This study explored the impact of different school factors on adolescent smoking behavior from the perspective of Hirschi’s social bond theory. Hirschi proposed that social bonds restrain adolescents from deviant behavior. According to this theory, four elements constitute social bonds: attachment, involvement, commitment, and belief. Hirschi suggested that these bonds, such as attachment to teachers and peers, involvement in conventional activities, commitment to higher educational levels, and belief in moral values, reduce the likelihood of rules being broken.
According to the Global Youth Tobacco Survey (Health Promotion Administration [HPA], 2020), the smoking rate among high school students in Taiwan was 8.4% (12.7% for boys and 3.7% for girls). The HPA and Taiwan’s Ministry of Education have been promoting the Campus Smoke -Free Campaign since 2009. However, during the 2018-2019 academic year, an increase in the use of cigarettes among high school students was observed (HPA, 2020). Problems associated with cigarette use, such as cancer and addiction, are not uncommon, particularly among Taiwan’s adolescents.
This study examined the influence of different school factors on smoking behavior. Accordingly, we applied a zero-inflated negative binomial (ZINB) regression model to determine the appropriate distribution of data.
Using the Taiwan Youth Project (TYP) survey data and ZINB regression model, we simultaneously analyzed the impact of school factors on expected smokers and the frequency of cigarette smoking among adolescent smokers.
The TYP is a longitudinal study, and the baseline interview was conducted in 2000, with follow-up interviews being conducted annually until 2019. Only data obtained from the sixth wave of the project (2005) were used for this research. The final samples involved 1,668 high school (including vocational school) students, with 837 boys and 831 girls.
Smoking behavior was the dependent variable, and it was assessed using a 5-point rating scale (0 = “no smoking,” 4 = “always smoking”). For attachment, teacher-student relationships, peer relationships, and school belonging were selected as indicators. Measures of commitment included indicators of educational aspirations. Involvement included absenteeism and participation in school leadership and competitions. Finally, measures of belief included belief in the moral value of smoking. Gender and school types were included as control variables.
Approximately 11.33% of the participants reported smoking at least once in the previous 30 days. A large percentage of the participants reported not smoking. The distribution for smoking data exhibited a high positive skew, with the variance exceeding the mean; this indicates that the outcome was overdispersed. ZINB models are designed for analyzing this type (excess zeros and overdispersion) of distribution under the assumption that the zeros originate from two latent subclasses.
The likelihood ratio for the full ZINB model was determined to be X2 (35) = 190.27 (p < .01; maximum likelihood = -770.76), indicating the significance of the overall model. Furthermore, the Vuong test for nonnested models supported the use of the ZINB model over a standard negative binomial model (z = 7.70, p < .01).
After controlling for the contextual effects of gender and school type, we observed that teacher–student relationships (β = .30, z = 3.16, OR = 1.35), educational aspirations (undergraduate degree: β = 1.06, z = 2.16, OR = 2.90; master’s degree: β = 1.35, z = 2.58, OR = 3.86; doctoral degree: β = 1.30, z = 2.52, OR = 3.68), and positive smoking moral values (β = 2.79, z = 2.05, OR = 16.29) had a significant positive influence on the zero-inflation probability (logit link). As expected, school absenteeism (β = -.41, z = -5.16, OR = 0.67) had a significant negative impact on the zero-inflation probability. However, for the negative binomial distribution (log link), we did not find evidence of school factors influencing the frequency of adolescent smoking behavior. Specifically, school factors were not significant predictors of the number of cigarettes consumed by the expected smokers.
From the perspective of Hirschi’s social control theory, adolescents do not inherently conform to rules, and adherence to rules is accomplished through the formation of a social bond between the adolescent and conservative society.
When an adolescent has a healthy and strong relationship with their teacher, an emotional bond exists between them, influencing adolescent smoking behavior. Teachers serve to prevent students from breaking rules. However, in this study, peer attachment and a sense of school belonging did not exhibit any effect on smoking. Adolescents aspiring to study at university and attain a higher level of education were observed to be less likely to smoke. Involvement in school activities and assuming a student leadership position also had no impact on smoking behavior. Nevertheless, we observed that smoking was significantly associated with school absenteeism. Adolescents who are frequently absent from school are less likely to be attached to their teachers and school, and they are thus more likely to be engaged in smoking because they lack conventional social bonds. Finally, adolescents with positive moral values were noted to be less likely to smoke. Conventional moral values serve as a means to regulate social interactions. The results of this study demonstrate that boys are more likely to engage in smoking than girls. Furthermore, vocational high school and junior college students are more likely to become cigarette users than high school students.
The study results highlight the impact of school factors on expected adolescent smokers and on the frequency of cigarette consumption among adolescent smokers. When school factors were tested solely against smoking behaviors, the results demonstrate that teacher-student relationships, educational aspirations, and positive moral values prevented adolescents from adopting smoking behavior. However, we observed school absenteeism to have a negative effect on smoking behavior. The effects of school factors on the frequency of cigarette smoking among adolescents were nonsignificant.
This study has several limitations that should be noted. First, the study relied on secondary data that were collected with a focus on traditional cigarettes. Using up-to-date data to conduct a comprehensive assessment of the effect of school factors on the use of e-cigarettes would benefit future research. Second, although this study demonstrated social bonds to be associated with smoking behavior, the causal association between them is still unclear. A longitudinal survey is required to examine the nature and direction of this association.
Sustained efforts to implement laws to control and prevent cigarette use are crucial for reducing and regulating smoking among Taiwan’s adolescents.