Data Equity and Democratization: Fixing Data Absenteeism (NORC)

COLLIDE Data + AI Conference 2023Data Equity and DemocratizationWatch the full session · 26 min

At COLLIDE 2023 in Atlanta, Ashani Johnson-Turbes, Vice President and Director of the Center on Equity Research at NORC at the University of Chicago, argued that data democratization should mean more than giving employees access to data. In a 26-minute talk, she proposed an expanded definition that also covers who appears in the data, who gets the skills to analyze it, and who shares in the resulting knowledge and power. She also offered six practical steps organizations can take.

What is data democratization, and what is missing from the usual definition?

Johnson-Turbes began with a common business definition: opening organizational data to as many employees as possible, within reasonable legal and privacy limits. That means sharing data, sharing data skills and sharing data responsibly.

Historically, she said, data was available to only a few, and unequal access to data meant unequal access to opportunities and resources. Democratization breaks down those walls by moving information, knowledge and ultimately power from a few people to many. In the context of inclusive and equitable research, she said, the definition must be expanded to democratize "data, knowledge and power," and to let non-specialists gather and analyze data without much outside help.

What is data absenteeism?

A central idea in the talk was "data absenteeism." Democratizing data, she argued, is not only about sharing it with people. It is also about making sure those people, particularly marginalized and minoritized groups, are present in the data.

If people are absent from a data set, she said, the analysis and conclusions are "less valid" and "less reliable." For her, that made inclusive data collection a matter of rigor, not only fairness.

What is equity science?

Johnson-Turbes, a political scientist by training who described herself as a methodologist rather than a data scientist, said her center launched in May 2022. Its job is to embed inclusivity and equity into social and behavioral science research across NORC.

She defined equity as fairness, justice and impartiality, distinct from equality, and science as the systematic, evidence-based pursuit of knowledge. Equity science combines the two. She presented four principles of inclusive and equitable research, of which she described three: be thoughtful about your own positionality and biases; center culture and community engagement; and share data back, whether with staff and stakeholders or with the very people under inquiry. Her image for embedding equity in the process was that it should be "the eggs in the cake" and not the frosting.

What lesson did a Chicago public housing job teach her?

Johnson-Turbes opened with a story from the mid-1990s, when she worked at the Chicago Housing Authority gathering data from public housing residents about childhood lead poisoning. She said she did not ensure diversity among the people she talked to, and did not share findings back with residents.

Had she done so, she said, she would have learned that residents wanted fans and air conditioning during a deadly heat wave, that mothers cared more about health care than lead abatement, and that older adults wanted social engagement. Her takeaway was a Maya Angelou line: "When you know better, do better."

What can organizations do?

She noted that companies such as Walmart, with a data cafe, and Airbnb, with a data university, pair access with education. She also pointed to the General Social Survey and its GSS Explorer tool, run by NORC, as an example of an effort to make data analyzable by more people.

Her recommended actions were:

• Convene a community advisory board of staff or stakeholders to advise on data access, how data is acquired and how it is analyzed.
• Ensure diversity in the data to avoid data absenteeism.
• Offer training and support, because putting data on a platform is not democratization without the skills to use it.
• Write clear guidelines for handling data responsibly.
• Provide tools for analysis.
• Create space for knowledge sharing, including "data parties," time-limited events of one to two hours where diverse staff or stakeholders collect and analyze data together.

She said she cared about the topic because equity is "not just a noun but a verb" that requires action, and because inclusion makes teams more innovative and findings more trustworthy.

Where to learn more

The AI Enterprise Conference 2027 takes place Tuesday, August 31, 2027, at Pier Sixty in New York City, and COLLIDE 2027 returns to Sandy Springs Performing Arts Center in Atlanta on Wednesday, October 13, 2027. Details are at /events/ai-enterprise-conference/ and /events/collide/. Registration is coming soon.

Frequently asked questions

What is data democratization?

Johnson-Turbes cited the business definition of opening organizational data to as many employees as possible, within legal and privacy limits, and argued it should be expanded to include knowledge and power.

What is data absenteeism?

It is the bias that arises when people, particularly marginalized groups, are absent from a data set, which she said makes analysis and conclusions less valid and reliable.

What is equity science?

She described it as the systematic, evidence-based pursuit of knowledge applied to advancing fairness and justice, including the experiences of historically marginalized populations.

What is a data party?

A time-limited event of one to two hours where diverse staff or stakeholders come together to collect and analyze data, an idea she credited to a colleague, Dr. Shandrea Jones.

This article is based on remarks by Ashani Johnson-Turbes, Vice President & Director of the Center on Equity Research, NORC at the University of Chicago at COLLIDE Data + AI Conference 2023 on October 3, 2023. More coverage: COLLIDE AI Conference.

Filed under: People, Culture & Leadership, AI Governance & Trust