Few Americans are deeply invested in consuming so-called fake news, Jacob L. Nelson, of Northwestern University, and Harsh Taneja, of University of Illinois Urbana-Champaign, write. The authors analysed the browsing data collected from one million Americans before and after the 2016 presidential elections.

Nelson and Taneja compared online traffic to established news websites (e.g. New York Times and Fox News) and fake news websites. They used a list of fake news websites maintained by the OpenSources project, according to which fake news are “sources that entirely fabricate information, disseminate deceptive content, or grossly distort actual news reports”.

OpenSources lists 126 fake news websites, but only 30 were visited frequently enough to be registered in the browsing data. In addition to these 30 fake news websites, the authors’ sample included 24 mainstream news websites. Comparison of the two groups shows that mainstream media websites attracted on average 28 million unique visitors per month. For fake news websites, this number was 675 000 visitors.

Mainstream news also appear to be more engaging. Readers spend on average twice as much time on real news websites than on fake news websites. Fake news readers are not very loyal, either: most visitors to fake news websites also visit real news websites. For example, over half of fake news readers visited Yahoo-ABC alone.

The use of fake news appears to correlate with overall internet use: heavy internet users consume more of everything, even fake news, the authors suggest. The problem, then, is not that the audience would be locked into “echo chambers”, but that a part of the audience is unswayed by real news. “This suggests that real news is failing to function as a corrective for the misperceptions perpetuated by fake news”, Nelson and Taneja conclude.

The article “The role of audience availability in fake news consumption” was published by the journal New Media & Society. It is available online on the publisher’s website (abstract free). A draft version of the article is available from the Social Science Research Network’s eLibrary (open access). More information on the OpenSources project is available from opensources.co.

Picture: Untitled by Pexels, licence CC0 1.0.