The effects of advising strength on users' trust/distrust beliefs in online recommendation agents : the moderating role of brand familiarity

Abstract

Online recommendation agents are Web-based intelligent consumer decision support systems, which provide consumers with product advice based their needs and preferences. Users' trust and distrust beliefs in recommendation agents influence their use and acceptance of these agents. However, while numerous studies have investigated users' trust beliefs, such as their antecedents and consequences, few of them have focused on their conceptual counterpart: distrust beliefs. This thesis focuses on both of them and conceptualizes users' trust beliefs as their beliefs in the agents’ credibility and benevolence, whereas their distrust beliefs as their beliefs in the agents' discredibility and malevolence. In this thesis, I investigate (1) how the strength of the way a recommendation agent recommends a product (i.e., advising strength) influences users' trust and distrust beliefs while taking into account their familiarity with the recommended product (brand familiarity, in particular); and (2) the distinct and asymmetric effects of trust and distrust beliefs on several behavioral intentions, including users' intentions to reuse recommendation agents, their positive word-of-mouth (WoM) and negative WoM about the recommendation agent. A laboratory experiment was conducted to testify the hypotheses I have proposed, and the results provided support for most of the hypotheses. The effects of advising strength used by a recommendation agent on users' credibility and discredibility beliefs in the agent are moderated by users' familiarity with the brand of the recommended product. In particular, when users are familiar with the brand of the recommended product, the strong advising strength leads to higher users' credibility beliefs (lower discredibility beliefs) than the weak one. Conversely, when users are not familiar with the product brand, the weak strength, compared with the strong one, results in higher users' credibility beliefs (lower discredibility beliefs). However, users' brand familiarity doesn't moderate the effects of the advising strength on their benevolence or malevolence beliefs. Furthermore, the results confirm (1) the stronger influence of users' trust beliefs on their cooperative behaviors (i.e., intention to reuse and positive WoM) than on their punishment behaviors (i.e., negative WoM) and (2) the stronger influence of distrust beliefs on punishment behaviors than on the cooperative behaviors.