Expert Commentary

Does media fragmentation contribute to polarization? Evidence from lab experiments

2014 study in Public Opinion Quarterly that suggests that political polarization cannot be explained by media diversity and the fragmenting of audiences.

From health care reform and global warming to marijuana legalization and same-sex marriage, voters are increasingly polarized, particularly along partisan lines. In this context, the level of fragmentation in the media landscape is assumed to be an important explanation for this polarization.

While the impact of the media on political beliefs and preferences is undeniable, the extent to which media affect voters’ choices and partisanship is still under study. In the era of “big broadcast,” prior to the Internet, scholars were concerned about the lack of media diversity and the fact that citizens were a “captive audience” potentially subject to mass manipulation. Thus, many hoped that the democratization of news media, particularly online, could have a positive impact on political participation, civic engagement and the production of information. However, in a digital era featuring media abundance, scholars now worry about what they call “selective exposure.” A more fragmented media environment may prompt citizens to seek out more like-minded news sources, contributing to the reinforcement of prior beliefs and opinions and exacerbating polarization. Thus, an “echo chamber” — Fox News and MSNBC are often cited as examples — can make certain views even louder and more pronounced among a group of like-minded individuals. These media dynamics are thought to be at work as we witness shifts in public opinion on topics such as climate change.

Behind this debate is a fundamental cognitive truth established by the behavioral science field: The majority of individuals have a confirmation bias. When individuals are open to evaluating arguments for a certain position, they are likely to be constrained by cognitive biases that make them reinforce their initial position, even in the presence of contradictory evidence.

A 2014 study in Public Opinion Quarterly, “The Informational Basis for Mass Polarization,” analyzes the effects of individuals’ information environment on opinion dynamics and the role of prior attitudes in moderating those effects. To study whether information choice serves as a foundation for polarization, the author, Thomas Leefer of Aarhus University, conducted two experiments in Northwestern University’s Political Science Research Laboratory involving information choice behavior at the micro-level on health care reform and U.S. military action in Libya. The experiments involved 176 undergraduates in the first study, while the second study involved a mix of students and Internet- recruited non-students, totaling 300 persons. The experimental procedures provided informational prompts and gave subjects the chance to search through and read different kinds of articles — constituting the “information environment,” and a way of “priming” subjects — before asking them about their views on the issues.

The study’s finding include:

  • Individuals reacted to the information environment by choosing more articles or news in the direction of the slant of the environment. Individuals in “pro” environments — in favor of a particular position — read more pro than con arguments; those in mixed environments also read more pro than con articles; those in con environments read the same number of pro and con articles.
  • Individuals’ opinions moved in the direction of the slant of the environment. On a scale from 0 to 1, the information environment was found to make a difference of 0.04 in opinions — the equivalent of 4 percentage points.
  • However, the role of high- and low- importance beliefs — those who have strong, dogmatic beliefs on a subject, versus those with little preconceived opinion — is key to understanding the link between search behavior, opinions, and information environment.
  • Regardless of the fragmentation of the media environment in the experiment, individuals with high-importance attitudes became more polarized toward their initial position. This is because they appear to evaluate information in a consistent way, regardless of the information they encounter. On the other hand, low-importance respondents’ opinions did not change much and actually moved in a similar fashion to a control group that was not subject to manipulations.
  • A pro-slanted environment makes individuals more supportive over time. However, the size of that effect will be highly conditional on the importance of prior attitudes: Individuals with low-importance prior attitudes will be more affected by the relative fragmentation of the media compared to those with strong attitudes.

“The results presented here suggest that information choice, at least among those with personally important opinions, does not appear to make those individuals or democracy better off,” Leefer concludes. “Freedom to choose one’s political news seemed to many scholars of the 1980s a much-needed component of democratic health, but the abundance of choice that has emerged in the ‘post-broadcast’ present is now being seen as democratically problematic…. This paper has shown that the implications of choice are highly conditional.” The experimental evidence suggests that “polarization therefore seems to require more than media fragmentation,” and therefore may be better explained by deeper societal patterns.

Related research: A 2013 study published in Public Opinion Quarterly, “Extreme Groups: Interest Groups and the Misrepresentation of Issue Publics,” looks to empirical evidence to help settle some of these debates, testing whether members of motivated groups are “giving voice” to wider public communities or pushing their own unrepresentative agendas. Also of interest is a 2013 study published in Psychological Science, “Political Extremism Is Supported by an Illusion of Understanding,” that explores a central paradox of our politically polarized era: How can people maintain such strong views on complex policy issues that they seldom understand with any sophistication? The researchers attempt to measure the degree of overconfidence people typically have in their own understanding of the mechanics of how systems and issues work, and to evaluate how the process of explaining their views might moderate extremism.

Keywords: news, cognition, communications, polarization

About The Author