Segregation In Social Networks Research Articles

What Education Cannot Equalize: Segregated Social Networks and Media in Stratified Reproduction

ABSTRACT In the United States, age at first birth is increasing and fertility rates are declining. In the absence of policy solutions that enable people to have the children they desire, individuals who have postponed childbearing may seek biomedical solutions in the form of fertility treatments. However, fertility care is marked by racial and socioeconomic inequalities in the United States. How and why such inequalities persist remains unclear, especially for populations with socioeconomic advantages. Using an intersectional and comparative approach, this article explores how access to health information related to in-vitro fertilization (IVF) among Black and White women in United States graduate programs may contribute to the observed inequalities in fertility care. Through qualitative analyses of semi-structured interviews, the results demonstrate social class can act as an equalizer of exposure to and knowledge about IVF. However, social network segregation and media portrayals of infertility construct social contexts that create implicit, structural barriers to Black women’s use of fertility treatments while encouraging use among White women. The findings provide new insights into the social structures that create stratified reproduction and demonstrate how resources garnered by social class cannot entirely ameliorate the effects of structural and systemic racism on health information.

Drop edges and adapt: A fairness enforcing fine-tuning for graph neural networks

The rise of graph representation learning as the primary solution for many different network science tasks led to a surge of interest in the fairness of this family of methods. Link prediction, in particular, has a substantial social impact. However, link prediction algorithms tend to increase the segregation in social networks by disfavouring the links between individuals in specific demographic groups. This paper proposes a novel way to enforce fairness on graph neural networks with a fine-tuning strategy. We Drop the unfair Edges and, simultaneously, we Adapt the model’s parameters to those modifications, DEA in short. We introduce two covariance-based constraints designed explicitly for the link prediction task. We use these constraints to guide the optimization process responsible for learning the new ‘fair’ adjacency matrix. One novelty of DEA is that we can use a discrete yet learnable adjacency matrix in our fine-tuning. We demonstrate the effectiveness of our approach on five real-world datasets and show that we can improve both the accuracy and the fairness of the link prediction tasks. In addition, we present an in-depth ablation study demonstrating that our training algorithm for the adjacency matrix can be used to improve link prediction performances during training. Finally, we compute the relevance of each component of our framework to show that the combination of both the constraints and the training of the adjacency matrix leads to optimal performances.

Controlling Segregation in Social Network Dynamics as an Edge Formation Game

This paper studies controlling <i>segregation</i> in social networks via exogenous incentives. We construct an edge formation game on a directed graph. A user (node) chooses the probability with which it forms an inter- or intra- community edge based on a utility function that reflects the tradeoff between homophily (preference to connect with individuals that belong to the same group) and the preference to obtain an exogenous incentive. Decisions made by the users to connect with each other determine the evolution of the social network. We explore an <i>algorithmic recommendation mechanism</i> where the exogenous incentive in the utility function is based on <i>weak ties</i> which incentivizes users to connect across communities and mitigates the segregation. This setting leads to a submodular game with a unique Nash equilibrium. In numerical simulations, we explore how the proposed model can be useful in controlling segregation and echo chambers in social networks under various settings.

Echo Chambers and Segregation in Social Networks: Markov Bridge Models and Estimation

This article deals with the modeling and estimation of the sociological phenomena called echo chambers and segregation in social networks. Specifically, we present a novel community-based graph model that represents the emergence of segregated echo chambers as a Markov bridge (MB) process. An MB is a 1-D Markov random field that facilitates modeling the formation and disassociation of communities at deterministic times, which is important in social networks with known timed events. We justify the proposed model with real-world examples and examine its performance on a recent Twitter dataset. We provide a model parameter estimation algorithm based on maximum likelihood and a Bayesian filtering algorithm for recursively estimating the level of segregation using noisy samples obtained from the network. Numerical results indicate that the proposed filtering algorithm outperforms the conventional hidden Markov modeling in terms of the mean-squared error. The proposed filtering method is useful in computational social science where data-driven estimation of the level of segregation from noisy data is required.

Network resources and educational outcomes among Mexican-origin youth

Despite schooling gains over the last two decades, Mexican-origin adults complete fewer years of schooling than adults from other ethnic backgrounds. Explanations emphasizing network resources suggest Mexican-origin adolescents have social ties that are more likely to be “closed” from adults with experience in higher education—and this, in turn, inhibits the transition to college. In this study, we draw on unusual network data measuring characteristics of students' peers and friends, as well as the socioeconomic background of peers' and friends' parents. We demonstrate that Mexican-origin adolescents are much less likely to have friends whose parents have college educations. 83% of non-Hispanic Asian students and 72% of non-Hispanic white students have nominated friends with college-educated mothers; about half of Mexican-origin students do. These patterns are the result of socioeconomic segregation in social networks both across and within schools. Within schools, we observe that the educational background of friends is predictive of schooling outcomes for non-Mexican students. We find evidence that this network resource shapes non-Mexican students’ educational expectations in high school and longer-run completed schooling as adults more so than it shapes the outcomes among Mexican-origin students.

Ethnicity And Risk Sharing Network Formation: Evidence From Rural Vietnam

The ethnic divide remains a persistent challenge for Vietnam. This paper explores one mechanism that could drive ethnic inequalities by analysing risk-sharing networks in rural communes. After finding substantial differences in size and composition between ethnic minority and Kinh household networks, we explore homophily patterns in link formation as drivers of these differences. In particular, we disentangle baseline homophily stemming from the different local distributions of ethnic groups from inbreeding homophily resulting from mechanisms such as ethnic-based preferences and/or biased matching processes and its effects on network structure. We find that ethnic segregation in social networks (inbreeding homophily) leads to ethnic minorities having smaller, less diversified networks than the Kinh majority. We show that inbreeding homophily is more frequent in Kinh networks than in other groups; and seems to be driven in part by segmentation of social interactions within rural communes. This pattern means that compared to other groups, Kinh could be more efficiently insured against covariant risks as they can rely on larger and more diversified risk-sharing network in terms of member occupation and location. Under this view, inequalities among ethnic groups in Vietnam appear to be partially rooted in the cultural and social distances between them.

Beyond Distance Decay: Discover Homophily in Spatially Embedded Social Networks

Existing studies suggest distance decay as an important geographic property of online social networks. Namely, social interactions are more likely to occur among people who are closer in physical space. Limited effort has been devoted so far, however, to quantifying the impact of homophily forces on social network structures. In this study, we provide a quantitative understanding of the joint impact of geographic distance and people’s socioeconomic characteristics on their interaction patterns. By coupling large-scale mobile phone, income, and housing price data sets in Singapore, we reconstruct a spatially embedded social network that captures the cell phone communications of millions of phone users in the city. By associating phone users with their estimated residence, we introduce two indicators (communication intensity and friendship probability) to examine the cell phone interactions among places with various housing price values. Our findings suggest that, after controlling for distance, similar places tend to have relatively higher communication intensity than dissimilar ones, confirming a significant homophily effect as a determinant of communication intensity. When the analysis is focused on the formation of social ties, though, the homophily effect is more nuanced. It persists at relatively short distances, whereas at higher distances a tendency to form ties with people in the highest social classes prevails. Overall, the results reported in this study have implications for understanding social segregation in cities. In particular, the physical separation of social groups in a city (e.g., residential segregation) will have a direct impact on shaping communication or social network segregation. The study highlights the importance of incorporating socioeconomic data into the understanding of spatial social networks.

The Missing Link: Network Influences on Class Divides in Political Attitudes

AbstractPrevious research provides a detailed picture of class differences in political attitudes. Less is however known about the social structures that enforce this political divide across social classes. This article contributes towards filling this gap by considering how the class profile of personal social networks influences political attitudes. We propose a general framework for incorporating an individual’s social network into class analysis of political preferences. Using Sweden as a case, we empirically evaluate our approach using a population survey with information about the respondents’ own employment situation, egocentric networks, and political attitudes in terms of redistribution and welfare chauvinism. We find that there is considerable class segregation in social networks as individuals tend to have more ties within their own and neighbouring class positions. Concerning political preferences, results show that: (i) a substantive part of the class–attitude relationship is shaped by a person’s social network; (ii) the class profile of networks influences attitudes over and above one’s own class position; (iii) class segregation in networks fortifies class divides in political attitudes. We thus conclude that social networks constitute a (hitherto) ‘missing link’ in class analysis of political preferences that merits careful consideration in theoretical models of contemporary politics.

Inequality is rising where social network segregation interacts with urban topology

Social networks amplify inequalities by fundamental mechanisms of social tie formation such as homophily and triadic closure. These forces sharpen social segregation, which is reflected in fragmented social network structure. Geographical impediments such as distance and physical or administrative boundaries also reinforce social segregation. Yet, less is known about the joint relationships between social network structure, urban geography, and inequality. In this paper we analyze an online social network and find that the fragmentation of social networks is significantly higher in towns in which residential neighborhoods are divided by physical barriers such as rivers and railroads. Towns in which neighborhoods are relatively distant from the center of town and amenities are spatially concentrated are also more socially segregated. Using a two-stage model, we show that these urban geography features have significant relationships with income inequality via social network fragmentation. In other words, the geographic features of a place can compound economic inequalities via social networks.

The \u201cSocial Brain,\u201d Reciprocity, and Social Network Segregation along Ethnic Boundaries

How does segregation along ethnic boundaries emerge in social networks? Human evolution resulted in highly social beings, capable of prosociality, mindreading, and self-control, which are important aspects of the “social brain.” Our neurophysiologically “wired” social cognition implies different cognitive goal frames. In line with recent developments in behavioral theory, the present study defines network ties as episodes of social exchange. This dynamic definition can account for shifts in goal frames during an exchange episode: whereas deliberate choice and hedonic or gain goals drive the initiation of a tie, given the opportunity structure, the normative goal frame activates a strong dynamic effect of reciprocity, which limits actors’ choice set and appears as “self-organization” at the network level. Longitudinal analyses of 18 birthday party networks comprising 501 students support the definition of network ties as exchange episodes, as well as the relevance of humans’ inherent tendency to reciprocate. However, reciprocation is much stronger in dyads of the same ethnicity than in dyads of different ethnicities. Network segregation along ethnic boundaries results from deliberate decisions during the initiation of an episode, but also from different commitments to reciprocity during the ongoing exchange process, depending on intra or interethnic dyadic constellations.

Quantifying segregation in an integrated urban physical-social space

Our knowledge of how cities bring together different social classes is still limited. Much effort has been devoted to investigating residential segregation, mostly over well-defined social groups (e.g. race). Little is known of how mobility and human communications affect urban social integration. The dynamics of spatial and social-network segregation and individual variations along these two dimensions are largely untapped. In this article, we put forward a computational framework based on coupling large-scale information on human mobility, social-network connections and people's socio-economic status (SES), to provide a breakthrough in our understanding of the dynamics of spatio-temporal and social-network segregation in cities. Building on top of a social similarity measure, the framework can be used to depict segregation dynamics down to the individual level, and also provide aggregate measurements at the scale of places and cities, and their evolution over time. By applying the methodology in Singapore using large-scale mobile phone and socio-economic datasets, we find a relatively higher level of segregation among relatively wealthier classes, a finding that holds for both social and physical space. We also highlight the interplay between the effect of distance decay and homophily as forces that determine communication intensity, defining a notion of characteristic 'homophily distance' that can be used to measure social segregation across cities. The time-resolved analysis reveals the changing landscape of urban segregation and the time-varying roles of places. Segregations in physical and social space are weakly correlated at the individual level but highly correlated when grouped across at least hundreds of individuals. The methodology and analysis presented in this paper enable a deeper understanding of the dynamics of human segregation in social and physical space, which can assist social scientists, planners and city authorities in the design of more integrated cities.

Sources of Segregation in Social Networks: A Novel Approach Using Facebook

Most research on segregation in social networks considers small circles of strong ties, and little is known about segregation among the much larger number of weaker ties. This article proposes a novel approach to the study of these more extended networks, through the use of data on personal ties in an online social network. We illustrate this method’s potential by describing and explaining the degree of ethnic and gender segregation on Facebook among a representative survey of adolescents in the Netherlands ( N = 2,810; ~1.1 million Facebook friends). The results show that large online networks are more strongly segregated by ethnicity than by gender. Drawing on the same survey data, we find that core networks are more segregated in terms of ethnicity and gender than are extended networks. However, an exception to this pattern is personal networks of ethnic majority members, whose core networks are as segregated by ethnicity as their extended networks. Further analysis suggests this exception is due to their larger population size and the ethnic segregation of their social settings. We discuss the implications of these findings for the role of structural opportunities, homophily, and balance.

Segregation in Social Networks: A Structural Approach

This paper studies racial segregation in schools using data from Add Health. I estimate a structural equilibrium model of friendship formation among students. Preferences depend on direct connections, but also indirect friendships and popularity. I find that students tend to interact with similar people. Homophily goes beyond direct links: students also prefer a racially homogeneous set of indirect friends. I simulate several counterfactual desegregation busing programs, showing that policies that transport minorities to other schools have nonlinear effects on within-school segregation and welfare. In some instances, these interventions increase segregation within schools.

Network Effects and the Black-White Wage Gap

We study the impact of segregation in social networks on the disparities in labor market outcomes between Black and White workers. We provide a dynamic search and matching model of the labor market where we incorporate two kinds of network effects: i) job finding through social contacts and ii) transmission of work norms by network neighbors. When considering the first type of network effect, an increase in the level of social segregation leads to higher unemployment rates and lower wages for both the Black and White worker groups. At the same time, inter-group disparities increase, however, social segregation can account for only a negligible share of the Black-White residual wage gap.When studying the norm transmission role of social networks, the model is able to explain the whole residual wage gap, however, predicts a negative relationship between the level of social segregation and the wage gap, which contradicts the empirical evidence. We show that combining the two kinds of network effects can account for both the empirically observed magnitude of the Black-White wage gap and it's increasing pattern in the level of social segregation. We use this latter model version to study the impact of policy changes on the inter-group inequality.

Measuring segregation in social networks

Network homophily is a pattern in which ties are more likely to exist between nodes similar to each other. It is frequently observed for various types of social relations. At the same time, segregation is often encountered in urban areas as a tendency of families to occupy neighborhoods inhabited by other families similar to them. In this paper we conceptualize both phenomena as in the language of networks of interlinked positions occupied by a population of actors characterized by some node-level attribute. We review existing indexes and approaches to measuring the extent of homophily/segregation in social networks. We pursue an approach of, first, specifying a set of properties that a generic segregation measure might possess, and which, in our view, are relevant in substantial contexts. Second, we check which measures satisfy which properties. The use of measures is illustrated with four empirical examples. Given the particular application and the need for some descriptive measure of segregation, the results presented in this paper can help in selecting an optimal measure for the task at hand. We conclude that the most crucial aspects for the choice of a particular segregation measure include (1) whether the network ties or actors’ attributes are assumed to be subject to change, and (2) how one should treat the presence of network isolates.

Religion, friendship networks and home visits of immigrant and native children

Using data from a school survey of N=1190 children at the age of 10 in N=20590 directed dyads and p* models for network data, we investigate the impact of religion on migrant and native children’s friendships and visits at home. Deriving hypotheses from the formation of religious in-groups, our analyses show that having the same or a different religious affiliation as well as regularly attending worship has an impact on having a tie in friendship networks. Visiting alter’s home depends more on similarity in worship attendance. These results indicate that religious diversity can be an additional factor increasing actual levels of immigrant–native segregation in social networks.

GROUP INEQUALITY

We explore the combined effect of segregation in social networks, peer effects, and the relative size of a historically disadvantaged group on the incentives to invest in market-rewarded skills and the dynamics of inequality between social groups. We identify conditions under which group inequality will persist in the absence of differences in ability, credit constraints, or labor market discrimination. Under these conditions, group inequality may be amplified even if initial group differences are negligible. Increases in social integration may destabilize an unequal state and make group equality possible, but the distributional and human capital effects of this depend on the demographic composition of the population. When the size of the initially disadvantaged group is sufficiently small, integration can lower the long-run costs of human capital investment in both groups and result in an increase the aggregate skill share. In contrast, when the initially disadvantaged group is large, integration can induce a fall in the aggregate skill share as the costs of human capital investment rise in both groups. We consider applications to concrete cases and policy implications.

Social networks and labor market inequality between ethnicities and races

AbstractThis paper analyzes the relationship between unexplained racial/ethnic wage differentials on the one hand and social network segregation, as measured by inbreeding homophily, on the other. Our analysis is based on both the US and Estonian surveys, supplemented with the Estonian telephone communication data. In the case of Estonia we consider the regional variation in economic performance of the Russian minority, and in the US case we consider the regional variation in black--white differentials. Our analysis finds a strong relationship between the size of the wage differential and network segregation: Regions with more segregated social networks exhibit larger unexplained wage gaps.

Identity-based learning and segregation in social networks under different institutional environments

To study the evolution of segregation in social networks across systems embedded in different institutional environments, we develop an identity-based learning model where segregation is stochastically conditioned by the initial distribution of the actor's attention to identity and the updating of this distribution over time. The updating process, which we call the process of mutual learning multiplier, is based on an actor's success and failure experiences in tying with the same-subgroup and cross-subgroup actors. Results from a Monte Carlo simulation of the model show that the mutual learning multiplier produces disproportional relationships between the initial distribution of identity attention and the level of segregation in social networks. We also find that those relationships are affected by the actors' attention to structural holes, rate of learning from experience, system size, and the identity heterogeneity of the system. Overall, the model provides insights into various dynamics of network structuration across time and space.

The Cost of Segregation in Social Networks

This paper investigates the private provision of public goods in segregated societies. While most research agrees that segregation undermines public provision, the findings are mixed for private provision: social interactions, being strong within groups and limited across groups, may either increase or impede voluntary contributions. Moreover, although efficiency concerns generally provide a rationale for government intervention, surprisingly, little light is shed in the literature on the potential effectiveness of such intervention in a segregated society. This paper first develops an index based on social interactions, which, roughly speaking, measures the welfare impact of income redistribution in an arbitrary society. It then shows that the proposed index vanishes when applied to large segregated societies, which suggests an “asymptotic neutrality” of redistributive policies.