On his first day in office, President Biden signed an executive order directing federal agencies and White House offices to examine barriers to racial equity and initiated several efforts to address equity for people of color and underserved communities. The executive order takes an unprecedented step by committing the government to actively pursuing more equitable engagement and outcomes across all agencies and policy areas.
The order not only clarifies the government’s essential role in advancing racial equity—it also denotes the role data play in tracking progress and holding government accountable.
The order calls for the development of an Equitable Data Working Group (EDWG)—an interagency effort reporting to the President through the Domestic Policy Council—to examine the existing federal data infrastructure and provide recommendations for improving it. A large part of this effort will be disaggregating data by race, gender, and other demographic variables so policymakers can better understand policy implications and develop targeted solutions to ensure more equitable outcomes.
Establishing the EDWG is an important step to improving federal data infrastructure. Between our new Racial Equity Analytics Lab, our data science work, and other efforts, Urban Institute researchers have been arguing for the need and benefits of more disaggregated data and better data infrastructure, and we suggest five actions for policymakers’ consideration.
1. Shoring up and modernizing the census
The quality of the decennial census, one of our most powerful sources for disaggregated race and ethnicity data, is at risk because of the COVID-19 pandemic, deferral to extend the census follow-up timeline, and uncertainty in how the new differential privacy measures will affect accuracy.
The census requires new, supplemental infrastructure to address population inequity issues. An Office of Data Equity could help apply an equity lens to collect data from and about diverse, underserved populations, report more accurate statistics on diverse subpopulations, and evaluate and assess the impact of data flaws that prevent equitable measurement of segments of our diverse society. It could also help the EDWG committee understand how these three tangled factors affect the accuracy and reliability of the data and how researchers and agency officials can correct for and interpret disaggregated data responsibly in their analyses.
The census continues to have pressing concerns around fairness and accuracy that should be addressed for the 2020 count, but an Office of Data Equity could ensure efforts led by the US Census Bureau embed an equity lens and improved disaggregation into future censuses and other data collections led or supported by the bureau.
2. Establishing better data sharing
Currently, many barriers exist (PDF) to interagency data sharing. Some of those barriers are statutory and may be difficult to address, but the working group could tackle others. Some government agencies spend months every year negotiating with other agencies for data access, only to have to renegotiate at the beginning of the next fiscal year, wasting significant time and resources.
The Office of Management and Budget could make some data sharing between agencies permissible, uniform, and transparent so agencies could spend their time producing new aggregated statistics instead of negotiating. Congress could allocate more funding to cash-strapped agencies to ensure they have the personnel necessary to share data and the expertise necessary to produce disaggregated statistics across agencies. Although the Foundations for Evidence-Based Policymaking Act represents a strong building block toward the executive order’s goals, it was not initially accompanied by the funding sufficient to achieve them.
With reduced barriers, agencies that don’t directly collect data on race or ethnicity could more easily link data with those that do. The College Scorecard, housed at the US Department of Education, already draws upon data within its own agency, the Internal Revenue Service, and the Census Bureau, but those partnerships could be expanded to better incorporate data on race and ethnicity.
3. Making the communities of focus part of the process
Including the communities being studied in every step of the research process helps ensure accurate findings and effective policy solutions, and, equally important, it empowers people and fosters civic engagement. Communities of color have expressed concerns about being overstudied (PDF), so the EDWG could ensure any effort to expand research among this group addresses these concerns and incorporates them as coinvestigators in the process.
Many federal agencies are not well equipped to conduct robust community engagement, so they can draw upon insights and processes already developed by local governments and other nonprofit researchers. Establishing partnerships with community leaders and organizations who can facilitate meaningful inclusion of community voices will be critical to the success of this work. Urban has learned firsthand how this methodology not only yields insights institutional researchers would otherwise miss but also increases community trust and mechanisms for accountability.
4. Examining new and creative methods to address data limitations
The working group could consider using new and innovative data sources and methods, building on existing efforts within and outside the federal government. For example, supplemental private-sector data (PDF) could provide more granular, real-time data to public statistics—a tactic the Census Bureau and other agencies are already employing. On the methods side, some agencies are imputing race and ethnicity on existing administrative data to improve their ability to disaggregate data for decisionmaking, efforts that could be expanded.
Here at Urban, we are working with agencies such as the Internal Revenue Service Statistics on Income Division to generate privacy-preserving synthetic data and validation server technology. Our evidence suggests this could significantly increase data access while respecting privacy concerns.
5. Taking privacy seriously
Ongoing debates over protecting personal data versus supporting the common good often overlook the groups most vulnerable to exposure of their private data: people of color and people with low incomes. If federal policymakers want to ensure equitable privacy protections and societal benefits, they should acknowledge and account for these unequal risks before relaxing data protections. Additionally, any attempts to tighten privacy restrictions should acknowledge the limitations of data accuracy for small disaggregated populations, whose numbers may need to be significantly distorted to preserve their privacy.
Additional federal funding for the statistical agencies is crucial but will take time to materialize. We are optimistic that foundations, local governments, the private sector, and federal research funding can step in to fill this need. These entities will be crucial to furthering progress on the executive order on a reasonable timeline. Engaging with agencies would also give these entities the opportunity to learn alongside the federal effort—insights that would benefit both public and private data infrastructure.
Getting this right is critical. The risks of inaccurate, incomplete, or bias-laden data could cause irreparable harm, but the potential gains could foster a new era for policymaking that finally leads to equity for all.