||Recent research has identified a positivity effect in consumers’ evaluations of agents. When evaluating an agent as a potential source of advice, consumers often rely on the extent to which they have agreed with the agent in the past (West, 1996; Gershoff and Johar, 2006). The agreements on loved alternatives are considered to there are be more informative by consumers than the agreements on hated alternatives, when evaluating similarity in taste with the agent, and suitability of the agent as a source of future advice. In this study, we examine whether the positivity effect is affected by the novel attribute, number of alternatives, and the revelation of the agent evaluation. Then, we further examine the addition of involvement as a moderator.
This basic proposition is tested in two pilot studies and three main studies. The pretests determine involvement (high: digital camera, low: tea) and novel attributes (digital camera: GPS, tea: Gymnema Sylvestre). In the main studies, the positivity effect is moderated by the novel attributes, number of alternatives, and revelation of the agent evaluation. Experience 1 is a 2 (agreement valence: like; hate) X 2(novel attributes: absence; presence) X2(involvement: high; low) between-subjects experimental design. Experience 2 is a 2(agreement valence: like; hate) X 2(number of alternatives: low; high) X 2(involvement: high; low) between-subjects experimental design. Experience 3 is a 2(agreement valence: like; hate) X 2(revelation of the agent evaluation: not revealed; revealed) X2(involvement: high; low) between-subjects experimental design. The subject are 240 students in college. This study found that that the positivity effect is attenuated when the novel attribute is present, number of alternatives is high, and the revelation of the agent evaluation. Moreover, we showed that the positivity effect will be different from the degree of involvements.