BIG DATA IS GETTING A BAD RAP THESE DAYS.
Virtually every day, there’s another story about how companies are using data to manipulate consumers, or how yet another organization has failed to protect the personal information of its users. Our eating habits, driving patterns and even our daily movements seem to be up for grabs by companies putting profit over privacy.
But data itself is neither good nor bad — it’s what you do with it that matters. Scandals over the misuse of data might be grabbing headlines, but at Weinberg College, social science researchers are using data to make the world a safer, healthier and more egalitarian place.
Professors like sociologist Andrew Papachristos, anthropologist Sera Young and economist Seema Jayachandranare using data to address some of the thorniest problems facing society — from gun violence to climate change to the protection of the global water supply. Working with Northwestern’s Institute for Policy Research, they are transforming raw data into actionable solutions to the most urgent social issues of our time.
A "Win-Win" for Climate Change
The uneducated eye sees only dots on a screen. But to Seema Jayachandran, the pixelated images spell out a step toward solving climate change.
A few years ago, the United Nations tapped Jayachandran, an award-winning development economist, to evaluate efforts to prevent deforestation in western Uganda, a region hit hard by poverty and environmental degradation.
The region’s forests are home to large groups of chimpanzees, which drive a lucrative tourist industry. The deforestation of such areas can erode the quality of the soil, eliminate many species, and decrease local water quality. It’s the second largest human-caused factor in climate change, which, as Jayachandran points out, hurts poor populations the most.
Private households own much of the primary forest in this region, and most owners were clearing their parcels for subsistence farming or to sell trees to timber and charcoal dealers — not to get rich, but to meet their most basic needs.
Jayachandran developed a policy that pays people for doing something good for the environment — an approach known in the field as “payments for ecosystems services,” or PES. First, Jayachandran determined how much money the owners earned from clearing their land. Then, she calculated what they would need to be paid to do something different
A local conservation nonprofit then asked interested forest owners to sign a contract that would give them $28 per hectare of primary forest if they did not clear the trees in it.
Then Jayachandran and her team pored over the high-resolution satellite images taken at the beginning and end of the two-year project to see how much of the forest remained.
“Scientists can now classify shapes in an image to see if each pixel is part of a tree or not,” she explains. “They create algorithms that look for green, irregular circles — treetops — and they know whether the region is covered by trees or not.”
They discovered that in the 60 treatment villages, only 4 percent of the forest had disappeared. That’s compared to 9 percent in the 61 villages in the control group. That 9 percent represents “a rapid rate of deforestation,” Jayachandran says. “That means that in 100 years, there won’t be much left.”
Given the results, Jayachandran thinks that the PES approach might succeed in other countries in Asia, Africa, Latin America and Central America that struggle with deforestation.
Funding such programs also offers richer countries a new way to reduce carbon emissions. “Contributing $1 million to a poorer country to adopt sound environmental strategies goes a lot farther than it would in, say, the United States,” Jayachandran says.
“This is not the entire solution to climate change, nor the entire solution to forest clearing,” she adds. “But it represents an opportunity for a win-win.”
Getting Ahead of Gun Violence
There are are many barriers to studying gun violence. And when it comes to the crimes caused by illegal guns, the hurdles are even higher.
You need a way to predict who will have guns, and who will get shot. But to develop that model, you would need to set up a randomized, controlled study — a logistically and ethically challenging task, given the nature of gun violence.
Qualitative methods, such as interviews, don’t provide enough information to develop broader statistical models, either. On top of that, the government severely limits the use of federal funds for gun research. Given those challenges, an academic approach to the investigation of gun violence seems virtually impossible.
But data science can get the job done. Sociologist Andrew Papachristos has spent the last several years looking at how network science — in this case, the study of how social relationships influence the behavior of people — can be used to understand the spread of crime and gun violence. He has completed several high-profile studies with a new approach that combines data sets from various sources.
“The data are hiding in plain sight,” he explains. “They’re sitting in administrative data sets — in arrest records, health departments, and trace information from the Bureau of Alcohol, Tobacco and Firearms.”
Papachristos and two co-authors mined this data to conduct a 2017 study that compared gun violence in Chicago to a contagion that should be treated as a public health crisis rather than as a policing issue. The researchers created an epidemiological network of more than 100,000 individuals at risk of getting shot between 2004 to 2016. Their models, which incorporated both social contagion and demographics, predicted future gunshot victims better than models based on social contagion or demographics alone.
Papachristos co-authored another study in 2018 that focused on the role that Chicago neighborhood networks play in gun violence. The research found that people within any neighborhood — from the highest-crime community to the lowest-crime community — could obtain an illegal gun within an average of “two to three handshakes,” even with Chicago’s strict gun laws.
“So that means you can get a gun from a friend of a friend,” Papachristos explained. “We found that gangs reduce that distance, which is what gangs are supposed to do — be there for protection. If they can do that, think how easy it would be to get a gun to commit a crime.”
The ultimate goal, Papachristos says, is to build models to prevent gun violence by predicting who can provide resources and innovations, and when and where to use them. For instance, information on the timing and pathways of gunshot events could enable street outreach workers to mediate conflicts before they become violent. “We’re trying to use science to save lives,” Papachristos says.
Getting HWISE to Water Insecurity
Sera Young was studying food insecurity when she made a surprising discovery. Young, an assistant professor of anthropology, was investigating how food insecurity and HIV affect mothers and their infants in western Kenya. To better understand the causes of food insecurity in the region, she asked the mothers to take photos of factors that influenced their ability to feed their babies.
Young was surprised when she saw the submissions. They weren't the pictures of scorched crops or rotten produce that she’d been expecting.
“A lot of them took pictures related to water — of traveling great distances over rough terrain or standing in long lines in the hot sun to reach a clean water supply,” she says.
To quantify her findings on food insecurity, Young asked the study participants to answer a standard, globally validated set of questions. But when she sought to explore the question of water insecurity, she soon realized that no similar measurement existed. There were a variety of scales that captured water insecurity at the macro level, but none that addressed the unique experiences of water-insecure individuals.
No wonder the issue took Young by surprise. “Our inability to quantify problems with water means that many of them have gone unrecognized,” Young says.
Indeed, water insecurity is a pervasive, complex problem. For at least one month each year, 4 billion people around the world suffer from severe water scarcity. In addition, 663 million people have no access to a clean water source, which puts them at a higher risk of drinking contaminated water.
Young set out to create a measure that would capture the multifaceted aspects of water and the many ways it is used. She gathered a team of interdisciplinary researchers, practitioners and local collaborators who collected data from more than 8,000 households across 28 sites in 23 countries.
The data collectors used a questionnaire designed to identify all the water problems experienced by households — for example, how frequently respondents had consumed unsafe water or had gone to bed thirsty in the previous month. Young and her team eventually narrowed that exhaustive list down to a final set of 12 questions that capture the universal experiences of water insecurity.
These items comprise the new Household Water Insecurity Experiences (HWISE) Scale.
A key element of the scale is its ability to measure water insecurity in a cross-culturally validated manner, meaning that the scale applies equally to any location in the world.
“We want the criteria to be the same for Pakistan or Colombia or Mexico, so we can share those data and confirm that the score is invariant,” notes Young, who won an Andrew Carnegie Fellowship in April to support her research in this area.
Young’s work has caught the attention of numerous national and multinational institutions, including the World Bank and USAID. UNESCO is also seeking to collaborate with Young to include the HWISE Scale in the 2020 Gallup World Poll, which is administered in more than 140 nationally representative surveys.
The results could be transformative. A baseline global assessment of the prevalence of water insecurity could help development agencies identify the most water-insecure communities, allocate limited resources more effectively and track progress over time.
“This would produce the big data that would drive water policy for the world,” Young says.