Marginalization Effects in Face-to-face Social Networks

By Fariba Karimi, Marcos Oliveira, and Markus Strohmaier

The act of marginalization—which Merriam-Webster Dictionary defines as “to relegate to an unimportant or powerless position within a society or group”—stems from various causes, including the inherent structure of organizations. Structural marginality refers to structural conditions that push certain groups towards a network’s margins, limiting their access to resources [4].

Despite this problem’s importance, researchers possess only a minimal quantitative understanding of its manifestation in networks and are thus unequipped to answer several urgent societal questions. For example, what underlying structural mechanisms drive marginalization? How can marginalized groups improve their social positions? Do hard (mathematical) limits exist on actions that would alleviate structural inequalities? As humankind grows increasingly interconnected and algorithms become an integral part of the decision-making process, scientists must tackle these pressing questions in order to achieve a just and equitable society.

Social networks are complex, time-dependent, and path-dependent systems that comprise heterogeneous individuals who interact within evolving, adaptive, and multilayer networks of friendship, collaboration, communication, and trade, among other categories [3, 8]. Much of the previous development in computational models of complex networks has focused primarily on universality, phase transitions, and the emergence of scaling [3]; as such, it has often overlooked the composition of social systems as individuals and groups with fundamentally heterogeneous attributes and historical precedents. For instance, women and people of color are historically disadvantaged in many academic fields and blue-collar jobs [5, 7]. The presence of such path dependencies and heterogeneities at the individual level can shape structural marginalization, which researchers must therefore approach from a complex system perspective.

We explore these issues in one of the primary forms of social communication: face-to-face gatherings [6]. Social gatherings are fundamental for the creation and preservation of meaningful and long-lasting relationships; in fact, this style of correspondence—which we inherited from our ancestors—is an essential mechanism for the transmittal and affirmation of culture [1, 2]. Unlike mediated communication, face-to-face interaction is particularly unique due to its spatiotemporal constraint: individuals must be at the same place at the same time in order to connect. This constraint is critical to interaction opportunities, especially in confined situations like conferences and workplaces. 

We used computational models of face-to-face interactions to consider the heterogeneity in individuals’ characteristics (e.g., race, gender, and ethnicity) and group mixing biases that establish the probability of interactions within groups. Given such conditions, can we computationally determine the effect of marginalization? When combined with spatiotemporal constraints, how do group size and group mixing preferences cause structural marginalization? And how should a marginalized group interact with other groups to increase its network visibility?

Figure 1. Degree (network connectivity) distribution of minorities and majorities in social gatherings in three scenarios: high intergroup mixing \((h = 0.2)\), no mixing preference \((h = 0.5)\), and high intragroup mixing \((h=0.8)\). Group mixing preferences that are based on the individuals’ characteristics can create network effects in degree visibility. Figure adapted from [6].

We found that the interplay between space and time constraints—together with group mixing and size imbalance—can produce structural marginalization, meaning that members of specific social groups have fewer social ties than average. When an individual utilizes an opportunity to interact with someone, fewer opportunities remain for interaction with other people. Depending on the way in which groups mix, this limitation can lead to inequality. For example, when a member of the majority interacts with someone else from the majority, there are less opportunities for this individual to interact with minorities — thus decreasing minority connectivity. This inequality emerges irrespective of individuals’ abilities. Even when minority members have high levels of intrinsic fitness (attractiveness or talent), they can end up at a network’s margins due to ingroup mixing bias.

What can minorities—or society—do to elevate their network visibility and overcome structural barriers in face-to-face interactions? Structural marginalization is challenging to mitigate without the help of the majority group. Our analysis indicates that majority group mixing accounts for most of the variance in connectivity inequality. The minority group alone can only slightly reduce inequality, and a successful strategy for reduction depends on group size. Depending on minority size, our model predicts that minority members must refrain from connecting with the minority group to increase their overall connectivity. We identified a tipping point that determines the way in which minorities should strategize their social interactions. When the minority group size is below a critical point, homophilic minority interaction decreases minority connectivity. Conversely, homophilic minority interaction helps increase minority connectivity when the minority group size is sufficiently large:

\[f_0^* = \frac{h_{11}}{2(h_{01}+h_{11})}.\tag1\]

The critical minority fraction \(f_0^*\) delineates two regimes wherein minorities’ average degree (the yellow distribution in Figure 1) may either increase or decrease with enhanced minority intragroup mixing \(h_{00}\), given fixed majority intragroup mixing \(h_{11}\).

Our social lives are rapidly expanding far beyond face-to-face gatherings. Digital spaces have transformed the way in which we connect with people and perceive information, and may potentially reinforce existing inequalities or shape new forms of marginalization. Additional computational network-theoretic research is necessary to shed light on this form of marginalization and the possible mitigation of its effects.

References
[1] Dunbar, R. (1992). Neocortex size as a constraint on group size in primates. J. Human Evol., 22(6), 469-493.
[2] Duncan, S., & Fiske, D.W. (1977). Face-to-face interaction: Research, methods, and theory. Mahwah, NJ: Lawrence Erlbaum Associates.
[3] Holovatch, Y., Kenna, R., & Thurner, S. (2017). Complex systems: Physics beyond physics. Euro. J. Phys., 38, 023002.
[4] Mehra, A., Kilduff, M., & Brass, D.J. (1998). At the margins: A distinctiveness approach to the social identity and social networks of underrepresented groups. Acad. Manag. J., 41(4), 441-452.
[5] National Center for Science and Engineering Statistics. (2015). Women, minorities, and persons with disabilities in science and engineering: Special report NSF 15-311. National Science Foundation. Retrieved from http://www.nsf.gov/statistics/wmpd/.
[6] Oliveira, M., Karimi, F., Zens, M., Schaible, J., Génois, M., Strohmaier, M. (2022). Group mixing drives inequality in face-to-face gatherings. Commun. Phys., 5(1), 1-9.
[7] Telles, E.E. (1994). Industrialization and racial inequality in employment: The Brazilian example. Amer. Sociol. Rev., 59(1), 46-63.
[8] Walby, S. (2007). Complexity theory, systems theory, and multiple intersecting social inequalities. Philos. Soc. Sci., 37(4), 449-470.

Fariba Karimi leads the Network Inequality Group at the Complexity Science Hub Vienna. Her research explores structural marginalization in social networks and algorithms with methods from complexity science. Marcos Oliveira is a lecturer in city science and analytics in the Department of Computer Science at the University of Exeter. His research interests lie in complex systems, human dynamics (including human mobility and face-to-face interactions), and self-organizing mechanisms. Oliveira is particularly interested in the use of network science to understand the emergence of social phenomena in cities—such as crime dynamics and social inequality—and to uncover the intricacies that drive the intelligence in swarm intelligence algorithms. Markus Strohmaier is the Chair for Data Science in the Economic and Social Sciences at the University of Mannheim in Germany. He is the Scientific Coordinator for Digital Behavioral Data at GESIS – Leibniz Institute for the Social Sciences and an external faculty member at the Complexity Science Hub Vienna. Strohmaier is interested in the development and application of computational techniques to research challenges at the intersection of computer science and the economic and social sciences.