The rise of the Duke Polarization Lab

Monday, December 2, 2019
Sunshine Hillygus and Polarization Lab

The Polarization Lab at Duke brings together scholars from the social sciences, statistics, and computer science to develop new technology to bridge America’s partisan divide. Before their first team publication, faculty and graduate students from the Department of Political Science began collaborating in the Lab, guided by Director Chris Bail.  Social media sites are often blamed for contributing to political polarization because they encourage people to segregate themselves from those with opposing political views. So in October 2017, the Duke Polarization Lab set out to test the question, "Does the status quo strategy to nudge polarized people into engagement moderate them?"  A team from the Lab formed, including Haohan Chen, to run an experiment.

They recruited a representative sample of 1,200 Republicans and Democrats who visit Twitter regularly to complete a survey about their political views.  One week later, the Lab offered to pay some of them to follow bots the Lab created that retweeted 24 messages each day from an opinion leader from the opposing political party.  But when Lab members measured the participants' political views one month later, they discovered that Democrats who followed the Republican bot became slightly more liberal and Republicans who followed the Democratic bot became substantially more conservative. 

As the Lab grew, Sunshine Hillygus from the Department of Political Science and Alex Volfovsky from the Department of Statistical Science became Co-Directors.  "I think it's important to have the perspective of a survey methodologist," Hillygus said, "big data gets a lot of hype, but we need to understand not only the data that we have, but also the data that we don't have."  One of the great unknowns about social media's effect on political polarization is to what extend Russian trolls influenced the 2018 U.S. elections?  At this point, Brian Guay, a research assistant with the Duke Initiative on Survey Methodology, joins the Lab and Chris Bail steers the team into position to study the impact of the Russian Internet Research Agency using unhashed, restricted-access data provided by Twitter's Elections Integrity Hub.  

What they discovered is possibly the most provocative null hypothesis of the year: "we found no substantial effects of interacting with Russian IRA accounts on the affective attitudes of Democrats and Republicans who use Twitter frequently toward each other, their opinions about substantive political issues, or their engagement with politics on Twitter in late 2017." Journalists and pundits are now scrambling to interpret the findings.  "We are not at all suggesting Russia’s attempt to interfere is not problematic," Sunshine Hillygus tells MarketWatch, "We are trying to be very careful in what the take-home messages are. We need to give credit to the American public: They are not entirely susceptible to propaganda."   

In the months to come, the Polarization Lab is turning towards building research tools to run experiments in social media.  Could we see cause and effect by turning on and off different components of social media in an experimental setting?  What should the criteria be for apps to make recommendations to learn more or recommend new friendships? How do level changes in online anonymity or social reputation systems effect behavior? "What would a better social media look like," Chris Bail asks, "how could social media make us better at connecting with each other, rather than worse, make us more civil, not less.  We're building our own platform, in order to figure out the secret sauce."  Visit the Duke Polarization Lab to learn more.