January 30, 2020 | By Natalie Evans Harris

Society uses data for just about everything. Every day we hear about different ways organizations collect data about us for marketing purposes, insurance decisions, and improved delivery of social services including housing, education, and mental health. We also hear about data being used to deny home loans, set outsized bail, and often exacerbate existing biases within our social systems. It’s no question that, good or bad, data drives decisions by large organizations, small nonprofits, government officials, and everyone in-between. 

Through this expansive approach to using data, many government agencies are also experiencing the pains of governing how that data is shared, resulting in practices that are unsustainable, ineffective, and not forward-thinking. There is a fundamental need to evolve these practices into a governance approach that balances the need to protect people’s data with the need to uncover opportunities to better serve communities through data. 

As we head into 2020, it is already clear that a shift in how we make decisions with data is underway. The Federal Data Strategy Action Plan makes data governance processes a top priority. The California Consumer Privacy Act went into effect Jan 1, 2020, requiring entities to fundamentally change how they handle data with data governance standards as a main focus. And Congress continues to work on national privacy legislation that influences data governance standards, including nearly 10 bills under consideration for regulating the collection and use of personal data, individual consent, and even defining what constitutes personal data. Simply put, new data governance strategies are being developed and policy improvements are driving this conversation.

At the Beeck Center, I spent the past year leading a research effort to gather best practices and lessons learned on data sharing. In partnership with The Rockefeller Foundation, I hosted collaborative discussions with multiple stakeholders and practitioners, conducted independent research with dozens of organizations, companies, and government teams, and drew on my nearly 20 years leading data practices. We are excited today to launch a new resource based on that research: Sharing Data for Social Impact: a Guidebook to Establishing Responsible Governance Practices.

cover of Sharing Data for Social Impact report

Thankfully, we aren’t starting from scratch as many government agencies have well-established use cases for sharing data in pursuit of improved social service delivery in areas such as K–12 education, public transportation, and healthcare. For example: 

  • In 2017, Florida’s Broward County saw the number of children moving into their Kinship Care Program — where kids live with grandparents or other non-parental relatives — had increased significantly. To improve services for these kinship providers, the county took data from a variety of sources — the public schools, Department of Children and Families, Department of Juvenile Justice, and others — and worked with the community to analyze the information, then put it into practice. This engagement of stakeholders and participants created community feedback loops on shared data between families and agencies, strengthening family outcomes through a decision making process that emphasizes collaboration, transparency, and shared interest in positive results.  
  • On the other side of the country, Los Angeles County wanted to study the effectiveness of a number of social service programs for people experiencing homelessness. While the data was available, it was trapped in individual agencies, making it difficult to understand if an individual used services outside a single agency. As a way to combat this siloing of data and link social service organizations, researchers created an integrated data system. This system, launched in 2015, “provided agencies with a comprehensive picture of the [homeless population] and their needs and helped these agencies consider different models for service delivery… The project was relatively easy to execute with the [integrated] data, but would have been impossible without it.” By not only linking data from several agencies but also outlining data-use practices and procedures for each agency, Los Angeles County is ensuring the delivery of vital services to a vulnerable population.

Another recent trend is private companies, governments, and nonprofits forming cross-sector data-sharing collaboratives in support of the social good, but these can be hampered by organizational rules restricting the availability of data to external actors. In an environment where data are only used for making funding decisions or to narrowly evaluate programs, this model can work well. But in pursuit of innovation or improved social service delivery, this model is less encouraging. I discussed the need to shift to a more equitable and sustainable governance process in a previous blog. 

As the amount of data and methods for collecting it increase, so have opportunities for drawing insights about society. Bringing together diverse data sources is crucial to ensuring that insights promote equitable growth. And as promising as data sharing is for improving societal outcomes, the analysis of integrated data (especially through predictive analytics) can easily repeat inequities learned from past service delivery. Contextualizing data analysis with methods used by social sciences and ongoing community engagement is crucial to ensuring data analytics do not replicate or worsen inequitable outcomes.

Through our research, we found three key phases critical to establishing equitable and sustainable data sharing governance practices for social impact. Our guidebook helps individuals and teams seeking a primer to better understand the key legal, technical, and cultural components to data sharing governance. The guidebook provides a holistic process detailing each phase and extensive resources to aid stakeholders.

Stages of Data Sharing Governance

Build the collective

Get everyone on board. Start with the policy problem. Identify stakeholders. Take stock of capacity, motivations, barriers, and potential data solutions. Demonstrate value and reduce uncertainty to generate buy-in. Establish a minimum viable coalition and enshrine your shared vision in a charter. 

Data Sharing - Build the Collective graphic

Define the operations

Get everyone in line. Create the governance framework tied to the charter. Design a feedback loop and integrate it into the governance framework. Formalize those two elements into a data-sharing agreement. Launch the operations of the minimum viable coalition.

Drive impact

Get everyone to improve and share. Re-evaluate assumptions, approach, and metrics. Survey impacted communities and stakeholders. Use feedback loops to enact iterative improvements to the governance structure. Repeat this process until feedback becomes minimal. Scale up. 

We recognize that many different actors will be involved in this process and that each one faces unique challenges, goals, motivations, and opportunities. This guidebook is for people looking to leverage data and data sharing towards evidence-based policy making. Moreover, it can be used by policy makers and organizations interested in giving agency to individuals over their data along with organizations interested in ethically and responsibly sharing data. 

Data Sharing Driving Impact graphic

While data use can sometimes lead to harmful outcomes, what will never change is that data can, should, and will be used for good. Because data plays such a large role in society, it is imperative that organizations and governments use and share it responsibly. While there are resources out there to do this, our Guidebook delivers the perfect framework with resources, advice and practical examples for tackling the complexities of data sharing going forward. We look forward to supporting organizations as they activate the lessons we captured and will continue recording and sharing good practices through that process.

Natalie Evans Harris is a Beeck Center Fellow and a sought-after thought leader on the ethical and responsible use of data after nearly 20 years advancing the public sector’s strategic use of data. Follow her on Twitter @QuietStormNat

 

November 20, 2019 | By Robert Roussel 

A great idea is a terrible thing to waste, and people do it all the time. 

In academia and policy institutions, research is often regarded as a key analytical asset. However, research alone has limited utility. Research needs to be resourced with practices and structures in order for that research to be activated, iterated upon, and deployed. Failing to do so shows a fundamental misunderstanding of the purpose of research—a failure to understand that research is a means and not an end. This approach begins with how we train students and I worry that, all too often, academic training has entrenched problematic approaches to teaching aspiring professionals what the value of research actually is.

I went to graduate school to learn how to evaluate and implement public policies, which, at the end of the day, was about translating statistics and analysis into writing. My peers and I considered writing both our biggest pain point and our most powerful asset. I was reminded of this recently as I sorted through old papers and memos to pick a writing sample when applying for a job that would, in part, pay me to write. I started remembering all of the topics and arguments I had so eloquently and passionately inked onto a digital page. Undeniably, my graduate program taught us how to write and think logically and persuasively but as I perused my hard drive for old papers, I could not shake the feeling that my new ‘ideal’ job at a think tank — the one I’ve been wanting for years — would relegate my writing to a fate similar to my academic exercises: gathering dust, and longing for eyeballs.

Photo by Martin Adams on Unsplash

To businesses and consultancies, the idea that research without applications is useless is obvious. In academia, this culture is not always the norm. At Georgetown, I had the privilege of practicing valuable quantitative and analytical skills through thoughtful exercises led by experts in their fields. Even at this institution, however, I couldn’t help feeling that some professors often seemed quite willing to ignore the need for a more applied approach to teaching, tacitly implying that an efficient division of labor within the policy-making ecosystem would translate our clever and thoughtful words into action.

Researchers in academic settings need to be rewarded for being consulted by policy makers, not cited by fellow researchers.

Researchers in academic settings need to be rewarded for being consulted by policy makers, not cited by fellow researchers. Even when professors were practitioners themselves, their in-class behaviors often failed to reflect that fact. It seems more likely than not that this approach is not just borne out of convenience, but a culture rooted in academic tradition. Though universities like to talk big about their cutting-edge research, often their approaches to pedagogy seem remarkably risk-averse. Academic culture is very slow to change, and the incentives for taking such a large departure are just not there. This culture shift will likely need to occur from the bottom-up, as students demand to be more involved in the activation of research—meaning that any education should be as much about writing words as it is about resourcing those words in clever intentional ways that help, rather than hope, words to translate to actions and actions translate to impact.

How to ‘Activate’ Research: A Brief Case Study

During my time at the Beeck Center, I noticed that the leadership was placing a lot of emphasis on activities other than research, such as convenings, workshops, interviews, or meetings. One of my projects was to design a framework and resource repository for establishing responsible data-sharing practices for social impact. Many third-party organizations have emerged to help facilitate data-sharing for social impact but the resources for sharers and the bandwidth of these facilitators are limited. From the start, it was clear this research was just a launching point and not the end product. We were also going to build a community around this research product that would help activate it and keep it alive, constantly open to change as new best practices and case studies emerge. To me, this approach — one that is both highly collaborative and constantly seeking input — is exactly the way we should be approaching public problems.

One of the most often given pieces of advice is that people should spend 99% of their time understanding the problem and that, if working this way, finding a solution should be so obvious that it takes just the remaining 1% to solve. If that is true, we need to be extra sure that we are solving the right problem or else our deployed solution might not be all that useful. The cleverness of keeping the Beeck Center’s data-sharing guidebook ‘alive’ was that it made the guidebook both a solution and a problem exploration process at the same time. Certainly, it aimed to create a solution to a problem but its openness serves as a way to constantly re-evaluate this solution. That malleability and that openness to collaborate is what will activate the research in the guidebook.

This is a smart approach to making sure that research is activated, but it might not go far enough. With the resources and bandwidth of stakeholders being limited, there are clearly gaps in capacity that limit the scalability of this project. While being careful of the hubris of applying a ‘there’s an app for that’ mentality to complex social problems, I proposed a solution that can help activate the guidebook and resource guide. This solution was borne out of a seemingly impossible trade-off between brevity and usefulness. A shorter guidebook would have recommendations that are more digestible but would have to be more generic, and by extension, not useful beyond a surface level. 

My answer, which is an answer I urge researchers facing problems of activation to consider, is a customizable tool (a ‘wizard’, if you remember Windows ’98) that creates unique guidebooks and resource-repositories for each ‘bin’ of users, reflecting the variety of resources, motivations, and barriers or different stakeholders. When it comes to translating research into action, this approach would significantly help constrained organizations that may not have the resources to discover new approaches wade through the literature and see how it might apply to them.

Advice for Policy Students and Researchers

Seen in the most generous of lights, writing academic papers and memos is training for conducting professional research in the real world. I fear, however, that many students will trip on these bad habits as they enter the professional research world—and that those worlds are comprised of ex-students with similar tendencies. With so many vital issues facing the social sector, we need to be sure that our research efforts build out our ability to generate actionable recommendations and tangible impact. A relevant internship or part-time job while in school could be one possible step forward here, but many students complain that their time is spent on passive class assignments and papers that remain all too often unread and unused. Opportunities that give students a real taste of what it is like to see research applied are lacking — and this is a role that academia needs to fill. In my opinion, applied graduate programs should be thoroughly experiential, matching students with real clients in real teams to solve real problems. Across the country, universities are experimenting with this model, but this new approach is a heavy lift and would require a major revamping of the tenure model — an unlikely proposition in many settings.

Unlike coursework, research doesn’t end when a paper is handed to a superior. If you believe that, by handing someone a memo, you have just handed that person everything they need to know for them to get the job done, you are likely mistaken. You dove deep and you need to be intimately involved in applying that research. The term policy maker is a catch-all term and its vagueness makes it rather difficult to understand when the research stops and the policy creation begins. I urge policy students to reject the suspicions borne out of the structure of their academic program that see a clear demarcation between these two fields. Only when we are deeply involved in a project from ‘start’ to ‘finish’ can we effectively suggest action that is researchable and create research that is actionable.

Don’t let your great ideas gather dust in the cloud; have them gather stardust.

 

Robert Roussel was a student analyst at the Beeck Center in Summer 2019. He is a 2019 graduate of the Georgetown University McCourt School for Public Policy and is currently working at Accenture Federal Services as a tech analyst.