Value Relevance of Social Media Sentiment: An Experiment with Steem using Association Analyses

Abstract

Most popular social media platforms support a voting mechanism to capture an assessment of how much the network values a particular post, where more votes or “likes” implies greater valuation among network participants. Such voting mechanisms are subject to confounding factors, such as relative popularity of the poster, as well as outright acts of manipulation to increase vote counts. We hypothesize that post sentiment plays a role in content valuation. We expect that participants will value posts with positive sentiment more than posts with negative sentiment. Further, we postulate that the degree of positive sentiment matters, such posts with a lesser degree of positive sentiment will be more highly valued than posts demonstrating a greater degree of positivity. We base our hypothesis on theories associated with risk aversion, where users are more interested in content that may signal a need to act to avoid potential negative consequences.