Case Study 4.1: Confirmation Bias in Cable News Consumption

How Media Environments Amplify What We Already Believe

In 2003, Shanto Iyengar and Sean Westwood published research confirming what many observers had suspected: Americans were increasingly sorting themselves into partisan media environments, and the media they consumed was increasingly confirming rather than challenging their political beliefs. By the 2010s, this pattern had become so well documented and so widely discussed that it had acquired a common name — the "echo chamber" — and a public policy profile that rivaled concerns about deliberate disinformation.

This case study examines what the research actually shows about partisan media and confirmation bias, how it differs from early popular accounts, and what it does and does not tell us about propaganda vulnerability.

What the Research Shows

Iyengar and Westwood (2019): In multiple large studies, Iyengar and Westwood documented that partisan animosity — hostility toward the opposing political party — has grown dramatically since the 1990s and now exceeds racial hostility in some measures of social distance. More relevant to confirmation bias: this polarization predicts media behavior. People who feel strong partisan animosity are more likely to seek confirming media and less likely to expose themselves to disconfirming sources.

Pew Research Center studies (2014, 2020): Pew's surveys documented that consistent conservatives and consistent liberals have substantially different media diets. Consistent conservatives were most likely to trust Fox News and least likely to trust MSNBC, The New York Times, and most other national news sources. Consistent liberals showed the opposite pattern. The two groups also reported strikingly different factual beliefs about the same real-world events — employment rates under different presidents, crime trends, immigration statistics — suggesting that media exposure differences were producing factual belief differences.

Pariser (2011) vs. later research: Eli Pariser's popular "filter bubble" concept proposed that algorithmic personalization would dramatically reinforce individual echo chambers. Subsequent empirical research has found a more complicated picture. Studies by Guess, Nyhan, and Reifler (2018) found that people's news consumption includes more cross-cutting content than the filter bubble theory predicted. The problem may be less that people are exposed only to confirming content and more that confirming content receives more cognitive processing and is more likely to shape belief.

The Mechanism: How Confirmation Bias Shapes Media Consumption

Confirmation bias operates at multiple stages of media consumption:

Selective exposure: People prefer media sources that generally share their political orientation, reducing the probability of encountering strongly disconfirming information.

Selective attention: Within any given media environment — even one that includes disconfirming information — people pay more attention to confirming items. Eye-tracking studies show that people spend more time looking at political headlines that confirm their views.

Selective processing: When disconfirming information is encountered, it receives more critical scrutiny. People identify methodological flaws in studies with conclusions they dislike and overlook flaws in studies with conclusions they prefer — a pattern well documented in research by Charles Lord and colleagues.

Selective memory: Confirming information is better remembered and is more likely to be retrieved when making subsequent judgments.

The result of these four stages is that even an information environment with relatively balanced content can be processed in ways that increase confirmation bias.

Case: Fox News Viewership and Crime Perception

One of the most thoroughly documented confirmation bias dynamics in American media involves the relationship between heavy Fox News viewership and political beliefs.

A 2012 Fairleigh Dickinson University study found that Fox News viewers scored lower than viewers of most other news sources — and lower than people who reported watching no news — on factual knowledge about domestic and international political events. This counterintuitive finding (more news consumption producing less accurate beliefs) is consistent with systematic misinformation rather than simply selective omission.

On crime specifically: multiple studies have found that heavy Fox News viewership correlates with substantially overestimated violent crime rates, even as FBI statistics showed declining rates. The mechanism is availability bias amplified by confirmation: the audience disproportionately receives vivid crime coverage, the vivid examples become cognitively available as a reference point, and subsequent assessments of crime frequency are calibrated to the media environment rather than to statistics.

What This Is and Is Not

This case is sometimes presented as evidence that Fox News viewers are uniquely manipulated or uniquely biased. This framing is inaccurate in two ways that matter analytically:

First, the mechanisms are media-general. Confirmation bias, availability heuristics, and selective processing operate across all ideological media environments. Studies of MSNBC viewers, of left-leaning partisan media consumers, and of social media echo chambers on both sides find similar dynamics. The specific misinformation differs; the cognitive mechanisms are the same.

Second, political polarization has structural causes that extend beyond media. Geographic sorting (people increasingly living in politically homogeneous communities), economic stratification, and racial demographic change all contribute to the conditions in which partisan media environments are more likely to form and more likely to be reinforcing.

The propaganda question is not whether Fox News viewers are uniquely biased. It is whether any media environment that consistently presents information calibrated to confirm prior beliefs, using techniques designed to exploit availability bias, constitutes propaganda — and whether the answer depends on whether the organization is aware it is doing this.

Discussion Questions

1. The case study documents that confirmation bias operates across ideologically diverse media environments. Does this finding undermine the working definition of propaganda (which requires intent), or does it support Ellul's structural model (which does not require individual intent)?

2. Research shows that people who consume more news from partisan sources can hold less accurate factual beliefs than people who consume less news overall. How should this finding affect the public discourse on "media literacy" as a remedy for misinformation?

3. If you learned that a media organization's internal research showed they understood that their crime coverage was causing viewers to dramatically overestimate crime rates — and they continued the coverage anyway because it maximized engagement — how would you classify this under the working definition of propaganda?

4. The "filter bubble" concept turned out to be empirically overstated — people encounter more cross-cutting content than Pariser predicted. Does this finding reduce your concern about partisan media effects, or does it shift the concern from exposure to processing? Explain your reasoning.