AIMS: The goal of this study was to examine social selection and social influence within reciprocal and non-reciprocal friendships, and the role of parents and siblings, as factors explaining similarity of smoking behaviour among adolescent friends. A new social selection-social influence model is proposed. DESIGN: Longitudinal design with two measurements. SETTING: Data were gathered among Dutch high school students in the control group of the European Smoking prevention Framework Approach (ESFA) study. PARTICIPANTS: The sample consisted of 1886 adolescents with a mean age of 12.7 years. MEASUREMENTS: The main outcome measures were the smoking behaviours of the respondents, best friends, parents and siblings. We tested the social selection-social influence model with structural equation modelling techniques. FINDINGS: Social selection and social influence both played an important role in explaining similarity of smoking behaviour among friends. Within non-reciprocal friendships, only social selection explained similarity of smoking behaviour, whereas within reciprocal friendships, social influence and possibly also social selection explained similarity of smoking behaviour. Sibling smoking behaviour was a more important predictor of adolescent smoking behaviour than parental smoking behaviour. CONCLUSIONS: Social selection and social influence both promote similarity of smoking behaviour, and the impact of each process differs with the degree of reciprocity of friendships. These insights may contribute to further refinement of smoking prevention strategies.
AIMS: The goal of this study was to examine social selection and social influence within reciprocal and non-reciprocal friendships, and the role of parents and siblings, as factors explaining similarity of smoking behaviour among adolescent friends. A new social selection-social influence model is proposed. DESIGN: Longitudinal design with two measurements. SETTING: Data were gathered among Dutch high school students in the control group of the European Smoking prevention Framework Approach (ESFA) study. PARTICIPANTS: The sample consisted of 1886 adolescents with a mean age of 12.7 years. MEASUREMENTS: The main outcome measures were the smoking behaviours of the respondents, best friends, parents and siblings. We tested the social selection-social influence model with structural equation modelling techniques. FINDINGS: Social selection and social influence both played an important role in explaining similarity of smoking behaviour among friends. Within non-reciprocal friendships, only social selection explained similarity of smoking behaviour, whereas within reciprocal friendships, social influence and possibly also social selection explained similarity of smoking behaviour. Sibling smoking behaviour was a more important predictor of adolescent smoking behaviour than parental smoking behaviour. CONCLUSIONS: Social selection and social influence both promote similarity of smoking behaviour, and the impact of each process differs with the degree of reciprocity of friendships. These insights may contribute to further refinement of smoking prevention strategies.