The Role of Gender in Scholarly Authorship

Authorship order in academic papers is a high-stakes currency. West, Jacquet, King, Correll, and Bergstrom put it plainly: first and sometimes last author publications are at least as important as raw publication counts for hiring, promotion, and tenure. The first author gets credit for doing the work, while the last author signals senior leadership — the principal investigator or the group leader. Middle authors, in many fields, are largely invisible. Therefore, two researchers with identical paper counts can have very different records of visible intellectual leadership depending on where their names appear. That gap has consequences, and it turns out the gap is gendered in ways that raw totals completely obscure. To find out how deep that inequality runs, West and colleagues built an analysis on a scale large enough to actually see the patterns. Their dataset came from the JSTOR archive — eight point three million documents spanning from 1545 through early 2011, including four point two million research articles. They then focused on roughly one point eight million documents that cite or are cited by other documents in JSTOR, creating what they call the JSTOR network dataset. This citation-linked subset allowed them to place papers into disciplines using a technique called the hierarchical map equation, which detects communities in citation networks and groups papers into domains, fields, and subfields. Assigning gender was the other major methodological challenge. They used United States Social Security Administration records of the top thousand names per year for about 296 million births from 1880 to 2010. They extracted full first names from the corpus, dropping anyone who listed only initials, and required that a name be associated with a single gender at least ninety-five percent of the time to make an assignment. Names that were genuinely ambiguous, like Leslie or Sidney, were left unassigned. The method yielded confident gender assignments for seventy-three point three percent of the two point eight million authorships with full first names. It is worth being clear about the limitations: this is binary inference only, it depends on first names, and it can misfire across national contexts — Andrea, for example, is typically female in the United States but male in Italy. Still, the scale is extraordinary. Across six thousand eight hundred seventy-nine unique first names assigned, women represented twenty-one point nine percent of the gender-identified authorships in the network dataset. That twenty-one point nine percent figure is the baseline. Now watch what happens when you disaggregate it. The time trend is real. Women made up ten point six percent of authorships in the 1960s and twenty-nine point two percent in the 2000s. Across the full corpus, the female share rises from fifteen point one percent before 1990 to twenty-seven point two percent from 1990 to 2012. Progress is genuine but it is not uniform, and it is not complete — and this is where the paper's central argument lives. Breaking authorship down by position reveals the persistent fault line. Female first-authorship has improved markedly: it rose from nine point two percent in the 1960s to thirty point nine percent in the 2000s. Since the 1960s, the difference between total female authorships and female first-authorships has stayed within two percent. Women are essentially keeping pace at the front of the author list. The last-author position is a different story. Female last authors were fifteen percent in the 1960s and climbed to just twenty-two point eight percent in the 2000s. The discrepancy between total female authorships and female last-authorships has remained above five percent throughout. In many fields — particularly the biosciences — last authorship is the prestige signal. It marks the person who ran the lab, secured the funding, and drove the intellectual agenda. Women are systematically underrepresented there. The field-level variation makes this even more concrete. In mathematics, women represented ten point six four percent of authorships from 1990 to 2011, across six thousand one hundred thirty-four authorships. Philosophy was twelve percent, economics thirteen point six eight percent across sixty-nine thousand one hundred forty-two authorships. At the other end of the spectrum, sociology reached forty-one point four one percent female across forty-four thousand eight hundred ninety-five authorships, demography forty-one point nine percent, and education forty-six point three five percent across twenty-eight thousand six hundred thirty-five authorships. In molecular and cell biology — which constitutes more than half of the authorships in the JSTOR network dataset — women held twenty-nine point two five percent of authorships across two hundred seventy-seven thousand thirty-two papers, but were overrepresented as first authors and underrepresented at last-author positions. In sociology, women were underrepresented at both ends. In mathematics, where author order is commonly alphabetical, women were neither over- nor underrepresented by position, though their overall frequency remained strikingly low. The inequality takes different forms in different disciplines, but it does not disappear in any of them. Then there is the single-authored paper, which is where one of the study's most striking findings lives. A sole-authored paper is an unambiguous signal of independent intellectual contribution. There is no coauthor to share credit with, no collaborator whose role is unclear. You either did it, or you didn't. Women account for twenty-one point nine percent of authorships overall but only seventeen percent of single-authored papers across the full dataset. Decade by decade: eight point seven percent in the 1960s, twelve point five percent in the 1970s, eighteen point seven percent in the 1980s, twenty-four point five percent in the 1990s, and twenty-eight point five percent in the 2000s. Since 1990, women represent only twenty-six percent of single-authored papers. The gap between overall authorship share and sole authorship share persists. It matters especially because multi-authored papers have become more common over time while single-authored papers have declined — so these rare, unambiguous credits are distributed unequally precisely as they become rarer. This is a different form of inequality than gendered position in a multi-author list, but it reinforces the same message: counting papers gives you an incomplete picture of standing in a field. West and colleagues are careful to say their data cannot pin down causes. However, they do outline the structural patterns consistent with what they observe. Collaboration networks matter: their analysis shows that even in disciplines with near-parity overall, men and women cluster in different subfields. Women are, in many scientific fields, less likely to participate in collaborative research, which reduces opportunities to occupy prominent authorship positions in team-based work. Career stage compounds this. Last authorship accumulates over time — you have to become a group leader before you start appearing as last author. The authors acknowledge a lag is expected, but they also raise the possibility of what they call a leaky pipeline, where women disproportionately exit academia after graduate or postdoctoral training, before they reach the seniority that generates last-authorship. The numbers are consistent with this: forty-eight percent of PhD recipients were women in 2004, but women held only thirty-nine percent of full-time faculty positions. Negotiation and evaluation also play a role. The paper cites broader findings that men negotiate and self-promote more, and points to experimental and field evidence that judgments in review and hiring can disadvantage women. Because last authorship is an explicit leadership signal that feeds directly into hiring and promotion decisions, underrepresentation there can perpetuate itself. Fewer women in senior positions means fewer women generating last-authored papers, which means fewer women with the publication profile that leads to senior positions. The core contribution of West, Jacquet, King, Correll, and Bergstrom is methodological and conceptual at the same time. Methodologically, they showed that analyzing eight point three million papers across centuries and disciplines — using name-based gender inference and citation-network community detection — can reveal patterns invisible in aggregate counts. Conceptually, they showed that the question is not just whether women are publishing, but where they appear in the author list and whether they are publishing alone. Raw publication counts can make inequality look like it has nearly closed when the more consequential forms of it persist. Their call, at the close of the paper, is direct: academics should be aware of the subtle ways that gender disparities can occur in scholarly authorship. The data are out there. The tools exist to look at the right metrics. The question is whether the people making hiring and tenure decisions choose to use them. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.