Are tweets biased by audience? An analysis from the view of topic diversity

Abstract

The emergence of blogs, and especially microblogs, has granted users the possibility of publishing and sharing ideas, news, opinions and any other kind of content with their audience. But this has also brought them the arduous tasks of self-censorship and adaptation of the content to an audience previously envisioned in order to keep, and even increase, their social influence. Taking into account the impossibility of knowing this imagined audience and using Twitter as a case study, we analyse if the diversity of topics chosen by users in their tweets is biased by the size of their audience. Considering the number of followers as the users’ audience and applying a methodology based on clustering the representative terms in tweets, we found that individuals with large audiences tend to deal with topics more diverse than those with small audiences. Understanding how audience size affects the range of topics chosen by a speaker have theoretical implications for sociological studies and even for the effective design of marketing campaigns.