Abstract

In order to test the three-level theory of attitude change in an extended area of application, the effects of source expertise and message quality (Experiment 1) and those of message quantity and quality (Experiment 2) on the changes and cognitive responses of job-relevant attitudes with low, moderate, and high involvement were examined by factorial design experiments. Except for an unexpectedly strong persuasive power of non-experts, most results of both the experiments consistently supported the hypotheses: As predicted by the three-level theory, (1) effects of message quantity as evidence of peripheral route processing were the largest in attitudes with low involvement; (2) those of message quality as evidence of central route processing were the largest in attitudes with moderate involvement; and (3) attitudes with high intrinsic involvement showed resistance to persuasion even under the combination of high quantity and quality of messages. As postulated in the three-level theory of attitude change, high involvement mentioned in the elaboration likelihood model (ELM) is in fact only a moderate level of involvement, and the resistance process of a real intrinsic high involvement was proven as an active biased central route processing, which is different from a simple peripheral route processing. Findings on cognitive response measures also supported the three level theory's predictions, so that indifference appeals were relatively more produced by attitudes with low involvement, pro-message responses were more by those with moderate involvement (expecially when given high quality message), and counter-message responses were relatively more generated by those with high intrinsic involvement.