Diversity in science can bring together people with unique backgrounds, perspectives and ideas to create something new. But achieving diversity requires a clear understanding of the global research enterprise.
Examining publication data spanning 20 years and 27 subject areas, Elsevier and its partners revealed significant gender differences among global researchers. In the US, for example, women make up 40 percent of researchers across disciplines but represent only 20 percent of all researchers in the fields of energy, engineering, mathematics, physics and astronomy. Women are less likely to publish their work or collaborate internationally than men, and are under-represented among inventors.
These challenges were revealed in Elsevier’s recent report Gender in the Global Research Landscape, which has generated intense interest, with mentions in Forbes, Scientific American, Nature, Science, The Chronicle of Higher Education and The Economist. Now the findings are being discussed in a series of report symposiums around the world, starting in the US and continuing in Brussels on Friday.
In Washington, DC, more than 100 researchers, funding agency representatives, journalists and leaders from business, government and academia convened at the National Press Club March 31. Speakers focused on various aspects of gender equality in science and the power of data and analytics to reveal challenges and opportunities for action.
In her opening remarks, Ann Gabriel, VP of Academic and Research Relations at Elsevier, explained the impetus for the report, which combines Elsevier’s focus on information analytics with the organization’s longstanding support for gender equity in science: “Elsevier has been undergoing a rapid transformation to become not only a publishing company but a data and analytics (provider), using innovative technology to extract new knowledge from the science we publish and the networks that help to create it.”
Congressman Beyer: “Whatever you measure, you change”
Congressman Don Beyer, whose district is home to the National Science Foundation, expressed hope that the US would succeed in maximizing the strength of women in science and positions of leadership in the research enterprise.
“Our economies and cultures are stronger when we elevate women to all parts of society,” he said, pointing out that equal participation of men and women in scientific research drives innovation and strengthens the economy.
Beyer praised the Elsevier research team on their efforts in analyzing gender representation in research as a powerful step toward improving the gender balance in the STEM fields, reminding the audience that “whatever you measure, you change.”
We need to promote research partnerships with equitable gender representation
Linda Sanford, former Senior VP of IBM and a board member of Elsevier’s parent company, RELX Group, said collaborative research benefits from diversity, but more needs to be done to promote research partnerships with equitable gender representation.She noted that the Gender in the Global Research Landscape report comes at a time when the world is changing rapidly and needs the active participation of women to achieve innovation.
Dr. Sanford described two key takeaways. First, that decision making relies on data and its analysis, and rational people will make rational decisions when they have the facts in front of them. “This report has brought to the forefront data that clearly outlines the gender-related issues facing today’s global research enterprise and the need for advancing gender diversity,”she said.
Second, she recognized the need for more collaboration across disciplines, between the academy and business, and across countries. Sanford described a 21st-century movement away from a “siloed, command-and-control organization” and toward more horizontal, peer-to-peer collaboration that breaks down barriers and asks team members to listen to each other, take the best ideas, and build something new.
“Consensus just means the least common denominator,” she said. “We need collaboration to take best ideas and make them possible.”
NAS leader: Data-driven analysis of gender inequality in research is critical
The keynote address was given by Dr. Rita Colwell, Chair of the National Academies of Sciences, Engineering, and Medicine Committee on Women in Science, Engineering and Medicine (CWSEM) and Professor of Cell Biology & Molecular Genetics at the University of Maryland at College Park. She described the work of the CWSEM, which seeks to understand the complex social and cultural factors that influence the equity of women in research globally and is working to remove barriers and create supportive institutional environments to achieve gender equity. Dr. Colwell described the recent decision by the CWSEM to conduct a consensus study on sexual harassment in the STEM fields to gain insight into its impact on attrition and to develop strategies to prevent it.
Dr. Colwell said data-driven analysis of gender inequality in research is critical for illustrating the economic, social, and global consequences to stakeholders. She said she hopes institutions will make use of Elsevier’s report to ask themselves some hard questions: What programs do we support that create opportunities for women? How are we helping move women into positions of leadership? What are the cultural, social, and economic barriers that are facing women?
Dr. Colwell stated that the value of gender equity in research needs to be calculated and communicated in a rational, fact-based way that can be understood by legislators and policymakers — in terms of the businesses that don’t get started, the discoveries that don’t get made. Women are underrepresented in the sciences everywhere but “countries benefit when they have the full participation of their citizenry,” she said.
Dr. Colwell called on support for the social and behavioral sciences to continue investigating the issues facing the advancement of women in science, and to generate the data needed to drive cultural and policy change for global gender equality. She quoted Dr. Marsha McNutt, Director of the National Academies of Science: “Science without policy is still science, policy without science is gambling. We can’t afford to gamble.”
When asked how to get more men engaged in discussions about gender equity, Dr. Colwell stated that men in CWSEM are fiercely supportive of the mission and enormously supportive of women in science. Men are willing to participate and do understand; they just need to be actively invited to join the conversation, she said. Dr. Colwell also encouraged women in science to participate at the community level, showing elementary and middle school-age girls that there are women in science and sharing their knowledge with community groups. Parents and teachers need to believe in girls and support them to continue in STEM.
“It may take time,” she said, “but our shared efforts are going to make a difference to ensure that women — our daughters, sisters and granddaughters — are equally represented in the STEM fields.”
Dr. Londa Schiebinger: 3 “fixes” to improve diversity in research
Prof. Londa Schiebinger, Director of the EU/US Gendered Innovations in Science, Health & Medicine, Engineering, and Environment project and Professor of History of Science and Stanford University, started her presentation by asking, “Does diversity enhance creativity?”
She noted that while the simple answer is “yes,” achieving that creativity also requires more than just a diverse workforce — creative management structures and teams with engaged managers must use diversified research methods to seek diversified knowledge outcomes.
Dr. Schiebinger presented three “fixes” to improve diversity in research:
- Fix the number of women doing the research.
- Fix the institutions and their policies to support diverse research teams.
- Fix the knowledge by integrating sex and gender analysis into research.
That third fix asks, “Can we harness the creative power of sex and gender analysis to take research in new directions?” She noted that efforts to increase women’s participation in research will not succeed unless sex and gender analysis is part of the research being done.
Dr. Schiebinger described work at Gendered Innovations to develop a tool to quantify sex and gender of participants in clinical trials and in health research more generally. Without knowing the sex of the research subject, she said, the data cannot be analyzed and cannot be included in meta-analyses, which equals money wasted. Dr. Schiebinger shared the costly example of the Apple Healthkit app, which had poor market uptake because it was not designed to track the menstrual cycle. She explained that case studies of sex and gender bias in research design, data analysis, and data collection methods are powerful ways to illustrate the economic and societal benefits of integrating sex and gender into clinical research and health-related product design.
What are the solutions? Dr. Schiebinger pointed to various stakeholders who are positioned to promote the integration of sex and gender into the research enterprise, from granting agencies and journal editors, to educators and industry leaders. She explained that integrating sex and gender into research and product design can open new markets, enhance global competitiveness, and reach more customers.
“It’s a matter of excellence,” she said. “When you leave a variable — sex and gender — out of research and design, the quality of the research, or the product to be developed, is lessened.”
Examining the methodologies behind the report
Dr. Holly Falk-Krzesinski, VP of Strategic Alliances for Global Academic Relations at Elsevier, began by presenting highlights of the report. For example:
- The US is the only comparator country in which the impact of papers is higher for women than men.
- There is no evidence that inequalities in the representation of women researchers across countries and fields and in their scholarly output affect how their research is read or built on by others.
- Among inventors, women are generally under-represented, though the percentage of patent applications that include at least one woman among inventors is increasing.
- When men appear as authors in engineering papers, they are more likely to take the first or corresponding author position; the opposite is true in the field of nursing.
- Women’s scholarly output is less likely to result from international collaboration than men’s
Then she addressed concerns about the varied use of the terms sex and gender, particularly in the field of medicine, suggesting that natural language processing might be useful to determine when each term is being used correctly. There were several questions regarding the author counting and gender disambiguation methodologies used in the report. Dr. Falk-Krzesinski talked about the need for expansion of the study to the Middle East, Africa and Asia using alternative gender identification methodologies.
“As a steward of world research, Elsevier has a responsibility to promote gender equality in the STEM fields and advance our understanding of the impact of diversity in research,” she said. “As a leader in information analytics in health and science, we drew upon our robust data and analytics expertise, and the input of global stakeholders, including the World Intellectual Property Organization, to create the … report.”
Researchers talk about gender disparities in contribution and recognition and how to address them
In a panel moderated by Ludivine Allagnat, Senior Academic Relations Manager at Elsevier, the findings of the report were discussed in context of ongoing research into gender equity in STEM.
Dr. Cassidy Sugimoto, Associate Professor of Informatics at Indiana University Bloomington, described her research into who participates in science and how they are perceived and rewarded. She found that globally, science productivity remains dominated by men, with the exception of countries in which men have left to seek opportunities elsewhere, which speaks to cultural mobility differences between men and women. In a particularly compelling line of research examining the use of the CRediT contributorship taxonomy, her group found that women are more likely to have gained authorship by performing the research, whereas men are more likely to contribute to designing and analyzing the research and providing reagents. The same division of labor held even for women-led papers and for large groups of authors.
Similar to the findings of Elsevier’s report, Dr. Sugimoto’s group found that the proportion of women researchers and inventors is increasing, while the “care” disciplines — the health professions and social sciences — are feminized. Her work also revealed that women are more likely to collaborate nationally, and men internationally, and that both men and women seek men as coauthors because they are more central to the collaborator network.
But while many of the findings of Dr. Sugimoto’s research are confirmed in Elsevier’s report, her group also found much greater and more persistent gender differences in scholarly publishing, which have profound implications for the impact of women’s work. She found that not only is the citation impact lowest for single-author papers by women, but the impact of collaborative papers with a woman as lead author are still lower than papers with a man as lead author. The gender disparity in impact is greatest for papers published in higher-tier journals. Dr. Sugimoto suggested that this persistent bias in the way that women’s work is cited and received may be related to differences between how men and women collaborate, which have not changed over time.
Dr. Sugimoto also shared other research from her group that suggests that social media may be a democratizing influence on impact in that women’s research in some disciplines is “tweeted” more often than men’s.
Representing a partner in the development of the report, Dr. Julio Raffo, Senior Economic Officer at the Economic and Statistics Division of the World Intellectual Property Organization (WIPO), described his group’s work in identifying the gender of inventors filing applications through the Patent Cooperation Treaty (PCT) system, published in 2016.
Mirroring the challenges faced in the world of scientific publishing regarding gender identification of authors, Dr. Raffo acknowledged receiving pushback from PCT member countries to the idea of having inventors identify their gender on their applications. For their report, WIPO developed a World Gender-Name Dictionary (WGND) to disambiguate the gender of inventors. The dictionary includes 6.2 million names-country pairs, covering 182 different countries, and was used to analyze the gender of 9 million inventors and individual applicants of PCT patents.
While the WIPO study found that gender participation in the intellectual property system is improving, with progress across countries, disciplines, and in both academia and industry, Dr. Raffo noted that “there is still a long way to go to achieve gender equity—if the pace of applications stays the same, a gender balance in the IP space won’t be achieved until 2080.” WIPO intends to expand their research into gender equity in IP to include industrial design patents and trademarks in the near future.
“Gender equality is not only a human right but a necessary foundation for a peaceful, prosperous, and sustainable world,” Dr. Raffo said.
Dr. Kjersten Bunker Whittington, Associate Professor of Sociology at Reed College, spoke about the application of network dynamics to examine gender inequities in collaborative research relationships and in career trajectories. Her work evaluates the production of science as a social activity rather than as an individual scientist’s efforts — as a web of relationships rather than a “hub and spoke” that starts with one scientist.
Positioning within this web of relationships is crucial: by bringing together disparate information, “strategic ties” within a collaboration network are more likely to lead to innovation, Dr. Whittington explained. In her most recent research, her group found that while women and men have comparable reach and numbers of ties within a network of life sciences inventors, women are in fewer strategic positions in the network, are linked to other women more often than men are linked to men, and receive less return from their ties in terms of productivity.
Dr. Whittington said Elsevier’s report was a major contribution toward using large data sets to understand the basis for gender inequality in research, but stated that “data analytics are only as useful as the explanations that stem from their use, and best practices will stem from addressing the underlying mechanisms.” To ensure that women derive the same benefits from collaboration as men do, there is a need to look at all dimensions of intersecting scientific activity — publishing, patenting and involvement in epistemic communities — and then find solutions.
“To achieve gender equity in research, we need to consider who is missing from collaboration networks and why,” she said. “The challenge is to understand the root causes of marginalization.”
Download the report
You can download the Gender in the Global Research Landscape report and associated infographics and read more about the report findings here. All the references used in the report, plus many more from the literature, are available in a public Mendeley group, serving as a strong evidence base and community resource.
Elsevier plans to continue building on the findings of the Gender in the Global Research Landscape report and advancing gender equality through data-driven initiatives both internally and externally.
This report is one of many Elsevier has created as a resource for the research community, policymakers and government funding agencies, who need reliable data and analytics to make informed decisions. Other reports include Cancer Research: Current Trends & Future Directions in support of the Cancer Moonshot initiative, and Sustainability Science in a Global Landscape, released in parallel with the UN Sustainable Development Goals.
“Our data and analytics capabilities have enabled us to develop groundbreaking reports,” said Ann Gabriel, VP of Academic and Research Relations at Elsevier.
Upcoming report symposia
The European Union report symposium will be in Brussels on May 12. Then there will be presentations at the Gender Summit Asia Pacific (GS10) in Tokyo May 24-26 and at the Gender Summit North America (GS11) in Montreal November 6-8. Details about these events are posted on Elsevier’s Gender and Science Resource Center.
Join the conversation on Twitter
The cluster analysis resulted in six clusters of workplaces with different patterns of gender equality. The clusters were named by the main characteristics of their gender equality pattern as presented in Figure 2. No clusters were gender-equal on all five indicators. Four of the clusters were gender-equal in one or more of the indicators (C2, C3, C4, C5) whereas two clusters were gender-unequal on all indicators (C1, C6). The results are structured according to the two aims of the study and presented below.
Patterns of gender equality at the workplaces (aim 1)
The patterns of gender equality in each cluster are described below in relation to the indicators (Figure 2). The total number of employees from the register data in each cluster and proportion for young employees are described in table 2. For a richer understanding of each clusters, the participants from the Northern Swedish Cohort employed in the clusters are described in terms of socioeconomic position, type of work and occurrence of psychological distress (Table 3).
The workplaces in C1, Unequal with higher scores for men, were characterized by a gender equality pattern where men used more days of parental leave and temporary parental leave than women. These workplaces had a majority of men as employees and the men had higher salaries and a higher educational level than women. This cluster had a larger proportion of young men compared to young women. Among participants in the cohort, the most common occupational sector was manufacturing for men and administration for women. For both women and men, the largest group of participants was blue-collar workers and those working with things. Forty percent of the women and 32 percent of the men reported psychological distress at age 42, which was equivalent to the proportions in the total population. Differences between women and men in this cluster were not significant which can probably be explained by the low number of women in this cluster (n = 5).
Cluster 2, Socioeconomic equality & majority of women, was the cluster with the highest number of workplaces and therefore represents the most common gender equality pattern in the material. The characteristic feature of the pattern was that women were in the majority and used more days of both types of parental leave than the men at the same workplace, while the salaries and educational level were equal. This was also the only pattern with a majority of women. A high proportion of the cohort participants in this cluster were upper white-collar workers and those working with people. In this cluster women more often worked in the health-care sector whereas men more often worked in scientific and artistic work. This cluster included the largest proportion of psychological distress among men and the second largest proportion among women. For women there was also a significant increase in psychological distress (16 percent units) between age 21 and age 42.
The workplaces in C3, Socioeconomic equality & more parental leave for men, were characterized by gender equality in salary and educational level, just like C2. However, in this pattern men were in the majority and used more days of both types of parental leave. Among the participants in the cohort, blue-collar workers were the largest groups among men whereas upper white-collar workers were the largest group among women. The majority of women worked with people, whereas men mainly worked with things. Compared to the other clusters, the psychological distress among cohort women in C3 was low.
The workplaces in C4, Unequal with equal representation, were characterized by a gender equality pattern with an equal number of women and men whereas all of the other indicators were unequal. Men had higher salaries and educational level and women used most days of both types of parental leave. Among the participants in the cohort, upper white-collar workers were overrepresented in this cluster. A majority of the women worked with people, whereas working with data was the most common type of work among men. Psychological distress was as frequent as in the total population.
The gender equality pattern of the workplaces in C5, Equal in divergent spheres, was characterized by equal salary and equal division of parental leave between women and men. However, women took moderately more days of temporary parental leave. Women were in the minority and had a lower educational level than the men. In spite of that we consider this cluster to be the most gender equal cluster as these workplaces were classified as gender equal in the divergent spheres of economy and parental leave. The age distribution was similar among women and men at these workplaces. In this cluster women and men in the cohort were strikingly similar in the distribution of socioeconomic position, type of work, occupational sector and psychological distress. No significant differences between women and men were found. The largest group of participants were upper white-collar workers, the most common type of work was working with data and the most common occupational sector was administration. In this cluster women and men had a lower proportion of psychological distress compared to the other clusters.
The workplaces in C6, Traditionally unequal, had a gender unequal pattern with a majority of men, higher salaries for men, lower educational level for men and fewer days of both types of parental leave for men compared to the women at the same workplace. Among the participants in the cohort, working with data was the most common type of work in this cluster. The proportion of men in the manufacturing sector was significantly higher than among women. There were also significant differences between women and men in socioeconomic position as a majority of women were upper white-collar workers whereas a majority of the men were blue-collar workers. The women in this cluster had the highest frequency of psychological distress compared to women in the other clusters. Men on the other hand had the second lowest frequency of psychological distress compared to other men. Differences between women and men in psychological distress within the cluster were also significant.
Associations to psychological distress (aim 2)
In the Northern Swedish Cohort (n = 715) psychological distress at age 42 was reported by 39 percent of the women and 27 percent of the men. This was a significant increase of 10 percent units for women and 4 percent units for men compared to psychological distress at age 21 (Table 3). Among the participants in the cohort, there were also significant differences in psychological distress between the clusters (data not shown). However, in separate chi-square analysis for women and men, the differences in psychological distress between the clusters were not significant (Table 3). Multivariate logistic regression analyses were performed with psychological distress as outcome. C5 was used as the reference category as this cluster was gender equal on two indicators in divergent spheres. For men, there were no significant associations with psychological distress in bivariate or multivariate logistic regression analyses (data not shown). For women, belonging to C6 was associated with higher odds for psychological distress in all models except those including psychological distress at age 21 (Table 4). C2 was also associated with higher odds for psychological distress among women adjusting for type of work (model 4) and in the full model (model 6).