Forget about news echo chambers

Most people are well-acquainted with the other side

In a fascinating study out of Oxford University, researchers have found that people who only get their news from highly partisan outlets are small minorities almost everywhere. In short, the “news echo chamber” isn’t really a thing because regardless of political affiliation, most people are exposed to centrist and contrary news sources.

The important exception to this rule is among left-leaning audiences in the US. We’ll get into that below, but first I want to talk about how the study was done, and therefore how far we can rely on its findings.

The study did not include Australia and New Zealand but was based on data that may be available for Australian audiences.

Before we dive into the specifics, I also want to point out an intriguing data offshoot of the study: it classifies different UK and US news outlets on a political spectrum. The results, based on the political makeup of audiences, are uncannily reflective of conventional wisdom on the leanings of the publications. Check out the table from the study below, which shows the major UK news publishers and their “slant” rating: a positive number indicates more right-wing, a negative number more left-wing.

Source: “How Many People Live in Politically Partisan Online News Echo Chambers in Different Countries?”, Journal of Quantitive Description, Vol. 1, 2021.

I encourage you to access the full paper for the US slant ratings, which to me seem equally accurate.

Please subscribe to Crawford Media if you don’t already. Free, weekly media industry analysis.

Just the facts

The paper was published in July in the Journal of Quantitative Description and is based on data from the 2021 Reuters Institute Digital News Report. All the papers’ authors work at the Reuters Institute. No surprises they had the data to hand.

The methodology went something like this:

  • In each news market, survey around 2000 people (representative of the general population) and ask them their political position on a 7-point left/right scale

  • Ask them what news outlets they use

  • Reverse-engineer the political leanings of the news outlets based on their audiences

  • Remove the dead-centre news outlets, then analyse how many people only consume news from subsets of outlets on one side of the political divide.

The definition here of an “echo chamber” is pretty broad: people who are not exposed to any news from outlets outside their broad political allegiance.

The results

The researchers analysed data from Austria, Denmark, Germany, Norway, Spain, the UK and the US. They provide average percentages for the number of people stuck in echo chambers in both the left and right for each country: for example, in Spain, 3% of people are in a left echo chamber, 2% on the right. This total of 5% chambered people is very small, even more so when compared to the more than 20% of Spaniards who don’t consume online news at all.

In every country except the US, the totals were around the same and always less than 10%. In the US, the total average of people in echo chambers was 13% (10% on the left, 3% on the right).

It is important to note that the researchers are not claiming anything about the actual news content or the relative slant of media in different nations. They are only measuring the tendency of audiences to stick to news of their own political leaning.

The American data is very interesting. It cannot be dismissed as a political interpretation, because how the news outlets are characterised is entirely dependent on their own audiences. It could be a function of how many left-leaning news outlets there are in the US (15 above a -.10, compared to 7 in the UK) combined with the lack of a centrist public broadcasting monolith like the BBC.

My interpretation

I have contacted the study’s lead author, Richard Fletcher, and I hope he will help me answer some questions. I do have a vague concern that the finding of very small echo chambers could be a mathematical artefact of the study method, rather than a substantive conclusion. It would be interesting to see an example using the same method where the echo chamber was big, say above 30%. How could that happen in a society? Would it require a degree of homogeneity within audiences and segregation between audiences that is impossible to imagine?

Those concerns aside, the conclusions seem highly plausible to me, and agree with other academic findings that filter bubbles and echo chambers are not as big online as many suppose them to be. What this study indicates is that we will have to go looking for another culprit if our aim is to investigate social and political division.