This dissertation investigates on how online social interactions and new ways of engagement that are all made available to users by the design of social media platforms impact the engagement behavior of users. This dissertation is mainly consisted of three essays, each one of which investigates a different aspect of the engagement dynamics and contributes to both theory and practice.

The first essay studies the impact of different ways of message framing on users’ engagement behavior regarding the brand posts on social media brand communities and to determine how users’ thumbs-up and reply moderates this impact. A panel data analysis was conducted on a panel with 11,894 observations on 850 unique brand posts from the Facebook pages of the world’s most valuable brands in Technology sector over a seven days window with two observations in each day. A system of equations was estimated using Ordinary Least Squares, Hausman-Taylor IV, and Seemingly Unrelated Regressions to test the proposed hypotheses. The empirical findings confirm that more positively and negatively-framed comments result in increased users’ engagement. Also, an increase in thumbs-up ratio for neutrally and negatively-framed comments results in less engagement. The reply ratio might also have a positive and negative moderation effect on the influence of neutrally and positively-framed comments on engagement behavior, respectively. This is likely to be the first study that examines the moderating effect of users’ thumbs-up and reply on the relationship between message framing and user engagement behavior.

The second essay studies the impact of consumers’ emotional expressions on user engagement behavior regarding the brand posts on social media brand communities. A panel data analysis was conducted on a panel with 65,941 observations on 4700 unique brand posts from Facebook pages of the world’s most valuable brands over a seven days window with two observations in each day. A semi-supervised affect analysis method is applied to mine the consumers’ comments to identify their affective state. A system of equations was estimated using Ordinary Least Squares, Hausman-Taylor IV, and Seemingly Unrelated Regressions to test the proposed hypotheses. The empirical findings confirm that six basic affective states including anger, disgust, fear, joy, sadness, and surprise that are embedded in consumers’ comments can potentially affect engagement behaviors such as liking, commenting, and sharing. This study provides an in-depth understanding of emotional expressions via WOM communications on Facebook pages.

Finally, the third essay studies the impact of a multimodal emotional reaction system that Facebook implemented on February 24, 2016 (a.k.a. Facebook Reactions) on the engagement behavior of users. Although, there are numerous views claiming about the increased engagement of users on Facebook after implementation of Facebook’s multimodal emotional reaction system, there is yet no scientific evidence to prove it. This study argues on why this case is an instance of “feature fatigue” phenomenon and how it has influenced user engagement behavior on Facebook. Using data from several active news outlets on Facebook and Twitter, and relying on the theoretical lens of “feature fatigue,” this study examines the impact of Facebook’s multimodal emotional reactions system on user engagement behavior. Findings of the quasi-experimental investigation show that users tend to engage less with news article related posts due to a feature fatigue caused by implementation of the Facebook’s multimodal reactions system.