to which smoking can shorten a smoker’s life. Although the majority of smokers (68 percent) and nonsmokers (79 percent) recognize that smoking shortens one’s life, close to 26 percent of smokers and 18 percent of nonsmokers responded that they did not know whether this was actually the case. When asked more specifically about the number of years that smoking can shorten a life span, 44 percent of smokers and 48 percent of nonsmokers correctly identified that smoking can shorten one’s life by 5 to 10 years; however, 28 percent of smokers and 19 percent of nonsmokers reported that they did not know. A high proportion of respondents also reported inaccurate assessments of the lethality of smoking compared with those of other behaviors. Many of those surveyed failed to recognize that smoking causes more deaths than gunshots and car accidents (42 percent) or alcohol and the use of other drugs (62 percent).
Another important finding from the Annenberg Tobacco Study concerns “optimism bias,” that is, smokers’ belief that the smoking risk is lower for themselves than for others engaging in similar behaviors (Weinstein 1989). In other words, an abstract understanding of the nature and the magnitude of smoking risks does not necessarily translate into a personalized appreciation of the hazards to oneself. In their analysis of the survey data, Romer and Jamieson found that, among smokers who correctly estimated that half of lifetime smokers die from smoking-related causes, 40 percent viewed their own smoking as less than “very risky” (the scale ranged from “very risky” to “not at all risky”). Among respondents who estimated that 60 percent or more of lifetime smokers die from smoking-related causes, 25 percent did not view their own smoking as very risky (Romer and Jamieson 2001a). Arnett reported similar results from a survey of both adolescents and adults (Arnett 2000). Arnett found that in both of these groups, smokers were more than twice as likely as nonsmokers to doubt that they would die from smoking, even if they were to smoke for 30 to 40 years. Moreover, a nontrivial proportion of adolescent smokers (29 percent) doubted slightly or strongly that they would die from smoking if they smoked for 30 to 40 years. Other studies have also shown that participants who reported that they smoked rated the chance that a negative health outcome as well as a negative social outcome (e.g., getting into trouble) would occur lower than did participants who did not smoke (Arnett 2000; Halpern-Felsher et al. 2004; Virgili et al. 1991). Furthermore, risk perceptions vary by level of smoking (Chassin et al. 2000; Halpern-Felsher et al. 2004; Soldz and Cui 2002) or stage of smoking (Pallonen et al. 1998; Prokhorov et al. 2002), with individuals who have smoked longer and more often perceiving fewer risks than those who have smoked for shorter periods and less often. Similarly, studies have found that perceived health and social risks are related to behavioral intentions and that these intentions are the most important and immediate determinants of behavior (Ajzen 1985; Distefan