BITSS and Berkeley Opportunity Lab
When a representative from the California Judicial Council attended the BITSS Annual Meeting last December, she asked the group an interesting question – if we want to be more transparent and share data, how can we do so in a responsible way that limits risk to those who share and use the data, while also maximizing its usability and impact? This question was asked in the context of the California Attorney General’s new Department of Justice initiative – OpenJustice – but also broader efforts to open up criminal justice data for many of the same reasons we advocate for transparency. For this group, and many researchers who work with this type of data, making communities safer means measuring our effectiveness in the criminal justice system with data and metrics. Sharing criminal justice data can help us understand how we are doing, hold ourselves accountable, and inform policy to improve public safety. Given tragic recent events, this really is a timely discussion.
To start answering this question, BITSS partnered with the Judicial Council, the Department of Justice and the recently established Berkeley Opportunity Lab (O-Lab) to develop and deliver a Data Sharing Forum. We held the Forum on June 8 to take advantage of some of the faculty in town for the BITSS Summer Institute. During the Forum, we engaged with 25 researchers, analysts, and key stakeholders from California State Justice Agencies, including the: Judicial Council, Department of Justice, Board of State and Community Corrections, and Department of Corrections and Rehabilitations. These agencies often have vast datasets with a wealth of information, but these resources lack organization and accessibility. The Forum was intended to start the conversation around best practices for sharing and using this data.
Discussions were led by O-Lab faculty members Steven Raphael and Justin McCrary who shared optimal practices for linking internal data systems and contextual data from other agencies, while discussing the challenges in accessing large datasets. Simson Garfinkel, of the National Institute for Standards and Technology, provided insights into challenges with data de-identification and the importance of thinking through how data may link to other data, thereby increasing risk of loss of confidentiality or other protection.
The informal, short presentations with open dialogue ensured that the content was engaging. According to participants, the information on data sharing and ways to promote transparency and reproducibility in data collection efforts was both relevant and informative, and gave a more holistic view of how to share data in a way that protects privacy. In addition to the training, the forum provided a valuable avenue for interacting with the academic community and learning about data collection and research across the criminal justice field and other related social services. The forum was particularly helpful in fostering an environment that encouraged participants to network with staff from other state agencies.
As the initial question raised signals, it is important for open data and transparency efforts to remember the need to protect confidentiality and privacy to the extent possible. Properly de-identifying data may be a vital step in secure data management. On the other hand, de-identification efforts may distort the data, limiting its usability and therefore its potential to drive real change and improvements in the system. Edward Miguel, the Faculty and Scientific Director for both BITSS and O-Lab, believes the solution is somewhere in-between. Creating public memorandums of understanding (MOU) from the beginning may make the process of releasing and accessing datasets more transparent and inherently simplify the process, while maintaining a level of restriction that requires clearance and reduces the risk of de-identification. It was an important first discussion, but left as many questions as answers. The good news is, there is a strong group of dedicated professionals who are willing to keep asking the tough questions in the hopes that movements like OpenJustice can be successful in their mission to improve learning and accountability in the criminal justice system.