To manipulate participation in the CRP, the respondents assigned to the referral condition read a scenario that described them giving a recommendation: while browsing TelStar's webpage, they were prompted to think of their friend Alex, who had just mentioned a desire for a new cellular provider. Because vividness contributes to the impact of recommendations on product judgments (Mangold, Miller, and Brockway 1999), we strove to make the recommendation articulation tangible and vivid. We asked participants to verbalize their recommendation in a text box (Appendix, Condition 2a), and all participants did, using sample recommendations such as, "I recommend TelStar because I am very satisfied with the service. They are friendly and the coverage is also good," "I recommend TelStar because I have been there a long time and never had any trouble," and "I recommend TelStar because they are cheap and good." The recommendations were similar across participants and typically consisted of one or two sentences. The scenario did not specify whether their CRP participation was successful (i.e., if the noncustomer made a pur-chase). Participants in the no-referral condition read that they would have recommended the service provider but could not think of anybody looking for a new cellular telecommunication provider (Appendix, Condition 2b). This information was explicit, so any differences in the dependent loyalty measure between groups cannot be attributed to the additional attitudinal information for the experimental group (Aronson et al. 1990).