Reviewed February 2021
Meredith L. Hale, Metadata Librarian
University of Tennessee, Knoxville
Data Feminism, by Catherine D'Ignazio and Lauren F. Klein, aims to demonstrate to readers how data can be used to “remake the world.” Building upon scholarship on intersectional feminism, data science theory, and examples of data projects, the authors formulate seven principles of data feminism aimed at exposing and addressing forces of structural oppression in data. Each chapter is centered on one of the seven principles: examine power, challenge power, elevate emotion and embodiment, rethink binaries and hierarchies, embrace pluralism, consider context, and make labor visible.
The targeted audiences for the publication are broadly “data scientists” and “feminists.” The authors also note that the work will prove helpful to anyone making data-driven decisions or those seeking to more fully understand and create data visualizations. Library and museum professionals who identify with one or more of these groups will find it a helpful guide that causes them to question their assumptions about data. It would also prove to be a helpful source for students, faculty and staff who are interested in these topics, regardless of their level of expertise. The authors welcome a broad audience to the text through clearly defining their terms, writing in a conversational tone, and consciously citing data examples from outside of the academy (including many examples from social media).
The text highlights pitfalls to avoid in creating, visualizing, cleaning, and sharing data in order to confront unequal power structures and develop counterdata that will spur society towards justice. One issue is that often people and their challenges go uncounted. This problem is featured through the story of Christine Darden, a Black mathematician at NASA whose life was recently dramatized in the movie Hidden Figures. She found that women were not promoted with the same frequency as men with similar educational backgrounds. A bar chart showing the disparity between gender and rank made this sexist practice visible and ultimately helped Darden reach the top rank in the federal civil service. Conversely, sometimes being counted has dangerous consequences. This is illustrated through the example of redlining practices in which neighborhood demographics documented in residential security maps rather than creditworthiness were used to determine whether individuals (especially those living in Black neighborhoods) were granted a home loan. Throughout the text, the authors address many misconceptions about data that contribute to negative outcomes. Because data sets are produced by people, data is never raw nor objective.
While the work is structured as a book, with citations and sequential chapters, its open availability online and special features distinguish it from a traditional tome. Data Feminism is an open access resource that has been assigned a Creative Commons Attribution 4.0 International license. Making it freely available online further supports the publication’s call for transparency and open sharing of data. Users can interact with the open access edition on PubPub, a community publishing platform supported by MIT’s Knowledge Futures Group, that emphasizes collaboration and iteration in the editing process. PubPub offers free accounts that allow readers to comment on the content of Data Feminism as they read. While comments were minimal in the “final” edited version, an initial draft of the work utilized this feature extensively as part of an open peer review process. As an online publication, Data Feminism differs from traditional books in that it does not have any page numbers to use for citations or sharing generally. The PubPub platform has made it possible to link out to particular paragraphs or images to solve this problem. One feature that could improve the user experience is direct links to figures mentioned parenthetically in the text. For those wanting to learn more about the code that runs PubPub or those that developed it (following D’Ignazio’s and Klein’s call to “make labor visible”), the code is openly available in the platform’s GitHub repository.
Beyond highlighting projects that exemplify the seven principles of data feminism in the written text, the online publication itself also reflexively enacts the principles it preaches. This is most prominently seen through the chapter on the author’s values and metrics. Here the authors demonstrate how they have attempted to “embrace pluralism” through including data projects or theory that were led or developed by women, people of color, transgender people, Indigenous people, nonacademics, and other minoritized groups throughout the text. While the authors did not meet all of their aspirational metrics, they consistently followed their intersectional feminist values and successfully shared a diverse range of projects highlighting social problems consistently overlooked by dominant groups. The curated examples and the principles they embody are a good starting point for challenging oppression through reimagining how we use data.