Testing the limits of optimistic bias: event and person moderators in a multilevel framework.

N. D. Weinstein (1980) established that optimistic bias, the tendency to see others as more vulnerable to risks than the self, varies across types of event. Subsequently, researchers have documented that this phenomenon, also known as comparative optimism, also varies across types of people. The authors integrate hypotheses originally advanced by Weinstein concerning event-characteristic moderators with later arguments that such optimism may be restricted to certain subgroups. Using multilevel modeling over 7 samples (N = 1,436), the authors found that some degree of comparative optimism was present for virtually all individuals and events. Holding other variables constant, higher perceived frequency and severity were associated with less comparative optimism, higher perceived controllability and stereotype salience with more comparative optimism. Frequency, controllability, and severity were associated more with self-risk than with average-other risk, whereas stereotype salience was associated more with average-other risk than with self-risk. Individual differences also mattered: comparative optimism was related negatively to anxiety and positively to defensiveness and self-esteem. Interaction results imply that both individual differences and event characteristics should jointly be considered in understanding optimistic bias (or comparative optimism) and its application to risk communication.