Using network analysis, Karimi examined why disparities persist despite equal qualifications and research quality, as reflected in lower citations and publications, among other factors. In collaboration with Jun Sun from the Leibniz Institute for the Social Sciences in Cologne, she published a study in Communications Physics analyzing 668,000 publications from journals of the American Physical Society (APS) between 1893 and 2020, along with 8.5 million associated citations. Although the proportion of women authors increased over time, they still lag behind in absolute terms.
Karimi and Sun identified two key structural factors influencing this trend:
First, homophily, the tendency for individuals to collaborate with others who share similar characteristics. In physics, as in many fields, established professional networks tend to reflect historical demographics, which can inadvertently limit opportunities for underrepresented groups. Second, the advantage of established networks—researchers with well-developed professional connections, who are often already in positions of influence, have greater access to collaboration, financial supports, visibility, and impact in scientific publishing. These dynamics can reinforce disparities over time.
Our models show that quotas by themselves will not lead to lasting change unless individuals are meaningfully included in professional networks.
"Complex systems such as science are partly path-dependent, meaning small initial differences can have long-term consequences," says Karimi. "Feedback loops across generations, where certain groups have more resources, contribute to structural inequalities that create barriers for others."
To address these disparities, Karimi and Sun developed a dynamic network model of adoption that closely matched the real-world data of APS publications, citation and collaboration networks.
So how can structural barriers in academia—affecting women and other historically underrepresented groups—be reduced? Karimi suggests that universities and research institutions implement stronger incentives for fostering diverse teams. These could include financial support and policies such as hiring goals, dual-career plans, and child-care support. Highlighting successful mixed research teams as role models may also be an effective strategy.
However, Karimi emphasizes that policy changes alone are not enough: "Our models show that quotas by themselves will not lead to lasting change unless individuals are meaningfully included in professional networks." Beyond institutional policies, senior researchers and those who are sitting on hiring committees play a crucial role in ensuring that academic networks are open and accessible to all.
