Effects of Pedagogical Agent-Delivered Persuasive Messages with Fear Appeal on Learners' Attitude Change

This dissertation study examined how pedagogical agent's credibility and fear arousing (threatening) message influence affective learning as well as cognitive learning. This study employed a 2 x 3 factorial design with the two independent variables: agent credibility with two levels (less and more) and threatening message with three levels (not threatening, moderately threatening, and strongly threatening). The design specifications for both agent credibility and threatening message were derived from the previous research and the design specifications were validated through a pilot study. A total of 332 undergraduates participated in the study and each of them was randomly assigned into one of the six experimental conditions. After they studied the instruction given according to their assigned experimental condition, they completed the dependent measures. The dependent variable, affective learning, was assessed with the attitude change measure and the dependent variable, cognitive learning, was measured with both recall and application tests. The results found the main and interaction effects for agent credibility and threatening message on cognitive learning. The more credible agents were more effective than the less credible agents in both recall and application test and the not-threatening messages more were effective than the strongly threatening messages in recall test. Regarding the interactions, the moderately threatening messages were more effective in recall test than the strongly threatening message given the less credible agent while there were no significant differences among the three message conditions given the more credible agent. However, no significant main or interaction effects were found in affective learning.