4 questions to ask before sharing constituent data

Americans are increasingly growing wary about how the government handles their personal data. In a time when virtually everything is online and seemingly everybody wants their information, people are becoming less certain about the ways in which it is used and shared. The number of Americans who say they are worried about government use of personal data has increased from 64% in 2019 to 71% today, according to a recent Pew Research Center survey.

These concerns come as state and local governments are offering more and more digital services and, as a result, collecting more and more data. Applications that once required basic information like name, address and Social Security number now tap into tax databases, school records, credit histories and sometimes even require biometric authentication. The concerns also come amid government efforts to break down data silos, streamline the data collection process and serve constituents more effectively and efficiently.

Without a data sharing and integration governance plan and process, agencies will struggle to make informed decisions, further eroding the community’s trust in government, said Amy Hawn Nelson, director of training and technical assistance at the Actionable Intelligence for Social Policy, or AISP, an initiative from the University of Pennsylvania that advocates for the responsible use of data in state and local government. 

To help officials enhance their data-driven decision-making, a new report from the group offers guidance on what data agencies are legally allowed to access and how they can weigh the risks and benefits of data sharing and integration. 

Before starting any analysis that depends on data sharing or data integration, the researchers advise agencies to ask themselves four questions: Is this legal? Is it ethical? Is it a good idea? Who decides whether the data use is appropriate?

When it comes to legality, agencies should check with their legal team to determine whether they have the authority to access and share certain data, such as health information. They need to strike the right balance between informed policymaking and federal, state or local privacy and confidentiality laws. 

Sometimes legal authority comes from executive orders mandating data sharing as a policy priority, the report said. For instance, Virginia Gov. Glenn Youngkin issued an order in May requiring executive agencies to participate in a data-sharing project aimed at addressing the state’s fentanyl crisis.

Managers should consider who has lawful access to an agency’s data because data restrictions may vary depending on who will access it. For instance, an agency that shares data with a research institution may require the data to be de-identified and aggregated to ensure individuals’ privacy is maintained, according to the report. But data released to social services for case management is likely not required to be anonymized. 

Government should also ask about the ethics of data sharing. Agencies must demonstrate to residents and stakeholders that they can responsibly manage data, and that means, in part, ensuring it is used in a fair and accurate way, said Sharon Zanti, an AISP fellow and report co-author. Using low-quality data or a small sample size can give policymakers an inadequate assessment of local issues and misguide their policies, programs and funding efforts. And for communities that have been historically underserved or underinvested, poor data-driven decision-making can worsen inequity. 

Another aspect of ethical data governance involves responsible use of administrative data. That information can inform social benefits programs, for instance, but reusing that same data or integrating it with other information for research purposes can “compromise the validity and reliability of the data,” Nelson said. 

Administrative data is collected for operational purposes; it “wasn’t designed to answer a specific question,” she said. That means sharing potentially inadequate administrative data could do more harm than good. 

One example of valid data sharing was North Carolina’s Pandemic Electronic Benefits Transfer, or P-EBT, program. It was designed to give residents food assistance during the pandemic so families with children who qualified for the free and reduced-price meals at school prior to the pandemic could buy groceries with an EBT card. To do that, the state’s health and human services department and education department linked school records showing which students were enrolled in assistive meal programs with food and nutrition service benefit records to determine which families qualify for P-EBT cards. The program helped families buy food during the pandemic, reducing the risk of children losing out on meals because of school closures.

While there was a risk in sharing student-level data, Shanti said, it was far outweighed by the benefit of supporting families during a situation as dire as the pandemic. 

Officials should also ask themselves whether a data sharing project is a good idea. With the limited funds, staff and resources that many state and local governments already face, a data sharing or integration project may not be a worthwhile investment, the researchers wrote. 

For instance, as more states look to regulate social media access and content, Nelson said some are considering linking student records with their social media accounts to identify harmful or threatening behavior among students. That raises a number of concerns, she said. 

Aside from the privacy aspect, any data insights from an integration like that may not even yield insights school administrators and other officials can act on. For instance, if an analysis flags 400 inappropriate social media posts in a school with a low teacher- or counselor-to-student ratio, there’s not much that the administration can do to address the issue, Nelson said. Instead of pouring financial and staff resources into the data side of issues, sometimes a better use of agencies’ time and energy is to put those resources toward appropriate staffing or intervention services.  

Finally, governments should have legal teams to determine a proposed project’s legality, skilled staff to manage the analysis processes and data governance committees that include data stewards, data owners and front-line program staff to determine if it is ethical and a good idea, the researchers wrote. 

Each group is crucial to the rollout of data sharing projects: The legal team can help agencies decide what data they can use and how, data staff can carry out the daily operations of data management, while a data governance group can review and oversee data sharing projects.