How much political news do people see on Facebook? I went inside 173 people’s feeds to find out
What do people see in their Facebook feeds? How much news do they encounter there — from legitimate outlets or from those known for sowing misinformation? Are most people’s accounts cesspools of fake news and garbage from disreputable sites, or do they see stories from Facebook’s trusted news partners?
Or…do they mostly see innocuous memes, baby pictures, and cute animal stories?
Debate over these questions has raged for years, especially in the weeks leading up to the highly fraught 2020 U.S. presidential election. Answers depend on who you ask and what tools you use.
Look at daily engagement data from the Facebook-owned CrowdTangle, which measures the reactions, shares, and comments a post gets, and you’ll see lists dominated by Donald Trump and far-right pundits.
Facebook claims its internal data paints a different picture. The company says that the posts with the highest engagement aren’t actually the ones that reach the most people, and that the average Facebook user sees little political content. But since Facebook doesn’t make its reach data public, researchers who want to get a sense of what the average Facebook user sees are generally forced to rely on anecdotes, company statements, or CrowdTangle.
I decided to approach the question in a different way — by peeking directly into people’s Facebook feeds. Between October 1 and 31, 2020, I surveyed the Facebook habits of, and got real News Feed samples from, 306 people aged 18 or older in the United States. I reached them using Amazon’s Mechanical Turk platform, asking them to send me screenshots of the first 10 posts in their Facebook feeds. (I used the images for classification purposes only. No identifying information is referenced in this story.) After cleaning the data and removing entries that were submitted incorrectly, I had data from 173 people — a total sample of 1,730 Facebook posts. (See more in the methodology section at the bottom of this post. You can see my spreadsheet here.)
In doing this project, I built on Nieman Lab research from three years ago, conducted by my then-colleague Shan Wang, who is now a senior editor at The Atlantic. (We miss you, Shan!) At the time, Shan found that people saw surprisingly little news in their Facebook feeds. Three years later, in one of the craziest news months in U.S. history, my findings were similar: People saw surprisingly little news in their Facebook feeds.
More than half the people in our survey saw no news at all.
October — which kicked off with Donald Trump revealing that he and the First Lady had contracted coronavirus — was a nonstop news month in the most relentless news year in decades. If there were ever a month that you’d expect people to see a lot of news at the top of their Facebook feeds, it would be October 2020.
Nope. Even using a very generous definition of news (“Guy rollerblades with 75-pound dog on his back“), the majority of people in our survey (54%) saw no news within the first 10 posts in their feeds at all. In Shan’s 2017 survey, that figure was 50%.
79 people — or 46% of the sample — saw at least one news article. Hard news — which I defined as serious and important stories about national politics, Covid-19, and so on — was the most common type in my sample. In total, out of the 1,730 posts I looked at, 178 were news stories; of those, I classified 138 as hard news. Of the people who encountered any news at all, 81% came across at least one hard news story. The other 19% saw only soft news (weather, elephants smashing pumpkins, “Irish court rules subway bread is not bread“).
National politics and Covid-19 were the most common news topics.
There were lots of stories about national politics or Covid-19 (or, sometimes, the intersection between the two). Trump’s Covid-19 diagnosis and its aftermath was a huge story that I coded separately from other news stories about Covid-19: 17% of the hard news articles in my sample were about Trump contracting Covid. (I put stories about other politicians contracting Covid-19 in the “general” Covid-19 category.)
What does this mean for the engagement vs. reach debate?
In recent months Kevin Roose, a technology reporter for The New York Times, has consistently used CrowdTangle and NewsWhip data to assert that Facebook is teeming with misinformation, based on high engagement figures (likes, shares, etc.) for news stories from partisan outlets. He runs a Twitter account, Facebook’s Top 10, that each day provides a list of the 10 top-performing link posts by U.S. Facebook pages. The top-10 list is almost invariably filled with posts from far-right sites and politicians, including conspiracy-driven sites like Gateway Pundit and Infowars.