Software for the Social Good: Where Are the Analysts?

Editor's note: The Digital Communities Special Report, which appears twice a year in Government Technology magazine, offers in-depth coverage for local government leaders and technology professionals. View links to the entire report here.

Like so many communities around the country, Marin County, Calif., is struggling with the fallout from an epidemic of opioid addictions. The county’s Public Health Department is trying to get a grip on the problem, and one way to do that is by pulling together data from local law enforcement and public safety agencies, as well as from the state and federal government to help county officials track everything from the number of opioid prescriptions given out to the number of emergency room admissions for drug overdoses.

But the county faces two problems. First, figure out a way to spend more time exploring what the data means, and second, find people who can make sense of it. The answer to the first question is to use one of the new digital tools that can visualize the data. Marin County uses LiveStories. Other local governments, such as Allegheny County, use Tableau, while Esri has an arsenal of tools for teasing out stories from data. As for finding people who are not public health experts to be able to interpret the data, Marin’s Public Health Department has had to consider more unconventional approaches, such as using volunteers to help make sense of the data and find correlations that might lead to better outcomes.

For Allegheny County, the situation is a bit different. With Carnegie-Mellon University’s Heinz College of Information Systems and Public Policy nearby, the county government has an easy-to-tap pipeline of talent for conducting analytics. More importantly, the director of DHS sees data analysts as a priority, according to Dalton. Hiring analysts is not an option, but an investment decision, she said.

Hiring More CDOs Alex Engler, program director of Computational Analysis and Public Policy at the University of Chicago, points to the role of chief data officer as a new position that is helping change how government tackles social issues. “The cities that seem to get the most done have a chief data officer,” he said. Engler points to Chicago’s Tom Schenk as someone who has been a trailblazer when it comes to data-driven policies and practices. But the role remains relatively new and is still uncommon, especially at the local level. New York City was the first local government to appoint a CDO in 2011. This year, there were a dozen CDOs in city government and three at the county level, according to Government Technology. Recognizing the growing importance of the CDO, in March 2016, Harvard Kennedy School’s Ash Center for Democratic Governance and Innovation announced that it had been awarded a grant by the Laura and John Arnold Foundation to establish and support a national peer network of urban chief data officers, who will collaborate on shared projects that advance the use of data visualization and predictive analytics in solving important urban problems related to economic opportunity, poverty reduction, and addressing the root causes of social problems. Since then, a handful of cities have appointed CDOs. But the role is expected to become more important in the coming years. Gartner is projecting growth of 1,600 percent in CDOs across all industry sectors over the next two years and expects by 2019 that 90 percent of large organizations will have one.

The explosion in data at the local level has led to a wide range of breakthroughs in the field of social services, according to Alex Engler, program director for computational analysis and public policy at the University of Chicago. He points to analytics in Chicago that allowed city officials to prioritize which homes needed to be inspected based on their likelihood of having lead paint. They were even able to single out at-risk homes prior to when a family with children moved into them. He also mentioned the city’s ability to predict where rat populations were likely to occur and curb outbreaks before they happened.

The point, he said, is that predictive analytics can happen when analysts repurpose data that was initially collected for another reason. Yet, Chicago is one of just a handful of local governments that has figured out how to analyze disparate data sets in a useful way, a dilemma that Goldsmith and others have warned about.

“This type of shift in how data can be used is a big problem for public social services, because there’s no one with the skill set to deal with it,” said Engler. “The number of computer science graduates who go into public policy is unbelievably low. The public sector is about 30 percent of the economy, yet less than 10 percent of computer science grads go into public service.”

Engler left a promising career as a data scientist working for Washington, D.C.-based think tanks and has returned to academia at the University of Chicago to recruit and train promising public policy students in the arcane world of data science and analytics. The education process centers around the practice of causal inference in public policy, which means looking retroactively at whether a program works, said Engler. As an example, he mentioned how changes to the Medicaid program can affect health-care prices, which can then impact consumer health care.



Philadelphia CDO Tim Wisniewski and San Francisco CDO Joy Bonaguro address a panel at Harvard’s Summit on Data-Smart Government in November.

---

“The tools you use to teach causal inference are statistical analysis tools, such as Stata and SAS,” he said. But a shift is taking place in academic data science, with more students learning analysis using open source tools, such as R or Python. At the same time, Engler believes the expanding universe of open data in government is helping to make possible more data-driven solutions for problems around poverty, homelessness and health. The universe of data that could be beneficial to a local government’s social problems keeps expanding, with more private data sets beginning to have an impact too.

5 Trends Driving Data-Driven Progress Chris Thomas, Esri’s director of state and local government marketing operations, has worked with the public sector for decades and has seen mapping technology evolve to solve myriad problems. Now, he’s seeing a similar transformation unfold in the field of social services. Here’s his list of five technology trends that are making data-driven solutions for social problems more effective: Open data. Once almost inaccessible, government data that is open and readily accessible is helping a growing number of communities better cope with homelessness and poverty. While most people perceive open data as something for the public to consume, some of the biggest users of open data are government agencies. Story maps. This new concept in data visualization helps communicate what’s going on and where the data is. It allows people to interact with data in a better way than having a single person try to explain what’s going on to a room full of policymakers and stakeholders. Mobile solutions. Collecting data in the field in a consistent manner so it can be shared easily has had a significant impact on tackling social problems for local governments. Dashboards. This tool has provided workers with situational awareness for monitoring how a change in policy is doing against the norm. Data analytics. Data-driven projects require analytics to support decision-making and to help predict outcomes. It can take a good project and make it into a transformative one.

“Today, we have to teach students how to use new tools such as Python and have them evaluate Yelp data for health inspections,” said Engler.

The new generation of statistical tools are getting better, according to Engler, “and a lot of the best tools are free.” But the tools are not getting dramatically easier to use. “There’s no point-and-click solution to any of these things,” he said. “You have to build an application that is going to integrate data coming in from various city services, as well as the private sector, and then build a platform for internal and external use.”

Where Engler sees progress is with local governments that have a chief data officer in place with a small staff who partners with the community. He cites Tom Schenk, Chicago’s CDO, as someone who has created a model data-driven strategy that is impacting a broad range of public policies, including social services. One thing Schenk has done to expand the impact of data on policies is use community volunteers to help sort through the vast amounts of data the city is collecting. Cities like Chicago are doing this as much out of necessity as for community engagement and improved outcomes.

Chicago, despite its size, struggles to find people who can interpret the data government collects and develop predictive models. In one project, the city turned to volunteers to pore over data that will predict with a high degree of accuracy when the city’s beaches would be affected by an E. coli outbreak. Another data project is helping the city accurately gauge how much rainwater runoff goes into the city’s sewers and how much can be diverted by more environmentally friendly methods.

Using volunteers to conduct data science utilizing dashboards and other tools does have risks. They might see some data that the government doesn’t want them to see. Or, they may make correlations and predictions that could end up as inconsequential or even distracting. But for local governments that can’t find or afford analysts with a background in public policy, using volunteers can be beneficial as long as the data is presented in the proper context and with the right parameters.