Exploring Expectancy Violations and Emotions in Computer-Mediated Communication: A Mixed Method, Hybrid Approach to Content Analysis of Online Mental Health Messages

Deanne C. Canieso

Major Professor: Kevin B Wright, PhD, Department of Communication

Committee Members: Xiaoquan Zhao, Gary L. Kreps

Online Location,
April 02, 2021, 11:00 AM to 01:00 PM


The purpose of this research was to explore expectancy violations and emotional features of online mental health messages utilizing a mixed-method, hybrid human coded and computer coded approach to content analysis. Scholarship that explores the genesis and prevalence of mental health illnesses typically investigate cognitive variables such as thwarted belonging, perceived burden, hopelessness, defeat, and entrapment (Witte, Fitzpatrick, & Joiner, 2005; Cornette, Abramson, & Bardone, 2000). However, emotional states have long been implicated in contributing to suicide risk, as well as clinical depression, anxiety, and other mental health-related mood disorders (Dekel, Goldblatt, Keider, Solomon, & Polliack, 2005; Figley, 1995). Moreover, involvement in computer-mediated communication (CMC) platforms encourages both empowering and disempowering processes among individuals coping with mental health illness, and the emotional experience is at the crux of these interactions prompting possible increased engagement. Given that CMC is a novel space to study social interactions, and the role of emotion in communication is still largely unknown in this context, studies investigating emotion message features and its impact on our online interactions is warranted. The research aimed to address two overarching questions: What are the expectancy violation and emotion features of mental health messages shared in computer-mediated communication and 2) How do emotion message features influence social media response behaviors? Mental health narratives disseminated online, and their associated Facebook posts, were analyzed and a series of content analyses studies were done employing Grounded Theory methodology, hierarchical regression, textual sentiment analyses and an unsupervised learning algorithm in the R programming language. An integrated theoretical model guided by the Emotional Broadcaster Theory, Expectancy Violations Theory, and the Elaboration Likelihood Model was used to explore and identify the factors of expectancy violations, emotion, and additional linguistic message variables that prompt social media engagement, thereby promoting information contagion of mental health messages. Results provided insight into the violation experiences of those suffering from mental health illness. Findings additionally enhance our current understanding of the impact of emotion features on computer-mediated communication.