In modern collaboration contexts, structural inequalities between men and women still persist. The “glass ceiling effect” describes the presence of an invisible barrier which seems to block women and other marginalized groups from accessing top ranks in their respective contexts. Its existence has been proven in a wide range of fields, ranging from women’s under-representation in boards and tenure track positions in STEM fields, to their coverage in music charts and in film production. Naturally, a substantial amount of research has tried to identify this phenomenon and its causes. Beyond other gender disparities, the presence of informal, social networks haven been identified as a potential explanation. These “old boy networks” are typically occupied by highly ranked, senior male professionals and function as social network through which crucial business information, such as job openings and referrals are propagated. Existing work has shown the existence of these structures, highlighted their importance, and dipped into understanding their structure and emergence.

A potential cause of such structural inequality are gender biases in individual in- teractions. “Homologous reproduction”, for instance, refers to the concept of men preserving their advantageous position by recruiting and promoting mostly other male applicants. Although such biases might occur unconsciously, they reinforce the glass ceiling phenomenon by excluding women from higher ranks. Existing literature has untangled many of these mechanisms qualitatively. A quantifying approach that spans various contexts, large temporal & population scales, and provides a thorough network perspective is still lacking, however.

To close this gap, Jan Bachmann will systematically study the emergence and persistence of old boy networks and their contribution on the glass ceiling effect in various collaboration networks. Temporally evolving collaboration networks enable the identification of structural inequalities, whereas rank dynamics of professional success pose a promising approach to quantify transparent barriers. The study will be carried out in four collaboration contexts: academia, software development, film and music production.

The first project focuses individual interactions as link formations in collaboration networks to provide a deeper understanding of old boy clubs. In these collaboration networks, individuals are represented as nodes connected by links if they collaborated on a common project, e.g., when publishing a scientific article together. On a macro- scale, the formation of highly connected node clusters surrounded by peripheral nodes is known as a “core-periphery structure”. In the first part, I will apply statistical tools to infer such structures, their temporal evolution, and participation of women therein. A systematic study of gender biases in link formation will then constitute the second part. In particular, “homophily”—the preference to interact with individuals similar to oneself—will be considered as a factor of homologous reproduction. Its symbiosis with “triadic closure”, i.e., the preference of people to connect to those whom they share common contacts with, could explain how this disadvantage is manifested by integration of new male collaborators into the existing old boy’s club. To this end, statistical analysis of “temporal motifs”, i.e., small building blocks of the network, will capture the formation of triangles, revealing gender biases in who collaborates with whom.

In the second project, biases in link formation and access to the old boy network will be coupled with how women are able to progress through rankings of professional success. I will utilize rank dynamics to categorize professional rankings and detect structural barriers, i.e., top rank regions that are rarely occupied by women. Moreover, identifying patterns in how men and women typically progress through such rankings in terms of speed, direction and jump distances will hint towards responsible mechanisms. A study of the co-evolution of old boy networks and rank dynamics will then conclude the second project. Multivariate lagged correlations between link formations and rank progression will indicate temporally predictive factors of the glass ceiling. Finally, synthetic models capturing these couplings will identify causal mechanisms and hint towards the effectiveness of mitigation strategies.

The contributions of the proposed studies are threefold: (i) they will provide em- pirical evidence of the glass ceiling phenomenon in multiple professional contexts, (ii) operationalize the identification of old boy networks, and (iii) provide a deeper un- derstanding on how glass ceilings and old boy networks develop and co-evolve. My findings will inform policymakers to design more efficient legal efforts to mitigate gen- der biases, diversify existing unbalanced structures and hence further deconstruct the glass ceiling to ensure better representation of women in influential positions.