What’s really needed for data-driven government
Connecting state and local government leaders
Data held by local government is a potential goldmine of insights into how to improve people’s lives and their communities, a recent report says. But fundamental changes and best practices are needed for agencies to effectively use it.
While poor data quality, the challenges of pulling and integrating data from legacy IT systems and the difficulty of persuading stakeholders to embrace open data can all be stumbling blocks to a data-driven government, there are proven ways local governments can improve data governance.
The Local Government Association and Nesta, both U.K.-based, studied the ways local governments were using data, describing challenges and projecting success factors. The entities worked together during 2016 to review and analyze data practices of local councils to see how local authorities could better use their data.
In their case studies, they found several common issues facing data projects in local governments and described ways to combat these issues. Some of these common issues were:
Lack of data standards (or poor data quality) hindering data integration
In every case study, some or all of the data was unusable. In some instances this was because of data entry errors, mainly because the staff members inputting the data were not the primary data users and did not understand the importance of accurate data. In other cases, data was collected in ways that made it hard to integrate with other information. Data was not routinely checked for quality.
“The case studies confirmed that poor data quality is likely to be a feature of most, if not all, local government data projects,” the researchers said.
Furthermore, data in legacy systems of one agency did not always follow the same data standards as at another agency, making integration difficult. In most cases, researchers said, the problem was due to “how values were represented, data categories and fields, or how the data is organized.” In some cases it was because data standards had not been adopted systematically.
Solution: Researchers recommended finding where problems with data quality originate and putting the onus on the data owners to improve the data quality. Dashboards also can help expose where poor data quality is linked to data input methods. As data quality and access increases, the number of people needed for data extraction and reporting is reduced.
Poor information governance and data sharing
Legal and cultural hurdles regarding information governance are unavoidable, researchers said. These issues should not be seen as an insurmountable barrier, though. In every project studied, team members found ways to legally share information in accordance with current legislation.
Solution: Researchers recommended embedding decisions about sharing data into new systems. Additional best practices include developing information governance protocols for responsible data sharing based on specific-use cases and creating a team manager to drive data integration/use projects. These data-sharing decisions should be based on a balanced risk assessment that “weighs privacy concerns against the risk to the organization or individual of not sharing,” they said. Data should not be collected where there is no apparent, immediate use for it. Instead, all data projects should start with a problem that can be solved by data, then determine which data and analysis might help solve it. The system should record who is using the data to better track if the data is being used inappropriately. There should also be a data inventory that is published as metadata.
Difficulty working with legacy IT systems
Pulling data from legacy IT systems is difficult, particularly when there is poor quality architecture and systems. Data from legacy systems may be in the same format as that in other systems, creating additional integration issues.
Solution: Researchers recommended software to “broker between multiple legacy systems” that can integrate data from several sources into a single data repository that can reduce duplicative efforts, facilitate processes and increase data accuracy. Organizations should also devote resources to work with legacy systems and create a data dashboard to more easily understand issues, improve prioritization, decrease decision-making bottlenecks and expose data quality issues.
Resistance to open data
Oftentimes teams were nervous about potential negative reactions from revealing the poor quality of their data. Additionally, once data was integrated and open, it was sometimes difficult to get some employees to use the data -- usually for cultural reasons.
Solution: Researchers recommended actively engaging potential data users, creating partnerships with interested communities to better solve challenges faced by the local public sector. Publishing full, raw datasets, creating an open-data portal with multiple datasets with a user-friendly interface are also recommended. The data should be machine readable in a standard open format and use an application programming interface wherever possible. Data should be routinely updated, preferably automatically. It should also be analyzed to determine if it is consistent across the enterprise and achieving its purpose of solving the issue for which it was intended. Historic data that is no longer used should be retired.
“The opportunities to get more value from data are increasing all the time,” the report noted. “Most councils are only just starting to get to grips with all the data they have, and all the ways they could use it to make improvements. The data held by the local government sector is a potential goldmine of insights into how to improve people’s lives and make our communities better.”