Echo chambers and polarization: how groups amplify confirmation bias

Guide · Media and society · Check it yourself Echo chambers: concentric circles around a single view and one person outside the circle
⏱ approx. 14 min read 📚 Category: Blog 🕓 Updated: October 13, 2026

Short answer: echo chambers (information bubbles) are environments in which we mostly hear voices that agree with our views, and this reinforces confirmation bias: the tendency to seek out and remember what fits our beliefs. Research shows that it’s not just algorithms but above all our own choices (clicks, friends, sources) that decide what we see. Getting out of an echo chamber is a practice: deliberately read the other side’s arguments in the original, check sources laterally and talk about facts calmly, because contrary to popular opinion, a correction rarely backfires.

This article focuses on the social aspect. We describe the psychological basis of the phenomenon separately in our article on confirmation bias.

What echo chambers and information bubbles are

An echo chamber is an information environment in which our views come back to us like an echo: from friends, from the accounts we follow and the sites we read. A filter bubble is a related concept, usually referring to algorithms choosing content for us based on what we’ve clicked on so far. In practice the two mechanisms mix: what we see is decided both by the system and by ourselves.

Confirmation bias, as described by psychologists (including Nickerson), is the tendency to seek out and interpret evidence in line with one’s own beliefs. An echo chamber is its social version: instead of looking for confirmation ourselves, we get it from those around us.

What research says: user choice vs. the algorithm

The most often cited study on the subject was published in Science in 2015. It covered about 10 million American Facebook users who had stated their political preferences.

Results of the 2015 study (according to an AAAS summary)
MetricResult
Friends from the opposing partyon average more than 20% of friends who declared party views
The effect of users’ own clicksa reduction in the share of cross-cutting content of about 4 percentage points
The effect of the news-ranking algorithma change in that share of about 1 percentage point
⚠️ Caveats: the sample included only people who had publicly stated their political preferences, so it doesn’t represent all users. A commentator on the study quoted in the summary, David Lazer, warns that a small effect today could become a big one tomorrow, so the influence of algorithms is still worth checking.

The takeaway for the ordinary user is practical: since in this study our own choices weigh more than the algorithm, we have real influence over what we see.

Polarization: the same evidence, ever more distant views

In the classic 1979 experiment by Lord, Ross and Lepper, 48 students — supporters and opponents of the death penalty — read two supposed studies on its effectiveness: one consistent with their view and one contradicting it. Both groups rated the study matching their own position as more convincing and methodologically better, and ultimately strengthened their initial opinions. The result was greater polarization, even though everyone saw the same, mixed evidence.

Scale this up to society: when millions of people read the same events through their own filters, each side feels that “the evidence is on its side.” An echo chamber adds a group effect to this: everyone around us confirms that we’re right.

How disinformation exploits echo chambers

In its series “How not to be manipulated,” the Polish fact-checking organization Demagog describes how disinformation campaigns prey on mental shortcuts. The most important ones include:

  • Confirmation bias. We consider information that matches our views important and look for content that reinforces it. We remember counterarguments less well, and once drawn into a false narrative, we find it harder to get out.
  • The availability heuristic. Easily recalled, emotional examples seem more important and more common than they really are.
  • Out-group homogeneity bias. We see people from another group as similar to each other, which breeds stereotypes, e.g., “all politicians lie.” Disinformation uses this to turn society against selected groups.
  • The halo effect. One positive trait or first impression makes us trust someone more and check them less.
  • The bias blind spot. We notice biases in others, but not in ourselves.

Case study: when a bubble amplified a wrong number

Case 1: crime statistics in Warsaw (July 2023)

According to Demagog, in July 2023 a message spread about crime in Warsaw more than doubling. The source was official city statistics that contained an error. Even so, some politicians considered the data credible and linked it to emotional stories about crimes committed by foreign drivers. The number fit an existing belief, so it didn’t arouse suspicion, and the emotional stories cemented its credibility. It’s an example of how confirmation bias and an information bubble together amplify a wrong number.

Case 2: two people, the same feed (illustrative scenario)

Two people with opposing views follow different authors and click on different headlines. After a few weeks, each sees a different picture of the same event, and each is convinced their picture is complete. According to the 2015 study, it’s precisely their own clicks that may weigh more here than the algorithm.

Case 3: a search engine instead of a feed (illustrative scenario)

The query “why X is harmful” tilts results toward confirming the harm, while a neutral “is X harmful and what does research say” gives you a chance at a fuller picture. The way you phrase a question acts like your own personal mini-bubble. We write more about phrasing queries in our guide to Google operators.

The backfire myth: do facts always make things worse?

There’s a popular belief that when we present someone with facts that contradict their views, they dig in even deeper (the so-called backfire effect). Wood and Porter conducted five experiments with more than 10,100 participants, covering 52 topics, including highly polarizing ones. No correction caused a backfire effect. The authors conclude that the evidence for such an effect is much weaker than previously thought, and that people usually accept factual information even when it clashes with their ideology.

That doesn’t mean changing views is easy. Facts are usually accepted, but they don’t always change an overall judgment or identity. Still, it’s worth correcting misinformation — and doing it calmly.

How to get out of an echo chamber: 8 steps

You can’t completely escape confirmation bias, but you can limit its effects. Below are steps based on research and fact-checkers’ recommendations.

1

Consider the opposite

Before you decide you’re right, ask: what evidence would support the opposite claim? In a 1984 experiment by Lord, Lepper and Preston, the simple instruction “consider the opposite” reduced biased evaluation of evidence.

2

Read laterally

Instead of staying on the page that makes a claim, open new tabs and check what others say about that source and that claim. This approach, recommended by fact-checkers, also works when you trust “your own” sources.

3

Write neutral queries

Ask “is X harmful and what does research say,” not “why X is harmful.” Your own wording decides what picture you get.

4

Slow down when emotions run high

Content that immediately triggers anger or satisfaction needs extra verification. Disinformation uses emotional stories because they’re easier to remember and accept as typical.

5

Diversify your sources

Follow several reliable media outlets with different profiles and read original documents, not just write-ups of them. Our practical tip: once a week, read the best argument from the side you disagree with.

6

Check your own feed

Look at whom you follow and what you click. The 2015 study suggests users’ choices may limit content diversity more than the algorithm — and that’s the part you can influence.

7

Separate facts from identity

Ask yourself whether you’re defending a claim or your group. When views become a badge of belonging, they’re harder to revise.

8

Write down what would convince you

Before checking, decide what evidence would change your mind. If nothing would, you know you’re not examining the matter but defending it.

How to talk to someone who sticks to their version

Since research doesn’t confirm the backfire effect, correcting errors makes sense. Our practical tip: start with what you have in common, ask about the source instead of attacking the person, give one strong piece of information instead of a list of accusations, and leave the other person room to change their mind without losing face. A public argument in the comments rarely convinces anyone; a one-on-one conversation more often does.

Frequently asked questions (FAQ)

It’s an information environment in which the same views are repeated and dissenting voices are rarely heard. It reinforces the group’s beliefs, because everyone around confirms them.

A filter bubble arises mainly through algorithms that select content. An echo chamber is a broader social phenomenon in which we ourselves choose people and sources that match our views.

A 2015 Science study of about 10 million Facebook users found that users’ own clicks reduced cross-cutting content by about 4 percentage points, and the algorithm by about 1. It’s one study on one platform, so it doesn’t settle the matter.

Usually not. In five experiments with more than 10,100 people, Wood and Porter found no backfire effect. Facts are usually accepted, though they don’t always change the overall judgment.

Views become a badge of belonging, and changing them risks losing your place in the group. On top of that comes confirmation bias, which makes us judge counterarguments more harshly.

If your feed rarely shows reasonable arguments from the other side, and people with other views seem uniform and unreasonable to you, that’s a warning sign. It’s worth reviewing the accounts you follow.

Yes. Demagog describes how disinformation campaigns play on confirmation bias, emotions and stereotypes about out-groups, because this makes false narratives easier to accept.

In our separate article on confirmation bias, which covers the psychological mechanisms, while this text focuses on the social side.

More from our blog

Sources: Nickerson (1998), Confirmation Bias; summary of Lord, Ross, Lepper (1979); Lord, Lepper, Preston (1984); Wood, Porter, The Elusive Backfire Effect; AAAS on the 2015 Science study; Demagog, How not to be manipulated? Cognitive biases. We describe the psychology of confirmation bias in a separate article.

Scroll to Top