Communication professor JungHwan Yang at University of Illinois Urbana-Champaign. (University of Illinois photo by L. Brian Stauffer)

Popular wisdom suggests that the internet plays a major role in shaping consumers’ political attitudes in the U.S., and some recent studies blamed partisan news outlets’ coverage for the increasing polarization of the nation’s electorate.

However, a study of 1,037 internet users during the 2018-19 U.S. midterm election found that online partisan media may have little direct impact on consumers’ political beliefs and activities.

Instead, the primary consequence of greater exposure to right- or left-leaning news media is the erosion of readers’ trust in the mainstream press, said communication professor JungHwan Yang of the University of Illinois Urbana-Champaign.

Yang is a co-author of the study, published in the Proceedings of the National Academy of the Sciences. The paper was co-written by politics and public affairs professor Andrew Guess, of Princeton University; computational political scientist Pablo Barberá, of the University of Southern California; and Simon Munzert, a professor of data science and public policy at the Hertie School.

The study participants, who were recruited from the data and analytics group YouGov’s Pulse panel, allowed the researchers to survey them multiple times and agreed to install passive metering software on their laptop and desktop computers or tablets so the researchers could track their online activities. The researchers collected data on more than 19 million of the participants’ website visits and their Twitter posts and follows.

The study was novel in its combined use of real-world experimentation and computational social science techniques, Yang said.

“Past studies that have shown links between partisan media and polarization mostly relied on small-scale controlled experiments or surveys,” Yang said. “So it was not only difficult to observe people’s online media use accurately, but also to disentangle whether participants were selecting news sources that aligned with their partisan predispositions or if the partisan media were making people’s views more extreme.

“In our study, we were able to track their online activities for an extended time period as well as assess their attitudes with surveys so we could actually see what information they were consuming and its political consequences.”

Yang said they used a “nudgelike” approach that subtly but naturally increased participants’ exposure to two partisan websites during their daily online activities to demonstrate the importance of basic digital “opt-ins” at structuring people’s information consumption.

For a month, one-third of the participants were asked to set the default homepage on their web browser to the conservative outlet Fox News while another one-third set theirs to the left-leaning outlet HuffPost. The remaining participants, who were not asked to change anything, were assigned to the control group. Participants in the Fox News and HuffPost groups also were asked to subscribe to affiliated newsletters.

The participants were interviewed seven times from July 2018-October 2019 and were asked about their news media consumption, levels of trust in the mainstream media, their approval of then-President Donald Trump and their opinions on a variety of foreign and domestic policy issues. Participants’ views on immigration were of particular interest to the researchers because immigration was a topic of contentious debate during the election.

Prior to the study, participants spent less than 34 minutes per week on news-related websites, according to the study.

During the first week of using their new homepages, people in the HuffPost group visited about one additional page on that site daily, amounting to nearly 50 seconds of additional browsing time. Their counterparts in the Fox News group visited three or four additional pages on that website each day, for a total of about two additional minutes.

By the eighth week of the study, people in the Fox News group visited the site an additional 3.7 times each day, while people in the HuffPost group made an additional 0.4 visits daily on average.

The increased exposure sparked no major changes in either group’s feelings about the political candidates or toward the parties, their voting behaviors or their perceptions of polarization, although the HuffPost users’ views about immigration became more liberal over time, Yang said.

However, both groups’ trust and confidence in the mainstream press significantly declined, an effect that emerged during the first several weeks of increased exposure and remained detectable a year later, the researchers found.

“We saw a lowering in their overall trust of the media and that can promote polarization by making people less receptive to information that challenges their beliefs,” Yang said. “If consumers do not believe the mainstream media, they will look for other news sources, and there are a lot of alternative sources these days. If that trend continues, over time they will have different information and a differing understanding of what’s true and what’s not. And that can have negative implications for democracy.”

A positive outcome of participants’ increased exposure to the news media was that they were better informed about current events and more adept at distinguishing events that actually occurred from fictitious events on the surveys, the researchers found.

The research was funded by the Volkswagen Foundation Computational Social Science Initiative, the Princeton University Committee on Research in the Humanities and Social Sciences, and the Center for International Studies at the University of Southern California.