Partisan Sorting in News Media Consumption: Yes, Actually, the US Is an Exception

The next speaker in this session at I-POLHYS 2024 is Ana Sofia Cardinal, and her interest is in (news) partisan sorting. This builds on digital trace data from the Web browsing practices of Internet users in several European countries and the US. This work is important given the suspected increase in political polarisation, the decrease in trust in the media, and the rise of far-right parties in several countries.

All of this is happening against the backdrop of a high-choice media environment characterised by an increasing number of partisan media; growing opportunities to personalise media diets; and greater partisan selectivity – and yet we also do not see any meaningful evidence for the emergence of echo chambers or filter bubbles. What are the patterns of partisan sorting in news consumption across France, Germany, Spain, the UK, and the US, then?

The project builds on several months of Web tracking data from late February to mid-June 2022 (which takes in the full-scale Russian attack on Ukraine, two rounds of presidential elections in France, and various other national and global political events). From these data it is able to extract information on news consumption, which is then combined with a media slant score for each individual ness outlet, and a selective exposure score for each individual study participant.

Media outlet lists take the 50 most popular online news outlets in each country (as assessed by ComScore) and combine them with the most visited news outlets in the dataset); the political slant of each outlet is then assessed and scored based on the ideological composition of its audience (as based on participant self-reporting of their political positioning, and using a weighted average based on the frequency of use), and the score represents the distance of each outlet from the midpoint for a given country. As with the previous presentation, the breadth of the spectrum is a great deal smaller in Germany than in the US, for example.

Finally, each user’s selective exposure score is calculated from the weighted media slant average of the media they consumed, divided by number of visits (hope I got this right) – and notably, only the US displays a clearly bimodal distribution where left- and right-wing audiences exhibit clearly divergent media diets. In the other three countries, the majority of users are located somewhere towards the centre, independent of their personal political preferences. This means that outside the US there is very little partisan sorting; in the US it is especially right-wing users who construct a very different media diet for themselves, and that media diet also correlates with substantially lower media trust.

This means either that the US is simply exceptional in its right-wing media ecosystem and its implications, or that the US is ahead of global developments and such a bimodal partisan sorting has yet to occur in other major countries. Also, of course, there may be media use practices that this approach to digital trace data from Web browsing might simply not capture (e.g. in-app exposure, etc.) – but it is difficult to see how else we might study such patterns at scale and in a timely fashion.