Fig. 2.

CID (% female) by author's role and field of study. The CID for a year estimates how much larger or smaller the change from the previous year is compared with past change to the previous year. Panel A: We measured the CID for the percentage of female authors broken down by authors’ roles in a paper: the first, the last, and the rest (others). This shows that the% change of female first authors drops most significantly in 2020 by 0.33% (orange bars), reversing the increase in trend up to 2019 (blue bars). Although less dramatic, we also observe decreases in the% change of female last authors, which is unobserved for other author roles. Despite the significant decrease of female authors in first and last authorship, CID for any authorship decreased by only 0.02%, which suggests that the majority of female authors play other roles than either first or last author. Panels B and C: We present bar graphs of CIDs in (B) the first and (C) last authors by 19 fields of study post-pandemic in 2020 (orange bars) and pre-pandemic in 2019 (blue bars). Except for four fields, CIDs for 2020 tend to decrease relative to CIDs for 2019. In other words, the percentage of female first authors began to decrease in 15 of 19 academic categories in 2020. Although less apparent than those of the first authors, more than half of CIDs for 2020 exhibit negative values indicating a decline in the percentage of female last authors in most fields. In short, what we observe in panel A is not attributable to a few fields but common across various fields (for comparing to CIDs between 2018 and 2019, fig. S2).