How AGU Used Data to Address Diversity in Science
Image credit: Rawpixel.com/Shutterstock
For 12 years, Paige Wooden had managed peer review in the publications department at the American Geophysical Union. In 2018, she moved into data analytics. She explains here how AGU has used its data to address diversity in science.
Over the past few years, the American Geophysical Union has taken a close look at the gender, age, geographical and ethnic diversity of its member scientists and their networks. The goals are to 1) ensure internal operations progress in an intentional direction and 2) we're educating and inspiring the scientific community to reflect on its practices and institutional policies.
AGU is a community of 130,000 Earth and space scientists and enthusiasts committed to building and supporting a diverse and inclusive research community that can address humanity’s grand challenges.
Our investigations on gender within the peer review processes and in author networks were based on AGU’s journal and annual meeting data.
AGU Journals Invite Too Few Female Reviewers
In one assessment1, my colleagues Brooks Hanson and Jory Lerback found that AGU journal editors were inviting fewer women to submit reviews across most age groups when compared to the pool of potential reviewers. Brooks and Jory suggested the potential reviewer pool should consist of the same proportion of female lead-authors of our journals.
The analysis found that papers from female corresponding authors have a higher acceptance rate than papers from their male counterparts. And that on average women were submitting fewer papers than men: overall, as a percentage, and per person,
Figure 1: Original analysis by Jory Lerback & Brooks Hanson.
Source: Jory Lerback & Brooks Hanson, Nature 541, 455–457 (26 January 2017)
© Macmillan Publishers Limited, part of Spring Nature. All rights reserved.
Permission obtained to re-use using Copyright Clearance Center’s RightsLink® service.
In 2017, in an attempt to address the disparity between invited reviewer and authors, AGU journals updated the manuscript submission form to ask submitting authors to provide the names of potential reviewers in underrepresented groups:
Evaluation of our journals’ peer review practices suggests that women were less likely than men to be asked to review. Please help us improve the diversity of our reviewer pool by including women, young scientists, and members of other underrepresented groups in your suggested reviewers (e.g., age, ethnic, and international diversity).
The journals' editors-in-chief have also made a concerted effort to increase the number of female editors and associate editors on their boards. As a result, we have seen an increase in female editors, associate editors and invited reviewers as shown in Figure 2.
Figure 2: Proportion of female corresponding authors, invited reviewers, members and editorial board members from 2014-2020.
Source: American Geophysical Union.
“Girl Power” in Peer Review
All journal manuscripts are assigned an editor. About 35 percent of the manuscripts are subsequently assigned to a subject-specific associate editor (AE) who will invite reviews and make a recommendation to the editor.
Of the manuscripts sent to an AE, papers submitted by women were more likely to be assigned to a female AE. If the editor was a woman, even more papers submitted by women – 32 percent – were assigned to a female AE. Compare this to when the editor and corresponding author were both men: only 19 percent of the papers were assigned to a female AE (see Figure 3).
Figure 3: Proportion of manuscripts assigned to female AEs by gender of corresponding author and editor in journal submissions sent to review from 2016-2019.
Source: American Geophysical Union.
Grouping women together in the peer review process produces, what I like to call the “girl power” effect. Contributing factors could be (1) certain fields of study have a higher proportion of women, (2) women in leadership roles (in this case an editor) may be more cognizant of the need to provide opportunities to female colleagues, and (3) a tendency to organize “like” things together.
Author Networks Show Gender- and Age-Affinity Biases
Differences between the professional networks of men and women are also likely to affect the way women are grouped together in the peer review process. In another analysis2 we looked at the composition of these networks. Using five years of annual meeting abstracts, we analyzed more than 400,000 unique author-author connections to construct co-author networks by age, gender and country.
We found that men’s networks included fewer women across most age groups when compared to women’s representation in AGU membership. Women’s networks included about the same or more women compared to their representation in our member population.
We also found that both men’s and women’s networks were weighted towards those in their own age group. The exception was those in their 20s, whose networks included more members who were older.
Our research also showed that women had fewer international co-authors than men did and, consequently, a higher proportion of co-authors from their own country. We found higher rates of this geographical insularity for women in all countries except Canada, Switzerland and the U.K. This shows that women are more likely to have a balanced network regarding gender, but fewer opportunities (either given or taken) to collaborate beyond their countries’ borders.
By looking at the demographics of our authors’ networks and of the peer review process, we hoped to unearth any biases and provide solutions for addressing those biases. Our findings show that it is important to be intentional about all the decisions – small or large – made during the peer review process, and to understand whether one option or another is actually moving towards or away from diversification goals.
As long as men’s networks are largely confined to other men and all professional networks include people of similar ages at disproportional rates, young women’s ability to expand their networks may be limited. Another limitation women face is the relative lack of opportunity for international collaboration.
At an enterprise level, it is important to periodically examine overarching trends using large datasets to identify any unconscious (or conscious) biases so we can be more intentional about our actions. We can also use these findings to set diversity and inclusion goals and periodically assess our progress.
Unconscious biases affect not only our own professional and personal lives, but also the lives of our colleagues, mentees and, in the case of the peer review process, scientists we have never met.
- Jory Lerback & Brooks Hanson. Journals invite too few women to referee. Nature 541, 455–457 (26 January 2017)
- Hanson, B., Wooden, P., & Lerback, J. (2020). Age, gender, and international author networks in the earth and space sciences: Implications for addressing implicit Bias. Earth and Space Science, 7, e2019EA000930.