Ajah goes to Coleridge Summit: we found our people!
Photo by Nathan Dumlao from Unsplash
As you might recall, the Ajah contingency - Michael Lenczner, Elle Gemma Gruver, and David Goodman -were eager to attend the Coleridge Convening Cross-Agency Collaboration for Evidence-Building at the end of March. The Coleridge Initiative focuses on building collaboratives for data sharing through initiatives like the Administrative Research Data Facility (ADRF) and Applied Data Analytics Training. We got to nerd out on administrative data at the conference, and spoiler alert, we loved it. It gave us a lot to reflect on and may have even renewed our hope in this work moving forward (maybe a little!)
It is a foregone conclusion that sharing administrative data has to happen at multiple levels to be effective. This presents challenges as different states - and often the agencies within them - are highly autonomous, with differences in staff and technical capacities and comfort around data sharing. Coleridge offered a unique space for states, agencies, and partner organizations to convene and collaborate. The aim was to share learnings and build on one another’s work - helping organizations and initiatives to stop reinventing the wheel. Below are some of the takeaways we’ve been thinking about since leaving the convening:
This work is about relationships and trust first - and technology second (a far second).
Relationships are critical to this work because data sharing is difficult, risky and even adversarial by nature, due to all of the variations and differences found in different data sets. The only way that we are able to overcome this barrier is developing trust. Centering trust before technology provides an opportunity to collaborate instead of duplicate. Actors can work to advance existing infrastructure instead of all redeveloping the same tools. This sounds easy, but in practice it requires working together across agencies in spaces that don’t foster cross-agency collaboration (how many times have you heard the word ‘silo’ in reference to the data/communication in your agency?). Building technology and infrastructure are easy, but they can’t and shouldn’t happen until we’ve built trust and found common value.
1. To get buy-in, offer clear use cases. Finding common value propositions will help you find conspirators in the work.
Building collaborations starts with a pitch. Agencies need to demonstrate what the value will be in creating collaborations to data owners and decision makers that they approach. One way to do this is creating “quick wins” to see value more quickly. This entails creating feasible goals at first and starting small. This also means creating conditions for the value of combined data to be greater than individual level data. Defining clear use cases will help to move away from creating more ‘data mausoleums,’ where data is locked away (covered in dust and spiderwebs, probably) and never used.
2. Governance is critical to this work, but how do we facilitate good governance when many states and agencies are still struggling to evolve and collaborate?
Data Governance is key to responsible, sustainable collaboration. You might think of data governance as archaic bylaws and pedantic sub policies, but proper governance can protect data and the use of the data, while building trust and understanding among participating orgs. Effective governance needs to be collaborative, flexible, and sustainable in order to ensure equity while ensuring that it can evolve and expand as needs and preferences change over time. Some examples we like include @KYstats, @Colorado DT, and @Kansas ECIDS data trust. We need to move away from proprietary governance policies that are limited in scope and time, or tied to individual people or isolated projects. We also need to learn to be more flexible and find ways to allow people to participate from where they are at - which might not be quite ready to share data, but still wanting to participate and learn.
3. The reality is that not all States and agencies have the necessary capacities to do this work. So how do we move forward?
We must invest in human capital - yes, even with the AI takeover. Organizations and agencies are reluctant to share data because they can be ashamed of the state of their data and what this means for their programming. Having more people devoted to ensuring data sets are clean, complete, and well managed can help avoid this “skeletons in the data closet” problem.
A caveat: don’t spend time cleaning the data if you haven't thought about who is going to use the data, and whether or not there is capacity to actually do things with the data (or the capacity to manage the operations and governance of data sharing). We have to be judicious in prioritizing resources for data that can actually be used and not just data we wish we could use.
4. States and counterparts in other States need to work together.
Our final point is an echo of what we’ve been saying all along - we are more capable when we work together. Sharing experiences and best practices are key to building state capacities and ensuring that data and tech can be used more efficiently. We shouldn’t be in competition - we should be working together and learning from one another. States and agencies need to collaborate with one another. We’ve been excited to do some work along these lines with the Investment Readiness Program for the Government of Canada. We are involved in work to support Economic and Social Development Canada and Statistics Canada to connect the Investment Readiness Program administrative grants data with the Business Registry to drive insights about the impact of the granting program on grantees. We’ll be presenting more about this work at the ANSER conference in May. We look forward to attending the next iteration of the Coleridge Convening and continuing to advocate around better data sharing work in general.