Abstract
In this paper, we examine the importance of two different peer effects as determinants in the adolescent’s decision whether or not to smoke. One is measured at the class level and the other reflects the smoking behaviour of the adolescent’s best friends. A nationally representative wave of Spanish data, collected in different state and private centres of secondary education and vocational training, and several linear probability models are used to estimate the role of peer effects. We find that a 10 % increase in the proportion of classmates is associated with a 3.6 points increment in the probability of smoking. Similarly, if the smoker’s friends go from “only some” to “the majority”, the probability of smoking increases by 39 points. Although both peer effects are significant if introduced separately, the class peer variable is not significant once the closer peer effect is introduced. Our work provides evidence to support the hypothesis that peer effects are important determinants of smoking among adolescents. This has implications for policy-makers, since the existence of peer effects would amplify the effects of interventions.