Last year Maryam Mirzakhani, a Professor at Stanford University who was born and raised in Iran, was the first woman mathematician to be awarded with a Fields Medal (a mathematician’s “Nobel Prize”), along with three other laureates: Martin Hairer (Austrian, based in the UK), Manjul Bhargava (Canadian-American, Princeton University), Artur Avila (Brazilian-French, Paris). Their names alone show how incredibly diverse and international science is today.
And yet, in every country of the world, considerable gender gaps remain in practically every scientific field.
To estimate the overall gender gap in science, we inferred the likely gender of about 1,000,000 scientists referenced in ORCID.org and thousands of the world’s most renowned researchers mentioned in HighlyCited.com.
ORCID is a community-driven registry of researchers, with a Board of Directors including members from MIT, Thomson Reuters, Elsevier, CERN and other important organizations. Its aim is to solve the author and contributor name ambiguity problem in scholarly publications: homonyms, name variants, typos, transliterations, use of initials vs. full scientist name… ORCID provides an API and releases a data dump every year for data mining purposes.
HighlyCited is a web site listing researchers who earned the distinction by writing the greatest numbers of reports officially designated by Thomson Reuters’ Essential Science Indicators as Highly Cited Papers — ranking among the top 1% most cited for their subject field and year of publication. The ratio of women in HighlyCited is a matter of initial demographics (ratio of women studying in a given discipline, evolution over time) and recognition of talent by the scientific community. But as in many professions, other social factors may play a part: attrition rate among women pursuing a scientific career, unconscious gender biases in the way scientists themselves accept a paper for publication or make citations, differences in career paths or grants allocations…
Our Findings – Highlights
Those are some of our main findings. Starting from a theoretical ratio of 50% female and 50% male world inhabitants, we estimate that:
- Among one million scientists in ORCID.org, 33% are women.
- In 2001 women scientists were 7% of all researchers mentioned in HighlyCited; the number grew to 13% in 2014.
- The gender gap in HighlyCited researchers is narrower for social sciences (31% women in 2014) compared to computer science (9% women in 2014).
- In some fields the gender gap is closing fast: HighlyCited women engineers were 11% in 2014, up from 1% in 2001; same for mathematics 11% in 2014, up from 4% in 2001.
- In other fields the gender gap is not closing (example: HighlyCited female researchers in Physics were 4% in 2014, compared to 5% in 2001).
Commenting on the results, Kaisa Snellman, Professor at INSEAD, said:
“I was positively surprised to see that the gender gap in science has narrowed over the past decade. I find it especially encouraging that the share of women among “highly cited” researchers has almost doubled since 2001. At the same time, when I look around in the hallway, most of my tenured colleagues are male. Across universities, women are more likely to get stuck in lecturer and instructor positions that are often part-time and provide no access to tenure track. Of course, the pay is much lower in these adjunct positions – not to talk about job security. Hopefully the narrowing gender gap among highly cited researchers means that in the future we will see more female professors climbing up the career ladder at the same pace as the male scientists.”
Prominent male scientists promote the idea that science is not just for men. In France, Cédric Villani, mathematician who was awarded a Fields Medal in 2010, participated to a discussion with Zena M. (@ScienceFilles) and said,
“I think I generally had more girls than boys in my class until I specialized in Mathematics, then the ratio dropped abruptly. When I studied at ENS (Ecole Normale Supérieure, Paris) in 1992, I was in class with four female students out of forty. But in 1994, there were either six or seven girls studying mathematics at ENS! A record year, which brought to academics some of our best current mathematicians in France: Laure Saint-Raymond, Nalini Anantharaman, Sylvia Serfati*”.
[* Laure Saint-Raymond received the Irène Joliot Curie 2011 « Young female scientist » prize;
Nalini Anantharaman, Sylvia Serfati were both awarded with the Prix Henri Poincaré 2012]
Elaine Filadelfo (Twitter Data), said:
“Those working in STEM fields have the potential to radically change the lives of many people around the world. Not having adequate gender participation in those fields can hinder the ability of these fields to adequately reach 50% of our population — whether through developing medical treatments for women’s health issues or the development of public infrastructure that better serves families or technology designed with a diverse user-base in mind. Through research into the gender gap, we can work to diagnose and treat the issues preventing women’s equal representation.”
We will curate readers’ comments to update this study for June 18th 2015, to coincide with our talk at InspireFest 2015. This international conference takes place in Dublin, Ireland, and focuses on science, technology and innovation, featuring leaders from the world of STEM. If you are in academics and you would like to share your experience and views, we’ll be grateful.
Questions / Topics of Interest:
- Are you surprised by these numbers? If so, how?
- In your opinion, what are the biggest challenges that female scientists face?
- Do you think it’s harder to be recognized as a woman scientist?
- What can be done to improve these numbers, leading to more gender balance?
- Do you have anything else to add to the discussion?
How it works
We used NamSor to infer the gender of scientists’ names, recognizing that Andrea Rossini is likely male, whereas Andrea Parker is likely female. The software is highly accurate for all international names, including names of Africa, India… and it recognizes also that the gender of Chinese and Korean names cannot be determined. The error rate on gender gap estimates is typically less than one percent.