Improving the visibility of minorities through network growth interventions

Leonie Neuhäuser, Fariba Karimi, Jan Bachmann, Markus Strohmaier, Michael T. Schaub

Introduction

In many social and professional networks, a "rich-get-richer" dynamic (preferential attachment) ensures that those who arrive early and connect well maintain their positions at the top of rankings. When combined with homophily—the tendency for individuals to associate with others similar to themselves—this creates a structural barrier for minorities. While institutions often use "group size interventions" (like quotas) to increase diversity, these measures frequently fail to translate into increased "visibility," or representation in top-degree rankings. This paper investigates the interplay between group size and social behavior to determine how minority groups can effectively break into a network's elite circles.

Methodology and Model Design

The researchers developed a two-phase network growth model based on an extension of the Barabási-Albert (BA) model. The process is divided into a pre-intervention phase (Phase 1) and a post-intervention phase (Phase 2).

Overview of the two-phase network growth model and intervention types.

In Phase 1, the network grows with an initial minority size ($min_1$) and homophily level ($h_1$). At a specific time $T$, parameters are adjusted to simulate an intervention:

  • Group Size Intervention: Adjusting the ratio of incoming minority vs. majority nodes ($min_2$).
  • Behavioral Intervention: Adjusting the homophily parameter ($h_2$), which dictates the probability of nodes connecting to their own group versus the other group.

The model was validated using synthetic data and a real-world case study of a co-authorship network from the American Physical Society (APS) journals (1940–1970), testing how different gender-parity interventions might have altered the visibility of female scientists.

Key Insights and Results

The study demonstrates that group size and social behavior are inextricably linked. A central finding is the coordination necessity: in highly homophilic settings, a high quota alone is often insufficient because minority nodes primarily link to each other, failing to capture the social capital held by the majority.

Impact of behavioral and group size interventions on minority visibility in top rankings.

As illustrated in Figure 3, the effectiveness of interventions varies significantly based on the type of social mixing. In homophilic settings, minority representation in the top ranks generally grows with the quota. However, in heterophilic (well-mixed) settings, increasing the quota can paradoxically decrease the relative visibility of the minority.

A critical "role change" threshold occurs at the 50% mark: if an intervention makes the minority the majority of newly arriving nodes, they should adopt homophilic behavior to solidify their position. Conversely, if they remain a numerical minority in the incoming stream, heterophilic behavior is more effective for rising in the ranks.

Furthermore, the timing of interventions is crucial. Because of preferential attachment, nodes from Phase 1 (the "old" nodes) have a massive head start. The APS case study revealed that interventions are far more effective when applied early in a network's life; once a network becomes "scale-free" and stable, structural inequality becomes much harder to dislodge.

Conclusion

Increasing the number of minority members in a network is often insufficient to improve their visibility in top rankings. Group-size interventions must be coordinated with behavioral changes—specifically how groups mix—to overcome the structural "glass ceilings" created by early-mover advantages and homophily. Future research should explore more complex social mechanisms and intersectional identities to refine these policy recommendations.

Summary updated

BibTeX
@inproceedings{neuhauser2023improving,
  author = {Leonie Neuhäuser and Fariba Karimi and Jan Bachmann and Markus Strohmaier and Michael T. Schaub},
  title = {{Improving the visibility of minorities through network growth interventions}},
  booktitle = {Communications},
  year = {2023},
  doi = {10.1038/s42005-023-01218-9},
}
Figure