Using Information Processing Strategies to Predict Message Level Contagion in Social Media

Abstract

Social media content can have extensive online influence [1], but assessing offline influence using online behavior is challenging. Cognitive information processing strategies offer a potential way to code online behavior that may be more predictive of offline preferences, beliefs, and behavior than counting retweets or likes. In this study, we employ information processing strategies, particularly depth of processing, to assess message-level influence. Tweets from the Charlottesville protest in August 2017 were extracted with favorite count, retweet count, quote count, and reply count for each tweet. We present methods and formulae that incorporate favorite counts, retweet counts, quote counts, and reply counts in accordance with depth of processing theory to assess message-level contagion. Tests assessing the association between our message-level depth of processing estimates and user-level influence indicate that our formula are significantly associated with user level influence, while traditional methods using likes and retweet counts are less so.