Does Being an Expert Make You More Negative? An Investigation of Subjective Expertise and Electronic Word-of-Mouth Communication

This dissertation is aimed at filling a gap in electronic word-of-mouth (eWOM) literature by examining how communicator’s subjective expertise impacts generation of eWOM. Based on a review of literature in psychology, marketing, consumer research, and communication, four underlying mechanisms (sense of power, dogmatic cognition style, hubristic pride, and emotions) were proposed to explain the effect of subjective expertise. Two distinctive features of eWOM platforms were incorporated into the study as potential moderators: anonymity and audience size of the review platform. Based on prior literature, this study proposed 14 hypotheses in total and 2 research questions. Two empirical studies were conducted. Study 1 was a field observation of eWOM communication in the natural setting. A sample of 470 Yelp users and 39,091 reviews produced by these sampled users were collected using the web crawling technique. Study 2 was a 2 (subjective expertise: high vs low) 2 (anonymity: anonymous vs real identity) 2 (audience size: large vs small) between-subjects experiment conducted online. Two hundred and sixty-two effective responses were collected via Amazon Mechanical Turk. Results of Study 1 revealed a negative correlation between eWOM valence (rating at the review level and aggregated rating at the user level) and length of expertise status. Results of Study 2 revealed a two-way interaction between subjective expertise and anonymity on numeric rating, and number of negative thoughts mentioned in one textual review. Moreover, results of conditional process modeling demonstrated that sense of power mediated such moderation effects. Specifically, a negative indirect effect of subjective expertise on eWOM valence through sense of power was discovered. Additionally, two dimensions of emotions (valence and dominance) mediated the effect of subjective expertise on eWOM valence, which was not influenced by either audience size or anonymity. Findings of this research provide theoretical contributions to eWOM research by exploring the influences of communicator characteristics and platform characteristics on eWOM generation.