Moderating Roles of Primary Social Influences in the Relationship Between Adolescent Self-Reported Exposure to Antismoking Messages and Smoking Intention

Abstract

This study explores moderating roles of primary social influences in the relationship between adolescent triers' and experimenters' self-reported exposure to antismoking messages and their smoking intentions. The theoretical arguments are drawn from primary socialization theory, group socialization theory, and the social development model, and the data are from the 2004 National Youth Tobacco Survey. The tobit regression models demonstrate that, as a primary social influence, peer smoking seems to be a strong risk factor for all of the adolescent segments' smoking intentions, whereas parental monitoring can be a significant counter–risk factor for middle-schoolers' smoking intentions. In addition, school intervention programs and parental monitoring against smoking appear to play a moderating role in the relationship between high-school triers' self-reported exposure to antismoking messages and their smoking intentions. The findings seem to suggest that campaigners should make more efforts to incorporate primary social influences to prevent adolescent smoking. The findings also suggest that campaigners should tailor antismoking programs to fit specific target audiences. In particular, middle-school experimenters deserve more attention from antismoking campaigners because they seem most vulnerable to future smoking.

Notes

¹The tobit regression model may be specified as follows (Long, 1997):

where ϵ_i ∼ N(0, σ², x's are independent variables observed for all cases, and β is a vector of regression coefficients. y* is a latent variable that is observed for values greater than τ and censored for values less than or equal to τ. The observed y is defined as y_i = y_i* if y_i* > τ, and y_i = τ_y if y* ≤ τ.

²The tobit coefficient can be decomposed with the following formula (Roncek, 1992). For cases above the limit (in this study, those who showed some intention to smoke for at least one item):

For the cases below the limit (i.e., those who answered “definitely not intend to smoke” across all the intention to smoke items),

where Y is the criterion, Xi is each independent variable, B_i is a tobit coefficient of Xi, F(z) is a proportion of cases above the threshold, z is a z-score of the distance between F(z) and a middle point (.50), and F(z) is the height of the normal curve, and sigma is the standard deviation of the error term.

³The findings related to the control variables can be summarized as follows. First, age is negatively related to smoking intention for high-school triers' and experimenters' smoking intention. The younger high-schoolers are, the more likely they intend to smoke. Second, gender tends be significantly related to middle-school and high-school experimenters' smoking intention. Females tend to have a higher smoking intention than males when they have already experienced smoking up to 99 cigaretters. Third, race is also significantly related to high-school triers and experiments in that African Americans tend to have lower smoking intention than other races. Fourth, perceived risk of smoking and smoking attitude are strong predictors of smoking intention for all four segments. The less they perceive smoking as risky and the more they think smoking is cool, the more likely they intend to smoke.