It is interesting that most research on unrealistic optimism has used the average person as the comparison standard (with the reference group to which this average person belongs varying widely among different studies). This approach has been appropriate because it is far easier in health campaigns to present information about others or "the average person" than about the numerous, separate individuals to whom the message is directed. However, if it turns out that the self is normally used as a standard for judging comparative risk, past research may have exaggerated the magnitude of optimistic biases by artificially changing the standard that people use.
We set out in this article to show how several social comparison factors -- direction of comparison; similarity, specificity, concreteness, and individuation of the target; attention toward own rather than others" behaviors; and standard of comparison -- influence the degree to which people are unrealistically optimistic. We reviewed some evidence that, consistent with some elements of social comparison theory, people choose to compare their risk with that of real or hypothetical individuals whose risk is greater. People are less likely to believe that their risk is lower than that of a comparison target when they do not hold a stereotype of a typical victim, and when the target is individuated, physically present, and perceived to be similar to themselves. Furthermore, unrealistic optimism appears to be greater when people fail to take the perspective of others and when others are used as the standard of comparison. Because social comparison research rarely examines the individuality of the target, the salience of participants" own and others" behavior, or the standard of comparison, these findings are theoretically significant. It remains to be seen whether these findings may be generalized to other social comparison situations. For example, when receiving negative performance feedback, will people be more likely to engage in downward comparison when the comparison target is an abstract, deindividuated person?
These findings have a number of implications for health promotion campaigns. Given that attention to one"s own behavior could exacerbate biases and that providing information about the risk of typical others is often ineffective, educational programs that rely on the dissemination of general risk information are unlikely to be successful. Instead, the use of similar, individuated, physically present others as comparison targets might lead people to exhibit less unrealistic optimism. Interventions that increase attention to these targets and that force the audience to compare these targets to themselves are also more likely to have an impact. We consider these recommendations in more detail in the final section of the chapter (see also Misovich, Fisher, & Fisher, chapter 4, this volume).