[Paper Review] Historical comparison of gender inequality in scientific careers across countries and disciplines

Junming Huang, Alexander J. Gates+2 more|arXiv (Cornell University)|Jul 9, 2019

scientometrics and bibliometrics researchDecision Sciences5 references73 citations

TL;DR

The study reconstructs 1.5 million scholars’ careers across 83 countries and 13 disciplines, showing that annual productivity is similar between genders but career length and dropout rates drive large gaps in total productivity and impact.

ABSTRACT

There is extensive, yet fragmented, evidence of gender differences in academia suggesting that women are under-represented in most scientific disciplines, publish fewer articles throughout a career, and their work acquires fewer citations. Here, we offer a comprehensive picture of longitudinal gender discrepancies in performance through a bibliometric analysis of academic careers by reconstructing the complete publication history of over 1.5 million gender-identified authors whose publishing career ended between 1955 and 2010, covering 83 countries and 13 disciplines. We find that, paradoxically, the increase of participation of women in science over the past 60 years was accompanied by an increase of gender differences in both productivity and impact. Most surprisingly though, we uncover two gender invariants, finding that men and women publish at a comparable annual rate and have equivalent career-wise impact for the same size body of work. Finally, we demonstrate that differences in dropout rates and career length explain a large portion of the reported career-wise differences in productivity and impact. This comprehensive picture of gender inequality in academia can help rephrase the conversation around the sustainability of women's careers in academia, with important consequences for institutions and policy makers.

Motivation & Objective

• Quantify long-term gender differences in scientific careers at scale across disciplines and countries.
• Decompose total productivity and impact into annual productivity and career length.
• Identify mechanisms (career length, dropout) underlying observed gender gaps to inform policy.
• Test robustness across multiple bibliometric databases to ensure generalizability.

Proposed method

• Reconstruct complete publication histories for 7,863,861 scientists from Web of Science (1900–2016) and identify genders for over 3 million authors.
• Focus on 1,523,002 scientists whose publishing careers ended by 2010 to enable full career comparisons.
• Replicate analyses in two additional datasets (Microsoft Academic Graph and DBLP) to validate results.
• Identify gender via a state-of-the-art gender inference method and remove self-citations and citation inflation in impact measures.
• Define and compute career length as time between first and last publication; measure annual productivity and citations at 10 years post-publication (c10).
• Employ matching experiments to control for country, discipline, affiliation rank, and exact career length to isolate effects of career length and dropout.

Experimental results

Research questions

• RQ1 What is the gender distribution of active authors across disciplines and countries from 1955–2010?
• RQ2 Do men and women have similar annual publication rates when controlling for career length and other covariates?
• RQ3 To what extent do differences in career length and dropout rates explain observed gaps in total productivity and citations?
• RQ4 Are there invariant quantities in performance between genders (annual productivity and per-work impact) across contexts?
• RQ5 How robust are findings across different bibliometric databases (WoS, MAG, DBLP)?

Key findings

• Women account for 27% of authors overall, rising from 12% in 1955 to 35% in 2005.
• Average annual productivity is nearly identical: women publish 1.33 papers/year vs. men 1.32.
• Men have longer careers on average (11.0 vs. 9.3 years), contributing to higher total productivity and impact.
• Total productivity gap is 27.4%; total impact gap is 30.5% overall, larger among top-productive authors.
• Career length differences explain a large portion of gaps; matching on career length reduces productivity gap from 27.4% to 12.4% and impact gap from 50.7% to 13.1%.
• Dropout rates are higher for women (10.8% vs. 9.0% annually), implying a 67% reduction in gaps when accounting for dropout; counterfactual simulations show substantial reductions in gaps under equal dropout assumptions.
• In a productivity-matched scenario (same total publications), the total impact gap nearly vanishes (1.9% in favor of women).
• Gaps persist across disciplines and countries, with Iran and Serbia as notable partial exceptions.