The purpose of this dissertation was to critically examine differences in risk perceptions among experts and lay people. In particular, this project aimed to address inconsistent definitions of "expert" found in the existing literature and to test the predictions of the psychometric paradigm in the context of communication. To examine the effect of message features and expertise on risk perceptions and evaluations of risk characteristics, this dissertation employed a 2 (emotional appeal: fear, anger) x 2 (message topic: nuclear energy, traffic accidents) x 4 (expertise: general risk assessors, traffic safety experts, nuclear energy experts, lay people) between-participants design.

The results replicated some findings of the existing research. First, in the main, experts reported lower risk perceptions than lay people. Second, expressed fear led to increased risk perceptions compared to expressed anger. This study also advanced theory regarding risk perception and risk communication in two critical ways. First, differences were found not only between experts and lay people but also among the various expert groups, and, even in the expert groups, these differences were influenced in meaningful ways by the messages viewed. Second, this study demonstrated the potential for messages to affect not only risk perceptions but also the evaluation of risk characteristics, a possibility not previously tested.

Specifically, the findings indicated that emotional appeals and message topic can affect evaluations of risk characteristics for risks both related to the message and unrelated to the message. The messages' effects on evaluations of risk characteristics were, in fact, more pronounced than the effects of the messages on general risk perceptions. The results suggest the factors argued to be predictive of risk perception (dread risk and knowledge risk), presented previously as inherent characteristics of risks rather than as targets for influence, can be altered through strategic communication. Both theoretical and applied implications of these results are discussed, and recommendations for future research are provided.