Editor’s Note: Today’s post is by Ana Heredia. Ana is a former researcher turned scholarly communications specialist. A biologist by training, after an MSc in Neurosciences, a PhD in Sciences, and two postdocs, Ana joined the STM publishing world, developing expertise in scientific information infrastructure, analytics, and publishing.

A recent article in Nature, Intersectional Analysis for Science and Technology, underscores the critical role of intersectional analysis in science and technology as a means to inform more effective social and environmental policies that promote global equity and sustainability. Intersectionality describes interdependent systems of inequality related to sex, gender, race, age, class and other socio-political dimensions. By focusing on the compounded effects of social categories, intersectional analysis can enhance the accuracy and experimental efficiency of science. The paper illustrates how quantitative intersectional analysis can be embedded throughout the research process — from setting strategic research priorities and shaping research questions to data collection, analysis, and interpretation. The outcome is a comprehensive set of recommendations designed to support researchers, peer-reviewed journals, and funding bodies in systematically incorporating intersectional perspectives. These recommendations are formalized in the Guidelines for Intersectional Analysis in Science and Technology (GIST) and are endorsed by the European Association of Scientific Editors (EASE).

Londa Schiebinger, co-author of the article and the Guidelines, kindly agreed to be interviewed for The Scholarly Kitchen. As a Professor of History of Science and the Founding Director of Gendered Innovations in Science, Health & Medicine, Engineering, and Environment at Stanford University (USA), Londa has been advocating for the importance of sex, gender, and intersectional analysis for innovation and discovery, considering that these approaches add valuable dimensions to research. She is also co-author of “Sex and Gender Analysis Improves Science and Engineering”.

What is the importance of adopting an intersectional approach in research and scholarly publishing, and why are the GIST necessary?

We need the GIST guidelines because intersectional analysis typically isn’t part of graduate training for natural scientists and engineers. This is a great fault. Scientists, engineers and medical people don’t learn sex analysis, gender analysis, intersectional analysis, and the AI people don’t learn the basics of social analysis that might help them avoid potentially harmful impacts in their work. The GIST guidelines are important now, but I’m hoping that we can put ourselves out of business rather quickly: My final goal is to have these types of analyses embedded in the curriculum, in the professional training of basic scientists, engineers, climate people, and medical people. That’s what’s needed. 

As an interim strategy, we can use the infrastructure of science to encourage the kind of thinking that produces excellence in science. At the beginning of the research process, funding agencies can ask that sex, gender, and intersectional analysis be integrated into applicants’ proposals. The European Commission does so already, and the US National Institutes of Health did until the recent Trump clampdown. Canada has been asking this both in the realm of health and at the Natural Sciences and Engineering Research Council (NSERC), their engineering science funder. These agencies encouraged (some required) applicants to integrate sex, gender, and/or intersectional analysis into the design of research, or to justify that it is not relevant. 

The idea is — and this is important — that if taxpayer money is to be used, the research should benefit everyone. And at the end of the process, peer-reviewed journals can require sophisticated sex, gender, and intersectional analysis when selecting papers for publication. In other words, peer-reviewed journals can hold people’s feet to the fire. As we say in our paper, by adopting the GIST guidelines, journals can signal the minimum expectations for rigorous, reproducible research. Intersectional analysis is important because it leads to better science; precision in science supports smart policy, and smart policy can best guide global solutions to climate change, public health, and AI safety issues. So that’s what we’re doing. Without integrating intersectional analysis, researchers risk amplifying existing inequities, both societal and environmental. There are many examples of how this approach can lead to better science. And that’s what we tried to spell out.

What methodology was employed in the GIST’s development, and how was this set of recommendations formulated and validated?

Our methodology and strategy for validation were quite clear. I assembled a global, multidisciplinary group of authors. It’s very hard to work with 12 co-authors from around the world, but that’s precisely what I considered necessary. We had two authors from Africa (Kenya and South Africa), and two from Asia (China and South Korea). We had several from Europe (Denmark, Italy, and Switzerland), as well as authors from the UK and the US. These authors were selected to ensure that we had expertise in the disciplines required. One author came from sociology, a discipline where intersectional approaches were developed primarily for qualitative work, and so this perspective was crucial to our work. In our efforts to expand these methods to the quantitative sciences, we also recruited authors from marine science, public health, climate change, planetary health, food and nutrition, environment, energy, computer science, and electrical engineering, as well as African American studies, information studies, race and digital justice, and history of science (that’s me). With this global, multidisciplinary group, we wanted to know if the intersectional factors we described, and the way we described and illustrated them through examples, made sense in their geographical location and discipline. For example, we know that race, when used as a category of analysis in the United States, is conceptually different from when it is used in South Africa. And some European countries don’t talk about race any more for historical reasons. So we wanted to be very careful with varying cultural contexts. 

Many of our authors also had practical experience we were keen to tap into. Our paper offers a set of guidelines for researchers, peer-reviewed journals, and funding agencies that facilitate systematic integration of intersectional analysis into relevant domains of science and technology. A number of us have worked with public funding agencies in the past, several of us with the European Commission, which, in many ways, forged the pathways for these kinds of questions in the grant funding process. One of our colleagues is from the National Research Foundation of South Africa, and she came with many insights. We had colleagues from the Global Research Council, which was important since they serve as an umbrella organization for research councils. Other authors are editors of major peer-reviewed journals, including Richard Horton, editor-in-chief at The Lancet. We were keen to have his perspective on whether what we are recommending is practical. Another author, Shirin Heidari, was instrumental in setting up the Sex and Gender Equity in Research (SAGER) Guidelines. SAGER have been highly effective and has now been adopted by thousands of journals. We wanted this experience; we didn’t want to start de novo. 

Through these processes, we defined and developed the guidelines to serve as a roadmap for quantitative intersectional analysis. We adopted the basic Gendered Innovations approach that guides researchers through the research process — from setting strategic research priorities and shaping research questions to data collection, analysis, interpretation, and dissemination. We used this process to organize our Nature paper. In each step along the way, we offered methodological strategies extracted from examples across various disciplines, with a focus, where possible, on machine learning/AI and climate change.

In this paper, we conceptualized intersectional factors at three interconnected, nested levels. At the center of the nest are the familiar socio-political dimensions, such as sex, gender, ethnicity, caste, religion, etc. These factors are embedded in contextual domains, such as the legal system, the healthcare system, the educational institutions, etc. They respond further to environmental conditions of air quality, water quality, and the like. It’s understanding the push and pull of these three basic levels — the socio-political dimensions, contextual domains, and the environmental conditions that, in my view, is important for intersectional analysis. Something we might do in the future is provide a web-based resource with current definitions of these factors from across cultures, to assist researchers who may not be familiar with their nuances. For instance, what does the American Heart Association say about gender? What does The Lancet say? And so forth. If the resource is web-based, it can be changed as societies and circumstances change.

What are the potential practical consequences of omitting the adoption of the GIST within the research process, including publishing?

Researchers don’t intentionally omit intersectional analysis. It’s just unconscious. Intersectional analysis may not have been part of their professional training, or they might be in a hurry to publish the next paper needed for promotion. The danger is that if you do not include all relevant variables, you get the research wrong. And there is a social cost to that. An iconic example that we report in our paper is the work Joy Buolamwini developed as a graduate student at MIT. She was developing facial recognition systems for her PhD when she realized that her own system could not “see” her—a beautiful Black woman–until she put on a white mask. So she switched the emphasis of her work to understand why. In Gender Shades, she and her co-author analyzed the intersection of race and gender in facial recognition systems and showed that they do not see Black women’s faces well. The gender analysis showed that the systems performed better on men’s faces than on women’s faces, while the skin-tone analysis showed that the systems performed better on lighter skin than on darker skin. But the intersectional analysis provided a fuller story: the system performed worse for Black women. Error rates were 35% for darker-skinned women, which is unacceptable, 12% for darker-skinned men, 7% for lighter-skinned women and less than 1% for lighter-skinned men, because they used themselves as models for developing the systems. Since facial recognition has many uses—opening your phone, security for international borders, etc., these systems need to work for everyone. 

What I like about Buolamwini’s work is that she also looked for solutions. She and her co-author developed a new data set that was more inclusive globally and that worked with commercial services to improve the performance of their systems. We call this an intersectional innovation! And that’s the goal: to create science and technology that is more precise and useful. 

What are the most significant positive outcomes resulting from the adoption and implementation of the GIST by scholarly editors and publishers in particular?

Let me choose one example from our paper, which highlights research on South Africa. This is an example showing that considering the intersections of gender, race/ethnicity, and geographic location improves the accuracy of results. In rural areas of South Africa, women, especially Black women, often suffer from energy poverty. When energy fails, girls are sent in search of cooking fuels, in this case, firewood. When so employed, they likely cannot attend school and therefore risk a lifetime of illiteracy. This approach shows the value of considering the interdependent dynamics of race/ethnicity, gender, and geographic location that policymakers need to take into account when seeking solutions. If we don’t get the full picture, NGOs may invest precious resources in the wrong places. 

What are the broader implications of both adopting and not adopting the GIST guidelines for promoting global equity and sustainability in scholarly publishing?

Another eye-opening example in our paper looks at the intersection of gender and religious preference. Interestingly, in this study, religious preference, and not socioeconomic status, plays a major role.  The study was about parents seeking care for their children. The researchers found a clear gender preference in this Muslim community – parents sought care for boys more quickly than for girls. The deciding factor was not so much the ability to pay as the community giving preference to boys’ health over girls’. And so a community-level variable helped to reveal the dynamics that could inform a culturally sensitive intervention to mitigate these disparities. Again, aid workers are better able to offer help when the key factors are identified. 

Would you outline a few specific actions that editors and publishers can undertake to effectively adopt and implement the GIST within their journal?

I very much appreciate EASE endorsing our guidelines. And they’ve invited us to write an article, which I’m working on now. Similar to the EASE publication on SAGER, we are developing a rapid checklist for journal editors and authors. You can’t prescribe how research is done, but the rapid checklist goes through the GIST guidelines in a systematic way that will help authors and editors understand what is expected in each step of the research process. For example, if the paper focuses on intersectionality, then that should be indicated in the title. If intersectionality is not the focus of the paper but is an important part of the analysis, point to that in the abstract. And so on. We hope this checklist will be a useful tool for both editors and researchers. 

Journal editors could also offer workshops for peer reviewers on the key points to look for in intersectional analysis, or “how-to” videos. I don’t know if journals do these sorts of things; funding agencies do. The issue is that, by-and-large, neither researchers nor reviewers know how to do intersectional analysis, and this sort of intervention by journals would proactively offer an incentive to learn more. In any case, as a next step, my group is developing our rapid checklist. 

Thinking from a researcher’s perspective, you mentioned some chemistry domains where intersectionality wouldn’t bring any new perspective. But for other domains, where intersectionality would not be traditionally considered, it may be important to keep some parameters in mind in different steps of the research cycle, as this improves the quality of the data and the quality of the research. So, what questions should we be asking ourselves when we conceive our hypothesis, or when we consider the experimental setup? In each of the steps of the research, what should we be thinking of? This is why the GIST are so valuable, as you’ve included as authors people from different research fields.

It’s not only some domains of chemistry where intersectional analysis is not relevant, but also areas of theoretical physics. Black holes, for example, have no sex, gender, or socioeconomic status. We don’t want to push a perspective where it is not relevant. However, researchers should think carefully about intersectionality before ruling it out. 

We started creating a method for the Intersectional Approaches for the Gendered Innovations website. In 2018-20, the European Commission supported a 25-person expert group to update Gendered Innovations — this resulted in Gendered Innovations 2: How Inclusive Analysis Contributes to Research and Innovation. For this update, Matthias Nielsen and others in the group created a method for intersectional analysis throughout the research process. There, we highlighted the kinds of questions you might ask when identifying the problem, designing the research, collecting data, etc. Now that we have finished our Nature paper, we should revisit this. We’ve long had well-developed methods for Analyzing Sex and for Analyzing Gender; we’d like to do the same for Intersectional Approaches in the coming year. 

This is already fantastic, by the way. I don’t think that researchers in my research domain (ants’ eco-ethology), for example, are concerned about intersectionality. Maybe those involved in research with direct implications for humans or society know about it. In any case, I agree with you that this perspective should be taught and be part of our formation as scientists. Because it’s not just about the research we do, it is also about how we interpret the papers we read, which is key for critical thinking in general.

In 2010, public funding agencies began asking for sex and gender analysis in applicant proposals; they put out videos on how to do this type of research, primarily in biomedical and clinical research. Canada’s Institutes of Health Research made excellent training videos, as did the US National Institutes of Health’s Office for Research on Women’s Health (though, sadly, those have now been taken down). It would be marvelous if we could create such training videos for intersectional analysis. I’ve long thought that funding agencies should require a certificate verifying that people have passed some sort of training in order to be eligible to apply for public funding. I don’t think that’s going to happen, but it’s an interesting idea. It would be fun to put some money together to make such training videos. I’m not the person to do that, but plenty of people have the skill to create interesting interactive online courses that begin to teach researchers how best to do this type of research. These types of measures are necessary until sophisticated methods of intersectional analysis are taught, as a matter of course, in natural scientists, medical researchers, and engineers’ professional, graduate training.