Sample and procedure

Sixteen Azorean schools accepted to participate in this crosssectional study. An online questionnaire was disseminated to students from 3rd cycle and secondary school levels by the teacher responsible for the class, through an internet link. The questionnaire was available online between November 2020 and February 2021. A total of 209 students responded, most of them lived on the islands of Terceira (51.7%) and São Miguel (40.2%).

The study and the questionnaire were approved by the Ethic Committee of the University of Beira Interior and by the Regional Direction of Education (DRE) – Azores Regional Government. Participation in the study was voluntary, and the anonymity of participants and confidentiality of data collected were guaranteed.

Questionnaire

The questionnaire was based on a review of the literature and earlier work about adolescent smoking behavior^16,18,20-22. A pilot of the questionnaire with Portuguese adolescents contributed to the selection of items and item formulation in Portuguese^18,23. Questions assess sociodemographic variables, beliefs about smoking (attitudes), social influence, self-efficacy, intention to smoke, and smoking behavior.

Sociodemographic variables were age, gender, year of schooling, participants’ household, the island of residence, weekly money available, and mother’s and father’s education level.

Fourteen questions measured the attitudes on smoking, answered using a Likert scale. Eight questions were about the advantages of smoking [pros of smoking, e.g. ‘If I smoke (or if I smoked), I find it pleasant’], with answers ranging: 0 = ‘I totally disagree’ to 5 = ‘I totally agree’. Six questions addressed the disadvantages of smoking [cons of smoking, e.g. ‘If I smoke (or if I smoked), I think it is bad for health’), with the possibility of answers ranging: 0 = ‘I totally agree’ to 5 = ‘I totally disagree’. Finally, the remaining two questions addressed the social advantages associated with smoking [s-pros of smoking, e.g. ‘If I smoke (or if I smoked), I think it would be easier to be part of a crowd’), with responses ranging: 0 = ‘I totally disagree’ to 5 = ‘I totally agree’. A factor analysis confirmed these three factors: pros of smoking (Cronbach α =0.85), cons of smoking (Cronbach α=0.88), and s-pros of smoking (Cronbach α=0.89). These three factors were handled in some analyses as scales (using their means).

Social influence was assessed through several questions. Some sought to assess how participants perceive relevant others’ norms on their behavior (subjective norms). Relevant others are mother, father, brother(s), sister(s), best friend, important people, people of the same age, and teachers (e.g. ‘My mother (or the woman responsible for me) thinks that I:’) was evaluated with answers ranging: 0 = ‘I smoke’ to 3 = ‘Does not apply to my case’. Others questioned whether parents or close people smoked – descriptive norms (e.g. ‘Does your father (or the man responsible for you) smoke?’), with answers ranging: 0 = ‘No’ and 3 = ‘I do not have or do not live in my home’. The remaining questions evaluated the pressure to smoke exerted by the same people mentioned above [e.g. ‘Have you ever felt pressured to smoke by your brother(s)?’], with answers ranging: 0 = ‘Often’ to 5 = ‘I do not have or do not live in my house’. A factor analysis revealed three factors, corresponding to the subjective norms (Cronbach α=0.98), the smoking influence by family members and teachers’ behavior (Cronbach α=0.92), and the smoking influence by the peer group behavior (Cronbach α=0.91). In some analyses were used as scales (considering their means).

Self-efficacy to refuse to smoke was measured using 12 questions. Some were designed to understand the individual’s ability not to smoke in certain social circumstances (e.g. ‘When you are with friends who smoke, are you able not to smoke?’), others when performing routines (e.g. ‘When you are watching TV, are you able not to smoke?’), and in situations of emotional instability (e.g. ‘When you feel down, are you able not to smoke?’). All answers to these questions were given on a Likert scale, ranging: 0 = ‘I am absolutely sure that I smoke’ to 6 = ‘I am absolutely sure that I do not smoke’. Factor analysis forced into three factors according to previous studies carried out with this questionnaire¹⁶, including in Portugal¹⁸, revealed the social self-efficacy factor (Cronbach α=0.97), routines self-efficacy factor (Cronbach α=0.99), and emotional self-efficacy factor (Cronbach α=0.98). These factors were used as scales in some analyses (considering their means).

Intention to smoke was evaluated by the question: ‘Do you intend to smoke in the next year?’, with answers on a Likert scale ranging: 0 = ‘Definitely yes’ to 6 = ‘Definitely not’.

Finally, smoking behavior was assessed using five combined questions^18,21,22. In this study participants were divided into two or three groups. Two groups were never smokers (those who had never smoked not even a puff) and already smokers (those who smoke or had smoked occasionally and regularly). Three groups were the previous never smokers, occasional smokers (who smoke less than a cigarette a week or had smoked occasionally and regularly) and regular smokers (those who smoke one or more cigarettes per week).

Statistical analysis

Statistical analysis was carried out at two levels, based on the items (smoking behavior, attitudes, social influences, selfefficacy, and intention to smoke) or based on the factors of the ASE variables (pros of smoking, cons of smoking, social attitude, social influence of peers, social self-efficacy, self-efficacy emotions, and self-efficacy opportunities).

Descriptive statistics of sociodemographic variables and relevant variables were performed. Differences between never smokers and already smokers were analyzed using chi-squared test and t-test.

After, a bivariate correlation was made to explore the relationship between smoking behavior (with three groups: never smokers, occasional smokers and regular smokers) and the other variables. Finally, a logistic regression was performed using smoking behavior with two groups (never and already smokers) as dependent variable and all the other variables as independent variables. All analysis were performed using SPSS v. 26 (IBM Corp., USA).

Smoking behavior and intention to smoke

According to the data presented in Table 2, regarding the smoking behavior of the participants, 68.9% reported never had smoked, not even a puff (never smokers); all others were classified as already smokers: 16.2% had smoked occasionally, but currently do not smoke; 3.3% very rarely smoked a few puffs; 3.8% smoked a few cigarettes a year; 1.9% smoked at least once a month; and 5.7% smoked regularly (at least one cigarette a week).

Most participants did not show an intention to smoke in the next year (82.8%) or in the future (78.0%) (Table 2).

The participants’ perception on the percentage of smokers in Portugal was illusory and inflated: 48.8% answered that 61–80% of the Portuguese are smokers (Table 2).

Comparison between never smokers and already smokers

Comparing never smokers with already smokers in the sociodemographic variables (Table 1), these two groups differed only in age: already smokers were older than never smokers (16.6 ± 1.4 vs 14.9 ± 2.0 years, p<0.05).

In the attitudes (Table 3), significant differences between never smokers and already smokers appeared in the items of beliefs that smoking is positive (0.4 ± 0.7 and 0.8 ± 1.0; p<0.05), pleasant (0.6 ± 0.9 and 1.6 ± 1.2; p<0.05), helps to be slim (2.8 ± 1.4 and 2.5 ± 1.2; p<0.05), is stupid (1.3 ± 1.6 and 1.5 ± 1.3; p<0.05), calms the nerves (0.8 ± 1.0 and 2.0 ± 1.3; p<0.05), makes one feel secure 0.7 ± 0.9 and 1.2 ± 1.2; p<0.05), makes one receive more attention (0.7 ± 1.0 and 1.1 ± 1.3; p<0.05), makes one more sociable (0.7 ± 1.0 and 1.4 ± 1.3; p<0.05), and integrate more easily (0.7 ± 1.1 and 1.5 ± 1.3; p<0.05).

In terms of social influence (Table 3), the values of both groups do not differ significantly, except in the case of a best friend who smokes, with never smokers reporting a higher percentage of best friends who never smoked. However, although differences were not significant, relevant others [e.g. mother, father, brother(s), sister(s), and best friend] for the already smokers had consistently a higher percentage of smoking behavior.

Regarding self-efficacy to refuse smoking (Table 3), the two study groups showed significant differences in all items. However, the differences were bigger in some items: ‘when you go out with friends’ (5.7 ± 0.9 and 4.1 ± 1.8; p<0.05), ‘when friends offer a cigarette’ (5.7 ± 0.9 and 4.5 ± 1.6; p<0.05), and ‘when with friends who smoke’ (5.7 ± 0.9 and 4.7 ± 1.4; p<0.05).

The results did not reveal significant differences between the two groups in the intention to smoke.

Correlations with smoking behavior

Regarding attitude scales (Table 4), the results indicated a significant correlation between smoking behavior and the pros of smoking (r=0.46). All self-efficacy to refuse smoking scales had a significant correlation with smoking behavior, especially social self-efficacy (r= -0.57) and emotional self-efficacy (r= -0.49). However, the correlations between behavior, social influence variables and intention to smoke next year were not statistically significant.

Logistic regression

A logistic regression was performed (Table 4). Regarding sociodemographic variables, only age (OR=2.14; 95% CI: 1.59–2.86) was significantly associated with behavior. Among the ASE variables, beliefs regarding the advantages of smoking (OR=4.34; 95% CI: 2.18–8.62), and self-efficacy to refuse smoking in social situations (OR=0.30; 95% CI: 0.15–0.60) were significantly associated with behavior. This model explained 44% of the variance in smoking behavior. Social influence variables and intention to smoke did not show a significant effect on behavior.

Strengths and limitations

This study has several limitations. First, this is a crosssectional study, which restricts the possibility of inferring the direction of the effects. Second, data resulted from a selfreport questionnaire, and the responses were not biologically validated. Third, the questionnaire was applied online to adolescents who responded on their own initiative, which may have been more appealing to never smoking people. Finally, the sample is small, with participants from only one region, limiting the results’ robustness and generalizability.

This study also has some strengths. It was carried out with a sample of Azorean adolescents. This population is often excluded from studies carried out at the national level and even more at the international level. Online data collection can have advantages, namely, enhancing the perception of responses’ confidentiality and participants’ anonymity.

Future directions

Finally, it is essential to mention the need to carry out more studies on smoking behavior in Azorean adolescents, namely, with higher sample sizes. Future studies will confirm or not the effects on smoking behavior of the pros of smoking and of social self-efficacy to refuse smoking and the absence of effects of cons of smoking, social influence and intention to smoke. Future studies could also compare subgroups (e.g. different ages, gender and socio-economic level) to improve the knowledge of smoking behavior and increase the effectiveness of smoking prevention activities in this population.