Brand Messages On Twitter: Predicting Diffusion With Textual Characteristics

Brand Messages On Twitter: Predicting Diffusion With Textual Characteristics Public Deposited

Last Modified

Creator

Abstract

This dissertation assesses brand messages (i.e. tweets by a brand) on Twitter and the characteristics that predict the amount of engagement (a.k.a. interaction) a tweet receives. Attention is given to theories that speak to characteristics observable in text and how those characteristics affect retweet and favorite counts. Three key concepts include sentiment, arousal and concreteness. For positive sentiment, messages appeared overly positive, but still a small amount of the variance in favorites was explained. Very few tweets had strong levels of arousal, but positive arousal still explained a small amount of the variance in retweet counts. Despite research suggesting that concreteness would boost sharing and interest, concrete tweets were retweeted and shared less than vague tweets. Vagueness explained a small amount of the variance in retweet and favorite counts. The presence of hashtags and images boosted retweet and favorite counts, and also explained variance. Finally, characteristics of the brand itself (e.g. the number of followers the brand had, the number of users it followed and its overall reputation of the brand) boosted retweet and favorite counts, and also explained variance.

Date of publication