Exploring intention to reuse recommendation agents from accessibility-diagnosticity perspective: The moderating effect of domain knowledge

Recommendation agents help users reduce information overload and improve decision quality. Yet, many online shoppers have negative reaction or have no motivation to use recommendation agents, since they have no idea of whether users can achieve their shopping goals with less effort. We think information is fundamental to using recommendation agents. This study develops a research framework from the accessibility-diagnosticity perspective and proposes explanation facility, perceived similarity and information diagnosticity are important determinants of users' intention to reuse RAs. We think explanation facility could persuade users of RAs' performance, similarity could move users to agree with RAs, and information diagnosticity could let users be capable of evaluating RAs. We also consider the moderating role of domain knowledge on relationship of similarity and information diagnosticity. This study conducted a 2^*2 factorial experiment for data collection. Results show that decision process and outcome similarity indirectly influence reuse intention by information diagnosticity and the effects of process and outcome similarity varies with degrees of users' domain knowledge. The influence of explanation facility on similarity is not obvious. The effect of "why" explanation facility on outcome explanation is significantly contrary to our expectation. Explanation facility may have to be utilized carefully. Implications are discussed.