The Challenges of Understanding Content Dissemination on Facebook

The final speakers in this Digital Methods plenary are Axel Maireder and Katrin Jungnickel, whose interest is in the uncertainties of the Facebook timeline. Facebook has continued to tinker with how the timeline is selected and presented for several years now, and this affects the flow of communication on the platform; what, then, are the factors which determine that flow?

This study combined content analysis and user surveys, but both these approaches have their drawbacks - it is impossible from the outside to track the content of users' timelines, for example, but surveys of users also suffer from self-reporting biases. In the end, the researchers asked users to copy the links they received through their timelines into an online survey, and to discuss the content of the URLs and the Facebook friends they received them from. Issues with privacy as well as the tedious nature of this approach also affect the results, however. Some 550 users participated in the study.

The categorisation of the links was more complex than expected, however. There is wide variation in the type of content being shared, and the specific content being shared is sometimes difficult to pinpoint if a shared URL contains multiple types of material (images, videos, audio, text, advertising, ...).

Other material (for example, links to TV livestreams) are also time-dependent - the content at the time of sharing may be entirely different from the content at the time of analysis. In the case of re-shared material, it also becomes difficult to determine who the producer of the content actually is - e.g. the musician performing in a music video, the TV station which aired it, or the YouTube user who shared the clip.

The same goes for the theme of the link. This may be situationally dependent: a consumer advisory could be shared because of its news value, for example, or for commercial or political reasons. And of course some links expire, or may be geo-blocked, so that an Austrian coder cannot access a music video shared in Germany.

This raises considerable methodological challenges, then. Nonetheless, some useful patterns can emerge from the data - mainstream media sources still remain important, for example, but ordinary users are key to the dissemination of links on Facebook. This suggests the existence of a multi-step flow across the site - but this can go in both directions, as some mainstream actors also share user-generated content links. Links shared by strong-tie contacts generated greater interest in the content shared, too.