Examining the consumption of radical content on YouTube

The Computational Social Sciences Lab’s (CSSLab) new dashboard, Mapping the (Political) Information Ecosystem, is a set of four data visualizations that highlight Americans’ media consumption habits, with a focus on echo chambers and the news. This is the second of a series of dashboards launched by the CSSLab, as part of its PennMAP (Penn Media Accountability) project, and focuses on expanding knowledge on the media by translating research papers into interactive, digestible content.

The dashboard is an extension of two research papers: “Evaluating the fake news problem at the scale of the information ecosystem” published in 2020 (with charts through 2018) and “Quantifying partisan news diets in Web and TV audiences” published in 2022 (with charts through 2019), both in Science Advances. The papers describe how misinformation and highly partisan political content fit into the broader context of media consumption.

This dashboard allows us to extend the time period of the charts to the present as well as allow for exploration of dimensions not available in the static version of the papers. To illustrate the usefulness of the dashboard, this post describes some findings that we were able to infer easily with the dashboard’s functionality. For these studies, the research team analyzed the television and online news consumption of thousands of Americans from 2016 to 2019 using data from Nielsen, whose representative panels measure the media habits of American households.

Overall News Consumption

Declining TV numbers: A Generational Divide

One of the more surprising findings from the original paper in 2020 was that TV was a more common mode for engaging with the news media than the internet, not only for older people, but also for younger people. While it is still true that younger people spend very little time with online news, the finding is less significant today because of a mass exodus of younger people from television. 

Since 2016, Americans younger than 45 cut their television viewing (not including streaming) by 75%.

Older Americans, on the other hand, still watch a lot of TV, with their total consumption declining by a relatively modest 25% in the past eight years. The average person 55 and older still spent more than 200 minutes per day watching TV in 2023.

Older Americans Still Dominate News Consumption

As one might expect, Americans 55 and older spend substantially more time than younger audiences engaging with the traditional news media on TV, desktop, and mobile devices. On TV, where the difference is the greatest, the oldest Americans spend 50 minutes per day watching national news (not counting local news) compared to only three minutes per day for people ages 18-24.   

Fake news makes up less than 1% of overall news consumption 

The main finding from the 2020 paper was that “fake news,” defined at the source level as demonstrably fake or hyper-partisan news websites, was a negligible part of news consumption. This is still the case, to the point that fake news is barely visible on the graph.

News Spikes with Community-related Events

The dashboard can show how news consumption corresponds to current events and specifically which groups are represented in the news. For example, in the middle of 2020 there was an especially prominent (greater than other groups) spike in TV news consumption amongst Asians. This increase may be linked to the early stages of the COVID-19 pandemic and media coverage of a surge in Anti-Asian hate crimes, suggesting that people are more likely to closely follow the news if it covers events which impact their communities.

Partisan Echo Chambers

Cable TV Fuels Partisan Echo Chambers

To get a better understanding of the problem of echo chambers, we analyzed media consumption in terms of partisan segregation. Here partisan segregation is the tendency for people to get the majority of their news content from partisan sources. Though this kind of thing is usually associated with echo chambers on social media, we see that cable TV plays a more significant role. Between 10-15% of Americans aged 55 and above get a majority of their news on TV from Fox News, with overall partisan segregation increasing after the 2020 US Presidential Election. However, only 5% of this age group experience online partisan segregation.

Trump Appears to Drive Engagement for Liberals

On another note, many speculated that Trump was driving ratings for the partisan cable news channels, discussing how engagement with those channels would decrease after he left office. We see evidence of a post-Trump dip in TV echo chamber (cable news) viewership, but it is much more pronounced on the liberal channels than on the right.

Black people do not watch fox news 

It is striking how few Black people are in the Fox News bubble: practically none. Although there recent evidence suggest that Black Americans’ historical allegiance to the Democratic Party is weakening, that trend has clearly not translated to any affinity for the conservative cable giant. This finding persists across all age groups. On the flip side, older White people are especially likely to be in the Fox News pubble and older Black people are especially likely to be in the liberal cable news bubble.

 

Echo Chambers Map– In a view that was unexplored in our original research, this map breaks down partisan segregation by state-by-state. The map allows users to examine geographic trends and if you click a state, timelines of partisan segregation on the state level appear.

TV Echo Chambers: more red states than blue states

In every state there is at least a modest rate of partisan segregation (>1%) for both right-wing and liberal cable news, but only 14/51 states (including DC) have more people in the liberal cable bubble; the remaining 37 have more people in the right-wing cable bubble. 

If we analyze state-level partisan segregation by the fraction of the population, as opposed to the overall count, the Washington DC Metropolitan area is the most liberal region in the US, whereas West Virginia has the highest rate of individuals in a right-leaning bubble.

Web Echo Chambers: more blue states than red states 

When switching from the TV to the Web view, there is a significant shift in the trend: 26 states lean more left.

Changing TV Audiences– The final figure is a network graph showing the evolution of partisan audiences on TV over time (2016-2024). Nodes represent “archetypes,” or people whose TV news consumption is dominated by a particular category of news (plus a “minimal news” group). The arrows between nodes show the net flow of people who have changed their behavior, over a given period of time, from one archetype to another.

TV Viewership is Declining and Becoming More Polarized 

A stunning fact is that since 2016, about 83 million people have stopped watching news on TV almost entirely (our threshold is 30 min per month, which is roughly one minute per day). Moreover, the Americans that continue to get news from TV are increasingly polarized and concentrated in the MSNBC and Fox News archetypes. That is, while the biggest trend visible in this figure is movement away from all archetypes to the “minimal news” category, the second biggest trend corresponds to people that had been getting news from classic broadcast news or Spanish language channels but have switched to MSNBC and Fox News. 

The dashboard’s data will automatically update every month, allowing people to explore and find new things that cannot be determined from static publications. We invite anyone interested in these topics of media and politics to interact with the data and form their own conclusions. 

This dashboard was built by: Baird Howland, Duncan J. Watts, David M. Rothschild, Yuxuan Zhang, Felippe Rodrigues, and Delphine Gardiner 

Disclaimer: All aggregation and analysis methods were conducted by the CSS Lab. Nielsen does not provide ideological labels to TV programs nor does it define what programs are considered right-wing or left-wing. The conclusions drawn from the Nielsen data are those of the research team and do not reflect the views of Nielsen.

 

AUTHORS

DELPHINE GARDINER

Delphine Gardiner

Communications Specialist