Bergendahl Olle and Rehn Andreas

Political filtering on Facebook?

ABSTRACT

In this bachelor degree thesis we’ve conducted an experiment on Facebook using fictive persons, also called personas. The experiment is based on the theory of the filter bubble and the filtering of information. Our main focus has been on answering the question “Does Facebook filter the information that reaches a user, based on who the user is, what political party the user sympathizes with, and what kind of friends the user has?”. Questions we’ve also brought fourth are: Can information filtering be used in a political way? What effects could filtering have on a user? What effects could filtering have on the society in general?

In the experiment we’ve created 17 personas with a strong political profile, separated into one left wing party and one right wing party. Thereafter the personas have been interacting with each other during a period of 16 days. At the end of the experiment we’ve extracted information regarding which posts have been visible and which posts that have been filtered out. We’ve also used the script Facebook Friend Rankings to further develop an understanding of the social connections amongst our personas.

Our results show that there - in an artificial environment - exists a political filter bubble on Facebook. A persona with left wing opinions is shown less posts from his right wing friends than from friends that share his political beliefs. Although the results are quite clear, it’s uncertain whether or not the results can be applied on a more general level. Further studies needs to be done, especially regarding the relationship between political filtering and the EdgeRank algorithm.

Keywords: Facebook, filter bubble, Eli Pariser, EdgeRank, Facebook Friend Rankings, censorship, political censorship, social connections, personas, Jeremy Keeshin,