Gender disparities in the dissemination and acquisition of scientific knowledge

Abstract

Recent research has challenged the widespread belief that gender inequities in academia would disappear simply by increasing the number of women. More complex causes might be at play, embodied in the networked structure of scientific collaborations. Here, we aim to understand the structural inequality between male and female scholars in the dissemination of scientific knowledge. We use a large-scale dataset of academic publications from the American Physical Society (APS) to build a time-varying network of collaborations from 1970 to 2020. We model knowledge dissemination as a contagion process in which scientists become informed based on the propagation of knowledge through their collaborators. We quantify the fairness of the system in terms of how women acquire and diffuse knowledge compared to men. Our results indicate that knowledge acquisition and diffusion are slower for women than expected. We find that the main determinant of women's disadvantage is the gap in the cumulative number of collaborators, highlighting how time creates structural disadvantages that contribute to marginalize women in physics. Our work sheds light on how the dynamics of scientific collaborations shape gender disparities in knowledge dissemination and calls for a deeper understanding on how to intervene to improve fairness and diversity in the scientific community.

Figures (4)

• Figure 1: Acquisition and diffusion fairness in the circulation of scientific knowledge.a Schematic of knowledge flow through the collaboration network. An author (node) starts sharing knowledge at a given year $t$, and the spreading continues following the author's connections (links). Once reached by the knowledge flow, informed nodes (circled in red) start diffusing to their uninformed neighbors. Authors are split into two groups according to their gender: Women (yellow) and men (blue). The network varies in time: At each year, new nodes enter the network and connections may change. b Number of authors over time. Both male and female scholars have a growing trend. Women are the minority in the network, although their percentage has increased over the years. c Acquisition and diffusion of knowledge. To quantify how women acquire and diffuse knowledge, we evaluate (i) the fraction of female authors reached when the spreading starts from a random (either male or female) author in the network (Acquisition), and (ii) the fraction of authors reached when knowledge spreads from a woman (Diffusion).
• Figure 2: Acquisition and diffusion fairness. Knowledge can spread either from a female (yellow) or a male (blue) author. After a time $\Delta t$, among the $N$ authors of the collaboration network, $I$ of them will be informed (red circled). For the acquisition, we compare the fraction of informed individuals that are women to the fraction of female authors. For the diffusion, we compare the time $\Delta t$ needed to reach a given fraction of the population when the spreading starts from a woman or a man.
• Figure 3: Fairness in the acquisition and diffusion of knowledge.a Example of women's acquisition of knowledge for a contagion process that starts from a man. The fraction of female scholars in the collaboration network, $n_f$ (yellow) is compared to the fraction of informed scientists that are women, $i_f$ (red). Women among the informed individuals are generally less than expected. b Example of how knowledge diffuses from a female author, compared to a process that starts from a male author. The process starting from a woman takes longer than that starting from a man to reach the same fraction of individuals. c Acquisition fairness as a function of the starting year of the knowledge flow. The curves indicate different values of the fraction of individuals reached. The dashed line represents the fair condition in which the fraction of women among those who acquired knowledge is equal to the fraction of women in the network. In general, the acquisition fairness is below the fair condition, meaning that the women reached by the spreading are less than expected. d Diffusion fairness as a function of the starting year for different values of the fraction of individuals reached. The dashed line represents the fair condition in which the average duration of the process is the same when starting either from a woman or a man. In general, the diffusion fairness is close but significantly below the fair condition, meaning that women take slightly longer than men to diffuse their knowledge.
• Figure 4: Determinants of the unfairness of the system. a Probability distributions of the number of coauthors for men (blue) and women (yellow). On average, female scholars have less collaborators than male scholars. b Evolution of homophily in the APS collaboration network. The tendency of both male (blue) and female (yellow) scholars to collaborate with scientists having their same gender has increased over the decades. See Supplementary Information for details on how to evaluate homophily. c-d The determinants of acquisition and diffusion unfairness. Bars represent the value of acquisition and diffusion unfairness for the null models (shades of blue) and the data (red), for three starting years of the spreading process. The "Balance" model shows no unfairness, i.e., $AF=1$ and $DF=1$ (black dashed lines). The "Late Arrival" model does not show unfairness, suggesting the gap in the number of women in the collaboration network does not impact the way women acquire and diffuse knowledge. The "Coauthors" is in good agreement with the data, hinting that the gap in the number of coauthors has a significant negative effect on acquisition and diffusion fairness.