Scholarly collaboration now drives the frontier of knowledge and complexity, hyperspecialization, and research funding. Given the professional advantages of collaboration, it is not surprising that the relative rate of collaborative research has increased over time and now dominates solo authorships in all areas except for the humanities, with collaboration rates rising even there.
As in other professional relationships, researchers choose collaborators based on instrumental considerations such as expertise and resources and social features such as respect, trust, personal chemistry, and friendship. Homophily—the principle that similarity breeds connection between individuals—has been found to structure professional and socio-emotional relationships. Gender homophily, in particular, creates profound divides in work environments, voluntary associations, and friendships.
A new study investigated the role of gender in collaboration patterns by analyzing gender-based homophily. To do so, scientists have come up with a novel method that explicitly accounts for the fact that the data comprises heterogeneous intellectual communities and that not all authorships are exchangeable.
Scientists mainly considered a broad scholarly landscape—the JSTOR corpus, an archive that includes documents across the physical sciences, social sciences, and humanities. While previous studies have also considered a broad corpus of documents, the novelty of our work lies in a method that allows us to exploit a previously developed hierarchical clustering of science, which partitions the scholarly landscape at varying levels of data-driven granularity.
Given the characteristics of each small intellectual community and how it is positioned in the hierarchical clustering, the new method accounts for the observed homophily, which could be attributed to variations in gender representation and authorship norms across sub-disciplines and scholarly fields.
In particular, they distinguish three phenomena that may affect the distribution of observed gender homophily in collaborations: 1. A structural component due to demographics and non-gendered authorship norms of a scholarly community, 2. A compositional component is driven by varying gender representation across sub-disciplines and time, and 3. A behavioral component is defined as the remainder of observed gender homophily after its structural and compositional components have been taken into account.
The method, which uses minimal modeling assumptions, allows scientists to test for behavioral homophily. They found statistically significant behavioral homophily across the JSTOR corpus. It shows that this finding is robust to missing gender indicators in our data. In a secondary analysis, The study reveals that the proportion of women representation in a field is positively associated with the probability of finding statistically significant behavioral homophily.
Scientists noted, “Scientifically, this result may seem counterintuitive on its face; however, it is not surprising from the perspective of homophily: as the representation of women increases, it becomes more likely that same-gender individuals who are sufficiently compatible along other key dimensions become available as prospective co-authors.”
“Surprisingly, we find that the ratio of the proportion of single-authored documents written by women to the proportion of women multi-authorships is not significantly associated with statistically significant behavioral homophily.”
Journal Reference:
- Wang YS, Lee CJ, West JD, Bergstrom CT, Erosheva EA (2023) Gender-based homophily in collaborations across a heterogeneous scholarly landscape. PLoS ONE 18(4): e0283106. DOI: 10.1371/journal.pone.0283106