The Effects of Social Media Tone on Engagement: Evidence from Indian General Election 2014
Posted: 5 Jun 2017 Last revised: 20 Jul 2018
Date Written: June 4, 2017
Individual celebrities or “human brands” in fields ranging from sports to art to politics use social media platforms to connect and engage with their audience. We analyze the effects of content generated by human brands (referred to as human generated content) in the popular social media platform, Twitter, on audience engagement. The first objective of this study is to examine the effect of tone (positive versus negative) of human generated content on the social media engagement measured by the number of retweets. Next, we investigate the role of popularity of human brand on engagement. Given that human brands are often associated with a group, our third objective is to build on the arguments of perception spillover effects and investigate spillover effects of the content generated human brands belonging to the same group as the focal human brand on the engagement of the focal human brand. Set in the context of Indian general election 2014, our analysis based on a mixed effects model that accounts for user and group level heterogeneity reveals that the tone of social media content created by human brands significantly affects audience engagement. Specifically, we find that the effect of negative toned content on social media engagement is significantly positive. In contrast, we find that the effect of positive toned content on engagement is greater only for popular candidates. Furthermore, our results also suggest that the popularity of human brand negatively moderates the relationship between negative tone content and engagement. Our findings provide evidence of interesting spillover effects of content created by related brands on the effectiveness of content created by a focal human brand. We find that whereas the effect of negative toned tweets of a focal human brand is lower when the tone of related human brands is positive, the effect of negative tone on engagement is greater when the tone of related human brands is negative. We supplement our core results with additional analyses that include alternative model specifications, estimation strategies, engagement metrics, and models that account for selection bias. Finally, we present implications of this study to practice as well as the literature on social media content.
Keywords: human brand generated content, social media analytics, engagement, tone of social media content
Suggested Citation: Suggested Citation