EJoseph FeldblumDuke Evolutionary Anthropology and Graduate Fellow at Kenan Institute for EthicsThe adaptive value of male relationships in the chimpanzees of Gombe National Park, TanzaniaAbstractIn many species of non-human primates, males form friendly social bonds while simultaneously competing with each other for high rank in a dominance hierarchy that determines mating access to females. While studies reveal a clear link between female bonds and fitness in several primate species, few studies have investigated such a relationship in males. Here, we investigate whether male social bonds, or position in cooperative networks, facilitate fitness benefits in one population of wild, free-ranging chimpanzees. We find that rank change was associated with social connectedness and betweenness in the network of coalition formation, but not with betweenness in the network of social relationships. Surprisingly, reproductive success was not associated with social connectedness or betweenness in coalitionary or social relationship networks, but grooming patterns strongly predicted reproductive success. Thus it appears that males who rise in rank have stronger bonds with group-mates, are central in the network of coalition formation, and groom others at a high rate (if they are low-ranking to begin with). However, males that successfully sire offspring are only those that groom others at a high rate. I will discuss the implication of these results for the applicability of social network analysis to the study of animal behavior.
2017-02-06, 12:45–2:00 PMGross Hall, Rm 230ECynthia RudinDuke, Computer Science and Electrical and Computer EngineeringTwo Mini Talks on Clustering and Feature SelectionAbstractI will present work in two mini-talks related to clustering in networks. In the first talk, there is a hidden network of crimes committed by the same individuals. In the second talk, the goal is to infer relationships between observations in machine learning. Both problems are (in a broad sense) subspace clustering problems. Mini Talk 1: Crime Series Detection via Subspace Clustering In crime series detection, the goal is to identify crimes that were committed by the same individuals. We cast this as a clustering problem with cluster-specific feature selection. It is joint work with my former student Tong Wang, and detectives Lt. Dan Wagner and Rich Sevieri of the Cambridge Police Department. Mini Talk 2: Bayesian Case Model The Bayesian Case Model (BCM) is a general framework for Bayesian case-based reasoning (CBR) and prototype classification and clustering. BCM brings the intuitive power of CBR to a Bayesian generative framework. The BCM learns prototypes, the “quintessentia” observations that best represent clusters in a dataset, by performing joint inference on cluster labels, prototypes and important features. This is joint work with Been Kim and Julie Shah.
2017-01-30, 12:45–2:00 PMGross Hall, Rm 230ETeague HenryUNC – Chapel Hill, Quantitative PsychologyModeling Heterogeneous Networks using Sender-Receiver Finite Mixture Exponential Random Graph ModelsAbstractModel-based inference on networks is made difficult by the interdependencies inherent in network data. The majority of model based inferential techniques for use on network data make an assumption of homogeneity, in that the data generating mechanism is identical for all edges and nodes within the network. However, failure to model potential heterogeneities can have wide-ranging effects on model misfit, both in terms of bias and efficiency, and these effects are made all the more problematic by the interdependency of the network. In this talk, we discuss heterogeneity in the framework of exponential random graph models, examine the consequences of leaving heterogeneity unmodelled, discuss the current modeling techniques to handle heterogeneity, and finally, introduce a novel method to handle heterogeneity in the form of the Sender-Receiver Finite Mixture Exponential Random Graph Model (SRFM-ERGM).
2017-01-23, 12:00–2:00 PMGross Hall, Rm 270Joint DUPRI and DNAC EventAbstractThe Duke Population Research Center (DPRC)/DUPRI and the Duke Network Analysis Center (DNAC) are looking for ways to bring together faculty interested in models of network influence, empirical studies testing these models and the use of network models for sampling populations. There are currently multiple network projects ongoing around the globe that are led by Duke researchers. If combined, this work would allow for unprecedented research on network-embedded social action. The returns to research would be new insights into how social networks shape health, development and strategic behavior around the globe. We are bringing together a group of faculty from Duke economics, political science, public policy, and sociology to meet and discuss their research featuring models of network influence and network data collection with the goal to identify mechanisms in which this work could benefit from common analysis tools, data infrastructure, and theory and spur collaborations.
2016-12-05, 12:45–2:00 PMGross Hall, Rm 230EKa Yuet-LiuUCLA, SociologyEmpirically Calibrated Simulation Experiment of Non-Medical Vaccine Exemptions and Disease Outbreak Potential in CaliforniaAbstractAn increasing number of U.S. children are entering schools without having received state-mandated vaccinations due to the rise of non-medical exemptions (NMEs). NMEs cluster spatially and create pockets of low immunization coverage. This study uses an empirically calibrated simulation model of all children under 18 in California to examine the effects of NMEs on outbreak potential. The model is calibrated with the observed spatial distributism of NMEs. We empirically calibrate the social contexts for children to come into contact with locational data (i.e., schools and shopping centers). Infection parameters are chosen to mimic that of measles. Such models help us gauge the relative contributions of the spatial clustering of NMEs and distributions of focal points to disease outbreak potential.
2016-11-21, 12:45–2:00 PMGross Hall, Rm 230EBlair SullivanNC StateAvoiding the Complexity Cliff in Network AnalysisAbstractHave you ever suddenly realized that every reasonable variant of the problem you’re interested in is NP-hard — dashing your dreams of ever having an efficient algorithm for analyzing that awesome new data set? If not, just wait – it’s almost guaranteed to happen (repeatedly) if you work with networks. What if there were an alternative to jumping head-first into the raging river of heuristics? In this talk, we discuss an approach to designing polynomial time algorithms that solve NP-hard problems by exploiting the non-uniformity of algorithmic complexity and inherent structure in the problem instance. We provide a gentle introduction to the main ideas of parameterized complexity, along with several concrete examples of how our group is employing these approaches to solve network analysis problems in computational biology and quantum computing.
2016-11-14, 12:45–2:00 PMGross Hall, Rm 230EDean KnoxMIT, Political ScienceA New Model for Path Data: Analyzing Sectarianism & Segregation on the Streets of BaghdadAbstractPath data describes the steps that an actor takes to get from point A to B. It offers researchers the opportunity to test theories about network navigation, e.g. in social and geographic networks. For example, path data can show whether individuals avoid out-group neighborhoods in their daily walking routes, resulting in societal inefficiencies and reducing inter-group contact. This data can also reveal how voters search social networks for political information, which may distort the information they ultimately receive. However, the sequential decision-making process in path data violates the underlying assumptions of existing models, which assume some form of conditional independence between observations. I propose a new random-path model (RPM) that explicitly captures this pathwise dependence, develop an estimation procedure, and demonstrate its properties. The RPM builds on a random-walk model, incorporating a realistic but difficult-to-analyze constraint to account for the fact that actors are purposefully navigating toward a destination. I validate the model in an analysis of the U.S. Interstate Highway planning process, where existing approaches fail to recover a known qualitative benchmark. Finally, the RPM is used to test two competing explanations of Baghdad’s recent segregation. Using smartphone-based behavioral data from Sunni and Shia participants in a field activity, I show that a need-based model of residential sorting—when families flee mixed neighborhoods to avoid political violence—is insufficient to explain participants’ walking routes alone. Instead, their choices reveal that conflict has also created significant taste-based aversion to out-groups in a city once known for its cosmopolitanism. These results suggest that societal preferences have shifted in a way that makes Baghdad’s eventual re-integration unlikely.
2016-11-07, 12:45–2:00 PMGross Hall, Rm 230EDaniel Della PostaCornell, SociologyClosure and Collaboration in the American MafiaAbstractHow do organizations obtain access to valued resources without diluting the loyalties and identities of their members? Network analysts suggest focusing on the boundary-spanning activities of “brokers” who bridge gaps in social structure. In many contexts, however, brokers are viewed with suspicion and distrust rather than rewarded for their diversity of interests. I examine organizations in which the theoretical deck is seemingly stacked against brokerage and toward parochialism: American-Italian mafia families. Using a historical network data set, I document a division of network labor in which a small number of brokers — often, surprisingly, ethnic outsiders excluded from formal membership — bridged otherwise disconnected islands of criminal activity to gain power within exclusive mafia circles. While social closure in solidary groups ensures a heavy premium on insider status, it can also paradoxically increase the returns to outsider brokerage, albeit only when taken up in a way that does not violate group norms.
2016-10-31, 12:45–2:00 PMGross Hall, Rm 230ERupert Freeman and Vincent ConitzerDuke, Computer ScienceFalse-Name-Proof Recommendations in Social NetworksAbstractWe study the problem of finding a recommendation for an uninformed user in a social network by weighting and aggregating the opinions offered by the informed users in the network. In social networks, an informed user may try to manipulate the recommendation by performing a false-name manipulation, wherein the user submits multiple opinions through fake accounts. To that end, we impose a no harm axiom: false-name manipulations by a user should not reduce the weight of other users in the network, and show that this axiom has connections to false-name-proofness. While it is impossible to design a mechanism that is best for every network subject to this axiom, we propose an intuitive mechanism Legit^+, and show that it is uniquely optimized for small networks. Using real-world datasets, we show that our mechanism performs very well compared to two baseline mechanisms in a number of metrics, even on large networks.
2016-10-24, 12:45–2:00 PMGross Hall, Rm 230EAnna MuellerU Chicago, SociologyWhy Networks Matter to Suicide: Examining the Structure & Content of Social Ties in a Suicide-Prone CommunityAbstractA large body of research suggests that exposure to suicide can increase an individual’s chance of experiencing suicidality; however, we know little about the mechanisms that confer this increased risk and even less about why suicide clusters form. Using qualitative data from an in-depth case study of a town with a significant history of repeated suicide clusters (N=110), we examine how both the structure of and culture embedded in social networks facilitate youth’s perception of suicide as an option. As with other social behaviors, we find that suicide is a social act replete with meanings for when, why, who, where, and how suicide happens. These meanings are (1) broadly shared in the community thanks to the highly-cohesive social networks and (2) somewhat unique to the community (compared to a reference group of respondents who do not live in the community). We also discuss evidence that these broadly shared understandings shape when youth see suicide as an option for themselves. Implications for the sociology of suicide and social networks research are also discussed.
2016-10-17, 12:45–2:00 PMGross Hall, Rm 230ERamon Lecuona Torras, Jonathan CummingsDuke, FUQUAThe impact of hierarchy on informal communication: Evidence from a field studyAbstractHow does hierarchy affect informal communication between co-workers? We empirically address this question by exploiting a shock that introduced (semi-random) variation in the communication costs between 105 employees of the headquarters of a Mexico-based multinational — i.e. the relocation of co-workers into a new office space where their physical location relative to other co-workers (including their direct boss) was determined by a lottery and not merely by the nature of work interdependencies. Our preliminary findings suggest that hierarchy ‘crowds out’ informal interactions between peers, even when communication is work-related. Namely, co-workers who are within closer proximity to their direct boss are less likely to sustain informal conversations with peers. We suggest that these effects are partly attributed to (often unintended) monitoring that is inherent to supervisor-coworker relations. We exploit the richness of our empirical setting to develop conceptual explanations for our findings and discuss the implications for managers.
2016-10-03, 12:30–2:00 PMGross Hall, Rm 230EVariousDuke UniversityFlash Talks Part 2AbstractCome and hear a variety of talks on research that is currently being working on! We hope these discussions help spark new ideas and future collaborations over the course of the semester!
2016-09-26, 12:30–2:00 PMGross Hall, Rm 230EVariousDuke UniversityFlash Talks Part 1AbstractCome and hear a variety of talks on research that is currently being working on! We hope these discussions help spark new ideas and future collaborations over the course of the semester!
2016-04-25, 12:30–2:00 PMGross Hall, Rm 270Ian McCullohJohns Hopkins, Applied Physics LabWhat is really online? Bias, sampling, and platform error in online networks.AbstractThere has been a recent increase in the use and study of online data, such as social media. Several scientists have explored the differences between online data and offline data that is traditionally used in social science research. This talk will review several of these differences, highlighting potential bias, sampling error, and platform error in online data. Despite the differences, bias, and error, online data provides valuable insight for applications ranging from sociology to marketing to organizational behavior. Management of social media fire storms (negative publicity) will be discussed. In addition, an application of online data is presented that promises to aid managers in dynamically moving their team between agility and efficiency as organizational requirements change. The goal of this talk is not to provide finished solutions, but rather to promote potential research threads, innovation, and identify collaboration opportunities.
2016-04-18, 12:30–2:00 PMGross Hall, Rm 270Joost van de Brake & Jonathon CummingsGroningen, Economics and Business & Duke, FuquaMultiple Team Membership and Team Performance: The Moderating Effect of Time FragmentationAbstractKnowledge-intensive work is increasingly carried out by teams whose members are also involved other teams. Multiple team membership (MTM) can increase a focal teams’ performance when members are able to access relevant knowledge from their other teams. At the same time, however, MTM can decrease a focal teams’ performance when members are unable to coordinate work within the team. We examine MTM in a longitudinal study of 5540 teams over 4 years in a large knowledge-intensive organization from the Netherlands. Employing a network perspective on teamwork, we focus on two distinct dimensions of MTM: the number of other teams in which members of the focal team are concurrently involved (breadth of external network), and membership overlap in other teams with members from the focal team (density of external network). In models controlling for characteristics of the teams, we find that membership overlap is positively related to team performance, while the number of other teams is not related to team performance. Furthermore, the effect of MTM on team performance is moderated by time fragmentation: the amount of non-overlapping time spent by members on the task (sparsity of internal network). When time fragmentation is low, both dimensions of MTM are positively related to team performance. When time fragmentation is high, neither dimension is related to team performance. These results have important implications for the design of effective teams and networks.
2016-04-04, 12:30–2:00 PMGross Hall, Rm 270Damon CentolaU Pennsylvania, CommunicationsSpontaneous Social Conventions: An Experimental Study of Cultural EvolutionAbstractSocial conventions are the foundation of social cooperation and productive economic activity, yet very little is known about how and when they form. Prominent theories argue that widely shared social conventions depend up on coordinating mechanisms, such as incentives for global coordination, aggregated information, and social leadership. We explore a competing ‘evolutionary’ theory of conventions, which hypothesizes that broad social coordination can emerge without any of these mechanisms. We use an Internet experiment to study the real-time evolution of endogenous collective behaviors from a ‘state of nature’ in which there are an infinite number of possible cultural conventions and no incentives for global coordination. Our results confirm our formal hypotheses, demonstrating that changes to network connectivity can generate the spontaneous formation of global social conventions. The results have unexpected implications for the evolution of culture in the expanding online domain.
2016-03-28, 12:30–2:00 PMGross Hall, Rm 270Örjan BodinStockholm Resilience CentreSocial-Ecological Networks — an emerging transdisciplinary approach to study social-ecological systemsAbstractThe network perspective is increasingly put forth as an analytical framework well suited to studying complex social-ecological systems. The underlying rationale is that the network approach as such is generic and allows the research to model any kind of systems as consisting of separated but interlinked components of different kinds. Hence, integrated social-ecological systems could be analyzed as social–ecological networks where interdependent ecological entities (e.g. species and habitat patches) and social entities (e.g. users, managers, agencies and NGOs) could be simultaneously incorporated in a common systems model. The last few years a number of studies have been conducted drawing on a social-ecological network approach. In this talk I will present some of these recent developments of this emerging line of research. I will demonstrate how a social-ecological network approach has been used to analyze complex patterns of actor/resource interdependencies in a small artisanal fishery in east Africa; the governance of fragmented forest patches in a rural agricultural landscape in Madagascar and in a large-scale biodiversity conservation initiative in Australia; and in land use planning of wetlands in the Stockholm County in Sweden.
2016-03-21, 12:30–2:00 PMGross Hall, Rm 270Joseph FeldblumDuke, Evolutionary AnthropologyCommunity fission, cohesion, and a “Four-Year War” in two populations of wild chimpanzeesAbstractIn 1973, researchers in Gombe National Park, Tanzania, observed the only apparent example of community fission in wild chimpanzees. Over the next four years, males of the northern daughter community killed all adult males and one female of the southern community, and eventually annexed their territory. However, an alternative hypothesis suggests that the two communities were always separate, brought together by human provisioning. Here, we test these competing claims, investigate potential causes of the hypothesized fission, and outline future comparative work to investigate the predictors of social cohesion in chimpanzee communities. Using male association, grooming and ranging patterns from 1967 to 1972, we employed modularity optimizing algorithms to identify the timing and nature of subgrouping, finding a dramatic increase in modularity beginning in 1971, two years before the complete fission. We then aligned this increase with candidate catalysts of the fission to determine likely proximate causes. Finally, we compare Gombe patterns with those in Ngogo, a large community in Uganda that some researchers believe may also be in the process of fissioning, from 2003 to 2008. Modularity in Ngogo rose slowly over the study period, but did not evidence the sharp increase seen in Gombe. I will discuss future work using more data from Gombe and Ngogo to investigate the general relationship between our proposed fission catalysts and modularity in Ngogo and later years in Gombe.
2016-03-07, 12:30–2:00 PMGross Hall, Rm 270Lu WangDuke, Statistical ScienceBayesian Network—Response RegressionAbstractIt is of increasing interest to learn how the human brain network varies as a function of continuous features, but flexible procedures to accomplish this goal are limited. We develop a Bayesian semiparametric model, which combines low-rank factorizations and Gaussian process priors to allow flexible shifts of the conditional expectation for a network-valued random variable across the values of a predictor, while including subject-specific random effects to improve prediction. We provide a simple Gibbs sampler along with procedures for inference, prediction, and goodness-of-fit assessments. The model is applied to learn changes in the brain network across intelligence scores.
2016-02-29, 12:30–2:00 PMGross Hall, Rm 270Andrei BoutylineUC at Berkeley, SociologyStudying the Structure of Attitudes with Belief Network AnalysisAbstractTheories of the structure of political belief systems typically conceive of them as networks of interrelated opinions, in which some beliefs are central and others are derived from these more fundamental positions. In this paper, we formally show how such structural features can be used to construct measures of belief centrality that are based on direct comparisons of relative positions of beliefs in a network of correlations. To demonstrate the usefulness of this method and contrast it with existing techniques, we examine belief networks we construct from the 2000 American National Election Study. While regression analyses of these data have been used to argue that political beliefs are organized around cultural schemas of parenting, our structural approach contradicts this interpretation. Instead, our results are broadly consistent with the conception of political identity as a heuristic device for acquiring attitudes. To search for possible heterogeneity, we then separately examine belief networks belonging to 44 different demographic subpopulations. These analyses indicate that belief systems of different groups vary in the extent to which they are organized, but rarely vary in the logic around which they are organized. While our analyses focus on political beliefs, techniques we introduce here can be applied to many other cultural domains.
2016-02-22, 12:30–2:00 PMGross Hall, Rm 270Ron BurtU Chicago, Sociology and StrategyIntegrating brokerage and closure in a general model of social capital: from bridges, to embedding, to guanxi
2016-02-15, 12:30–2:00 PMGross Hall, Rm 230ECharles SeguinUNC at Chapel Hill, SociologyNaming the Gender Binary: a Machine Learning Approach to Analyzing Gendered AestheticsAbstractNew children’s names are constantly introduced, and old names are continually rising or falling in popularity, yet these names continue to maintain a rigid separation between genders. This gender binary in children’s names exemplifies a more general puzzle: the specific content of boundary markers is constantly shifting, yet the symbolic boundary itself often remains. I use computational, or “big data,” techniques to develop a quantitative measure of names’ gender aesthetics — whether a name shares aesthetic features with predominantly boys’ or girls’ names. I then apply this measure to a dataset of US names from 1880-2009, in order to analyze the mechanisms which reproduce the gender boundary in children’s names. Results suggest that 1) the symbolic boundary in children’s names is maintained largely through aesthetic heuristics, such as girls being given names ending in a schwa, or short vowel sound (e.g. the final syllable in Emma), rather than more contemporaneous social influence processes; 2) the gender aesthetic is remarkably durable, and even novel names conform to prior aesthetics; 3) when aesthetic boundary crossing does occur, it is much more likely with girls’ than boys’ names; 4) boys’ names with girls’ aesthetic properties generally require some exogenous signal of their gender; 5) as naming practices become more diverse, they also conform more tightly to a gender aesthetic. These results suggest one way in which culture can be both stable and dynamic: invention and innovation within individual cultural objects mostly takes place within the aesthetics of existing social and symbolic boundaries.
2016-02-08, 12:30–2:00 PMGross Hall, Rm 270Alex VolfovskyHarvard, StatisticsTesting and estimation for relational dataAbstractRecent years have seen a dramatic rise in social media, networks, and other settings in which the relationships and interactions between individuals, countries or objects are observed. These types of relational data are often represented as a square matrix, the entries of which record the relationships between pairs of objects. Statistical methods for such data range from network regression where entries are frequently assumed to be independent to latent space methods that assume some degree of similarity or dependence between objects in terms of the way they relate to each other. However, formal tests for such dependence have not been developed. First, we provide a test (based on the observation of a single relational data matrix) for such dependence using the framework of the matrix normal model, a type of multivariate normal distribution parameterized in terms of row- and column-specific covariance matrices. Second, we develop an estimation procedure (still based on the observation of a single relational data matrix) that captures the variability in such data by leveraging the identical index sets of the rows and columns.
2016-02-01, 12:30–2:00 PMGross Hall, Rm 230EZack NealMichigan State, PsychologyAre cohesive, integrated neighborhoods possible?AbstractThere is a wealth of empirical evidence that there is a negative relationship between neighborhood spatial integration and social cohesion. The most segregated neighborhoods tend to be the most cohesive, while more integrated neighborhoods tend to be more socially fragmented. In this presentation, I use an agent-based simulation model of neighborhood network formation to understand why this pattern exists. I then extend the model to explore possible solutions that involve changing behaviors and changing neighborhood spaces.
2016-01-25, 12:30–2:00 PMGross Hall, Rm 230EKieran HealyDuke, SociologyTopic Areas, Gender, and Citation in Philosophy, 1993-2013
2015-11-09, 12:30–2:00 PMGross Hall, Rm 230EStephen VaiseyDuke, SociologySymbolic Boundaries and Peer Influence on Alcohol UseAbstractPeer influence is a fundamental process in social life. Network scientists generally conceptualize influence as an additive vector of forces, with increasing exposure to the behavior or attitudes of others associated with a greater likelihood of adopting them. Based on cognitive theories of symbolic boundaries, we propose an alternative model that regards peer influence as conditional on membership in the same symbolic group. We test this model with data from two large, nationally-representative datasets and find that, for religious adolescents, having a co-religious tie who also drinks is more strongly predictive of drinking behavior than exposure alone. Although the results do not clearly indicate a causal relationship, they are consistent with the idea that, at least for some behaviors, peer influence can be modeled as the violation of symbolic group boundaries.
2015-11-02, 12:30–2:00 PMGross Hall, Rm 230EChris GlynnDuke, StatisticsBayesian Analysis of Dynamic Linear Topic ModelsAbstractIn dynamic text analysis, the proportion of a document characterized by a semantic topic may depend on the time trend of that topic’s overall prevalence and covariates of the document itself. We extend the Dynamic Topic Model of Blei and Lafferty (2006) by explicitly modeling document-level topic proportions with covariates and dynamic structure that includes time trend and periodicity. A Markov Chain Monte Carlo algorithm that utilizes Polya-Gamma data augmentation is developed for posterior inference. Conditional independencies in the model and sampling are made explicit, and our MCMC algorithm is parallelized where possible to allow for inference in large corpora. To address computational bottlenecks associated with Polya-Gamma sampling, we appeal to the Central Limit Theorem to develop a Gaussian approximation to the Polya-Gamma random variable. This approximation is fast and reliable for parameter values relevant in the text-mining domain. Our model and inference algorithm are validated with multiple simulation examples, and we consider the application of modeling trends in PubMed abstracts.
2015-10-26, 12:30–2:00 PMGross Hall, Rm 230ESolomon MessingPew Research Center’s Data LabsIdeological diversity in news and opinion on FacebookAbstractExposure to news, opinion, and civic information increasingly occurs through social media. How do these online networks influence exposure to perspectives that cut across ideological lines? Using deidentified data, we examined how 10.1 million U.S. Facebook users interact with socially shared news. We directly measured ideological homophily in friend networks and examined the extent to which heterogeneous friends could potentially expose individuals to cross-cutting content. We then quantified the extent to which individuals encounter comparatively more or less diverse content while interacting via Facebook’s algorithmically ranked News Feed and further studied users’ choices to click through to ideologically discordant content. Compared with algorithmic ranking, individuals’ choices played a stronger role in limiting exposure to cross-cutting content.
2015-10-19, 12:30–2:00 PMGross Hall, Rm 230EPeter HoffUniversity of Washington-Seattle, Statistics and BiostatisticsTensor regression for dynamic network dataAbstractA fundamental aspect of relational data, such as from a social network, is the possibility of dependence among the relations. In particular, the relations between members of one pair of nodes may have an effect on the relations between members of another pair. In this talk I describe a type of regression model to estimate such effects in the context of longitudinal and multivariate relational data. I will also discuss an extension of this model to accommodate ordinal network data, such as counts of international relations events between countries.
2015-09-28, 12:30–2:00 PMGross Hall, Rm 230EDan SussmanHarvard, StatisticsStatistical Inference for Networks via Spectral EmbeddingAbstractIn this talk we will discuss using spectral methods for network analysis. We will show how a spectral embedding provides consistent estimates for latent positions in the random dot product graph model. This leads to accurate subsequent inference for a variety of tasks. An application to diffusion tensor MRI will be briefly discussed before overviewing some more practical aspects of spectral embedding.
2015-09-21, 12:30–2:00 PMGross Hall, Rm 230EKaoru IrieDuke, StatisticsFox News Network Data Analysis: Bayesian Dynamic ModelingAbstractWe propose a Bayesian approach to analyze data on Internet traffic flow among Fox News websites. The observations are time-varying counts (non-negative integers), so the straightforward application of existing Gaussian-type state-space models is not available. It is a Big Data problem, with many different types of articles, raising scalability issues; however, sparsity can be exploited in both modeling and computation. These features of the data motivate use of dynamic versions of count data models (Poisson-Gamma models and Multinomial-Dirichlet models), and lead to fitting an interpretable Gravity model that is an equivalent to two-way ANOVA. The conjugacy of this model enables use of Forward Filtering and Backward Sampling to obtain the posterior distributions. In addition, the Gravity model reveals the underlying structure of traffic networks across websites, allowing the detection of significant flows and flow dynamics, and enabling computational advertisers to better target their ad campaigns. This is the joint work with Xi Chen, David Banks, Mike West (Duke), Robert Haslinger and Jewell Thomas (MaxPoint).
2015-09-14, 12:30–2:00 PMGross Hall, Rm 230EBill ShiUniversity of Chicago, Computation InstituteCan We Agree on Science? Measuring the Ideological Alignment of Science with Book Co-purchase Data
2015-04-20, 12:30–2:00 PMGross Hall, Rm 230ESinan AralMIT, ManagementThe Dynamics of Social Influence and Relational ReputationAbstractIdentity and reputation drive some of the most important relational decisions we make online: Who to follow or link to, whose information to trust, whose opinion to rely on when choosing a product or service, whose content to consume and share. Yet, we know very little about the dynamics of social influence and relational reputation and how they affect our decision making. Sinan will describe a series of large scale experiments that explore the population level behavioral dynamics catalyzed by social influence, identity and reputation online. He will explore some of the implications for bias in online ratings, the foundations of social advertising and the ability to generate cascades of behavior through peer to peer influence in networks. Sinan will argue that new research on social influence and relational reputation could help guide our platform design and social policy decisions in light of the rising importance of peer effects and reputation online.
2015-04-13, 12:30–2:00 PMGross Hall, Rm 230EJacob FisherDuke, SociologySocial space diffusionAbstractSocial networks represent two different facets of social life: (1) stable paths for diffusion, or the spread of something through a connected population, and (2) random draws from an underlying social space, which indicate the relative positions of the people in the network to one another. The dual nature of networks creates a challenge — if the observed network ties are a single random draw, is it realistic to expect that diffusion only follows the observed network ties? This study takes a first step towards integrating these two perspectives by introducing a social space diffusion model. In the model, network ties indicate positions in social space, and diffusion occurs proportionally to distance in social space. Practically, the simulation occurs in two parts: positions are estimated using a latent space model, and then the predicted probabilities of a tie from that model — representing the distances in social space— or a series of networks drawn from those probabilities — representing routine churn in the network — are used as weights in a weighted averaging framework. Using a school friendship network, I show that the model is more consistent and, when probabilities are used, the model converges faster than diffusion following only the observed network ties.
2015-04-06, 12:30–2:00 PMGross Hall, Rm 230EOtti D’HuysDuke, PhysicsDynamics of Autonomous Boolean NetworksAbstractAutonomous Boolean networks are known to display complex dynamics, originating from the absence of an external clock, internal time delays and the non-ideal behaviour of the logic gates. We study experimentally such networks on a field-programmable gate array (FPGA). We briefly discuss some dynamical phenomena that can be observed in our experiments. In particular, we show how networks consisting of logic elements can produce long transients, due to an interplay of noise, asymmetry and network structure. Moreover, we demonstrate how these transients can be used to successfully perform reservoir computing, a new machine learning paradigm.
2015-03-30, 12:30–2:00 PMGross Hall, Rm 230ETiantian YangDuke, SociologyForged in the Heat of Battle: New Organizations as Business IncubatorsAbstract(paper by Tiantian Yang, Duke University; Howard Aldrich, University of North Carolina-Chapel Hill; Frederic Delmar, Lund University) Freeman’s (1986) provocative idea that entrepreneurs arise from existing organizations and are thus organizational products represented a watershed moment in research on entrepreneurship. His path-breaking idea called attention to the intersection of organizations and entrepreneurship, spurring scholars to investigate how organizations differ in their employees’ propensities to leave and launch new businesses (Audia and Rider 2005, Brittain and Freeman 1980, Sørensen and Fassiotto 2011). In our paper, we build on previous research by examining the organizational conditions under which employees who exit wage labor jobs subsequently create their own businesses, in a national context which many scholars view as particularly unfavorable to entrepreneurship: Sweden (Andersson and Klepper 2013, Delmar and Davidsson 2000, Henrekson 2005, Lerner and Ta ÌŠg 2013). In the process, we illuminate the importance of taking national institutional context into account in understanding conditions facilitating or impeding entrepreneurship generated by existing firms within a country. To conduct a rigorous analysis of the causal mechanisms of entrepreneurial spawning in organizations, we adopt a research design that permits us to effectively address the remaining endogeneity bias that may arise from individuals’ self-selection after taking into account the contextual mechanisms that affect sorting. We use data from the Longitudinal Integration Database for Health Insurance and Labor Market Studies (LISA) in Sweden that tracks every employee and every organization in the private sector from 1989 to 2002. Rich information about the complete career histories of individuals and their employing organizations allows us to apply a Shrinkage/Empirical Bayes method to effectively address the endogeneity bias that may arise from individuals’ self-selection even if information about individuals’ self-selection is incomplete. We believe that a theoretical framework with a clearer conceptualization of the institutional context and a rigorous analysis will enable us to more thoroughly investigate the causal effects of organizational context, and deepen our understanding of how social conditions affect individuals’ entry into entrepreneurship.
2015-03-23, 12:30–2:00 PMGross Hall, Rm 230Ethe groupVisualisation DayAbstractThe goal of this workshop is to puzzle through some aspects of data analysis/manipulation common to network analysis. We focus on how to visualize network data to help answering particular questions of interest. We will also share what people come up with as best practices in doing so, talking through the methods used, the logic behind them and so forth. If you have a network that has been a challenge to visualize, contact James Moody or Achim Edelmann—ideally, include the data (properly anonymized, etc.) and, importantly, the questions you seek to gain traction on with a visualization. We will then post those requests to the DNAC site and invite visualization attempts.
2015-03-16, 12:30–2:00 PMGross Hall, Rm 230EJohn Levi MartinU Chicago, SociologyPersistence and Re-Formation of Close Personal Ties over a Long, Long, Long Time.AbstractPersonal relationships are embedded in both spatial and relational contexts. Using data on 60 intentional communities from the Urban Communes Data Set, we examine how such embedding is related to the persistence and re-formation of close personal ties over a thirteen year period. We find that the precise ways in which local structure affects contact are bound up with the distance between dyad members. We also find asymmetries in these processes that other studies have been unable to uncover—that change in contact is not the same as change in friendship; that processes that lead ties to be dropped are not the same as those that lead them to be renewed; that increases in local embeddedness are not opposite of decreases. Finally, there is surprising evidence of hierarchical effects influencing the retention of friendships more than twenty-five years after most respondents left their groups.
2015-03-02, 12:30–2:00 PMGross Hall, Rm 230ESaray ShaiUNC, Department of MathematicsAttacks of modular networksAbstractModularity is a key organization principle in many systems around us. Social, technological and biological systems are organized into cohesive groups of elements, called modules. The relatively sparse interactions between the modules are critical to the functionality of the system, and are often the first to fail, as for example the case in neuronal networks where aging and schizophrenia could result in a damage to the interconnected nodes. Here we quantify the implications of such failures to the resilience of multi-scale modular systems. We find analytically a “tipping point”, which distinct between two regimes. In one regime, the modules remain functional but become disconnected, while in the other regime the modules themselves are damaged causing the system to collapse. Our model provides insights into the role modularity plays in real world systems, while considering advanced types of attacks that address the multilevel nature of the system.
2015-02-23, 12:30–2:00 PMGross Hall, Rm 230ELuke MaierDuke, Public PolicyCo-evolving Networks of International Conflict and CooperationAbstractNational security scholars and practitioners have an ongoing need to anticipate future conflict and ways to avert it using non-violent means. The three majors schools of international relations offer alternative predictions about how states behave in the face of uncertainty and insecurity, and political scientists are increasingly applying computational methods to operationalize these bodies of theory. Scholars in the liberal school emphasize that economics, transnational organizations, and diplomacy can play strong mediating and information-sharing roles in averting (or causing) interstate conflict. Indeed, since 1949, the number and connectivity of international organizations and the world economy have burgeoned, while large-scale interstate wars seem less frequent and less global. Noting this prima facie correlation, this project’s goal is to model conflict probability at two levels: at the aggregate network-level and that of individual states. For the network level, we will assess the covariance between trade and institutional connectivity and conflict incidence in the overall system. For individual states, we will use their relationships with other actors and those dyads’ network statistics to model their dyadic conflict propensity. For both levels of analysis, the liberal school would generally predict mutual membership in international institutions and strong trade ties negatively correlate with conflict propensity. **We would like feedback on how to operationalize this to test causality. So far, the project has focused on quantifying changes in the international system in a descriptive sense. But, we would like to explore the liberal school’s hypothesis to the extent that network analysis allows, and perhaps move towards an instrumental forecasting model.
2015-02-16, 12:30–2:00 PMGross Hall, Rm 230E~ cancelled due to snow storm ~
2015-02-09, 12:30–2:00 PMGross Hall, Rm 230EYuli Patrick HsiehRTI, Digital Technology and Society ProgramCheck the phone book: Testing information and communication technology (ICT) recall aids for personal network surveysAbstractHow can medical practitioners at a local clinic encourage residents to come in for cancer screening? How can politicians identify the opinion leaders and information gatekeepers among their constituency? Have Americans become more socially isolated than they were thirty years ago? Social scientists have employed the name generator procedure to collect self-reported information about individuals’ personal networks to examine the implications of interpersonal environments for people’s attitudinal and behavioral changes. However, self-reports on personal networks of close friends and acquaintances often result in significant underreporting of such information, which impairs the accuracy of findings and inferences drawn from personal network research. In this talk, I will present the findings of my randomized survey experiment that examined the effect of information and communication technology (ICT) recall aids for the name generator procedure from the General Social Survey. I will show how the study participants produced more comprehensive reports on their personal networks and how the quality of such data were evaluated. I will then discuss the design recommendations for network survey instruments as well as the broader implications of the use of ICT recall aids for personal network research in social science fields.
2015-02-02, 12:30–2:00 PMGross Hall, Rm 230ESusan AlbertsDuke, BiologySocial relationships and survival in a wild primate populationAbstractSocial integration and support can have profound effects on human survival. The extent of this phenomenon in non-human animals is largely unknown, but such knowledge is important to understanding the evolution of both lifespan and sociality. This topic represents a current area of synthesis between the biological and social sciences. I present a brief overview of this topic through the lens of each of these distinct disciplines, then summarize the state of knowledge about the topic as a consequence of these bodies of work. I then report evidence from my research project that levels of affiliative social behavior (i.e., ‘social connectedness’) with both same-sex and opposite-sex conspecifics predict adult survival in wild female baboons. In the Amboseli ecosystem in Kenya, adult female baboons that were socially connected to either adult males or adult females lived longer than females who were socially isolated from both sexes—females with strong connectedness to individuals of both sexes lived the longest. Female social connectedness to males was predicted by high dominance rank, indicating that males are a limited resource for females, and females compete for access to male social partners. To date, only a handful of animal studies have found that social relationships may affect survival. This study extends those findings by examining relationships to both sexes in by far the largest data set yet examined for any animal. Our results support the idea that social effects on survival are evolutionarily conserved in social mammals
2015-01-12, 12:30–2:00 PMGross Hall, Rm 230EYong-Mi Kim & Jonathon CummingsUniversity of Michigan, School of Information & Duke, Fuqua School of BusinessWho uses enterprise collaboration systems?AbstractWhile public social networking sites, such as Facebook and Twitter, have garnered a lot of attention from academic researchers, private social networking sites within corporations are less well understood. Often referred to as enterprise collaboration systems, these “social” tools in the workplace enable employees to set up online communities, share information and documents, and connect through wikis, instant messaging, and video conferencing. We study an enterprise collaboration system in a Fortune 500 company, and focus in particular on the online communities that employees use to collaborate with one another electronically. Analyzing HR data from 121,849 employees and usage data from 15,535 communities between 2010-2014, we explore (1) how community users differ from non-users, (2) how frequent community users differ from infrequent community users, and (3) how diverse the communities are relative to the company as a whole (e.g., gender, highest degree earned, company tenure, geographic location, organizational unit, and hierarchical level).
2015-01-12, 12:30–2:00 PMGross Hall, Rm 230ERobin Dodsworth & Richard BentonNCSU, English & SociologySchool Co-Attendance Networks and the Southern Vowel ShiftAbstractMany linguistic features that have defined the Southern United States for at least a century have recently been retreating in urban areas. In particular, the traditional Southern vowel system is shifting toward a non-regional system, largely as the result of proximity and exposure to migrants from other regions. Because vowel systems are mostly established by late adolescence, contact among children from distinct regions in school is a driving force in the retreat from Southern vowels. The present study explores how the Southern vowel shift is moderated by the school co-attendance network–a social network constructed from speaker-to-school ties. We apply this framework to a sample of 147 speakers in Raleigh, North Carolina. We employ quadratic assignment procedure (QAP) models to show that similarity in speaker birth-year predicts similar linguistic features but that this pattern is moderated by school co-attendance. We demonstrate that social network analysis provides a useful analytic framework for investigating linguistic variation.
2014-12-08, 12:30–2:00 PMGross Hall, Rm 230ECrystal Wiley Cené & Laura ShebleUNC (School of Medicine) & UNC (Center for Health Equity Research)Systems Science Methods and Health Part 1: System Dynamics and Network AnalysisAbstractWe will present a work-in-progress from a larger project/series of papers focused on the use of systems science methods (Systems Dynamics, Individual-based Modelling, and Network Analysis) in health using bibliometrics. We will present a few results from the Systems Dynamics search, as well as from the Network Analysis search (which is ongoing). This will be a very interactive session and we are looking for input/feedback from the group! We are also interested in identifying potential collaborators, so please come out and share your thoughts/input.
2014-11-24, 12:30–2:00 PMGross Hall, Rm 230EYing ShiDuke (Public Policy)Success by Degrees: Adolescent Popularity and Future EarningsAbstractAre there labor market returns to high school popularity? One can reasonably argue that friendship nominations contain valuable information about an individual’s social skills and social capital that matter for future employment. We define popularity using both local (degree) and global (Bonacich) centrality measures that vary the radii of individuals’ influence in school-based peer networks. We furthermore construct measures based on the alters’ attributes, namely, distinguishing between nominations from friends of the same vs. opposite sex. Using longitudinal and network data from Add Health, we examine how education, beauty, and other contextual factors moderate the influence of popularity on future earnings. Evidence shows that moving from the 10th to the 90th percentile of both local and global popularity distributions is associated with a 10 percent earnings premium for individuals in their late 20s and early 30s. ~ TBA ~
2014-11-17, 12:30–2:00 PMGross Hall, Rm 230EWeihua (Edward) AnUniversity of Indiana (Statistics)Health Surveillance through Social NetworksAbstractWe propose a network-based method to monitor health behaviors and point out the general conditions for it to work effectively. The method helps to identify effective informants for monitoring future health behaviors and to triangulate self-reports of sensitive health behaviors. We demonstrate the method by studying the smoking behaviors of over 4,000 middle school students in China. Using students’ observations of their schoolmates smoking, we construct smoking detection networks and examine the patterns of smoking detection. We find that smokers, optimistic students, and popular students make better informants than their counterparts. We also find that using three to four (or the 3rd quartile of) positive peer reports can uncover a good number of under-reported smokers while not producing excessive false positives.
2014-11-10, 12:30–2:00 PMGross Hall, Rm 230EMalgorzata TuralskaDuke (Physics)Networks of influence: transmission of information in systems of cooperative decision makers.AbstractThe surprising social phenomena of the Arab Spring and the Occupy Wall Street movement posit the question of whether the active role of committed groups may produce political changes of significant importance. Under what conditions are the convictions of a minority going to dominate the future direction of a society? To address this issue, we study a cooperative decision making (CDM) model, adopting a minimal fundamental assumption that people make decisions by imitating the behavior and actions of others. The CDM model generates consensus among the individuals within a model society through a phase-transition process. However, the global consensus state is not permanent and times of crisis occur when there is an ambiguity concerning a given social issue. Surprisingly, the instances of crisis are characterized by an increased correlation between the members of the society, which facilitates the transmission of the opinion of a small committed minority, leading to substantial changes in social consensus. To further explore the consequences of the presence of correlations extending across the entire system, we study the conditions under which network of individuals responds to an external source of information. We observe that the most efficient information transmission is closely connected to the criticality phenomenon. Finally, we discuss how the influence that the society exerts on an individual can be analytically approached with the help of fractional calculus.
2014-10-27, 12:30–2:00 PMGross Hall, Rm 230ERon BreigerUniversity of Arizona (Sociology)Multivariate Analysis as a Network ProblemAbstractThere was a time when network analysis was concerned exclusively with who-to-whom (“one-mode”) data. Much of the history of network research however has been written as the result of an expanded vision as to what constitutes a network (consider for example: affiliation networks, multi-mode formulations, and McPherson’s ecology of organization types based on overlaps among typical members within an innovative conceptualization of multivariate space). Regression modeling and its many generalizations aim to study networks among variables; relations among the cases are, for the most part, rendered invisible. However, David Melamed (U. of South Carolina) and I and other members of my research group have recently been formulating a dual to regression modeling that I will present and illustrate in this talk. We seek to use the variables to learn about the cases. Building on existing results, bringing them together in new ways and adding a bit, we show how the regression coefficients produced in conventional analyses may be usefully understood as sums across cases and clusters of cases (a two-mode formulation). Predicted values on the outcome variable in logistic (and other) regression models may be seen to be produced from a particular (one-mode projection) network among the cases. Among the gains of our approach: aggregating regression coefficients over an entire sample may mask systematic variability that our approach helps to sort out (some sets of cases may be associated with strong positive effects while others exhibit strong negative effects). We use an analysis of clustering among the cases to help us uncover statistical interactions among variables. We show that standard regression models (and generalizations) can be understood from the perspective of sociological field theory. Rather than “transcending” general linear reality, we seek to get more out of it.
2014-10-20, 12:30–2:00 PMGross Hall, Rm 230ENishant MalikUNC (Mathematics)Social environment and social clustering in spread of opinions in co-evolving networksAbstractTaking a pragmatic approach to the processes involved in the phenomena of collective opinion formation, first we investigate two specific modifications to the co-evolving network voter model with multiple opinions. (i) We replace the rewiring probability parameter by a distribution of probability of accepting or rejecting opinions between individuals, accounting for the asymmetric influences in relationships among individuals in a social group. (ii) We modify the rewiring step by a path-length-based preference for rewiring that reinforces local clustering. We have investigated the influences of these modifications on the outcomes of the simulations of this model. Our findings indicate that reinforcement of clustering in a voter model can have significant ramifications. To further explore the consequences of the presence of non-trivial clustering in voter model on co-evolving networks, we have studied a simplified version of the above model. Here, we consider only two opinions with a new parameter for dynamic updating of clustering. Simplicity of this model allows us to study it analytically as well. A comparison of our simulations with analytical results will also be presented.
2014-10-06, 12:30–2:00 PMGross Hall, Rm 230EJukka-Pekka OnnelaHavard (Biostatistics at School of Public Health)Cell phones, social networks, and public healthAbstractCell phones are now ubiquitous: it is estimated that the number of phones in use exceeds the size of the global population in 2015. Our recent and ongoing work uses call detail records (CDRs) to investigate the structure of large-scale social networks and their relationship to underlying geography. A parallel approach, one that generates richer data, is based on instrumenting a cohort of subjects with a customized smartphone application. This enables the collection of both “active data” (such as implementing traditional surveys on the phone) as well as “passive data” (such as deriving behavioral data from phone sensors). I will talk about some of our work that utilizes these two approaches, what types of insights they may yield to the study of social networks and behavior, and how this work interfaces with public health. Finally, I will highlight some results from our ongoing work with psychiatric outpatients.
2014-09-29, 12:30–2:00 PMGross Hall, Rm 230EBrad FultonDuke (Sociology)Bridging and Bonding: How Social Diversity Influences Organizational PerformanceAbstractAlthough many organizations aspire to be socially diverse, diversity’s consequences for organizational performance remain unclear. Social bridging theories argue that diverse organizations will perform better because they have access to more social resources via their members’ diverse networks. Social bonding theories, on the other hand, argue that diverse organizations will perform worse because they are less cohesive by virtue of their members differing socially from each other. When scholars test these competing theories they often (mis)specify social bridging and social bonding as being the inverse of each other. This study specifies them as distinct mechanisms and measures them independently—bridging as the diversity of an organization’s social composition and bonding as the intensity of its members’ social interaction. Then it assesses their effect on performance by analyzing data from a national study of organizations. The first analysis indicates a consistent positive relationship between social interaction and performance and a mixed relationship between social diversity and performance. The second analysis indicates that social interaction positively moderates that relationship between diversity and organizational performance. This finding suggests that being diverse is not enough. In order to fully realize the performance benefits of diversity, members of diverse organizations need to meet regularly and talk about their relevant differences. Overall, this study finds that organizations can improve their performance by having socially diverse members who interact often and in ways that engage their differences.
2014-09-22, 12:30–2:00 PMGross Hall, Rm 230EClare BarringtonUNC (Public Global Health)Social networks, migration and HIV in a new immigration destinationAbstractLatinos in the US are disproportionately affected by HIV and are more likely to present with a late diagnosis compared to non-Latinos, which creates delayed engagement in HIV care and treatment. Social networks are a central driver of Latin American migration to the US and have been used extensively in health promotion efforts with Latinos, but there is a need for more in-depth understanding of these networks in a new immigrant destination to improve such approaches. I will present findings from analysis of life histories (n=15) and in-depth interviews (n=17) with Latino men and transgender women in NC about their social networks, migration, and HIV experiences. We find that social networks generally get smaller over time and are impacted by migration and the context of a new immigrant destination. Salient contextual factors shaping networks include geographic spread, participants’ work demands, oppressive immigration policy, and racial/ethnic discrimination. These factors fuel processes of “othering” and hinder the formation and maintenance of close networks, challenging the assumption that Latinos naturally have large, supportive networks that can be used for both social support and social influence approaches to health promotion.
2014-09-15, 12:30–2:00 PMGross Hall, Rm 230ETony RiveraNDU (International Security Affairs)Honor as Social Power and Factions as Networks: Understanding the Iranian Political EliteAbstractIranian strategic decision-making remains largely misunderstood by the West. The system in place does make the Supreme Leader the ultimate decision-maker, and it does give the president constitutional authority, and, as in any system, there are multiple institutional prerogatives that drive decision-making as well. However, in Iran it is the informal institutions that drive decision-making and it is the factional competition that determines the structure of the decision-making system. Factions compete for power and position and it is this factional competition that enables and constrains policy choices. Further, the main mechanism used to advance factional competition is honoring and humiliating opponents. This paper examines the role of honor as social power in the factional competition, conceptualized and operationalized as social networks, that shapes Iran’s strategic policy preference formation. I do so by tracking the rise and fall of former president Mahmoud Ahmadinejad from grace using network analysis and honor as social power. To date, no one has attempted a thorough network analysis of the factions that make up the Iranian political elite. Further, the role of honor, a social construct, recently gaining renewed importance in the IR literature, has yet to be operationalized in a way that is compatible with either analytical or computational methods. This paper gives the background of the political elite, charts the factions that comprise that elite, conceptualizes honor as social power, and then uses a social network analysis method to demonstrate the salience of honor in the factional competition that shapes the foreign and security policy of Iran.
2014-09-08, 12:30–2:00 PMGross Hall, Rm 230EJames WilsonUNC (Statistics & Operations Research)A testing based approach to identifying statistically significant communities in social networksAbstractAn important problem in the study of networks is how to divide the vertices of a given network into one or more groups, called communities, in such a way that vertices of the same community are more interconnected than vertices belonging to different ones. A large number of community detection methods assume that every vertex belongs to a well-defined community; however, in many applications, networks contain a significant number of non-preferentially attached “background” vertices that do not belong to any distinct community. In these applications, contemporary detection methods can over fit the network and provide misleading results due to false discovery. To address this issue, we incorporate a criterion for statistical significance based on p-values that measure the strength of connection between a single vertex and a set of vertices by comparison to a reference distribution derived from the configuration random graph model. We propose and investigate a testing based community extraction procedure that identifies statistically significant communities while distinguishing background vertices. We evaluate the performance and potential use of our method through its application to the Enron email network as well as the author’s Facebook network. Optimality properties of the extraction method will also be discussed.
2014-04-15, 12:30–2:00 PMGross Hall, Rm 230EDane TaylorSAMSI and UNC Department of MathematicsComplex contagion on noisy geometric networksAbstractThe study of contagion on networks is central to our understanding of collective social processes and epidemiology. However, for networks arising from an underlying manifold such as the Earth’s surface, it remains unclear the extent to which the dynamics will reflect this inherent structure, especially when long-range, “noisy” edges are present. We study the Watts threshold model (WTM) for complex contagion on noisy geometric networks— a generalization of small world networks in which nodes are embedded on a manifold. To study the extent to which contagion adheres to the manifold versus the network, which can greatly disagree on notions such as node-to-node distance, we present WTM-maps that embed the network nodes as a point cloud for which we study the geometry, topology, and intrinsic dimensionality. Interestingly, this work bridges several research disciplines by aligning the pursuits of network science and epidemiology with those of manifold learning and dimension reduction.
2014-04-08, 12:30–2:00 PMGross Hall, Rm 230EAnthony KosnerForbesDesign Everywhere: Natural Machines and the Success (and Failure) of Emergent NetworksAbstractDesign is a universal principle present in all things, animate and inanimate, as explained by Adrian Bejan’s constructal law. At every scale of matter, time and organization there are design processes at work that recognize differences and exploit those differences to accomplish constructive work. These design processes are embodied as natural machines that do the work of intelligent construction at micro scales that become visible at macro scales as patterns, terrains and networks. Alex Wissner-Gross has conceived of intelligence as “an engine for maximizing future freedom of action.” His theory posits that two factors contribute to the quality of intelligence in a given system, the length of time frame and the number of entities within the system that are considered as part of this maximization. There is a bridge between Wissner-Gross’s intelligent agents and the evidence of design in Bejan’s constructal law in the form of emergent networks that maximize flow. Human technology has become better and better over time at creating natural machines of its own. Particularly in the areas of software applications and social networks it is apparent that these human domains have become areas of inquiry relevant to all areas of animate and inanimate natural phenomena. Artificial intelligence, machine learning, natural language processing, social media as well as the rendering of software applications themselves on diverse environments of devices all provide instructive analogies that illuminate these fundamental design principles. From this perspective, heat transfer is an app that has users that include snowflakes. Not all design processes are equal and the consideration of the failures of natural machines is as instructive as their successes. Computational linguistics, for instance, can quantify miscommunication as well as predict patterns of language usage. The degree of success of a design process, of a natural machine, should be discernable through the shape and configuration of the networks it engenders.
2014-04-01, 12:30–2:00 PMGross Hall, Rm 230EBailey FosdickDuke – SAMSI, statisticsTemporal latent space network models for dynamic grooming interactions in baboon troopsAbstractBaboon troops are intriguing social populations as they have strict social hierarchies and about once every fifteen years a given troop will fission into two new troops. Often this occurs according to matrilineal or patrilineal lines, but once in a while, neither of these familial patterns is exhibited. In these cases, primatologists are greatly interested in understanding the severance process and determining whether temporal data on baboon grooming activities may foreshadow the fission event and eventual troop memberships. Current network models are inadequate for modeling such data since they cannot readily account for variable intensity of observation across time and baboons, and they lack a natural mechanism that allows for social distancing over time and prediction of future grooming. In this talk, we present a dynamic latent space network model that addresses these issues. We demonstrate our methodology on data from a baboon troop in the Amboseli National Reserve in Kenya. This is joint work with Yingbo Li, David Banks, and Susan Alberts.
2014-03-25, 12:30–2:00 PMGross Hall, Rm 230EMalte DoehneZU Friedrichshafen – SociologyFrom attribute to quality signal – and back again? The adoption of screwcaps on fine wines, 1970-2012AbstractMore than 400 million bottles of wine are adversely affected by faulty cork closures each year. Premium winemakers are well aware of the problem and a solution long exists in form of the screwcap. Used on table wines since the 1930s, it eliminates two of cork’s major potential flaws: cork taint and premature oxidation. In recent years, a premium screwcap for use on expensive wines has been gaining market shares in some winemaking regions but not at all in others. The explanation commonly offered for non-adoptive behavior is that consumers associate screwcaps with wines of low quality and that premium winemakers must therefore use corks to signal the high quality of their product. While intuitive, this explanation fails to address the conditions under which adoption behaviors change and upon whose initiative such changes take place when they do. In this paper, I address these questions and their implications for understanding the diffusion of innovations on markets. I develop my argument as an empirical application of Harrison White’s ‘Markets from Networks’-approach. Drawing on a dataset of 35,000 wines made by 650 winemakers in 10 winemaking regions of Germany, I examine the adoption of screwcaps across multiple market contexts. I find evidence that variation in screwcap adoption is indicative of underlying differences in both local market structure and the relative market position of adopters. In presenting my findings, I demonstrate how ‘Markets from Networks’ can be made fruitful for empirical analysis.
2014-03-18 (probably 12:30–2:00 PM)Gross Hall, Rm 230EKristian LumVirginia Tech-Virginia Bioinformatics InstituteAn agent-based epidemiological model of incarcerationAbstractWe build an agent-based model of incarceration based on the SIS model of infectious disease propagation. Our central hypothesis is that the observed racial disparities in incarceration rates between Black and White Americans can be explained as the result of differential sentencing between the two demographic groups. We demonstrate that if incarceration can be spread through a social influence network, then even relatively small differences in sentencing can result in the large disparities in incarceration rates. Controlling for effects of transmissibility, susceptibility, and influence network structure, our model reproduces the observed large disparities in incarceration rates given the differences in sentence lengths for White and Black drug offenders in the United States without extensive parameter tuning. We further establish the suitability of the SIS model as applied to incarceration, as the observed structural patterns of recidivism are an emergent property of the model. In fact, our model shows a remarkably close correspondence with California incarceration data, without requiring any parameter tuning. This work advances efforts to combine the theories and methods of epidemiology and criminology.
2014-02-25, 12:30–2:00 PMGross Hall, Rm 230ECharles SeguinUNC – SociologyCultural Superstardom from Multiple Mechanisms: Two Mathematical Models of Cultural Object PopularityAbstractThe popularity of cultural objects such as music recordings, baby names, or novels is characterized by a large number of relatively unpopular “flops” as well as a few superstars that are several orders of magnitude more popular than the average. Despite these large ex post differences in popularity, ex ante it is very difficult to predict which objects will become hits, and which will flop. Scholars have proposed two major theories about the mechanisms leading to these outcomes. The first is based on cumulative advantage (CA), or rich-get-richer processes, wherein the success of cultural objects breeds future success. The other is based on convex returns (CR) and suggests that small differences in the talent of artists, or qualities of cultural objects, lead to large differences in popularity. I study mathematical models of both CA and CR processes, and derive their distributional implications. I first validate these models on experimental data from Salganik’s Music Lab project. I then apply the models to the distribution of US baby girls names, showing that CR is a better fit to those data. I end with a discussion on the models’ implications for theories of cultural consumption
2014-02-18, 12:30–2:00 PMGross Hall, Rm 230EJosh SocolarPhysicsQuantifying the complexity of Boolean networksAbstractWe study two measures of the complexity of heterogeneous extended systems, taking random Boolean networks as prototypical cases. A measure defined by Shalizi et al. for cellular automata, based on a criterion for optimal statistical prediction [Shalizi et al., Phys. Rev. Lett. 93, 118701 (2004)], does not distinguish between the spatial inhomogeneity of dynamically ordered phases and the more irregular behavior of dynamically disordered behavior. We consider a modification in which complexities of individual nodes are calculated, where individual nodes with high complexity are the ones that pass the most information from the past to the future. The complexity of a node depends in a nontrivial way on both the Boolean function of a given node and its location within the network. A network of prisoner’s dilemma players is used as an illustrative example. This talk will focus on the conceptual aspects of the work rather than the technical details.
2014-02-11, 12:30–2:00 PMGross Hall, Rm 230EDavid BanksDepartment of Statistical ScienceText NetworksAbstractThe dynamics of the Wikipedia, political blogs, and computational advertising are all situations in which the analyst can draw upon two kinds of data: information on text in webpages, and network connectivity structure between pages. In principle, each kind of information can inform the joint analysis; for example, latent Dirichlet allocation analysis can identify topics in text, and the extent to which a particular node participates in a topic may be a covariate used in forecasting the formation of edges. Reciprocally, one may use connectivity patterns to sharpen inference on topic memberships. This talk describes several forays into this area, and points up some of the emerging challenges in joining the recent field of network modeling with text mining.
2014-02-04 (probably 12:30–2:00 PM)(probably Gross Hall, Rm 230E)Kamesh MunagalaComputer ScienceOn the Precision of Social and Information NetworksAbstractThe diffusion of information on online social and information networks has been a popular topic of study in recent years, but attention has typically focused on speed of dissemination and recall (i.e. the fraction of users getting a piece of information). In this paper, we study the complementary notion of the precision of information diffusion. Our model of information dissemination is “broadcast-based”, i.e., one where every message (original or forwarded) from a user goes to a fixed set of recipients, often called the user’s “friends” or”followers”, as in Facebook and Twitter. The precision of the diffusion process is then defined as the fraction of received messages that a user finds interesting. On first glance, it seems that broadcast-based information diffusion is a “blunt” targeting mechanism, and must necessarily suffer from low precision. Somewhat surprisingly, we present preliminary experimental and analytical evidence to the contrary: it is possible to simultaneously have high precision (i.e. is bounded below by a constant), high recall, and low diameter! We start by presenting a set of conditions on the structure of user interests, and analytically show the necessity of each of these conditions for obtaining high precision. We also present preliminary experimental evidence from Twitter verifying that these conditions are satisfied. We then prove that the Kronecker-graph based generative model of Leskovec et al. satisfies these conditions given an appropriate and natural definition of user interests. Further, we show that this model also has high precision, high recall, and low diameter. We finally present preliminary experimental evidence showing Twitter has high precision, validating our conclusion. This is perhaps a first step towards a formal understanding of the immense popularity of online social networks as an information dissemination mechanism. Gross Hall, Rm 230E
2014-01-28, 12:30–2:00 PMGross Hall, Rm 230EJames FowlerUCSD – Political Science and Medical GeneticsFriendship and Natural SelectionAbstractMore than any other species, humans form social ties to individuals who are neither kin nor mates, and these ties tend to be with similar people. Here, we show that this similarity extends to genotypes. Across the whole genome, friends’ genotypes at the SNP level tend to be positively correlated (homophilic); however, certain genotypes are negatively correlated (heterophilic). A focused gene set analysis suggests that some of the overall correlation can be explained by specific systems; for example, an olfactory gene set is homophilic and an immune system gene set is heterophilic. Finally, homophilic genotypes exhibit significantly higher measures of positive selection, suggesting that, on average, they may yield a synergistic fitness advantage that has been helping to drive recent human evolution.
2014-01-21, 12:30–2:00 PMGross Hall, Rm 230EDavid EagleSociology DepartmentMethodological Considerations in the Use of Name Generators and InterpretersAbstractWith from the Clergy Health Initiative Longitudinal Survey, we investigate several methodological issues surrounding the use of name generators/interpreters to uncover a respondent’s core discussion network. First, we offer evidence that telephone surveys are less prone than face-to-face surveys to systematic variation by interviewer. Second, we discover that the inclusion of a large number of discrete boxes on a web form inflates the size of network generated. Third, we find that small changes to question presentation reduces the size of the network. Finally, we uncover evidence of significant levels of panel conditioning when name generators are implemented longitudinally. These findings have important implications for the design of surveys that utilize name generators/interpreters.
2014-01-14, 12:30–2:00 PMGross Hall, Rm 230ESteve McDonaldNCSU – SociologyIntra-Organizational Job Mobility Networks and Wage InequalityAbstractDespite the increasing prevalence of job mobility across work organizations, internal job moves remain a common yet relatively understudied aspect of work careers. The movement of individuals across jobs within an organization could therefore help to explain persistent and growing intra-organizational wage inequality. We analyze directed network data based on movement of individuals among jobs within 11 distribution centers of a private sector firm over the course of 6 years. The data are used to examine how the global and local network characteristics are associated with wage variation a) within the distribution centers and 2) within each job in each distribution center. Overall, the results are consistent with a relational inequality perspective, in that network structure and positioning appear to be consequential for the claims making process involved in wage determination.
2013-12-03, 12:30–2:00 PMGross Hall, Rm 230ETed MouwUNC – Chapel HillSampling a hidden population without a sampling frame: A practical application of Network Sampling with Memory.AbstractMouw and Verdery (2012) show that it is possible to increase the efficiency of sampling from a hidden population by collecting network information as part of the survey. They propose a new method, “Network Sampling with Memory” (NSM) that information on network members from the survey instrument to uncover the sampling frame for the target population. In this paper, we present a practical application of NSM that reduces the cost of data collection by collecting contact information on up to three referrals from the current respondent, which eliminates the need to re-contact prior respondents to ask for referrals. We test the accuracy of this modified method using simulated sampling on 215 school and university social networks, and we test the use of bootstrap methods to calculate the sampling variance. In addition, we report results from a pilot study using NSM, the 2013 Chinese African Health Study (CAHS) which sampled Chinese immigrants living in Tanzania, and we provide a step-by-step description of how to conduct an NSM-based survey in the field.
2013-11-26, 12:30–2:00 PMGross Hall, Rm 230EBrian SouthwellRTI International, Duke, & UNC-Chapel HillMultilevel Constraint on Social Interaction Regarding ScienceAbstractCommunication researchers are increasingly drawn toward social network concepts in explaining public understanding of science. Some of the resulting work is theoretically underdeveloped, however, as pundits and scholars look for “viral” information spread without sufficiently integrating program evaluation and other research into broader, multilevel conceptualization of communication systems. Southwell will discuss both his new book from Johns Hopkins University Press — described briefly at bit.ly/16xm74i — and solicit conversation about how we might integrate various approaches to better understand how social interaction relates to public opinion about science and health.
2013-11-12, 12:30–2:00 PM(probably Gross Hall, Rm 230E)Daniel HeardDepartment of Statistical ScienceNetwork Analysis Techniques for a Hard-to-reach PopulationAbstractThis work examines a community of men who have sex with men (MSM) in southern India with a high HIV rate. The data is from an ego-centric study and includes attributes such as religion, age, marital status, caste, sex position, religion, and whether an individual was a sex worker. The data also included self-reported attributes of interviewed individuals.We use a Bayesian mixed effects model (Hoff 2011) to investigate characteristics of individuals within the network and how these characteristics related to sexual behavior. To investigate the formation of ties and determine if any latent structure existed within the network, we used a mixed-membership stochastic block model (Airoldi 2008). This information gives insight into possible medical interventions to mitigate the HIV epidemic in the community.
2013-11-05, 12:30–2:00 PMGross Hall, Rm 230ELauren BrentCenter for Cogntive NeuroscienceGenes, Networks, and Babies: The Causes and Consequences of Social Behavior in Nonhuman Primates
2013-10-29, 12:30–2:00 PMGross Hall, Rm 230EKathleen CarleySchool of Computer Science – Carnegie Mellon UniversityThe Power of Meta-Networks for Rapid Ethnographic AssessmentAbstractMeta-Networks are high dimensional interlinked multi-mode and multi-link networks that can vary by time and space. They can be rapidly extracted from media data such as on-line news and social media. When this is done, the resultant analysis provides key insight into the evolving cultures and provides a rapid ehtnographic assessment. Key advances in in big data analytics and visualization, coupled with text-analytics make such analysis possible. This process is described and results shown for two cases – the Benghazi Consulate Attack and the current state of Myanmar.
2013-10-22, 12:30–2:00 PMGross Hall, Rm 230EKrista GileUMass Amherst – StatisticsNew methods for inference from Respondent-Driven Sampling DataAbstractRespondent-Driven Sampling is type of link-tracing network sampling used to study hard-to-reach populations. Beginning with a convenience sample, each person sampled is given 2-3 uniquely identified coupons to distribute to other members of the target population, making them eligible for enrollment in the study. This is effective at collecting large diverse samples from many populations. Current estimation relies on sampling weights estimated by treating the sampling process as a random walk on the underlying network of social relations. These estimates are based on strong assumptions allowing the data to be treated as a probability sample. In particular, existing estimators assume a with-replacement sample with an ideal initial sample. We introduce two new estimators, the first based on a without-replacement approximation to the sampling process, and the second based on fitting a social network model (ERGM), and demonstrate their ability to correct for biases due to the finite population and initial convenience sample. Our estimators are based on a model-assisted design-based approach, using standard errors based on a parametric bootstrap. We conclude with an application to data collected among injecting drug users, including extension to observable features of the sampling process.
2013-10-08, 12:30–2:00 PMGross Hall, Rm 230EIlya ZaliapinUniversity of Nevada, Reno – Department of Mathematics and StatisticsRandom self-similar trees: models, statistical inference, and applicationsAbstractHierarchical branching organization is ubiquitous in nature. It is readily seen in river basins, drainage networks, bronchial passages, botanical trees, lightening, and snowflakes, to mention but a few. Notably, empirical evidence reveals a surprising similarity among natural hierarchies — many of them are closely approximated by so-called self-similar trees (SSTs). This talk will focus on the Horton and Tokunaga self-similarity that provide easily parameterized constraints on random tree graphs. The Horton self-similarity is a weaker property that addresses the principal branching in a tree; it is a counterpart of the power-law size distribution for system’s elements. The stronger Tokunaga self-similarity addresses so-called side branching; it ensures that different levels of a hierarchy have the same probabilistic structure (in a sense that can be rigorously defined). The Horton and Tokunaga self-similarity have been empirically established in numerous observed and modeled systems. This hints at the existence of a universal underlying self-similarity mechanism and prompts the question: What basic probability models can generate Horton/Tokunaga self-similar trees with a range of parameters? We review the existing results and present recent findings on self-similarity for tree representation of branching and coalescent processes, random walks, and white noises. In particular, we establish the equivalence of tree representation for selected coalescent processes and time series models. We also describe a statistical framework for testing self-similarity in a finite tree and estimating the related parameters. Our results suggest at least a partial explanation for the omnipresence of Tokunaga self-similar structures in natural branching systems. The results are illustrated using applications in hydrology, seismology, and billiard dynamics. This is a joint work with Yevgeniy Kovchegov (Oregon State U) and Alejandro Tejedor (U of Minnesota).
2013-10-01, 12:30–2:00 PMGross Hall, Rm 230EAshton VerderyUNC Chapel Hill – Dept of SociologyUsing multiple data sources to evaluate Respondent Driven Sampling (RDS) and explore new approaches to studying hidden and hard to reach populationsAbstractThis presentation summarizes a set of recent papers on RDS and discusses new approaches to the study of hidden populations. We summarize work from an empirical RDS study conducted among female sex workers in China. In this newer work we evaluate the bias and efficiency of sample level mean statistics across multiple RDS estimators, substantive variables of interest, and seeding and recruitment assumptions. We also discuss recent simulation work that explores biases in variance estimation in RDS. Finally, we conclude with a discussion of newly proposed approaches to surveying hidden populations and preliminary results from a survey conducted this summer using such methods in Tanzania.
2013-09-24, 12:30–2:00 PMGross Hall, Rm 230EDavid HinesDepartment of Biology and Marine Biology – UNC WilmingtonAn Application of Ecological Network Analysis to Predict the Impacts of Seawater Intrusion on an Estuarine Nitrogen CycleAbstractMicrobial processes in estuarine sediments transform and remove biologically available forms of nitrogen from estuarine ecosystems. These nitrogen cycling processes can be coupled such that removal processes depend on transformation processes to supply a substantial portion of the reaction substrate. Seawater intrusion into the freshwater portions of estuaries results in changes to environmental chemistry that may alter the relationships among these biogeochemical processes. This work employs ecological network analysis to evaluate the potential impacts of seawater intrusion on the estuarine nitrogen cycle. We 1) present two comparative mass-balance network models for nitrogen budgets at an oligohaline and a polyhaline site in the Cape Fear River Estuary, NC, 2) estimate process coupling in each network, 3) evaluate the certainty of results using a Monte Carlo approach, and 4) use multiple hierarchical levels of analysis to further examine the results. This work presents a practical application of network analysis to highlight techniques that can be applied to a variety of systems.
2013-09-17, 12:30–2:00 PMGross Hall, Rm 230ENan Lin and Hang Young LeeDuke SociologyAssessing Measurements of Social Capital: Investigation of Four Typical Social Capital MeasurementsAbstractSocial capital has been one of the most salient concepts in social science. Although it conceptually delivers a coherent idea of resources embedded in social networks, several different empirical measurements have been employed across the disciplines. Four measurements have been most widely used: micro-level social capital is typically measured by the Position (PG) and Name Generator (NG), while macro-level social capital by generalized trust and voluntary organization participation. Nevertheless, little is known about whether those four measurements capture similar or different aspects of social capital. Using the 2008 SC-USA data specifically designed for the comparison of social capital measurements, we examine the relationship across those four measurements. The results show that each measurement captures different aspects of social capital. At the micro level the PG more captures resources accessed through weak ties. By contrast, the NG more measures them accessed through strong ties. At the macro level generalized trust is not associated with voluntary organization participation. Contrary to social capital theory, people who participate in various voluntary organizations do not show a high level of generalized trust. Intriguingly, a cross-level examination shows a selective affinity between two micro and two macro social capital measurements. Generalized trust is associated only with the NG, while voluntary organization participation is only with the PG. This finding suggests that strong rather than weak ties enhance generalized trust and that voluntary organization participation facilitates the development of weak than strong ties.
2013-09-10, 12:30–2:00 PMGross Hall, Rm 230ENeal CarenSociology – DukeRefCliq: The Networked World of Academic CitationsAbstractThis projects presents a method for depicting academic fields and subfields and identifying the major players in those research areas based on a network analysis of article citations. I examine which references are commonly cited together and what works are most central to those clusters. For example, what are the major research areas in the study of climate change, social networks, or in sociology, and who are the central researchers in each? Article citation networks provide a valuable, but underutilized, resource for understanding and creating academic networks. This method is of interest both to scholars studying how scientific fields are constructed and evolve and to researchers interested in learning the contours and primary works in a new field. The session will be focused on implementation of the RefCliq (https://github.com/nealcaren/RefCliq) program in Python, which utilizes community detection with data collected from co-occurrences of cited works.
2013-09-03, 12:30–2:00 PMGross Hall, Rm 230EVincent ConitzerComputer Science and Economics – DukeUsing Social Networks for More Accurate Voting and MarketingAbstractSocial network graphs are increasingly directly accessible to us and our algorithms. For example, Facebook has direct access to the graph on its users. To what ends can we use such direct access? I will discuss three topics. First, in voting (or rating), our opinions are often affected by those of our neighbors. If we know the social network, should this affect how we count the votes in the election? (I prove a result indicating that it shouldn’t.) Second, when an entity such as Facebook holds a vote among its users (as it has in the past), we may worry about the use of fake accounts to cast extra votes. We show how the incentive to do so can be removed by computing which parts of the graph are “suspect.” Third (time permitting), I will discuss a marketing application, where a seller can sell to nodes in a network one at a time and the price a node is willing to pay depends on which of its neighbors have previously adopted the product. The three topics correspond to the following papers. Vincent Conitzer. Should Social Network Structure Be Taken into Account in Elections? Mathematical Social Sciences, Special Issue on Computational Foundations of Social Choice, Volume 64, Issue 1, 2012, pp. 100-102. Vincent Conitzer, Nicole Immorlica, Joshua Letchford, Kamesh Munagala, and Liad Wagman. False-Name-Proofness in Social Networks. In Proceedings of the Sixth Workshop on Internet and Network Economics (WINE-10), pp. 209-221, Stanford, CA, 2010. Sayan Bhattacharya, Dmytro Korzhyk, and Vincent Conitzer. Computing a Profit-Maximizing Sequence of Offers to Agents in a Social Network. In Proceedings of the Eighth Workshop on Internet and Network Economics (WINE-12), pp. 482-488, Liverpool, UK, 2012.
2013-04-02, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Cosma ShaliziStatistics Department – Carnegie Mellon UniversityHomophily, Contagion, Confounding: Pick Any ThreeAbstractIndividuals near each other in a social network tend to behave similarly; you can predict what one of them will do from what their neighbors do. Is this because they are influenced by their neighbors (“contagion”), or because social ties tend to form between people who are already similar (“homophily”), and so act alike, or some of both? We show that observational data can hardly ever answer this question, unless accompanied by very strong assumptions, like measuring everything that leads people to form social ties. Most observational studies therefore provide no evidence at all about the existence or strength of contagion effects. We also suggest some possible constructive responses to these results.
2013-03-26, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Dong-Ju LeeHarvard University – Department of SociologyInternational Linkages and Liberalization of Abortion: Competing Institutional Logics and International Organization NetworksAbstractAbortion laws, despite their critical roles in political and social debates, have received little attention from comparative sociologists. This paper studies the worldwide liberalization of abortion laws among 202 countries during 1920-2007, focusing on the effect of countries’ embeddedness in world polity. Rights to legal abortion have been incorporated in the international regimes of human rights, but their legal status is still contested in ways that other human rights are not. Building on neo-institutional studies of policy diffusion, this study expands the literature’s focus on isomorphic diffusion and explores why there are varying modes of adoption across different policies and across different regions of the world. I pay attention to two international advocacy networks, namely international birth control movements and international women’s movements, and explore how each constructs and diffuses rationales for the liberalization of abortion. In doing so, I employ network analysis to measure the extent to which an individual country is exposed to the ideas of abortion rights through its central position in those networks of international organizations, and I explain how the linkages to each policy network lead to different consequences concerning the pace and pattern toward liberalization. Event history analyses reveal that countries having central positions in international women rights networks are more likely to liberalize their abortion laws while central countries in international birth control networks are not.
2013-03-19, 12:30 PMRoom Soc/Psych Bldg, Rm 329 (McKinney Rm)Adrian BejanDepartment of Mechanical Engineering and Material SciencesConstructal Law of Design and Evolution in NatureAbstractThe reoccurring patterns of nature have long puzzled even the most devoted proponents of chance and Darwin’s theory of evolution. But the Constructal Law changes the terms of this debate, and shows that a single law of physics governs the “design” behind everything that moves—whether animate or inanimate. According to the Constructal law, shapes and structures arise because they facilitate movement, in animal design, river basin design, traffic patterns, social dynamics, and technology and sports evolution.
2013-03-05, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Tyler McCormickUniversity of Washington – Department of Statistics, SociologyLatent space models for multiview network data: An approach to understanding social structure in Twitter
2013-02-26, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Stuart BorrettUNC, Wilmington – Systems Ecology and Ecoinformatics LaboratoryThroughflow centrality reveals important species in ecosystems and environmental impacts of shrimp trawling in Core Sound, NC.AbstractCentrality is a common tool for characterizing node importance in network science, but it is rarely used in ecology. Here, I introduce throughflow centrality as a global indicator of node importance for the energy-matter flow dynamics ON networks. I then present two ecological applications. First, I characterize the distribution of throughflow centrality in 45 empirically-based trophic ecosystem models. Consistent with theoretical expectations, this analysis shows that a small fraction of the system components (80%). Second, I use throughflow centrality to characterize the ecosystem impacts of shrimp trawling in Core Sound, NC. This highlights the wide ranging impacts of the fishing activity, not all of which are negative.
2013-02-12, 12:30–2:00 PMSoc/Psych 329 (McKinney Room)Jason PriemSchool of Information and Library Science, UNC-Chapel HillTelling a fuller story of research impact with altmetrics and ImpactStoryAbstractIn growing numbers, the workflows of scholars are moving online. As that happens, important parts of the scientific process, once hidden, are being exposed. Conversations, arguments, recommendations, reads, bookmarks–the stuff of day-to-day science–is leaving traces in places like Mendeley, Twitter, blogs, Faculty of 1000, and many others. Mining these traces can give us faster, more diverse, and more accurate data of scholarly impact. These alternative metrics or “altmetrics” could predict later citations, reveal impacts on diverse audiences like practitioners and clinicians, uncover impacts of diverse products like datasets, blog posts, and software, and reward researchers making subtle but vital contributions that the citation record ignores. After reviewing the growing research around altmetrics, we’ll discuss how these data sources can be of practical use for researchers today, focusing on ImpactStory, an open-source web tool that gathers and contextualizes altmetrics.
2013-02-05, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)John ScottDepartment of Public Policy – University of North Carolina, Chapel HillInfluence, Selection, and Activity: Social Structure and Processes in Medicare Lobbying and Agenda Setting.AbstractBy what process do interest group representatives select the issues on which they lobby the Congress? Any one policy area often has dozens or even hundreds of proposed bills and equally as many interest groups and stakeholders. In a crowded and competitive environment, lobbyists may look to other lobbyists when selecting issues for monitoring and lobbying in order to lower the costs of policy advocacy. And despite the number, diversity, and fluidity of interest types and for-hire lobbyists, Medicare policy has a core of representatives who are connected to each other through client relationships. The presence of this core implies an interconnected community in which information flows freely. The broad research question of this paper is whether social processes matter in the formation of interest group agendas. Specifically, do the choices of one lobbying organization affect the choices of another organization? To answer this question, I use longitudinal lobbying data about the legislative choices of interest groups from the Medicare & Medicaid policy domain. To capture the social interdependence of lobbying, I study the evolution of a network consisting of lobbyists and their clients choosing (or not choosing) legislative proposals on which to lobby. I analyze these selections by using stochastic actor-based dynamic model of network change with a focus on legislative choices that are conditioned on the choices of other organizations. The results suggest that a selection or ‘bandwagon’ effect in which organizations choose bills that are already popular and a social influence effect in which choices by a lobbying organization are likely influenced by another organization when the two organizations have overlapping agendas.
2013-01-22, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)David Banks and Justin GrossDepartment of Statistical Science; University of North Carolina at Chapel Hill – Department of Political ScienceTopic Modeling in Blog Networks
2012-12-04, 12:30 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Christina PrellUniversity of Maryland – Sociology DepartmentGaining Knowledge Expertise Through Social NetworksAbstractAs such, there is a subtle difference between the motivations described by Lin and those by Burt. Lin emphasizes a rational actor using knowledge of embedded resources (resources associated with other actors in the network) and Burt emphasizes rational actors pursuing a particular network position, or structure. Burt’s approach can be called a ‘brokerage’ network strategy, i.e. pursuing ties that would place a focal actor in a broker position; and Lin’s approach can be called an ’embedded resource’ networking strategy, i.e. pursuing ties with actors who have control over resources that interest a focal actor. In this paper, I compare actors’ use of a ‘brokerage’ networking strategy versus a ’embedded resource’ strategy as actors pursue the goal of increasing their own stock of knowledge . Which strategy yields the best pay-off, in the form of knowledge gains, and which strategy is the most costly? To explore this comparison, I simulate a great number of different networks, making use of Snijders’ Evaluation Function for guiding the rules of agents, and including a statistic to model ‘learning’ that is not originally found in his model. My preliminary findings suggest that pursuing brokerage and pursuing ties with knowledge experts each lead to very similar pay-offs: both strategies yield, on average, the same amount of ‘learning’ across the network. However, brokerage is a much less costly networking strategy. Finally, I compare both these strategies against a few other models (e.g. pursuing ties with others holding a similar knowledge level as oneself) for further discussion.
2012-11-06 (probably 12:30–2:00 PM)Soc/Psych Bldg, Rm 329 (McKinney Rm)Greg ApplebaumDuke Institute for Brain SciencesNeuroRhetoric: Mapping the semantic structure of cognitive neuroscience
2012-10-23 (probably 12:30–2:00 PM)Soc/Psych Bldg, Rm 329 (McKinney Rm)Jeffrey A. Smith, M. Giovanna Merli, James Moody, Jing Li, Sharon WeirSociology Department, Immunology; University of North Carolina at Chapel Hill – School of Public HealthNetwork Sampling Coverage in RDS: How Much of the Network Do We See?
2012-10-02, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Jeff SmithSociology DepartmentMeasuring Social Change as Categorical Change: Race and Education in AmericaAbstractSociologists often depict demographic categories as socially constructed, non-essential, and fluid. In practice, however, social trends still typically reflect the changing outcomes(e.g. health, income) of fixed, exogenous demographic categories. The goal of this dissertation is to bridge the gap between rhetoric and practice by offering a new framework for measuring social dynamics in a population. The proposed framework makes the categories themselves the key measure of social change.Here, demographic categories are seen as proxies for social locations, describing where individuals are allowed to go and who they are allowed to interact with. Different categories may come to represent the same behavioral limitations, or social locations, at different time points, and I use these changes to measure social change. Formally, I use interaction data and models of social space and social distance to place demographic categories into social locations. I then use those locations to equate the categories over time. Empirically, I examine changes in education and race using the proposed approach and census marriage data from 1940-2000. I explain how changes in educational categories vary by (social) racial locations, and how this racial structuring of education has varied over time. I thus use changes in the educational categories to characterize the racial stratification system-where different racial groups play different (or no) roles in determining the social meaning of education.
2012-09-25 (probably 12:30–2:00 PM)Soc/Psych Bldg, Rm 329 (McKinney Rm)James Moody, Peter Mucha and S. Joshua MendelsohnSociology Department; University of North Carolina at Chapel Hill – Department of MathematicsRemoval Centrality: A Comparative Evaluation of System-Influence Centrality Measures
2012-09-04, 12:30 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Brad FultonSociology DepartmentUncommon Collaborators: How Network Ties Enable and Constrain Organizational Action
2012-04-24, 12:00–01:30 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Karen JoyceWake Forest – Department of Biomedical EngineeringAn Agent-Based Model of the Human BrainAbstractAgent-based modeling is a great utility for studying complex systems, where the comprising components are typically very simple, but the assembled whole often exhibits sophisticated emergent behavior. Agent based modeling is a “bottom-up” modeling approach, where simple rules are used to drive the low-level interactions between system components, and higher level system behaviors are allowed to emerge. We use this approachable modeling technique to create an agent-based brain model, built using any functional brain network. While the nodes of the network follow simple instructions, the behavior of the model can be extremely complex. We use genetic algorithms to drive the system to produce specific behaviors, and also demonstrate that this model is capable of solving computational tasks.
2012-04-10, 12:00–01:30 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Peter ArcidiaconoDepartment of EconomicsTerms of Endearment: An Equilibrium Model of Sex and Matching
2012-04-03, 12:00–01:30 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Yanjie BianUniversity of Minnesota – Department of SociologyCorporate Social Capital in Chinese Guanxi CultureAbstractWe present a conceptualization of corporate social capital within the context of Chinese guanxi culture. Sociologists have defined corporate social capital as “processes of forming and mobilizing social actors’ network connections within and between organizations to gain other actors’ resources” (Knoke 1999: 5; Brass 1999). The point of departure for our conceptualization is that the formation and mobilizations of corporate social capital are culturally and institutionally contextualized. Building upon a relational approach to corporate performance, we examine culturally contextualized properties of Chinese guanxi networks and compare guanxi social capital with non-guanxi social capital. We then explain why guanxi-based corporate social capital is of growing significance to Chinese transition economy in an era of increasing market competition and institutional uncertainty. We conclude this essay by proposing a research agenda about the roles that guanxi-based corporate social capital plays in corporate performance in the Chinese context.
2012-03-20, 12:00–01:30 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Giovanna Merli, James Moody, Robin Gauthier, and S Joshua MendelsohnDuke UniversityShanghai Sexual Mixing: Are sexual contact patterns in Shanghai compatible with an HIV/AIDS epidemic?
2012-02-21, 12:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Steve McDonaldNorth Carolina State University – Department of SociologyDual Embeddedness and Institutional Transference: Network-based Job Finding and Macro-Institutional Dynamics in Germany and the United StatesAbstractThis project explores how differences in institutional context (across space and time) impact social relations, with a specific focus on network-based job finding behavior–finding jobs through personal contacts. First, cross-national survey data are used to explore differences in job finding in the United States and Germany. In the U.S., loosely regulated and hierarchical employment relations lead to extensive patterning of informal (networked) hiring across social groups, as well as network dominance in specific economic sectors. In Germany, informal hiring is more common but also more randomly distributed across individuals and jobs, due to presence of coordinated market relations, tight employment regulations, and an extensive social insurance system. Second, the study examines how the institutional transformation that accompanied the transition from state socialism to capitalism in East Germany influenced network-based job finding. The transference of economic institutions from West Germany to East Germany in the early 1990s led to a rapid decline in informal hiring in the East and a convergence in job finding behaviors across the two regions. The results show how the institutional environment can shape network relations and transform job mobility regimes.
2012-02-14, 12:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Brian SouthwellUniversity of North Carolina, Chapel Hill – School of Journalism and Mass CommunicationLeveraging social networks for health promotion: The promise and perils of peer-to-peer information sharing
2012-02-07, 12:00–01:30 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)James MoodyDepartment of SociologyReconstructing the Ship of Theseus: Groups, Roles & Trajectories in Early Adolescent Friendship Networks
2012-01-31, 12:00–01:30 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Peter MuchaUniversity of North Carolina, Chapel Hill – Department of MathematicsPeter Mucha will be discussing research on network clustering
2011-11-28, 12:45–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Rebecca WillettScalable Tracking of Dynamic Networks
2011-11-17 (probably 12:30–2:00 PM)Duke University (probably Soc/Psych Bldg, Rm 329)Steven DurlaufUniversity of Wisconsin – Dept of EconomicsLinear Social Networks Models
2011-11-08, 02:50–03:50Hudson Hall, Rm 125Jacob FosterUniversity of Chicago – Department of SociologyNovelty, metaknowledge, and models of discovery
2011-11-07, 12:45–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Jesse BlocherUniversity of North Carolina at Chapel Hill – Kenan-Flagler Business SchoolContagious Capital: A Network Analysis of Interconnected Intermediaries
2011-10-24, 12:45–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)David SparksDepartment of Political ScienceIdeological Extremity and Primary Success: A Social Network Approach
2011-10-04, 12:45–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)David BanksDepartment of Statistical ScienceMining Text Networks
2011-09-19, 12:45–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Jeff SmithDept of SociologyInferring Global Networks from Local Samples
2011-09-12, 12:45–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Balachander KrishnamurthyAT&TInternet Privacy: It is not getting betterAbstractThe leakage of privacy on the Internet continues unabated. I will present results from a number of studies over the last five years that show increasing aggregation of user-related data by a steadily decreasing number of entities. I will present details of leakage of personally identifiable information via Online Social Networks (traditional and mobile OSNs) as well as via popular non-OSN sites. I will discuss various technical and non-technical approaches to addressing this serious problem and why I am pessimistic about most of them.
2011-09-05, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)DNAC AffiliatesDuke Network Analysis CenterGeneral Meeting
2011-04-25 (probably 12:30–2:00 PM)Soc/Psych Bldg, Rm 329 (McKinney Rm)Philip BenfeyDept of BiologyDevelopment rooted in interwoven networksAbstractDetermining the identity of different cells is central to the development of both plants and animals. In an effort to understand the networks that control cell identity, we have analyzed the gene expression in all cell types as well as in multiple developmental stages within the Arabidopsis root. To acquire global expression profiles we developed technology that uses sorted marked populations of cells with subsequent hybridization of the labeled RNA to genome-wide expression microarrays. We are using experimental methods to identify networks functioning within different cell types and developmental stages under normal laboratory conditions and under different environmental stress conditions. A key aspect of a plant’s ability to explore its below-ground environment is the production of branch or lateral roots. Lateral roots form as repeating units along the root, however the developmental mechanism regulating this process is unknown. We found that cyclic expression pulses of a reporter gene mark the position of future lateral roots by establishing prebranch sites and that prebranch site production is periodic. Further analysis revealed two sets of genes oscillating in opposite phases at the root tip. Genetic studies show that some oscillating transcriptional regulators are required for periodicity in lateral root formation. Finally, we are analyzing the dynamics of growth of physical root networks using novel non-invasive imaging methods and developing mathematical descriptors of root system architecture.
2011-04-18, 12:30–2:00 PM(probably Soc/Psych Bldg, Rm 329)Savithri NageswaranWake Forest Baptist Medical CenterUsing network analysis to study care-coordination system for children with complex-chronic conditionsAbstractChildren with complex chronic conditions (CCC) receive care from diverse medical, educational and social service providers through various agencies for a prolonged period of time. Coordination of care between providers serving these children is lacking in most communities, resulting in gaps and inefficiencies in care. Very little information exists on how agencies collaborate to provide care for children with CCC. We used Social Network Analysis to evaluate the extent of collaboration that currently exists among agencies that serve children with CCC and to identify factors that improve collaborative relationships between agencies that serve children with CCC.
2011-04-11, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Ryan LightUniversity of Oregon – Dept of SociologyModeling the content of science: The case of HIV/AIDS research, 1990-2008AbstractThe story of scientific change is typically told in one of two ways: a historical focus on the great scientists or a more sociological approach that focuses on the networks of production through co-authorship or credit through co-citation. The former often introduces the content of science attributable to only a handful of practitioners, while the latter casts a wider social net at the expense of content. In this paper we offer a different tactic by examining structural change in the content of a HIV/AIDS research. Using data from two core journals, we build topic models developed by computational scientists for the organization of large sets of unstructured data. Initial findings delineate the maturation of HIV/AIDS research following the deployment and analysis of more successful treatment regimens, but with lower interdisciplinarity than anticipated given the diversity of the HIV/AIDS field. We conclude by discussing the strengths and weaknesses of computational sociology or the attempt to make sociological sense out of unstructured text data.
2011-04-04, 12:00–2:00 PM(probably Soc/Psych Bldg, Rm 329)Mason PorterUniversity of Oxford – Dept of MathematicsSocial Structure of Facebook NetworksAbstractWe study the social structure of Facebook “friendship” networks at one hundred American colleges and universities at a single point in time, and we examine the roles of user attributes – gender, class year, major, high school, and residence – at these institutions. We investigate the influence of common attributes at the dyad level in terms of assortativity coefficients and regression models. We then examine larger-scale groupings by detecting communities algorithmically and comparing them to network partitions based on the user characteristics. We thereby compare the relative importances of different characteristics at different institutions, finding for example that common high school is more important to the social organization of large institutions and that the importance of common major varies significantly between institutions. Our calculations illustrate how microscopic and macroscopic perspectives give complementary insights on the social organization at universities and suggest future studies to investigate such phenomena further.
2011-03-28, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Regina SmardonUniversity of Virginia – Institute for Advanced Studies In CultureThe Microsociology of Interdisciplinarity: Graduate students as networked social actors and cultural objects in motionAbstractEmpirical network analysis has begun to take seriously the ways in which cultural meaning can structure social networks (Lizardo 2006; Yeung 2005). We propose that social actors must conform to emergent rules of social interaction and the constraints and affordances of their network position, while cultural objects must conform to the rules of cultural motion (Urban 2004). The meaning of cultural ties within a network of cultural objects becomes distinctive through metacultural processes. Our research is based on a one-year ethnographic study of an interdisciplinary cancer research center which in turn was part of a larger mixed methods NSF funded study investigating the social organization of innovation. We explore the role of graduate students in a network of interdisciplinary cancer researchers to see how the forces of cultural motion and the rules of social interaction shape interdisciplinary collaboration. Implications for analyzing social and cultural networks are explored as well as implications for science policy research on interdisciplinary collaboration.
2011-03-14, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Rick DurrettDept of MathematicsVoter models in the age of Facebook, iPads, and Sarah Palin
2011-02-28, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Ashton Verdery and Ted MouwUNC SociologyEstimated Sampling Variance in RDS
2011-02-21, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)James MoodyDuke SociologySimulation Models for Diffusion over Multirelational Dynamic Networks
2011-02-07, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Heather RackinDuke SociologyHow Much Does it Cost to have a Baby? Differences in Perceptions of the Cost of Childbearing
2011-02-07, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Robin GauthierDuke SociologyThe Structure of Consensus: Cohesion and Hierarchy in Peer Networks
2011-02-07, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)S Joshua MendelsohnDuke SociologyLocal Cities, Global Influence
2011-01-31, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Jeff SmithDuke SociologyMacrostructure from Survey Data: Generating Whole Systems from Ego Networks
2011-01-31, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Yanlong ZhangDuke SociologyMarkets or Networks? Rural Households’ Borrowing Choices in Western China
2011-01-24, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Rachel KrantonDuke EconomicsStrategic Interaction and Networks
2010-12-09, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Skyler CranmerUNC Political ScienceLongitudinal Analysis of Networks: Temporal Exponential Random Graph Models, their Estimation, and Applications
2010-11-18, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Tyler McCormickColumbia StatisticsUsing network structure to estimate latent features in hard-to-reach populations
2010-11-11, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Mauro MaggioniDuke MathematicsMultiscale Analysis on Graphs
2010-11-04, 12:30–2:00 PMSoc/Psych Bldg, Rm 329 (McKinney Rm)Joshua SocolarDuke PhysicsDynamics of Boolean Networks