As the world struggles to achieve the UN’s Sustainable Development Goals (SDGs), the need for reliable data to track our progress is more important than ever. Government, civil society, and private sector organizations all play a role in producing, sharing, and using this data, but their information-gathering and -analysis efforts have been able to shed light on only 68 percent of the SDG indicators so far, according to a 2019 UN study.
To help fill the gap, the data science for social good (DSSG) movement has for years been making datasets about important social issues—such as health care infrastructure, school enrollment, air quality, and business registrations—available to trusted organizations or the public. Large tech companies such as Facebook, Google, Amazon, and others have recently begun to embrace the DSSG movement. Spurred on by advances in the field, the Development Data Partnership, the World Economic Forum’s 2030Vision consortium, and Data Collaboratives, they’re offering information about social media users’ mobility during COVID-19, cloud computing infrastructure to help nonprofits analyze large datasets, and other important tools and services.
But sharing data resources doesn’t mean they’ll be used effectively, if at all, to advance social impact. High-impact results require recipients of data assistance to inhabit a robust, holistic data ecosystem that includes assets like policies for safely handling data and the skills to analyze it. As tech firms become increasingly involved with using data and data science to help achieve the SDGs, it’s important that they understand the possibilities and limitations of the nonprofits and other civil society organizations they’re working with. Without a firm grasp on the data ecosystems of their partners, all the technical wizardry in the world may be for naught.
Companies must ask questions such as: What incentives or disincentives are in place for nonprofits to experiment with data science in their work? What gaps remain between what nonprofits or data scientists need and the resources funders provide? What skills must be developed? To help find answers, TechChange, an organization dedicated to using technology for social good, partnered with Project17, Facebook’s partnerships-led initiative to accelerate progress on the SDGs. Over the past six months, the team led interviews with top figures in the DSSG community from industry, academia, and the public sector. The 14 experts shared numerous insights into using data and data science to advance social good and the SDGs. Four takeaways emerged from our conversations and research:
1. Support the hidden work that makes effective data science possible.
Data scientists spend nearly half their time making sure their data is reliable, clean, and well organized. But too often, social sector organizations don’t have the resources to invest in essential steps like standardizing or digitizing data, which are critical to ensuring the quality of their data science projects.
Their struggles are understandable. Depending on the institution, preparing the “right data” for the task at hand could range from digitizing paper records to synchronizing knowledge management systems across different teams. This behind-the-scenes work can also involve the difficult task of developing and using standards for interoperability and data sharing. Some organizations are trying to make this easier. Carl Elmstam and Roderick Besseling, digital development experts from the Swedish development agency Sida and the Dutch Foreign Ministry, referenced the International Aid Transparency Initiative (IATI) to create a global standard for publishing data about humanitarian funding and projects. Such standards are important for coordinating complex, highly collaborative projects. For example, effective global research on COVID-19 requires harmonizing case data related to the disease.
The experts we spoke with said that tech companies can help civil society organizations struggling with this initial stage of a big data project by:
- Elevating awareness about the importance of organizing and standardizing data.
- Requiring funded groups to adhere to standards like those put forth by the IATI.
- Offering financial resources for social sector organizations to do this behind-the-scenes work.
2. Help partners build their data culture and digital transformation strategies.
A digital transformation strategy is an action plan to help an organization improve how it leverages digital tools and techniques, including data, to advance its goals more efficiently and effectively. By taking a big-picture view across the often deeply entrenched silos within social sector organizations, these strategies can help put the people, priorities, and incentives in place to cultivate a data culture of people who responsibly and proactively engage with data to solve problems. The global pandemic has underscored the importance of such strategies: Organizations have needed to adapt to working digitally and remotely, while nonprofits facing existential financial challenges have been forced to use data to invest limited resources where they’re most needed.
However, only 23 percent of nonprofits have a long-term vision for technology, according to a recent survey by Salesforce, and experts we spoke with shared similar observations.
“Organizations are trying to deal first with technology in itself,” said Catalina Escobar, a co-founder of the Colombian social change group MAKAIA. “They’re thinking about starting a digital transformation strategy.”
Some organizations are more advanced. The global NGO CARE developed a Responsible Data Maturity Model to help institutions craft a coherent, long-term data strategy. Partnerships such as the Digital Impact Alliance (DIAL) and New York University’s Open Data Policy Lab help coordinate successful digital and data-transformation strategies in the global development sector through frameworks like the principles for digital development.
If tech companies want their data or computing power to achieve their full potential in the DSSG movement, the experts we interviewed recommended:
- Working with partners that already have a digital transformation strategy.
- Helping organizations to design digital transformation strategies that will enable the meaningful use of data.
- Funding or contributing to efforts, such as those of CARE or DIAL, to support digital transformation at scale.
3. Provide skilled data scientists to work directly with nonprofits.
Tech companies often provide data or infrastructure for nonprofits, but too few nonprofits have the in-house experts (such as data scientists or data product managers) to make use of those digital assets. To fill this gap, some organizations, such as Zindi, put together competitions for online communities of data scientists to solve social problems. And initiatives such as Data Science Africa or Google’s AI Impact Challenge provide funding, training, and data science talent to nonprofits.
Although blueprints exist for tech companies to boost nonprofits’ data science chops, the experts we interviewed had mixed feelings about the success and sustainability of such efforts. The problem is that they can push nonprofits toward depending on tech firms for specific projects for limited periods, rather than building long-term independent capabilities.
“Anything that temporarily increases capacity at a nonprofit is not leading to more data-driven work,” said Claudia Juech, founding CEO and a board member of the Cloudera Foundation. “Once these people pull out, it will be very hard for nonprofits to keep it going and make it a part of their work.”
Some people argue we should experiment with more direct methods to strengthen in-house data capacity in the social sector. But Groups like DataKind have found success with a model they’ve developed for embedding data specialists into nonprofits to solve problems together. Their approach involves two important components: comprehensively understanding a nonprofit’s data problems and, secondly, training data scientists to appreciate the complexities, constraints, and incentives of working effectively inside nonprofits. DataKind’s senior director of product, Caitlin Augustin, and Afua Bruce, the organization’s chief program officer, said it is important to recruit the right people for such projects—they need to be highly skilled but humble, and able to communicate well with non-experts.
Nick Hamlin, a data scientist at Global Giving, agrees with the approach, further suggesting that—given many nonprofits’ struggles with limited resources—the data scientists chosen for such projects possess a “willingness to exist in the presence of uncertainty.”
To help civil society organizations’ address skill gaps in data science and to ensure tech firms’ digital contributions are used to their full potential, our experts suggested:
- Training data scientists, at scale and in a coordinated fashion across companies, on the work cultures, structures, and constraints of nonprofits.
- Continuing the funding of proven models to lend data science talent to nonprofits in longer-term engagements.
- For shorter-term engagements, deploying data scientists who are good communicators and trainers so they leave nonprofit teams with the skills to do the work on their own.
4. Support training for nonprofits to make responsible and ethical decisions based on data.
Social sector leaders need to be mindful of the potential biases that can seep into the collection, analysis, and presentation of data, or they risk taking inappropriate actions based on the information.
There are many educational and training resources tech companies can use to train their own teams and nonprofits in the critical thinking skills that ensure data projects are ethically executed. For example, the Johns Hopkins Center for Government Excellence and New York University’s GovLab provide training on interpreting data or understanding the risks when sharing data. TTC Labs, a project initiated and supported by Facebook, helps people across industries explore how to manage trust and transparency as they work with digital products and data. The global development NGO IREX helps leaders build soft skills for data-informed decision-making , while TechChange’s “Gender Data 101” course helps people be mindful of gender biases when working with data for social impact. To further bolster the creation and sharing of resources like these, our interviewees suggested that tech firms:
- Invest in training and educating their own employees on ethics and decision-making related to data.
- Sponsor training for nonprofits to build critical thinking, interpretation, and decision-making skills.
- Partner with non-traditional learning institutions that build critical thinking skills for using data ethically, and support similar efforts within the formal education system.
Avoid Going It Alone
As tech companies release datasets, donate cloud computing resources, or lend data scientists to the social sector, the experts we spoke with emphasized the importance of couching the efforts within existing systems, strategies, and knowledge. Coordination among all participants in the DSSG ecosystem—government, civil society, and private sector—will be critical, and numerous networks are already set up to do just that: 2030Vision, the Global Partnership for Sustainable Development Data, Data Collaboratives, the Data Culture Project, and others. Working through cross-industry cooperatives like these will help avoid redundant and siloed investments—and ensure we move more quickly toward achieving the SDGs as their deadlines loom.