School connectedness and daily smoking among boys and girls: the influence of parental smoking norms

Abstract

Background: The objective was to test whether an association between school connectedness and smoking exists among Danish school children, and if so, to examine whether parental smoking attitude and parental smoking behaviour influenced this association. Methods: Data were collected by the Danish contribution to the cross-national study Health Behaviour in School-Aged Children (HBSC) 1998. Analyses were performed on questionnaire-based data from 1537 students at grade nine from a random sample of schools in Denmark. Results: An independent inverse association was found between school connectedness and smoking among both boys and girls. Parents' attitude to their children's smoking significantly modified this association among boys. Among girls the modifying effect was less marked. Neither among boys nor girls did parental smoking behaviour significantly modify the association between school connectedness and smoking, although a modifying tendency was observed among girls. Conclusions: The smoking behaviour of Danish adolescents may be influenced by complicated interactions of varying sets of experienced smoking norms, and any research project or preventive programme focusing on the influence of school life on adolescent smoking behaviour needs to consider the family smoking norms. Additionally, the results stress the important role of gender by indicating that the smoking behaviour of girls may be more sensitive to restricting social influences than the smoking behaviour of boys.

The relationship between psychosocial ties to school and students' smoking behaviour has been examined several times. Although such ties have been labelled and defined in various ways, the general observation has been that poor psychosocial ties to school are associated with being a smoker.^1–6

Social control theories have been applied to understand this association.^7,8 These theories emphasize socialization and integration of persons into the family and community. They point to the consequences of weak ties to conventional society, its values, institutions and socializing forces, including school. The theories state that adolescents who have weak ties to society will not internalize conventional standards for behaviour. These adolescents are therefore more likely to adopt deviant behaviour, contrary to adolescents whose ties to conventional society make them internalize standards for conventional behaviour.

The core theoretical assumptions of the social control theories have been applied on a range of deviant behaviours such as theft and violence,⁹ drug use,^10,11 alcohol use¹² and cigarette smoking.^1,2,4,6 Still, social control theories have been exposed to criticism and some studies have shown its predictive value to be low.^13,14 The general assumption that important norm-setting arenas in young people's life (e.g. family, school, religion) all hold the same conventional norms has been challenged, and this assumption may not always be valid.^15,16

Many preventive efforts to reduce students' smoking have been implemented within the structure of the elementary school. Most elementary schools have a written smoking policy¹⁷ that indicates a general interest in reducing smoking behaviour among students. We therefore find is reasonable to assume that student smoking is behaviour against conventional norms and that this societal smoking norm is experienced by Danish students during everyday school life.

However, students participate in several social arenas and the norms of these different arenas may not be identical. Besides school, family is another important socializing arena and young people internalize many parental norms.¹⁸ In terms of smoking, the parental smoking norm may not always be consistent with the conventional societal smoking norm experienced at school.

Social norms have often been operationalized by measurements of attitude or behaviour within the social environment. For adolescent smoking, attitude and behaviour are generally treated separately, since the predictive value of parental smoking attitude and parental smoking behaviour varies. Many studies show that parental smoking attitude predicts adolescent smoking,^19–21 while the predictive value of parental smoking behaviour varies from low or moderate^19,21,22 to high.^23,24

One type of psychosocial tie to school that is associated with adolescent smoking is school connectedness. No general definition of school connectedness exists. In the present paper the applied definition of school connectedness is based on the definition given by Karcher and Lee.²⁵ We state that school children's school connectedness is a feeling of relatedness to significant others at school and a general feeling of belonging at school. The effect of parental smoking norms on the association between school connectedness and adolescent smoking is unknown.

The objective of this study was (i) to test whether school connectedness and smoking was associated among male and female Danish school children, and if so (ii) to examine whether this association was influenced by parental smoking attitude and parental smoking behaviour.

Methods

The study is part of the cross-national study Health Behaviour in School-Aged Children (HBSC). The overall objective of the HBSC study is to increase our understanding and knowledge of health and health-related behaviours of young people in their social context.^26,27 The HBSC study comprises repeated cross-sectional questionnaire surveys targeted at 11-, 13- and 15-year-olds. The applied standardized questionnaire included a core of questions on sociodemographic factors, health behaviours, health, social relations, perception of self and well-being. The present analyses used data from 15-year-old students from the Danish survey in 1998.

Sample

The study population was all students in classes at grades five, seven and nine (11-, 13- and 15-year-olds) in a national, random sample of schools. To ensure a sufficient number of pupils from large cities the sampling of schools was stratified according to urbanization. Of the 64 selected schools, 55 agreed to participate in the survey. The response rate was 99% of the students present on the day of data collection, corresponding to 87% of the pupils formally enrolled in these 55 schools (n = 5205). The participating schools included 1578 students at grade nine (mean age 15.8 years, standard deviation 0.38).

Prior to the data collection the school board, headmaster and student's council of every participating school gave informed consent and the school nurse was informed. All students in the relevant grades present on the day of the data collection were asked to complete the questionnaire following standard instructions from the teacher. In order to protect the students' anonymity they returned their questionnaire in a sealed envelope.

Measurements

Two ways of coding of determinants were tested. First, all indecisive responses (missing, don't know, don't have/see) were left out of the analyses. Secondly, all determinants were coded conservatively, i.e. indecisive responses were kept within the reference group. The conservative coding kept the estimates of the association with daily smoking as close as possible to 1.0. The two coding approaches showed the same direction of the observed associations and similar conclusions regarding the combined effect of school connectedness and parental smoking attitude and behaviour. In order to avoid over-estimation of associations the conservative coding was applied.

Smoking behaviour was measured by the item ‘How often do you smoke tobacco at present?’ (‘every day’, ‘at least once a week but not every day’, ‘less than once a week’, ‘I do not smoke’). Students reporting smoking every day were classified as daily smokers.

School connectedness was measured by a sum scale constructed from the following three items. A: ‘How do you feel about school at present’ (‘I like it a lot’, ‘I like it a bit’, ‘I don't like it very much’ ‘I don't like it at all’); B: ‘Our school is a nice place to be’ (‘strongly agree’, ‘agree’, ‘neither agree or disagree’, ‘disagree’, ‘strongly disagree’); and C: ‘I feel I belong at this school’ (responses as for B). The Spearman coefficients between the three items varied between 0.39 and 0.53, Cronbach's coefficient α was 0.75. The school connectedness scale ranged from 3 to 15 and showed approximately the same distribution among boys and girls. The participants were subsequently categorized into tertiles: high (3–6), medium (7–8) and low school connectedness (9–15).

Parents' attitude toward student's future smoking was measured by two items: ‘Do you think your father/mother minds if you smoke two years from now?’ (‘a lot’, ‘a bit’, ‘not very much’, ‘not at all’, ‘don't have/see him/her’). Strict parental smoking attitude were defined as at least one of the parents minding ‘a lot’, and lenient smoking attitude were defined by both parents mind ‘a bit’, ‘not very much’ or ‘not at all’. Indecisive response (don't have/see him/her, missing) were categorized as strict smoking attitude in accordance with the principle of conservative coding of determinants.

Mother's, father's and best friend's smoking behaviour was measured by the following items: ‘Do any of the following people smoke: father, mother, best friend?’ (‘smokes daily’, ‘smokes sometimes’, ‘does not smoke’, ‘don't know’, ‘don't have/don't see this person’). Responses ‘smokes daily’ or ‘smokes sometimes’ identified smokers, while the remaining responses identified non-smokers. Missing responses were categorized as non-smokers. Two categories of parental smoking behaviour were constructed: (i) non-smokers, no parent smokes; and (ii) smokers, at least one parent smokes.

Parents' labour market attachment and best friend's smoking behaviour were included in the models as co-variables, as they are well-documented predictors of adolescent smoking behaviour.^28–31

Data on parents' labour market attachment were derived from the items ‘What is your father's (mother's) job? Please describe exactly what he (she) does, for example shop assistant, farm worker, lorry driver, dentist, hairdresser, teacher. You can write ‘don't know’ or ‘has no paid job at the moment’ or ‘unemployed’. In the present analysis we categorized families where at least one parent had a job as ‘employed’ and families where no parent had a job as ‘economically inactive’. Among students in ninth grade, 94% provided sufficient information to categorize into ‘employed’ or ‘economically inactive’. Missing responses were categorized as ‘employed’.

A combined variable of school connectedness and parental smoking attitude was constructed. This variable included the six possible combinations of high/medium/low school connectedness and strict/lenient parental smoking attitude. Likewise, a combined variable of school connectedness and parental smoking behaviour was constructed including the six possible combinations of high/medium/low school connectedness and non-smoking/smoking parents.

Data analysis

All analyses were performed using SAS version 8.2. Homogeneity in smoking behaviour among boys and girls was tested by χ²-test. For the multivariate analyses multiple logistic regression analysis was applied (PROC GENMOD). The students were sampled in clusters with the school as the primary sampling unit. The presence of clustering would be expected to result in higher standard errors compared with a similar size of sample obtained using simple random sampling. Therefore, odd ratios with 98% confidence limits were estimated. Forty-one students had missing responses on either smoking behaviour or school connectedness or on both variables. These students were excluded, and the final sample therefore consisted of 1537 students (752 boys and 785 girls).

Results

The prevalence of daily smoking among girls at grade nine was 22% and among boys was 15% (χ²-test for homogeneity: P = 0.001). Table 1 presents the prevalence of daily smoking among girls and boys by school connectedness, parental smoking attitude, parental smoking behaviour, best friend's smoking behaviour and parents' labour market attachment.

Model 1 in table 2 presents the crude associations between school connectedness and daily smoking for boys and girls estimated by logistic regression analysis. Model 2 constitutes the full model. The full model included parental smoking attitude, parental smoking behaviour, best friend's smoking behaviour and parents' labour market attachment. The association between school connectedness and daily smoking remained significant for both boys and girls. Among boys, significant main effects were also seen for best friend's smoking behaviour and parents' labour market attachment, while rather strong but non-significant associations were seen for parental smoking attitude and parental smoking behaviour. Among girls significant main effects were observed for parental smoking behaviour, best friend's smoking behaviour and parents' labour market attachment, while the effect of parental attitude was moderate and non-significant.

When the interactions school connectedness × parental smoking attitude and school connectedness × parental smoking behaviour were included in the model the only significant interaction was between school connectedness and parental smoking attitude among boys (P = 0.049). Sensitivity analyses were conducted testing for interactions by alternative cut-off points of parental smoking attitude and parental smoking behaviour. No important interactions were found (data not shown). As the objective of the present paper is highly analytical, the general recommendations by Rothman and Greenland³² were followed. Although most of the tested interactions were statistical insignificant the estimates were assessed. These may still hold information on mechanisms underlying behavioural patterns.

The effect of parental smoking attitude on the association between school connectedness and daily smoking is shown in table 3. Among boys who experienced a strict parental smoking attitude, the strength of the association between school connectedness and daily smoking increased with decreasing school connectedness and we found a marked effect of having low school connectedness compared with high and medium school connectedness. Smoking among boys experiencing a lenient parental smoking attitude showed no marked association with school connectedness. For girls who experienced a strict parental smoking attitude the strength of the association between school connectedness and daily smoking was weak, while girls who experienced a lenient parental smoking attitude had a pronounced effect on their smoking behaviour of having low school connectedness compared with having high and medium school connectedness.

The effect of parental smoking behaviour on the association between school connectedness and daily smoking is shown in table 4. For boys, school connectedness and daily smoking was inversely associated regardless of parental smoking behaviour. The same pattern was seen among girls, although the association among girls with non-smoking parents was less marked than among boys. Among girls with smoking parents there seemed to be a pronounced effect of having low school connectedness compared with having high and medium school connectedness.

Discussion

Our study confirmed the inverse association between school connectedness and adolescent smoking found in other studies.^1–6 Sometimes school connectedness has been considered an individual attribute,^6,25 while others have used school connectedness as indicator of a contextual factor, school climate.^33,34 This study focuses on the modifying effect of school connectedness regardless of the origin of school connectedness.

In accordance with previous studies we found that adolescent smoking behaviour was strongly associated with parents' labour market attachment^28,29 and best friend's smoking behaviour.^30,31 Although we found an effect of parental smoking attitude on adolescent smoking behaviour, this was non-significant and weaker than expected from the literature.^19–21 From previous studies no general conclusions can be drawn concerning the association between parental smoking behaviour and adolescent smoking.^{19,20,22–24} We found a strong association between parental smoking behaviour and smoking behaviour among girls and a moderate, non-significant association among boys. Our findings therefore also confirm the results reported by other studies that the effect of parental smoking behaviour on smoking behaviour is stronger among girls than among boys.^35–38

Especially among boys, we found a modifying effect of parental smoking attitude on the association between school connectedness and smoking behaviour but no modifying effect of parental smoking behaviour. This observation of varying modifying effects of parental smoking attitude and parental smoking behaviour is in line with previous studies identifying varying direct effects of parental smoking attitude and parental smoking behaviour on adolescent smoking.^19–24 The present analyses suggest that parents' smoking-related attitude and behaviour influence adolescent smoking behaviour differently whether the influence occurrs directly or by modification. Our results therefore confirm the need for separate analyses and interpretations of the effect of parental smoking attitude and smoking behaviour on adolescent smoking.

Among boys, the well-known association between school connectedness and smoking only existed when the parental smoking attitude was against smoking. Among girls the association between school connectedness and smoking mainly seemed to exist when they experienced a parental smoking attitude in conflict with the conventional norm, although the interaction was statistically non-significant. For both boys and girls the interaction between school connectedness and parental smoking behaviour was statistical insignificant. However, among girls an association between school connectedness and smoking only existed when they had smoking parents.

Social control theory assumes that important norm-setting arenas in young people's life all possess the same conventional norms. The present study showed that the effect on adolescent smoking behaviour of being connected to school was affected by parental smoking norms. Whenever social control theories are applied on the specific association between psychosocial ties to school and adolescent smoking one therefore needs to consider that young people are simultaneously exposed to conventional as well as non-conventional smoking attitudes and smoking behaviours in school and other significant norm-setting arenas. The presented results illustrate that adolescent behaviour may be influenced by complicated interactions of varying sets of norms.

The results presented here indicate some interesting differences between boys and girls in terms of the combined effect of school connectedness and smoking attitude in the family. Low school connectedness was not a major risk factor among girls who experienced a strict parental attitude, while low school connectedness constituted a risk factor among boys experiencing strict parental smoking attitude. Low school connectedness constituted a risk factor among girls only in combination with lenient parental smoking attitude. Boys who experienced lenient parental smoking attitude were at risk despite being connected to school. The same conclusions concerning gender differences can be drawn in terms of the combined effect on smoking of school connectedness and parental smoking behaviour, although high school connectedness seemed to be a more protective factor among boys than among girls who have smoking parents. Still, the results suggest that smoking behaviour of girls may be more sensitive to restrictive social influences than the smoking behaviour of boys.

Some strengths and shortcomings of the study may have affected the results of the analyses. Information bias: prior to data collection the full questionnaire was validated several times by focus group interviews and full-scale pilot tests in the classroom setting. Any validity problem identified was adjusted to minimize information bias. Additionally, several studies reported a relatively high validity of self-report measures of adolescent cigarette smoking.^39,40 Misclassification: all covariates were conservatively dichotomized. Therefore, the potential bias due to misclassification tends to underestimate the associations between the covariates and smoking behaviour. Selection bias: five of the nine non-participating schools were small schools from the city of Copenhagen, and this selective non-participation may have affected the results. The proportion of non-respondents was low. A detailed analysis of non-respondent schools or students has not been done due to the complete anonymity of the study.

These findings may have implications for future smoking preventive programmes. First, the results suggest that interventions should be directed towards adolescents' school connectedness. However, any such approach needs to consider the family smoking norms of the targeted young people. This is further complicated by the fact that family smoking attitude and behaviour seem to influence adolescent smoking behaviour differently. Secondly, the gender-specific differences probably reflect different mechanisms behind boys' and girls' smoking behaviour.^35,36,41,42 The general significance of school and family may differ between boys and girls at grade nine, and our results generate the hypothesis that at this time of life the norm setting roles of school and family may vary between boys and girls. Subsequently, this may cause gender differences in norm internalization.

In conclusion, the results presented here demonstrate that the understanding of adolescent smoking behaviour is complicated, and that the tasks faced by future smoking preventive programmes may appear boundless. Still, the harmful effects of adolescent smoking on general public health are highly considerable, and the hard work to improve future smoking preventive programmes targeted at young people needs to continue.

The study was financed by the Danish Research Council and the Danish Health Insurance Foundation.

References