Using Data Feminism Principles to Create Better Data Infrastructure and Informed Policies

May 15, 2020

In April, Data2X hosted a book talk with Catherine D’Ignazio and Lauren F. Klein, the authors of Data Feminism. We were unable to answer the below questions from the audience during the book talk, so we turned them into a blog. Their answers have been edited for clarity and length. This blog is one of a two-part series; read the first blog here.

Your book outlines seven principles of data feminism. How can we incentivize governments, donors, and other organizations to implement these principles in their own statistics/data analysis?

Education is key. Incorporating principles of data feminism means unlearning and relearning previously held perspectives in order to create better data science, products, and projects. This also means listening to the communities you’re focused on; if you don’t listen to their needs, you miss an opportunity to help them in the way that works best for them and creates lasting impact.

What does the economics of data feminism look like? How can we build the case that it is a valuable investment for building better data infrastructure?

The economics of data feminism is not a one-size-fits-all approach. There are many different economic cases for different situations, but the bottom line is that without resources, you can’t do anything. The institutions that are influencing large-scale data infrastructure often have the resources; it’s just a question of what their priorities are. When investing in a project, if you’re not investing in mitigating the effects of inequalities, then there’s a much higher risk that the data infrastructure will fail to provide accurate answers and effect legitimate change.

For example, in 2017, the Boston public school system wanted to redesign the start times of their schools and bus routes to accommodate those new start times. They used an algorithm in their design and decision process but the algorithm had blind spots and officials didn’t get public buy-in from parents and students. As a result, officials received enormous pushback on the changes they proposed; many parents couldn’t manage this time shift with work schedules and other obligations. What the school system thought was a cost-saving method became a more expensive situation because they had to spend a lot of money mitigating the damage done. If the school system had been more transparent and considered the context around the data they used [Principle #6 of data feminism], the result might have been more successful.

So by investing resources upfront to build data infrastructure that incorporates data feminism principles, you can build data infrastructure that reduces risk and creates more successful outcomes in communities.

Can you speak to the vast amount of data that is currently available (as more governments, companies, etc. are transitioning towards making their data more accessible) but isn’t being used? How do you see the best way to translate the data that already exists into actual policy action?

This is a complex question because there’s so much data out there. However, usually by the time it arrives in a spreadsheet, it has been separated from the context that produced it. That’s a fundamental gap with information infrastructures such as the open data movement. There’s a lot of open data but it doesn’t always translate to actual policy changes because not everyone has the skills to deduce what the data mean. Datasets should always be published with robust metadata (what the data contains, what it doesn’t, and additional analysis and context) and narrative description (such as data user guides to datasets) so that newcomers can understand the large swaths of data.

From a citizen standpoint, the question is: how do we scale up data literacy so people know where to look, how to do data analysis, and how to work with activist organizations to mount a campaign? From a policy standpoint, the question is about how we bring data to policymakers’ purview and inform policy? Data intermediaries (people who translate between the data and people who want to work with it) have a huge responsibility in shaping conversations between these two sides so that data leads to concrete policy action.

What are the implications of not bringing data science and intersectional feminism together?

The reproduction of a status quo that is patriarchal, heteronormative, racist, colonial, and deeply unequal. Work that is produced from unequal conditions will reflect those unequal conditions. On the positive side, the value-add from incorporating data feminism principles yields more informed, more impactful projects.

What are the implications of unequal gender and power relations within international development-focused research organizations who collect and/or analyze data? There’s a lot to think about in the development community around data flows. Specifically, how data is extracted, who it’s extracted by, who the data is about and if they were consulted, and how knowledge transfers (if any) are done. For example, you can have well-resourced Global North countries extracting data from countries in the Global South with purposes of helping but what happens to the local community they worked with and how does the knowledge get transferred back? Who does the work? Who owns the data? Where is the data stored? Who has access to it? These kinds of questions are critical for understanding and upending the power dynamics that perpetuate inequalities.

Watch the book talk.

Catherine D’Ignazio is an Assistant Professor of Urban Science and Planning in the Department of Urban Studies and Planning at MIT.

Lauren F. Klein is an Associate Professor of English and Quantitative Theory and Methods at Emory University.

Stay in touch. Sign up for gender data updates.