A note to readers: We created this document as we began the writing of this book and included it as part of the manuscript draft that was posted online as part of the open peer review process. We were prompted to write it because of work on a prior project—the Make the Breast Pump Not Suck Hackathon—with equity consultant Jenn Roberts of Versed Education, and because of the values statements (and related statements of principles) published by groups such as the University of Maryland’s African American History, Culture, and Digital Humanities Initiative (AADHum) and the University of Delaware’s Colored Conventions Project.1 From these projects, we saw how statements of shared values can become important orientation points, guiding internal decisions at challenging junctures and making ethical commitments public and transparent. The idea to accompany our values with a set of metrics was also proposed by Jenn Roberts for the breast pump hackathon. We discuss that project, and the uses and limits of metrics for accountability, in chapter 4. The metrics below were calculated two times: first on the basis of the draft posted online, and second on the basis of the copyedited book manuscript. Aside from the addition of the second set of metrics, and a short reflection on our successes and failures, the language of the document remains unchanged from the version posted as part of the manuscript draft.
Feminism has always been multivocal and multiracial, but the movements’ diverse voices have not always been valued equally. The women’s suffrage movement largely excluded Black women and the abolition of slavery from its agenda. In the 1970s, lesbian feminists were called “the purple menace” by straight feminists. But feminism fails altogether if it is only for elite, white, straight, Christian, Anglo women. The work of activists and scholars, particularly Black feminists, over the past forty years insists on a feminism that is intersectional, meaning it looks at issues of social power related not just to gender, but also to race, class, ability, sexuality, immigrant status, and more. It does so, moreover, by looking to collectives as well as individuals, to structural issues as well as specific instances of injustice.
Equity is both an outcome and a process. Future justice must account for an unjust past in which some groups’ knowledges have been valued and others have been subjugated, as Patricia Hill Collins teaches us. In the process of achieving equity, those of us in positions of relative power must learn to listen deeper and listen differently—with the ultimate goal of taking action against the status quo that benefits us at the expense of others. For this reason, we listen and give priority in the text to voices who speak from marginalized perspectives, whether because of their gender, ability, race, class, colonial status, or other aspects of their identity.
As Kimberly Seals Allers, women’s health advocate, says, “Whatever the question, the answer is in the community.” People in a community know its problems intimately, and they know which phenomena go uncounted, underreported, or neglected by institutions in power (or, conversely, who is overly surveilled by institutions in power). They also know what interventions will work to solve those problems. In this book, we try to prioritize voices with closer and more direct experience of issues of injustice over those that study a data injustice from a distance.
We recognize that the transformation of human experience into data often entails a reduction in complexity and context. We further acknowledge that there is a long history of data being “all too often wielded as an instrument of oppression, reinforcing inequality and perpetuating injustice,” as the group Data for Black Lives explains. We keep these inherent constraints in mind as we write, attempting to introduce context and complexity whenever possible, and acknowledge the limits of the methods we discuss, as well as their strengths.
Acknowledging that our knowledge is shaped by our own perspectives and limitations (see more in the About Us section ahead), we strive to be reflexive, transparent, and accountable for our work. We are on a journey toward justice, and that inevitably involves making mistakes. We are grateful to those who have shown us generosity in letting us learn up to this point. And we respectfully say to our future teachers that you will find in us open listeners: we recognize direct and critical words as a generous offer and a vote of confidence in our ability to hear and be transformed by you.
To that end, we have an evolving table of explicit metrics (table A.1) that will guide us in auditing our citations and the examples that we elevate in the book. We note, here, that our foregrounding of race and racism reflects our location in the United States, where the most entrenched issues of inequality and injustice have racism at their source.
Feminist standpoint theory recognizes the value of situated knowledge—acknowledging the perspectives and experiences of the knower and how those have shaped the knowledge they produce. Accordingly, we situate ourselves and the learning contexts in which we work.
Catherine D’Ignazio is an assistant professor at Massachusetts Institute of Technology, a private research university in Cambridge, MA. Before moving to MIT, she worked at Emerson College, a private college in Boston focused on communications and the arts. From a middle-class, Italian-American and Scotch-Irish background, she grew up primarily in the American South, with some formative years spent in Latin America and Europe. She is a mother, an experience that has sharpened her understanding of how women’s bodies are stigmatized and underserved by mainstream institutions. Working mainly in urban New England, she experiences significant privilege from her whiteness, ability, institutional affiliation, and education, among other things, and experiences some oppression based on her gender. With decades of professional work in software programming, art/design, and digital media education, she comes to data feminism based on a commitment to democratize information and include more people and professions in contemporary conversations about data and power.
Lauren F. Klein is an associate professor at Emory University, a private university in Atlanta, Georgia, in the Southern United States. Before moving to Emory, she worked at Georgia Tech, a large public research university also in Atlanta. From a middle-class New York Jewish family, she grew up in suburban New Jersey and lived in New York for much of her adult life, with some time spent in Boston. Like D’Ignazio, she is also a mother. Working in the US South, she experiences significant privilege from her whiteness, ability, education, and institutional affiliation, among other things, and experiences some oppression based on her gender. She worked in web development before becoming an academic and comes to data feminism through her desire to convert theory into practice and to create more opportunities for humanities research (and researchers) to enter into conversation with communities, activists, organizers, and others working toward justice.
Table 1: Aspirational, draft, and final metrics and the structural problems they address
Aspirational metrics to live our values for this book
Draft metrics (open peer review)
Final metrics (copyedited manuscript)
• 75 percent of citations of feminist scholarship from people of color
• 75 percent of examples of feminist data projects discussed led by people of color
Scholarship: 36 percent from people of color
Projects: 49 percent led by people of color
Scholarship: 32 percent from people of color
Projects: 42 percent led by people of color
• 75 percent of all citations and examples from women and nonbinary people
67 percent of citations and examples from women and nonbinary people
62 percent of citations and examples from women and nonbinary people
• Center trans perspectives in discussions of the gender binary
• Use transinclusive language throughout the book
• Example or theorist in every chapter from a transgender perspective
Three of ten chapters feature transgender example and/or theorist
Nine of nine chapters feature transgender example and/or theorist
• Resist assumptions about family structure and gender roles
• Example or theorist in every chapter that illustrates the power of communal (vs. family) support networks
Ten of ten chapters feature communal example and/or theorist
Nine of nine of ten chapters feature communal example and/or theorist
• Challenge the dominance of visualization in the presentation of data
• Example or theorist in every chapter that employs nonvisual methods of presenting data
Nine of ten chapters feature nonvisual example and/or theorist
TK of ten chapters feature nonvisual example and/or theorist
• 30 percent of projects discussed come from the Global South
• Example or theorist in every chapter about Indigenous knowledges and/or activism
Projects: 8.5 percent from the Global South
Five of ten chapters feature indigenous example and/or theorist
Projects: 7 percent from the Global South
Seven of nine chapters feature indigenous example and/or theorist
• Acknowledge that data science, as a field, is premised on economic, educational, and technological privilege
• 50 percent of feminist projects discussed come from outside the academy
• Example or theorist in every chapter that demonstrates how the ideas can be applied without expensive technology and/or formal training
Projects: 88 percent from outside academy
Ten of ten chapters feature nonacademic example and/or theorist
Projects: 78 percent from outside academy
Nine of nine chapters feature nonacademic example and/or theorist
• 50 percent of feminist projects discussed feature and quote people directly impacted by an issue (vs. those who study or report on the phenomena from a distance)
Projects: 49 percent feature people directly impacted
Projects: 34 percent feature people directly impacted
There are interesting shifts to note between the metrics of the first draft and those of the final version. Some measures landed closer toward our goals in the final version. For example, we were able to meet our goal of including a trans voice in every chapter, and we included more Indigenous perspectives in the final version than in the first draft. However, we were not able to achieve all of our aspirational metrics in the final draft, and in fact, the proportional representation of women, nonbinary people, and people of color decreased from the first draft to the copyedited manuscript. The final manuscript also had proportionally fewer projects from the Global South, more projects from the academic realm, and less work from people who were directly impacted by each issue.
What explains this outcome that runs counter to our values and stated goals? In our view, there are two explanations. The first has to do with process: we did not keep precise tabs on our citational metrics as we were revising. We hadn’t wanted to give the aspirational metrics so much presence in our revision process that we would choose people and projects simply because of their gender, race, or other markers of identity. That seemed tokenizing and like “gaming the system” we had created. We did keep the overall gaps between our draft metrics and our aspirational metrics in mind, and attempted to close those gaps as we introduced new examples and removed old ones And we did in fact introduce more overall references to, for example, scholarship and projects led by people of color (243 vs. 92), and scholarship and projects based in the Global South (41 vs. 20). But we also introduced more references to scholarship and projects led by white people (475 vs. 136), as well as more scholarship and projects based in the Global North (933 vs. 278). In retrospect, since we didn’t calculate the second set of metrics until the end of the revision process, we were not aware of the changing proportions of projects and citations, and therefore had no way of rebalancing them after the fact.
An additional pass through the manuscript with this rebalancing in mind would have addressed the issue in the book. Indeed, we wish we had planned for this final pass in our revision process. But the root cause of the issue would have remained unresolved. Put simply: this cause is ourselves and our positions as scholars in the world of higher education, a world dominated by white people and especially white men. To some readers, this answer may seem obvious, but our path to it is worth taking the time to explain.
Many people who participated in our peer review process (both online and anonymously) noted that we should back up our assertions with citations. We agreed with this point, and felt that additional citations would both credit prior work and add legitimacy to our claims. Between the draft and the final version, we added many endnotes and additional references. As a result, the final version of this book is significantly more scholarly than the draft we posted online. But when looking at the history of engagement with a particular idea, or when asking ourselves which notable person in a particular field we should name, we thought less about our values for the book and more about what we already knew about those areas. In so doing, we inadvertently reproduced the biases of academia—ironically, through a mechanism very similar to the privilege hazard we name in the book.
We thought about not including the endnotes in our final accounting; that might have yielded “better” metrics. But that would have misrepresented the contents of the book—and, more importantly, the extent of the work that remains to be done. Instead, our final metrics offer a quantified reminder that “legitimate knowledge” has a race and a gender, as well as a class and a geographic location. These characteristics are inherited from the matrix of domination, and sustained by the matrix of domination as well. Although we challenge that fact in our writing, we readily admit that we fell back on learned habits in our citational practices, particularly when we felt that our scholarly credibility was on the line.
Our takeaway from this process is something we already knew but learned once again: values are not enough. We have to put those values into action and hold ourselves accountable time and time again. This constant emphasis on accountability is not easy, and it is not always successful (case in point). It also takes time. Our final metrics are uncomfortable but in some ways constructive: they serve as evidence of the distance between our ideals and our actions, they help us locate the help we need to bridge those gaps, and they help us persist.