The role of information and belief in adolescents' smoking behavior



Journal Title

Journal ISSN

Volume Title



In spite of extensive public campaigns stressing information about the dangers of cigarette smoking, smoking is increasing among junior high school children. A possible explanation for the apparent failure of traditional attempts to prevent smoking among school children may lie in the children's failure to attend to or to believe the information presented. To explore this possibility, a social psychological investigation of smoking information, belief, and behavior among 493 sixth, seventh, and eighth grade subjects was undertaken within the context of a large longitudinal study. Subjects in a separate sample responded to questionnaire items measuring their recall of smoking-and-health information and their belief in that information. Before providing self-reports of smoking behavior, subjects produced saliva samples, some of which were to be analyzed for nicotine content. Earlier research (Evans, Hansen, & Mittelmark, 1977) has indicated that subjects' awareness of the possibility of verification of self-reports of smoking behavior increases the validity of those self-reports. Extensive awareness of and belief in negative health consequences of cigarette smoking were found, indicating that the apparent failure of traditional smoking prevention attempts cannot be attributed to subjects' failure to attend to or to believe smoking-and-health information. In spite of the high reported frequency of belief, about 22% of the 176 eighth graders surveyed were classified as smokers (subjects who reported smoking every day or a few times a week), and another 37% were classified as experimenters (subjects who reported smoking a few times a month or trying smoking a few times). Only 40% of the eighth grade subjects reported that they had never tried smoking or that they had tried it only once. Weak, but statistically significant correlations between eighth graders' reported beliefs and their reported smoking behavior suggest low belief-behavior consistency. Only about 30% of the variance in eighth graders' reported smoking behavior was accounted for using a discriminant function or a multiple regression function based upon the variables studied. Three variables made significant contributions to prediction in both the discriminant and multiple regression analyses: number of subject's parents who smoke, belief that harmful chemicals can get in subjects' bodies if they smoke, and belief that breathing cigarette smoke is harmful to nonsmoking subjects. These three variables also loaded on the same factor with reported smoking behavior when a factor analytic technique was used. The low observed belief-behavior consistency indicates that the variables studied were inadequate as predictors of smoking behavior. Interpretation of the results from the perspectives of three similar theoretical orientations suggests plausible explanations for the low consistency between smoking beliefs and behavior found here, and for the failure of informational attempts at preventing smoking among adolescents in spite of their demonstrated success at establishing antismoking beliefs. Reiman's (1974) dynamic model of attitude formation and change, the health belief model of preventive health behavior (Becker, 1974; Rosenstock, 1966), and Fishbein's model for the prediction of behavioral intentions each raises considerations relevant to the low consistency finding. Subjects may not perceive that they may personally experience negative health consequences of cigarette smoking. Subjects may not believe that each instance of cigarette smoking results in negative outcomes. Beliefs about health consequences may be less important than other beliefs (e.g., beliefs about social consequences) in determining adolescents' smoking behavior. Subjects' evaluations of different possible outcomes have not been considered. In light of the many plausible explanations for the low levels of belief-behavior consistency found, the rejection of such a potentially valuable concept is premature. Further research is recommended to identify classes of beliefs that may predict behavior reliably, and to identify factors that influence belief-behavior relationships.