5 Q’s for Stefan Heeke, Executive Director of SumAll.org
The Center for Data Innovation spoke with Stefan Heeke, the executive director of SumAll.org, a non-profit organization focused on civic data analytics. Heeke discussed the opportunities for organizations to use data to drive meaningful social change.
Daniel Castro: Please tell me about SumAll.org. What does your organization do?
Stefan Heeke: SumAll.org harnesses big data to create tangible, real-world solutions to pressing social problems. Through civic analytics, impact measurement and data-driven intervention strategies, SumAll.org partners with the tech sector, nonprofits and philanthropy, defining new approaches to support vulnerable populations. SumAll.org then creates scalable implementations that can be applied globally, extending reach and creating measurable social impact.
Castro: You’ve worked on a series of fascinating projects: tracking homelessness in New York City; analyzing student performance in Haiti; tracking deaths in the Syrian conflict. What projects have been the most impactful?
Heeke: We see most impact when we have a clear scope, case level data, and partners who are able to change their processes to implement improved, data-driven interventions. This has been the case with CAMBA, a homelessness prevention non-profit in New York City which truly embraced the power of homelessness predictions for its prevention services. We all worked together to insert prediction knowledge into the process of social workers and could get immediate impact by targeting families most at risk of becoming homeless. Results suggest that we were able to help 50 percent more qualified families receive prevention services relative to traditional outreach methods, which translated into 70 families within 3 month in Community District 303 (Bed-Stuy).
The impact of our investigative analytics work is less measurable, of course. However, we see potential long term value in using statistics to assign responsibility for war crimes or reveal institutional bias, e.g., in prostitution arrests in New York courts which show inconsistent application of the law and a bias against women in sentencing. The measurability, or “lack of,” is probably comparable with investigative journalism where impact events may occur much later.
Castro: There are always so many civic problems to choose from. What types of civic problems are best suited for addressing with data?
Heeke: We focus on helping partners and change makers able to drive social change, and we focus where we feel we can make a transformative contribution by using data in new ways. A lot of work goes into scoping projects properly and making sure the data is able to support good results. We are fairly open to different issues, human trafficking, social justice, education, and poverty being recurring themes.
Castro: Making an impact is hard. What kinds of unique challenges do data scientists face when trying to use data to create social change?
Heeke: We realized that data analytics alone is not enough. We are looking to be involved in intervention design and push for implementations in pilot projects to get a sense for results. Rapid prototyping is very much favored by our funders who are tech entrepreneurs.
On the other hand we had to learn that impact is not always in our hands, and we’ve realized that impact takes time and cannot always be attributed to a single project. Being patient and following through after completion is a big part for us. Often, the advocacy and communication of results requires more resources than the project work itself.
Castro: Many government agencies already evaluate the effectiveness and side effects of programs, but often do so without using the latest analytical techniques or technologies. As agencies begin leveraging “big data” analytics, what lessons or tips would you offer them from your own experiences?
Heeke: Our big takeaway of last year is that organizations and stakeholders need to be willing and able to change themselves in order to drive social change with data, otherwise analytics work cannot really be transformative—or worse, data will be selectively used to justify the status quo.