A path model of smoking behaviour among senior high school students in Taiwan

Abstract

Objective:

The purpose of this study was to explore the ways in which social smoking expectations mediate the relationship between adolescent smoking behaviour and the smoking behaviour of family and peers.

Design:

Descriptive, cross-sectional survey.

Setting:

Taiwan, Republic of China.

Method:

The participants were 921 senior high school students who completed a questionnaire to measure smoking status, family and peer smoking and social smoking expectations. The survey was conducted from November to December in 2009. Structural equation modelling was performed to test a mediation model. Analysis was performed using PRELIS 2 and LISREL 8.7 software.

Results:

After adjusting for students’ gender, age and school type, it is clear that social expectations mediate the relationship between sibling smoking, peer smoking and adolescent smoking. The path coefficients of sibling and peer smoking on social expectations in the mediator model were moderate, β = .15–.41, p < .05, but the coefficient of social expectations on adolescent smoking was strong, β = .95, p < .001. The total indirect effects between exogenous latent variables and adolescent smoking were .54, indicating a strong association that, in this model, is explained by the mediation of social expectations.

Conclusion:

Determining the mechanisms that influence adolescent smoking is of scientific interest because knowledge of these mechanisms is a prerequisite for the development of recommendations aimed at prevention. With this knowledge, health professionals can develop smoking-prevention policies and education programmes based on reducing the mediating effect of social smoking expectations on the relationship of family smoking and peer smoking with adolescent smoking.

Keywords

Adolescent family/peer smoking smoking social smoking expectations

Introduction

Cigarette smoking is the leading preventable cause of disease and death worldwide (World Health Organization [WHO], 2009). The smoking rate among young people in Taiwan has been gradually increasing, rising from 10.7% in 1994 to 14.7% in 2011 (Bureau of Health Promotion, 2012; Department of Health, 1996). Smoking is a learned behaviour, and the natural history of smoking usually starts between adolescence and adulthood (Van De Ven et al., 2010). Social learning theory (Bandura, 1986), with its emphasis on social influences, has been used extensively to study adolescent smoking. Parent and peer smoking behaviours are well-documented risk factors for adolescent smoking (Huang et al., 2011; Selya et al., 2012). Jøsendal and Aarø (2012) have shown that the social dimension in outcome expectancies is the strongest predictor of adolescent smoking. Determining the precise mechanisms that influence adolescent smoking is of great scientific interest because knowledge of these mechanisms is a prerequisite for the development of interventions aimed at smoking prevention.

A number of social risk factors for adolescent smoking initiation have been established, including parental, sibling and peer smoking (Loke and Wong, 2010; Peng et al., 2007). However, findings across multiple studies have not been definitive in establishing an association between family smoking and adolescent smoking. While Kabir and Goh (2014) demonstrated that in South Asia, parental tobacco use is a predictor of adolescents’ tobacco use behaviour, a previous study that was designed to use a social contextual perspective to understand the development of cigarette smoking among youth did not include sibling factors in the family contributions to smoking behaviour (Ennett et al., 2010). In an earlier study, sibling and peer smoking were highly associated with adolescent smoking, but parent smoking displayed weak and inconsistent associations (Avenevoli and Merikangas, 2003). Therefore, a sibling smoking variable was included in this study to examine the overall effect of family smoking on adolescent smoking behaviour. According to social cognitive theory (Bandura, 1986), adolescents’ perceptions are important factors in understanding how their observation of the behaviour of significant others may directly affect their own behaviour.

Jøsendal and Aarø (2012) showed that the social dimension of outcome expectancies was a strong predictor of smoking behaviour. They defined ‘social expectancies’ as the ‘positive expected social consequences of smoking’. Because many adolescents and young people are exposed to family and peers who smoke, they may develop positive smoking expectancies and consequences (Blanton et al., 1997). Therefore, it was hypothesised in this study that exposure to significant others who smoke would be associated with increased social smoking expectancies, which would be associated with adolescent smoking (Figure 1). Positive and negative expected smoking outcomes have been identified as important exploratory factors involved in adolescent smoking motivation (Anderson et al., 2002; Tyas and Pederson, 1998). Positive outcome expectations have a stronger and significant association with adolescent smoking than negative outcome expectations (Dalton et al., 1999). The more adolescents endorse these positive expectancies, the more likely they are to be susceptible to smoking (Dalton et al., 1999). Positive outcome expectancies mediate the association between peer and parent smoking and adolescent smoking, but not sibling smoking (Tickle et al., 2006).

Figure 1.

Mediation model of family smoking, peer smoking, social smoking expectations and adolescent smoking.

The concept of smoking expectancies is still being debated. In a study by Dalton et al. (1999), positive outcome expectations were defined as personal attitudes about smoking and were measured using a 7-item scale, namely, enjoying smoking, giving me something to do when I am bored, helping me address problems or stress, helping me to stay thin, helping me to feel more comfortable at parties, being relaxed and making me look more mature (p. 462). These items emphasised the positive expectation of self-empowerment, rather than social facilitation. Another study used an 8-item scale to measure social consequences such as enjoying parties by smoking, feeling comfortable and appearing person friendly (Lewis-Esquerre et al., 2005). This study focused only on social situations.

The other definition of positive reinforcement was measured using a 5-item scale consisting of a cigarette tasting good, enjoying the taste sensations, the pleasant taste, enjoying the flavour of a cigarette and enjoying the feeling of a cigarette on my tongue and lips (Myers et al., 2003). These items represented and correspond with taste/sensorimotor reinforcement more than they did social reinforcement. Because this study focused mainly on exploring the mechanism of social effects on adolescent smoking, the measures of smoking expectancy used were specific social expectancy concepts for smoking. Previous research suggested that social and empowerment reinforcement occurs as adolescents use tobacco to pursue social acceptance, to facilitate social situations, to develop a smoker identity and to acquire a sense of belonging (Johnson et al., 2005). Therefore, positive smoking expectancies were defined in this study as social smoking expectations, including social facilitation and self-empowerment as a result of smoking.

Parental influences have an impact on adolescent smoking, and this impact varies in different ethnic groups (Shakib et al., 2003). A previous study showed that in the USA, ethnic variation exists in social influences on adolescent smoking among Black, Hispanic and White teenagers (Dornelas et al., 2005). The mechanisms underlying adolescent smoking and the effects of social influences on adolescent smoking may therefore be culturally determined. In Chinese society, very little is known about the mediating pathway through which these factors may act. Although smoking-prevention interventions have effectively targeted psychosocial factors contributing to the initiation and maintenance of adolescent smoking (Botvin and Griffin, 2007; Thomas and Perera, 2013), little research has been performed regarding the mechanisms underlying adolescent smoking to delineate the varying influence and valence of family factors.

Gender, school type and age have been identified as important individual characteristics in adolescent smoking (Department of Health, 2011; Lai et al., 2004). The Department of Health (2011) has shown that the adult smoking rate in the Taiwanese population is 35.4% for men and 4.2% for women. For this reason, this study measured parent smoking using the separate variables of father and mother smoking. The Bureau of Health Promotion (2012) reports that the smoking rates for male and female adolescents are 20.3% and 8.1%, respectively. The smoking rates of general school and vocational high school students are 3.6% and 16.5%, respectively (Bureau of Health Promotion, 2012). The smoking rates of senior school students aged 15, 16 and 17 years are 12.2%, 13.0% and 15.7%, respectively (Bureau of Health Promotion, 2012). Previous studies have shown age to be positively associated with smoking prevalence among adolescents (Bureau of Health Promotion, 2012). Numerous researchers have examined the individual characteristics for adolescent smoking. These relevant covariates of smoking behaviours were therefore controlled for in the mediating analyses.

Multiple social factors are related to adolescent smoking, including peer smoking, attitude towards smoking, parent factors, alcohol consumption and participation in physical activity (Kristjansson et al., 2008). However, multiple levels of social influence should also be considered, including smoking expectations and parental, sibling and peer influences. Moreover, in Chinese society, very little is known about the mediating pathway. The theoretical model explored in this study suggested that the parental, sibling and peer smoking behaviours that expose adolescents to examples of cigarette smoking would encourage adolescent smoking, and that these effects may be mediated by the social smoking expectations. Therefore, this study examined a theoretical framework specifying the direct and indirect relations between adolescent smoking, social smoking expectations and the smoking behaviours of family and peers.

Methods

Participants and procedures

In this study, we used a cross-sectional design with a questionnaire in conjunction with convenience sampling. Data were collected from November to December in 2009. Nine high schools in the southern region of Taiwan were invited to participate in the study. For each school, three classes were selected randomly by grade. The study was approved by the ethics review board of the Chang Jung Christian University (CJCU-98-007). The research assistant (H.-W.H.) sent a passive informed consent form to the parents. One week before the questionnaire was to be completed by the students, an information sheet was given to all parents to explain the nature of the study.

The participants were asked to complete the confidential questionnaire in their classrooms within one class session. All participants were informed both verbally and in writing that they could withdraw from the study at any time and that their names would be encoded to ensure anonymity. Each participant received a gift of stationery. A total of 921 participants were obtained, with a response rate of 89%. In all, 92 students were excluded because 82 parents did not want their children to participate in the study, and 10 participants were out of school during data collection. The final sample was composed of 921 high school students (519 boys and 402 girls) between the ages of 14 and 19 years (mean age = 16.7 years, standard deviation (SD) = .9 years). All participants completed a questionnaire that included demographic data; parents’, siblings’ and friends’ smoking behaviours; social smoking expectations; and self-reported smoking behaviours.

The response variable

Smoking behaviour

Students’ level of experimentation with smoking was the outcome variable and was assessed through self-reporting with the question, ‘What is your smoking experience in life?’ Response categories were 1 (never smoked, even a few puffs), 2 (smoked, but not in the past year), 3 (smoked in the past year, but not in the past month), 4 (smoked in the past month, but not weekly), 5 (smoked weekly in the past month) and 6 (smoked daily in the past month). In line with the national youth tobacco surveys (Department of Health, 2006) and Piontek et al. (2008), smoking behaviour was determined from responses to this question. Smoking status was defined by the following six levels: never, ever, occasional, experimental, weekly and daily smoking.

The predictors

Parental smoking

Two items were included to measure parental smoking: ‘Does your father smoke?’ and ‘Does your mother smoke?’ A father or mother who had never smoked in his or her lifetime was defined as a ‘Never smoker’; those who had smoked a puff of a cigarette but had not smoked in the previous month were defined as an ‘Ever smoker’ and those having smoked in the previous month were defined as a ‘Current smoker’ (Can et al., 2009; Kleinjan et al., 2009).

Siblings’ smoking

Participants responded to questions related to their siblings’ smoking on a 5-point scale ranging from 1 (none of them smoke), 2 (<25% smoke), 3 (between 25% and 50% smoke), 4 (between 51% and 75% smoke) to 5 (all of them smoke).

Peer smoking

Participants responded to questions related to their peers’ smoking on a 5-point scale ranging from 1 (none of them smoke), 2 (<25% smoke), 3 (between 25% and 50% smoke), 4 (between 51% and 75% smoke) to 5 (all of them smoke).

The mediating variable

To measure social smoking expectations, the study used the concept of social reinforcement developed by Johnson et al. (2005). Positive expectations were presented on an 11-item scale containing items such as ‘Smoking makes me feel popular’ and ‘Smoking helps me feel in control of my life’. Each item was rated on a 4-point scale (strongly disagree, disagree, agree, strongly agree). Total scores ranged from 11 to 44. The Chinese version was tested in a previous study, and the psychometric properties of the scale were found to be satisfactory, including internal consistencies (Cronbach’s α values .87) and exploratory factor analysis (Lu et al., 2009).

Statistical analysis

Confirmatory factor analysis (CFA) was used to assess the factor structure and item performance of the Chinese version of the Social Reinforcement Scale. Structural equation modelling (SEM) was used to test the model for the mediation effect of social reinforcement on the relationship between parental, sibling and peer smoking and current smoking status. The magnitude of mediation was estimated from the proportion of the mediated effect as the direct effect, the indirect effect and the total effect. The sample size, greater than 500, was stably estimated using the effect size of mediation (MacKinnon et al., 1995).

Analysis was performed using PRELIS 2 and LISREL 8.7. PRELIS can provide an estimate of the asymptotic covariance matrix of different measures (continuous, censored and ordinal variables). Such data transformation can be used in LISREL to perform powerful, accurate analysis. The SEM was performed with robust maximum likelihood parameter estimates with standard errors and chi-square test statistics (Boomsma and Hoogland, 2001). The study used the cut-off criteria for relatively good fit recommended by Hu and Bentler (1999): normed fit index (NFI) ≥ .95, non-normed fit index (NNFI) ≥ .95, comparative fit index (CFI) ≥ .95, root-mean-square error of approximation (RMSEA) ≤ .06 and standardised root-mean-square residual (SRMR) ≤ .08.

Results

Smoking behaviour and characteristics

Table 1 presents the descriptive statistics for the sample characteristics, the mediator variable and the outcome variable. With respect to father smoking, 25.6% of respondents reported that their father had never smoked in his lifetime, 19.4% reported their father had smoked a puff of a cigarette but had not smoked in the previous month and 55% reported that their father had smoked in the previous month. For mother smoking, 89.6% of the respondents reported their mother had never smoked in her lifetime, 2.8% reported that their mother had smoked a puff of a cigarette but had not smoked in the previous month and 7.6% reported that their mother had smoked in the previous month.

Table 1.

Smoking behaviour and characteristics (N = 921).

Variables	N	Percentage of total
Gender
Male	519	56.3
Female	402	43.7
School type
High school	279	30.2
Vocational high school	642	69.8
Smoking status
Never smokers	709	77.0
Ever smokers	92	10.0
Occasional smokers	23	2.5
Experimental smokers	17	1.8
Weekly smokers	14	1.5
Daily smokers	66	7.2
Father smoking
Never smoking	216	25.6
Ever smoking	179	19.4
Current smoking	506	55.0
Mother smoking
Never smoking	825	89.6
Ever smoking	26	2.8
Current smoking	70	7.6
Sibling smoking
Almost none	712	77.3
<25%	142	15.4
Between 25% and 50%	38	4.1
Between 51% and 75%	20	2.2
>75%	9	1.0
Peer smoking
Almost none	230	25.0
<25%	360	39.1
Between 25% and 50%	201	21.8
Between 51% and 75%	99	10.7
>75%	31	3.4
	Mean	Standard deviation (SD)
Social smoking expectations	15.4	6.7
Age (years)	16.7	0.9

For sibling smoking, 77.3% of the participants responded that they did not have a sibling who smoked, while 22.7% reported having at least one sibling who smoked. Only a quarter of the sample had no peers who smoked. The average score for smoking social reinforcement was 15.4 (SD = 6.7). The categories of smoking status were never smokers (77.0%), smokers (10.0%), occasional smokers (2.5%), experimental smokers (1.8%), weekly smokers (1.5%) and daily smokers (7.2%).

Table 2 shows the relationship between the predictor variables, the mediator variable and the outcome variable. Father, mother, sibling and peer smoking had a significantly positive correlation with social smoking expectations (r = .10–.39, p < .01). Father, mother, sibling and peer smoking and social expectations had a significantly positive correlation with adolescent smoking (r = .36–.71, p < .01).

Table 2.

Correlation matrix between parent, sibling and peer smoking; social expectations and adolescent smoking (N = 921).

	1	2	3	4	5	6
1. Adolescent smoking	–
2. Father smoking	.37**	–
3. Mother smoking	.36**	.33**	–
4. Sibling smoking	.49**	.29**	.31**	–
5. Peer smoking	.71**	.23**	.33**	.44**	–
6. Social expectations	.53**	.10**	.17**	.29**	.39**	–

p < .01.

Mediation analysis

Figure 1 shows the final model and path coefficients. The structure of the social smoking expectations scale was supported with CFA in the present sample, χ²(44) = 76.95, NFI = 1.00, NNFI = 1.00, CFI = 1.00, RMSEA = .03, SRMR = .02. The fit indices of the mediation model revealed a satisfactory fit, χ²(136) = 277.68, NFI = 0.99, NNFI = 1.00, CFI = 1.00, RMSEA = .03, SRMR = .07. Higher levels of social expectations resulted in higher levels of adolescent smoking. The model showed that sibling and peer smoking had a significant indirect effect on social expectations, but parental smoking did not. Adding direct paths from sibling and peer smoking to adolescent smoking was not significant and resulted in the same model fit, χ²(134) = 268.90, NFI = 0.99, NNFI = 1.00, CFI = 1.00, RMSEA = .03, SRMR = .07, indicating that sibling and peer smoking were related to adolescent smoking through social expectations. All path coefficients were significant except for the path from parental smoking to social expectations. The results of the regression analyses showed that the proportion of variance in adolescent smoking determined by the full model was 92%. Moreover, 24% of the variance in the mediator variable was explained by exogenous latent variables (i.e. sibling, parental and peer smoking).

The association between predictors and adolescent smoking explained by the mediator variable (i.e. social expectations) was also estimated. Table 3 presents standardised estimates of indirect effects in the mediator model. The total indirect effect between exogenous latent variables and adolescent smoking was .54, indicating a strong association that could be fully explained by the mediation of social reinforcement. Moreover, 72% of the association between sibling smoking and adolescent smoking was explained by social expectations. The path coefficients of sibling and peer smoking on social expectations in the mediator model were moderate, β = .15–.41, p < .05, and the coefficient of social expectations on adolescent smoking was strong, β = .95, p < .001.

Table 3.

Mediating model when controlling for gender, school type and age (N = 921).

Dependent variable	Predictor	β/γ	SE	p	Indirect effect	% explained of total effect
Social expectations	Peer smoking	.41	.07	<.001	.39	72.2
	Parent smoking	−.01	.09	.442	.01	1.9
	Sibling smoking	.15	.07	.023	.14	25.9
Adolescent smoking	Social expectations	.95	.06	<.001	.54
	Covariates
	Gender	−.02	.02	.342
	School type	.03	.03	.293
	Age	−.04	.02	.080

SE: standard error.

Total indirect effect: exogenous latent variables (sibling, parent, peer smoking) → adolescent smoking (.54).

Indirect effect: sibling smoking → social expectations → smoking (.14); parent smoking → social expectations → smoking (.01); peer smoking → social expectations → smoking (.39).

Discussion

In Taiwan, adolescent smoking is prohibited under the Tobacco Hazards Prevention Act, yet in 2011, the prevalence of smoking behaviour among adolescents increased to 14.7% for senior high school students (Bureau of Health Promotion, 2012). Parental, sibling and peer smoking have been found to be the most important factors in the development of adolescent smoking (Avenevoli and Merikangas, 2003; Wen et al., 2005). However, little emphasis has been given to studying the pathway by which these factors influence adolescent smoking. As hypothesised, high levels of sibling and peer smoking were each associated with increased endorsement of social expectancies about smoking which was related to adolescent smoking. In our proposed model, social expectations totally mediated the relationship between sibling and peer smoking and adolescent smoking. Contrasting parental and sibling smoking in relation to social expectancies and adolescent smoking, sibling smoking seems to be more important because adolescents need to expand their social skills and develop a unique identity (Erikson, 1980). According to Bandura’s (1986) social cognitive theory, adolescents observe the behaviour of significant others and thereby search for information or learn behaviours from references. In this way, they can develop positive outcome expectancies in regard to smoking. The proportion of variance in the mediator variable determined by the smoking behaviour of significant others (R² = .24) and in smoking status determined by the full model (R² = .92) was moderate and high, respectively. Moreover, the total indirect effect was .54, which seems to provide evidence to validate the theory.

While a quarter of the adolescents in the cross-sectional sample had at least one sibling who smoked, and half had at least one parent who smoked, nearly three-quarters had at least one friend who smoked. Family smoking seems to have had an influence on smoking stage in our sample. This result was consistent with previous studies that indicated parent smoking to be an important factor in adolescents’ smoking behaviours (Szabo et al., 2006; Wen et al., 2005). We found several positive associations between family and peer variables on cigarette smoking that are comparable with the findings of the Taiwan Youth Tobacco Survey (Chen et al., 2009). The percentages of fathers and mothers who smoked were somewhat similar in this study to those found in the 2009 Youth Tobacco Survey (Department of Health, 2009).

In the correlation analyses, family and peer smoking was positively associated with smoking social expectations and adolescent smoking. Additionally, social expectancies were strongly associated with adolescent smoking. The findings are similar to those of other studies. The importance of the smoking behaviour of significant others (i.e. parents, sibling and peers) for rendering adolescents vulnerable to smoking has been established (Piontek et al., 2008; Wiium et al., 2006). A previous study showed social, psychological and environmental factors, such as parental and peer smoking, acceptance of cigarette use, media tobacco exposure and tobacco control policies inside and outside of schools, to be associated with adolescent smoking in China (Cai et al., 2012). Social connection, social function, self-identity in social situations and social gain seem to have an essential influence on adolescent smoking (Baillie et al., 2005; Johnson et al., 2003).

Although the Taiwan government has adopted the WHO Framework Convention on Tobacco Control to mandate public health legislation – such as restricting tobacco advertising, promotion and sponsorship; introducing smoke-free areas and health warnings; and increasing taxation on cigarette sales – the smoking rate of senior high school students has increased even as adult tobacco use has fallen (Department of Health, 2011). A community-based intervention to modify the students’ environment has been proposed and may have a marked effect on smoking prevalence among adolescents (Leiva et al., 2014). The use of new media such as Facebook and Twitter, as well as online advertisements, is suggested as a way of promoting social mobilisation on tobacco issues (Hamill et al., 2013). Therefore, the modification of the students’ community and the use of new social media such as Line or WeChat for tobacco control intervention may be employed to reduce adolescents’ smoking social expectations.

According to an ecological perspective (Moos, 1979), family, peer and school contexts represent microsystems that may interact with individuals’ expectations about smoking. In Iceland, an effective substance prevention programme focused on establishing a partnership between and empowerment of parents and schools (Sigfúsdóttir et al., 2010). Smoking-prevention programmes implemented in homes and schools, and community intervention programmes, can reduce cigarette smoking among youth (Chen et al., 2012; Wilson et al., 2012). To reduce social smoking expectations and such interactions between these social factors, we therefore suggest strengthening both the supportive and collaborative role of family and school, and the network of neighbourhood communities around adolescents. To reduce positive smoking expectations, smoking-prevention efforts should pay special attention to increasing adolescents’ social interaction and confidence through socio-ecological programmes and social media campaigns, as well as addressing their need for social acceptance and a sense of belonging at school and at home. A further study with intervention designs is needed to expand our insights into such prevention programmes in health behaviour research.

The effects of the path coefficients in the mediator model were moderate or strong. In agreement with previous modelling testing by Tickle et al. (2006), our findings indicate that peer smoking and social expectancies about smoking are relatively strong predictors of adolescent smoking. Sibling and peer smoking are associated with adolescent smoking through social expectations. In a previous study, parental and peer smoking, but not sibling smoking, were associated with positive expectations about smoking (Tickle et al., 2006). The effects of family factors on social expectations and adolescent smoking may vary in different ethnic groups. Although parental smoking did not mediate adolescent smoking in this sample, father and mother smoking increased social expectancy about smoking. Parental and sibling influences in this study were inconsistent with recent research, perhaps because the measure of social expectations differed from the positive expectancies used in the Tickle et al. study. Future research may explore other possible mediators of parental smoking on adolescent smoking behaviour (e.g. identification as a smoker and normative beliefs about smoking) by using additional measures and diverse Chinese populations.

Although parents’ current and former smoking affects the likelihood of adolescent smoking (Otten et al., 2007), parental smoking did not contribute to the social and empowerment aspects of positive smoking expectations in senior adolescents of this study. The reasons for this are unclear, and the issue should be examined in future research. This study provides an important first step in outlining the possible mechanism of social expectations that can be addressed in ongoing attempts to find ways to decrease the influence of sibling and peer smoking on impressionable adolescents.

Limitations of study

To the best of our knowledge, this is the first study to confirm the mediating effect of social reinforcement about smoking on the relationship between sibling and peer smoking and adolescent smoking, after controlling for gender, school type and age. However, the study has several limitations related to the generalisation of results. First, as the study involved a cross-sectional design, a causal effect could not be inferred; hence, further research should include longitudinal studies to infer any causal relationship between social influences and adolescent smoking. Second, the population was limited to senior high school students in the southern regions of Taiwan, making it impossible to generalise the findings to all adolescents. Third, because parental smoking did not have a significant effect on social expectations in influencing adolescent smoking, future studies should explore other factors that may be more important than social expectations. The mechanism for parent smoking influencing adolescent smoking may be different from the effects of sibling and peer smoking through social expectations. Fourth, gender difference is an important factor in adolescent smoking in Taiwan (Department of Health, 2011). Because genders affect the outcome variable (smoking behaviour), this factor is controlled in the proposed model. Whether the gender interaction effects of family and peer smoking on adolescent smoking are mediated by social expectations will be examined in future research. Finally, the unique environmental influences of Taiwan may limit the generalisation of study results, although the social and cultural contexts are similar to those in mainland China and some other Asian countries. Studies should be conducted in other countries or with different ethnic groups in the future, and the results should be compared with those reported here.

Footnotes

Funding

This research was funded by Changhua Christian Hospital, Taiwan (Grant No. 99-ERCCH-CJCU-01).

References

Anderson

Pollak

Wetter

(2002) Relations between self-generated positive and negative expected smoking outcomes and smoking behavior: An exploratory study among adolescents. Psychology of Addictive Behaviors 16(3): 196–204.

Avenevoli

Merikangas

(2003) Familial influences on adolescent smoking. Addiction 98(Suppl. 1): 1–20.

Baillie

Lovato

Johnson

. (2005) Smoking decisions from a teen perspective: A narrative study. American Journal of Health Behavior 29(2): 99–106.

Bandura

(1986) Social Foundations of Thought and Action. London: Prentice Hall.

Blanton

Gibbons

Gerrard

. (1997) Role of family and peers in the development of prototypes associated with substance use. Journal of Family Psychology 11: 271–288.

Boomsma

Hoogland

(2001) The robustness of LISREL modeling revisited. In: Cudeck

du Toit

Sörbom

(eds) Structural Equation Models: Present and Future: A Festschrift in Honor of Karl Jöreskog. Lincolnwood, IL: Scientific Software International, 139 pp.

Botvin

Griffin

(2007) School-based programmes to prevent alcohol, tobacco and other drug use. International Review of Psychiatry 19(6): 607–615.

Bureau of Health Promotion (2012) Global Youth Tobacco Survey, GYTS in 2011. Available at: https://olap.hpa.gov.tw//search/ListHealth1.aspx?menu=1&mode=13&year=100 (accessed 15 July 2012).

Cai

. (2012) Social, psychological, and environmental-structural factors associated with tobacco experimentation among adolescents in Shanghai, China. International Journal of Environmental Research and Public Health 9(10): 3421–3436.

10.

Can

Topbas

Oztuna

. (2009) Factors contributing to regular smoking in adolescents in Turkey. Journal of School Health 79(3): 93–97.

11.

Chen

Huang

Chao

(2009) Susceptibility to initiate smoking among junior and senior high school nonsmokers in Taiwan. Preventive Medicine 49(1): 58–61.

12.

Chen

Ren

Lin

. (2012) Exposure to school and community based prevention programs and reductions in cigarette smoking among adolescents in the United States, 2000–08. Evaluation and Program Planning 35(3): 321–328.

13.

Dalton

Sargent

Beach

. (1999) Positive and negative outcome expectations of smoking: Implications for prevention. Preventive Medicine 29(6): 460–465.

14.

Department of Health (1996) Taiwan Public Health Report. Taipei, Taiwan: Department of Health.

15.

Department of Health (2006) Taiwan Tobacco Control Annual Report 2006. Taipei, Taiwan: Department of Health.

16.

Department of Health (2009) Smoking behavior online research system. Available at: https://olap.hpa.gov.tw//search/ListHealth1.aspx?menu=1&mode=12&year=98 (accessed 23 August 2012).

17.

Department of Health (2011) Taiwan Tobacco Control Annual Report 2011. Taipei, Taiwan: Department of Health.

18.

Dornelas

Patten

Fischer

. (2005) Ethnic variation in socioenvironmental factors that influence adolescent smoking. Journal of Adolescent Health 36(3): 170–177.

19.

Ennett

Foshee

Bauman

. (2010) A social contextual analysis of youth cigarette smoking development. Nicotine & Tobacco Research 12(9): 950–962.

20.

Erikson

(1980) Identity and the Life Cycle. New York: Norton.

21.

Hamill

Turk

Murukutla

. (2013) I ‘like’ MPOWER: Using Facebook, online ads and new media to mobilise tobacco control communities in low-income and middle-income countries. Tobacco Control. Epub ahead of print 13 December. DOI: 10.1136/tobaccocontrol-2012-050946.

22.

Huang

Chen

Huang

(2011) Smoking and cigarette dependence among Taiwanese adolescent smokers: Factors of influence. Hu Li Za Zhi 58(1): 28–36.

23.

Bentler

(1999) Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling 6(1): 1–55.

24.

Johnson

Bottorff

Moffat

. (2003) Tobacco dependence: Adolescents’ perspectives on the need to smoke. Social Science & Medicine 56(7): 1481–1492.

25.

Johnson

Ratner

Tucker

. (2005) Development of a multidimensional measure of tobacco dependence in adolescence. Addictive Behaviors 30(3): 501–515.

26.

Jøsendal

Aarø

L E

(2012) Adolescent smoking behavior and outcome expectancies. Scandinavian Journal of Psychology 53(2): 129–135.

27.

Kabir

Goh

(2014) Determinants of tobacco use among students aged 13–15 years in Nepal and Sri Lanka: Results from the Global Youth Tobacco Survey, 2007. Health Education Journal 73(1): 51–61.

28.

Kleinjan

Engels

van Leeuwe

. (2009) Mechanisms of adolescent smoking cessation: Roles of readiness to quit, nicotine dependence, and smoking of parents and peers. Drug and Alcohol Dependence 99(1–3): 204–214.

29.

Kristjansson

Sigfusdottir

Allegrante

. (2008) Social correlates of cigarette smoking among Icelandic adolescents: A population-based cross-sectional study. BMC Public Health 8: 86.

30.

Lai SR, Wu DM, Chiou SY, et al. (2004) Parental prevention strategy and adolescent’s smoking knowledge, attitude and behavior. Journal of Health Education 22: 41–62.

31.

Leiva

Estela

Torrent

. (2014) Effectiveness of a complex intervention in reducing the prevalence of smoking among adolescents: Study design of a cluster-randomized controlled trial. BMC Public Health 14: 373.

32.

Lewis-Esquerre

Rodrigue

Kahler

(2005) Development and validation of an adolescent smoking consequences questionnaire. Nicotine & Tobacco Research 7(1): 81–90.

33.

Loke

Wong

YPI

(2010) Smoking among young children in Hong Kong: Influence of parental smoking. Journal of Advanced Nursing 66(12): 2659–2670.

34.

Lin

Chen

. (2009) Psychometric testing of the Chinese version of the dimensions of Tobacco Dependence Scale. Journal of Clinical Nursing 18(17): 2470–2477.

35.

MacKinnon

Warsi

Dwyer

(1995) A simulation study of mediated effect measures. Multivariate Behavioral Research 30(1): 41–62.

36.

Moos

(1979) Social ecological perspectives on health. In: Stone

Cohen

Adler

(eds) Health Psychology: A Handbook. San Francisco, CA: Jossey-Bass, 523 pp.

37.

Myers

McCarthy

MacPherson

. (2003) Constructing a short form of the smoking consequences questionnaire with adolescents and young adults. Psychological Assessment 15(2): 163–172.

38.

Otten

Engels

van de Ven

. (2007) Parental smoking and adolescent smoking stages: The role of parents’ current and former smoking, and family structure. Journal of Behavioral Medicine 30(2): 143–154.

39.

Peng

Wang

Tung

. (2007) Factors influencing high-school students’ smoking behavior from the perspective of problem behavior theory. Journal of Evidence-Based Nursing 3(3): 236–245.

40.

Piontek

Buehler

Rudolph

. (2008) Social contexts in adolescent smoking: Does school policy matter? Health Education Research 23(6): 1029–1038.

41.

Selya

Dierker

Rose

. (2012) Risk factors for adolescent smoking: Parental smoking and the mediating role of nicotine dependence. Drug and Alcohol Dependence 124(3): 311–318.

42.

Shakib

Mouttapa

Johnson

. (2003) Ethnic variation in parenting characteristics and adolescent smoking. Journal of Adolescent Health 33(2): 88–97.

43.

Sigfúsdóttir

Thorlindsson

Kristjánsson

. (2010) Substance use prevention for adolescents: The Icelandic Model. Health Promotion International 24(1): 16–25.

44.

Szabo

White

Hayman

(2006) Can home smoking restrictions influence adolescents’ smoking behaviors if their parents and friends smoke? Addictive Behaviors 31(12): 2298–2303.

45.

Thomas

Perera

(2013) School-based programmes for preventing smoking. Cochrane Database of Systematic Reviews 4: CD001293.

46.

Tickle

Hull

Sargent

. (2006) A structural equation model of social influences and exposure to media smoking on adolescent smoking. Basic and Applied Social Psychology 28(2): 117–129.

47.

Tyas

Pederson

(1998) Psychosocial factors related to adolescent smoking: A critical review of the literature. Tobacco Control 7(4): 409–420.

48.

Van De Ven

MOM

Greenwood

Engels

. (2010) Patterns of adolescent smoking and later nicotine dependence in young adults: A 10-year prospective study. Public Health 124(2): 65–70.

49.

Wen

Tsai

Cheng

. (2005) Role of parents and peers in influencing the smoking status of high school students in Taiwan. Tobacco Control 14(Suppl. 1): i10–i15.

50.

Wiium

Torsheim

Wold

(2006) Normative processes and adolescents’ smoking behaviour in Norway: A multilevel analysis. Social Science & Medicine 62(7): 1810–1818.

51.

Wilson

Avila Tang

Chander

. (2012) Impact of tobacco control interventions on smoking initiation, cessation, and prevalence: A systematic review. Journal of Environmental and Public Health 2012: 961724.

52.

World Health Organization (2009) WHO report on the global tobacco epidemic, 2009: Implementing smoke-free environments. Available at: http://www.who.int/tobacco/mpower/en/index.html (accessed 23 April 2014).