Partisan misinformation on the Internet is at least partially a chicken and egg problem. On the one hand, there is the offer, which comes in part from politically motivated websites that are indistinguishable from propaganda, but also from some who only try to make money by earning clicks with the wrong headlines. But there is also a requirement to consider – these people are insatiable to quench their hunger for biased outrage. To understand the extent of the problem, it is helpful to consider both sides of this relationship. What misinformation is floating around and who is consuming it?
In a new study, Andy Guess, Brendan Nyhan and Jason Reifler used survey data to track the web histories of around 2,500 people the month before the 2016 US election. Combined with some demographic survey data on things like their favorite presidential candidate, the researchers were able to find out who read which articles.
Who publishes what?
The researchers relied on the list of “untrustworthy” locations from a previous study. This included several hundred that could well be described as falsification, but also more than one hundred that violated fact-checkers and were determined not to have editorial standards. This list includes conspiracy sites like InfoWars and Natural News, non-partisan sites like Ben Shapiros Daily Wire, and even some tabloids like The Express.
Overall, almost half of the participants visited at least one article from a website on this list during the study period. However, these articles made up only 6 percent of all messages read. However, these numbers were not evenly distributed across the political spectrum. About 57 percent of Trump supporters in the group visited an untrustworthy site at least once, which is about 11 percent of total message consumption. For the Clinton supporters in the group, 28 percent of visitors to at least one article corresponded to 1 percent of their total news consumption.
If you delve deeper into the data, a relatively small group of people is responsible for most visits to untrustworthy websites. The researchers categorized people according to the ideological focus of their “news diet”, from those whose reading is dominated by liberal websites to those who only read conservative websites. The 20 percent of people furthest from the conservative end of the spectrum made up almost two-thirds of the untrustworthy items.
A variety of sources
But even in this group, junk food misinformation didn't necessarily dominate the diet. It made up about 20 percent of their news recording. This is partly because people who read most news articles from more reputable websites – the most insatiable news consumers – are also most likely to have found at least one untrustworthy article.
Enlarge /. On the left, the proportion of people who have visited at least one untrustworthy article, broken down by the liberality or conservatism of their entire media diet. On the right is the number of untrustworthy articles in your media diet.
The next obvious question is whether people can find these untrustworthy websites through Facebook or another platform. The study was only able to assess this in a slightly cumbersome manner by checking whether Facebook, Twitter, Google or web-based emails are displayed directly in front of the desired URL in the browser history. Not surprisingly, Facebook appeared most frequently in about 15 percent of the cases. Webmail came in second with 10 percent, while Google and Twitter were under 5 percent.
The researchers also looked for articles to verify facts in these browser histories and found that about a quarter of people visited at least one. This number includes about half of those who have visited untrustworthy articles, which means that they are more likely to check their facts than a general reader. At least some of their facts. Of the 111 people who had read an article from a list that the researchers knew had been fact-checked, only three had read the relevant fact-checking.
The researchers say that this provides more evidence that fact checks largely haven't reached the people who need to see them the most. However, it is worth noting that this data lies ahead of many efforts by platforms like Facebook to highlight fact-checking for users who interact with an article that has been mis-rated.
Overall, the researchers concluded that "widespread speculation about the proliferation of untrustworthy websites was overrated". Of course, not everything is recorded in their data set, such as content that is only displayed on Facebook, or the impact of misinformation on the broader information ecosystem. But it is a unique study that supports what others have found – a relatively small proportion of the public consume a lot of what the researchers call “factually dubious content”.
Nature Human Behavior, 2020. DOI: 10.1038 / s41562-020-0833-x (About DOIs).