"More and more, scientists are realizing that computational methods can serve as an unparalleled lens for examining human activity."
Yet asking such thought-provoking questions is the essence of intellectual inquiry, and CSSI aims to model and analyze social phenomena in innovative ways, combining insights from the new field of computational social science—an approach that could yield significant revelations about such central human concerns as health, harmony, and work. (And as it turns out, yes, bacteria do network, just not via Snapchat.)
The campus’s efforts in computational social science began in 2010, when an interdisciplinary group of faculty led by computer scientist Andrew McCallum initiated the hiring of a handful of faculty members with expertise in computer science, statistics, and the social sciences. Two years later, James Kitts came from Columbia University to join McCallum as a founding codirector of the institute. Now, with 68 affiliates in 23 departments and nine colleges across the university, it has grown to become the largest, most diverse academic institute of its kind.
“CSSI is shaping an unrivaled population of experts into a cohesive scholarly community through programmatic support for research and training,” says Kitts. The research under its aegis explores a wide range of questions: Is obesity contagious? Does a diversified computer science faculty attract a more diverse pool of graduate students? Can computational analysis be used in innovative ways to help identify discriminatory employers?
“More and more, computer scientists are thinking of social organizations as computational systems,” explains Associate Director David Jensen. “When we click on a link in a friend’s Facebook post to read the latest presidential campaign news, we’re using that social media channel and that friend as an information-processing tool. On the other hand, more and more scientists are realizing that computational methods can serve as an unparalleled lens for examining human activity. Rather than being confined to using traditional social science methodologies, such as questionnaires and lab experiments, computational social science applies new means to examine the dynamics of complex social behavior as it plays out in real time for large populations or groups.”
Take, for instance, the question of whether one can “catch” obesity. Nationally, the number of obese children has more than doubled since 1980. In Massachusetts, nearly a third of children between grades 1 and 10 are overweight or obese. School-based interventions have had only limited success. Recent research suggests that friends may influence one another’s weight gain or loss. But which comes first, the obesity or the friendships? Through CSSI, a team of experts from kinesiology, nutrition, sociology, and statistics proposes to use innovative statistical methods to study peer influence in diet and exercise among Massachusetts middle schoolers. The next step will be using computer models to help design better school-based interventions to promote healthful behavior.
In addition to supporting interdisciplinary research projects, through its seminar series, CSSI fosters the cross-pollination of ideas. By bringing in speakers throughout the academic year, the seminars introduce the work of scholars at UMass and elsewhere to the campus community and, like the institute itself, encompass diverse areas of inquiry. This fall’s lineup includes UMass Amherst Professor Kimberley Geissler, whose public health research found that patients enrolled in high-deductible health plans use about 7.5 percent fewer diagnostic tests such as MRI, X-rays, and CT scans than patients without such plans. Geissler says that her findings were unexpected. “I think what we found most surprising,” she says, “are the large reductions in imaging use among people with high deductibles. We had hoped to find that patients were reducing the use of low-value imaging; instead, we found that they reduced all use similarly. It seems that patients are not informed enough to discern which tests are more optional and which are medically necessary.”
According to Kitts, surprising findings are not uncommon in CSSI research. “The interdisciplinary nature of the projects allow for broader perspectives and a more complete picture of what needs to be addressed,” he says.
CSSI also hosts “mixers” for researchers who investigate similar scientific questions in different disciplines. Organized around a public health theme, last spring’s event brought together researchers and administrators from 14 departments. Each researcher gave a two-minute “slam” presentation and a poster-discussion session on topics including access to malarial drugs, predicting and preventing epidemics, and distributing health information on social media. The event was a huge hit, stimulating animated conversations about possible collaborations. It also inspired researchers to organize a mixer to promote similar cross-fertilization between faculty at UMass Amherst and at UMass Medical School in Worcester.
All of these activities prompt new questions and provoke “aha!” moments. The I-never-thought-of-it-that-way response can be the first step toward finding new ways of seeing, making new connections, and creating new solutions. The combinations of disciplines CSSI nurtures are already shedding light on important scientific questions, and the institute is poised to play an important role in computational social science in the coming years.