OUP user menu

Socio-economic position and adolescents' health in Italy: the role of the quality of social relations

Alessio Zambon, Patrizia Lemma, Alberto Borraccino, Paola Dalmasso, Franco Cavallo
DOI: http://dx.doi.org/10.1093/eurpub/ckl051 627-632 First published online: 9 May 2006


Background: The quality of social relations in adolescence is possibly one of the major determinants of habits that can influence the health of young people, and it may also be one of the mediators of the effect of social position on health. In this paper we propose to test these hypotheses for Italian adolescents, in order to suggest interventions aimed at improving their health. Methods: The Italian data of the HBSC (Health Behaviour in School-Aged Children) survey 2001–02 have been analysed, and the distribution of the perceived quality of social relations has been described, stratified by age, gender, and economic well-being. Logistic models have been fitted using health behaviours as dependent variables and economic well-being and social relations as determinants. Results: The quality of relations with adults seems to decrease consistently from age 11 through age 15, while the relation with peers improves. The relation with the father seems positively correlated with economic well-being. Difficult relations with adults are associated with higher probability of smoking, drinking alcohol and using cannabis; difficult relations with peers are associated with lower physical activity and lower probability of having used cannabis. Conclusions: Even if the relations with adults become less important in adolescence, they are still associated with health behaviours. Our results fit the framework of socialization theories and can be used for planning adequate health education interventions.

  • adolescence
  • health behaviours
  • lifestyles
  • social relations
  • social support

Among the aspecific risk factors for young people's health and health behaviour, the quality of social relations plays a very important role. The association in adolescence between a high-quality network of social relations and a good state of health, and positive health behaviours, is well established.13

Since there is a strong correlation among many unhealthy behaviours,4,5 many authors have proposed to investigate the link between the low quality of social relations and such patterns of unhealthy (and possibly deviant) behaviours.6

Hypotheses on the mechanisms through which the quality of social relations influences health are summed up by Geckova.1 These authors agree with Pratt2 in identifying four main groups of mechanisms: (i) emotional support can enhance self-esteem and self-efficacy, which in turn are major recognized determinants of healthy behaviours and lower stress7; (ii) emotional support can help in converting intentions into actions (instrumental support); (iii) informational support provides a wider range of coping styles; and (iv) informational support allows for a higher possibility of being influenced and motivated in the adoption and development of healthier behaviours (appraisal support). Different types of social support also seem to be different for males and females.8

Far more established is the fact that social class, as well as age and gender, plays an important role in determining the perceived quality of young people's relations with peers and adults, particularly parents and teachers.1,2,7,9

The main objective of this paper is, thus, to verify whether the socio-economic position of Italian early adolescents is associated with the adoption of some basic healthy behaviours, and whether the perceived quality of their social relations mediate this relation. An underlying goal is to provide indications for planning strategies targeted at improving the supportive quality of young people's social relations. It starts, therefore, from the description of the perceived quality of interpersonal relations in a representative sample of the Italian adolescent population and analyses subsequently its variation with age, gender, and social class, in order to test whether this variable can be considered to have a mediating effect between social class and health.


The sample

The sample comprises 11, 13, and 15-year-old Italian boys and girls, chosen through a systematic sample of school classes divided into five geographical strata (North-West, North-East, Centre, South, Islands). The total sample number is 4386 youngsters (51.55% female; 34.75% 11 year olds, 37.23% 13 year olds, and 28.02% 15 year olds). Self-administered questionnaires were sent to 314 school classes, with an average response rate of 77.4%. The methods of data collection and the questionnaire itself (besides a few additional items) follow the HBSC study's international research protocol,10 which guarantees high-quality data. The international collaboration has been going on successfully since 1982; each item on the questionnaire has been chosen and defined on the basis of the existing scientific literature and/or has been previously piloted, as has the questionnaire as a whole. Research teams are allowed to participate in the international network only after proving their ability to carry out such a survey; a process of translation and independent back-translation guarantees the consistency of national questionnaires and cross-country comparability of data.

Assessment of quality of social relations

For father, mother, and best friend, we asked the children how easy it is for them to talk to that person about really troubling issues, response categories being: ‘Very easy’, ‘Easy’, ‘Difficult’, ‘Very difficult’, ‘I don't have or see this person’. For the relation with teachers, we used a five category Likert scale: for the statement ‘Teachers treat students fairly’, response categories were: ‘Strongly agree’, ‘Agree’, ‘Neither agree nor disagree’, ‘Disagree’, ‘Strongly disagree’.

Measurement of social class

The measurement of social class presents several problems, principally deriving from the complexity of the very concept of social class. Available measures in the HBSC questionnaire were the children-reported parental occupation and the highest level of education achieved (which both showed a very low reliability, at least in the Italian data) and the Family Affluence Scale10 (FAS) based on the purchasing capacity of the family. We decided, therefore, according to the HBSC international research protocol,11 to use the FAS index and aggregate it in three ordinal categories: low, medium, and high.

Assessment of health behaviours

Health behaviours were assessed through self-reported frequency items, in order to check how often young people adopt each of the considered behaviours. For this study we were interested in smoking, alcohol use, frequency of physical activity, and history of cannabis use. The items were already tested in previous surveys, or validated within the international HBSC study. Brener et al.12 reassures us on the employment of such questions in adolescence surveys: the HBSC research protocol follows the recommendation of this author to maximize their validity and reliability.


The quality of interpersonal relations has been described, stratified by age, gender, social class, and geographical area. The association between health behaviours and FAS has been tested by the ordinal by ordinal gamma index.

To test the hypothesis that health behaviours may be associated with social relations, a logistic model has been fitted, where health behaviours were taken, one at a time, as dependent variables and social relations as independent variables. Age, gender, and socio-economic position have been used as control variables. Dependent variables were dichotomized as follows: ‘Smoking daily’ versus ‘Less often or not smoking at all’; ‘Drinking any alcoholic beverages once a week or more’ versus ‘Less often’; ‘Being physically active for at least 1 h at least 2 days a week’ versus ‘Less often’; ‘Having tried cannabis’ versus ‘Not having tried cannabis’. Independent variables were also dichotomized: for father, mother, and best friend, talking easily or very easily were combined, and used as a reference category against the aggregation of the others. For teachers the aggregation of ‘Strongly agree’, ‘Agree’, and ‘Neither agree nor disagree’ (to the statement ‘Teachers treat students fairly’) were used as a reference category, against the aggregation of ‘Disagree’ and ‘Strongly disagree’.


Quality of interpersonal relations as declared by the interviewed children, stratified by age and gender, is summarized in table 1.

View this table:
Table 1

Percentage of boys and girls in different age groups declaring easier or more difficult relations with father, mother, best friend, and teachers

11 year olds (%)13 year olds (%)15 year olds (%)11 year olds (%)13 year olds (%)15 year olds (%)
Relation with father
    Very easy or easy74.3964.6660.2260.6043.3938.33
    Difficult or very difficult23.3032.0136.6537.0753.0156.62
    Don't have or see this person2.323.333.132.333.615.05
    Total (N)100.00 (777)100.00 (781)100.00 (543)100.00 (731)100.00 (832)100.00 (673)
Relation with mother
    Very easy or easy85.3179.0073.7481.9478.8670.35
    Difficult or very difficult11.8719.9524.5812.0819.9228.30
    Don't have or see this person1.821.041.680.971.211.35
    Total (N)100.00 (769)100.00 (772)100.00 (537)100.00 (720)100.00 (828)100.00 (668)
Relation with best friend
    Very easy or easy83.4088.0287.1487.5291.9992.37
    Difficult or very difficult14.9910.009.609.115.703.97
    Don't have or see this person1.612.973.263.372.313.66
    Total (N)100.00 (747)100.00 (740)100.00 (521)100.00 (713)100.00 (824)100.00 (655)
‘Teachers treat students fairly’
    Strongly agree, agree or neutral89.9086.3979.0493.3191.2077.96
    Disagree or strongly disagree10.1013.6120.966.698.8022.04
    Total (N)100.00 (772)100.00 (786)100.00 (539)100.00 (732)100.00 (841)100.00 (676)

It appears evident that girls have more difficult relations than boys with their fathers: only 60.6% of 11-year-old girls have easy or very easy relations with their fathers, against 74.4% of 11-year-old boys. As the age increases, we observe a reduction in these percentages: 64.7% for 13-year-old boys and 60.2% for 15-year-old boys, while for girls percentages are, respectively, 43.4 and 38.3%. The relative gap seems to increase with age: in fact, at 11, the percentage of boys finding it difficult or very difficult to communicate with their fathers is 14 points less than for girls, while at 15 the difference reaches 20 points.

No major difference was observed between boys and girls in the quality of the relationship with the mother; however, we observe a worsening in this relation as age increases, the perception of an easy or very easy relation dropping from 85.3% for 11-year-old boys and 81.9% for 11-year-old girls, to 73.7 and 70.4%, respectively, at the age of 15. This worsening effect appears to be less marked than that for fathers, especially for girls. In any case, in both sexes, the relation with the mother is easier than with the father.

The relation with the best friend shows, on the contrary, an improvement: those considering it easy or very easy at the age of 11 are 83.4% among boys and 87.5% among girls, while at 15 percentages are, respectively, 87.1 and 92.4%. Thus girls show even higher percentages than boys, and these remain stable as age increases.

As far as the relation with teachers is concerned, it seems better for girls, but only in middle school; during high school this difference disappears. Students in middle school who believe they are treated quite fairly are 89.9% among boys and 93.3% among girls. These percentages are lower at age 13 where we have 86.4% among boys and 91.2% among girls, dropping to 79.0 and 78.0%, respectively, at age 15.

Regarding social class (table 2), we can observe that the relation with the father is influenced by FAS in a statistically significant way (P < 0.01): 50.7% of low FAS adolescents rate the quality of this relation as easy or very easy, against 60.9% of those with high FAS score, this variation being slightly more evident for girls and for younger adolescents (results not shown). No significant association appears between FAS and quality of the relation with mother and teachers. FAS seems to influence instead the quality of the relation with the best friend: a positive association with the category ‘very easy’ (from 48.9% of those with low FAS score to 57.8% of those with high FAS score), and a negative one with the category ‘easy’ (from 37.5% of those with low FAS score to 31.7% of those with high FAS score) can be observed. These patterns are the same for all ages and genders.

View this table:
Table 2

Percentage of boys and girls in different FAS groups declaring easier or more difficult relations with father, mother, best friend, and teachers (stratification by age and gender is not shown since there are no significant differences in these variables for FAS effect)

FAS (%)
Relation with fathera
    Very easy or easy50.6858.0160.91
    Difficult or very difficult44.0538.6337.86
    Don't have or see this person5.273.361.23
    Total (N)100.00 (1101)100.00 (2022)100.00 (1141)
Relation with mother
    Very easy or easy77.8479.5680.91
    Difficult or very difficult20.6019.1917.85
    Don't have or see this person1.561.251.24
    Total (N)100.00 (1092)100.00 (2006)100.00 (1126)
Relation with best friend
    Very easy or easy86.4288.4889.57
    Difficult or very difficult10.008.937.89
    Don't have or see this person3.582.592.54
    Total (N)100.00 (1060)100.00 (1970)100.00 (1103)
‘Teachers treat students fairly’
    Strongly agree, agree or neutral84.4086.1985.81
    Disagree or strongly disagree16.6013.8114.19
    Total (N)100.00 (1113)100.00 (2028)100.00 (1142)
  • a: Ordinal by ordinal gamma index shows a significance for alpha = 0.01 for this sub-table

Table 3 presents the results of the logistic models fitted with respect to the health behaviours taken as dependent variables.

View this table:
Table 3

OR (95% CI) for the effect of the relations with father, mother, best friend, and teachers on four unhealthy behaviours (smoking daily, drinking alcohol once a week or more often, doing physical exercise for 1 h twice a week or less often, having used cannabis), controlled by gender, age, and FAS

Smoke daily (only 15 year olds)Alcohol useLow physical activityCannabis (only 15 year olds)
Male (ref.)
Female0.92 (0.65–1.29)0.38 (0.32–0.45)1.62 (1.41–1.86)0.34 (0.23–0.50)
    11 year olds (ref.)1.001.00
    13 year olds2.32 (1.88–2.87)0.96 (0.82–1.13)
    15 year olds4.48 (3.61–5.57)1.59 (1.34–1.89)
High FAS (ref.)
Medium FAS0.84 (0.58–1.23)0.64 (0.53–0.76)1.29 (1.09–1.53)0.76 (0.50–1.15)
Low FAS0.99 (0.63–1.56)0.72 (0.58–0.90)2.05 (1.70–2.47)0.96 (0.57–1.59)
Father: easy relation (ref.)
Father: difficult relation1.13 (0.79–1.63)1.14 (0.96–1.36)1.03 (0.89–1.20)1.83 (1.22–2.74)
Mother: easy relation (ref.)
Mother: difficult relation1.47 (1.02–2.12)1.57 (1.30–1.90)0.96 (0.81–1.15)1.57 (1.06–2.34)
Best friend: easy relation (ref.)
Best friend: difficult relation0.78 (0.44–1.39)0.76 (0.59–0.98)1.45 (1.18–1.78)0.43 (0.21–0.89)
Teachers: treat us fairly (ref.)
Teachers: treat us unfairly1.62 (1.12–2.34)1.63 (1.32–2.01)0.95 (0.77–1.16)1.77 (1.17–2.66)
  • Figures in bold indicate a statistically significant OR

From these data we can deduce the associations between indicators of the quality of social relations and four health behaviours strongly influencing the health of young people.

While the relation with the father seems to be associated only with cannabis use [the estimated OR (odds ratio) for having tried it is 1.83 for those with a more difficult relation with respect to those with an easier relation], a more difficult relation with mother and teachers is also associated, besides cannabis use, with alcohol consumption and smoking habits, with an OR almost always >1.5.

Therefore, in summary, an easier relation with adults, in particular parents and teachers, correlates with the considered health behaviours, except for the amount of physical activity.

The relation with friends shows different associations. First of all it is the only one associated with amount of physical activity: those who have a difficult relation with their best friend show a 1.45 higher chance of not exercising at least three times a week. For substance abuse, an easy relation with the best friend is often related with the use of substances such as marijuana. Marijuana use is less likely among those not having an easy relation with their best friend (57% less), as is the weekly use of alcohol (24% less).

As for the relation with mothers and teachers, these are the only ones associated with an increased risk of daily smoking.

A higher use of alcohol is associated with a worse relation with mother and teachers and with a better relation with the best friend. Girls show a lower risk of alcohol consumption than boys, and teenagers in privileged social classes use alcohol more frequently.

Concerning frequency of physical exercise, we can observe that it is associated with an easy relation with the best friend and with belonging to a privileged social class. Girls have a 62% higher chance of doing little physical activity than boys.

The fact of having tried marijuana at least once is much lower among girls (66% lower with respect to boys) and is associated with poor relations with adults. This behaviour is not associated, in the Italian population, with the FAS score.

Among the control variables, gender is most important: girls have a much lower risk of drinking alcohol and having tried marijuana (62 and 66% less, respectively), while there is a higher risk (62%) of little physical exercise and no major influence on smoking habits. With respect to age, we observe an increased risk for higher alcohol consumption and low physical activity (no observations are possible for smoking or for the use of marijuana, as we have only data concerning 15 year olds).

Finally, a disadvantaged social position appears to involve less physical activity and higher alcohol consumption but has no significant association with smoking or marijuana.


One important limitation of our study is that while the HBSC survey is school based, the 10th grade is not compulsory in Italy, so the population of 15-year-old adolescents in our study is only representative of those who chose to go to high school. Another limitation of our study is that we could not control for parents' habits, which we know to be very important in shaping their children's personal identity13 and healthy habits.14,15 Given these limitations, we will try now to explain the phenomena we observed, in terms of the peculiarity of the population we studied and its similarities with others found in literature.

As for the perceived quality of interpersonal relations in this sample of Italian adolescents, the observed worsening of the relations with parents and teachers with increasing age, and, correspondingly, the improving relation with peers, are consistent with results reported in international literature and with the more accepted/recognized developmental theories. At this age (11, 13, and 15 year old) adolescents are becoming more and more independent and are preparing to enter a new social network. Their self-esteem is no longer based entirely on the family of origin, but new sources for identity are sought, with many difficulties, in the new relations with peer groups and sexual partners.16 The present and past relations with parents are, nevertheless, still active in many ways, linking these two periods of life: the influences received from parents (values, attitudes, capabilities, etc.) also partially determine the kind of friends chosen by boys and girls.5,17

Adolescents' socio-economical position is also associated with the group of variables measuring the quality of social relations: relations with the father and with peers seem to be positively correlated with social class. This is not so for the relation with the mother and teachers, for which it has not been possible to find a social gradient. As an explanation of this fact, which shows similarities with previous findings,1,3 we can hypothesize a lower capability of establishing supportive social relations for adolescents of lower social class. This phenomenon was investigated in the 1960s and 1970s,1820 while recent literature on this issue is hard to find. These scholars, and also Bourdieu and the sociologists in his tradition, showed how the communication capabilities of children of lower social classes might be negatively affected by a limited access to the official culture and to more elaborated linguistic codes. Elaborated codes are those used with people external to the narrower social environment, and, therefore, requiring a higher level of abstraction. In this way the family, and its social class in particular, also influences the choice of peers and the quality of relations with them.5,17,21

Our data also show that relations with different kinds of people (adults, peers) are associated with different health outcomes. In general we can say that the mechanisms through which the influence of different persons is exerted can vary. Duncan5 highlights, for the relations with parents, the fact that inept parental monitoring and parent–child conflict can hinder the development of self-control skills, and encourage behaviours expressing contrast with the family, with rebellion against others acted out through self harm (smoking and other dangerous behaviours). From this standpoint, it is clear that in Italy older adolescents and girls (given their worse relation with the father) are at a disadvantage and more at risk of developing attitudes of this kind. In more recent years, Duncan5 suggests that time spent within the family and involvement in organizing family life could be important protective factors contrasting the mechanisms leading to unhealthy behaviours. A good quality of social support within the family emerges as one of the most important protective factors; the quality of family relations is independent from family composition, which does not seem important.5,14

Teachers are the first important adults with whom children have relations outside the family, so they should receive proper training and support. It is true that the relations within the family, being the first and most formative ones, influence the future of the children's social life,18,19 as also shown by the research on problem behaviours,22 but we should be aware that relations external to the family can have an independent impact. Geckova et al.,1 for example, underlines the fact that teachers' support is especially important for those children reporting low family support: they look for other adults' support, but difficult family communication also makes it difficult to communicate with other adults leading to a vicious circle of low general support. Anyway we have seen that the relation with teachers also has an independent role in several of the considered behaviours, so we can assume that prevention should also act on this dimension. It is especially important in Italy where, until only a few years ago, middle-school and high-school teachers had no relational training at all, so the quality of their relations with their pupils were totally left to individual skills and training.

Peer deviance is often considered one of the main determinants of unhealthy behaviours, and we saw an independent, negative effect of a higher quality relation with the best friend on health behaviours (table 3). The time spent with friends is not necessarily a risk factor; if friends are not deviant, time is spent in constructive activities,4 and the family has a big role in the process of peer selection.5 In adolescence self-esteem is very fragile, since young people lose part of parental support and strive to establish a new base in the peer groups. Hence being accepted by friends becomes very important, and it may appear that adopting dangerous, self-harming behaviours is an important way to gain their approval. In this context we want to point out that the old view of ‘bad company’, as the only important determinant of substance abuse and dangerous behaviours, is quite misleading. This view is also adopted by some scholars23: they view peer modelling, availability of substances in the group, peers' norms and expectations as the main determinants of alcohol abuse. Much more convincing are other suggestions17,24: in their research they use network analysis to check a two-stage model: at first peer selection and the acquisition of a new social context are focal, then influence by peers, and conforming to the peer group, becomes the main issue. The first stage seems to be influenced by the amount of time spent with the family and by positive school values: those who score low on these dimensions choose friends who smoke more than themselves (but no such effect is observed for alcohol use). On the other side, conforming is positively influenced by high peer acceptance and high friendship quality.

The process of selecting friends is also important in another way: not everyone may be able to gain the friendship of those they wish to be friends with, and some adolescents have more opportunities for finding more different friends, a situation that gives more opportunities for behavioural change. Thus adolescents with a broader network, and higher relational skills (mostly learnt within the family), will probably be able, in the future, to resist conformation to unhealthy behaviour. The fact that association with deviant peers derives from poor social skills and rejection by other peers is also shown by others.25,26

We have observed three behaviours (smoking, drinking alcohol, and using cannabis) that are mostly experimented in order to gain peers' acceptance: all these show very similar models for the influence of social relations, and ORs for acquiring that behaviour (even if not significant in some cases) are very similar. For example, the relation with the mother has, for the three behaviours, ORs equal to 1.47, 1.57, and 1.57, respectively; on the contrary, the relation with the best friend scores 0.78, 0.76, and 0.43. This suggests the idea, which needs to be explored further, that all these behaviours have a similar social function, and we can only hypothesize that this function is related to the search for acceptance, especially for young people who did not receive enough support (leading to a more autonomous perception of self-value) within the family. Physical activity is, in Italy, much less related to acceptance by other people, and it follows different patterns; only the relation with peers seems to correlate with this variable, and in fact physical activity is something that is performed mostly together with peers. We can also observe that physical activity is influenced more than any other behaviour by socio-economic position.

The observed association between the group of variables measuring social relations and the group related to health behaviours indicates that, in order to promote better health among young people, it is very important to plan interventions aimed at improving the quality of their social relations. On the basis of our results, and comparing them with the international literature, it is possible to elaborate some recommendations for social and pedagogical policies aimed at improving young people's health and ‘health potential’. Interventions should be as systemic as possible, involving above all the families and their relational skills, and also teachers and the school environment, while peer interactions should be enhanced in a collaborative and constructive way, rather than being based on refusal of specific (deviant) peers. This last statement contrasts with theories based on values such as ‘intolerance of deviance’ or ‘traditional attitudes’.27,28 Such theories, in fact, seem to propose a restraint, rather than a development, of social skills: the widely used ‘resistance skill’ concept29 seems to really be a ‘victim blaming’ attitude, in that it leads to the condemnation of friends belonging to the same social environment. A real mastering of social skills, developed through family, and basic trust and self-esteem are more difficult to tackle, but probably offer stronger protection against self-harm. Rather than excluding peers, interventions should be based on the construction of collaborative and constructive interactions.

The situation of females and of young people of low social class should be an important specific target of prevention strategies, to be addressed through an in-depth and broadly based needs analysis among these groups, with wider use of qualitative research techniques.

Key points

  • Study question: Is the socio-economic position of Italian adolescents associated with the adoption of some important health behaviours? Is the quality of their social relations a possible mediator of this association?

  • Main results 1: The quality of the relations with adults tends to decrease between the ages 11 and 15. Girls have a worse relation with their fathers than boys.

  • Main results 2: A high socio-economic position is associated with more physical activity; the quality of relations with peers seems to mediate this association.

  • Main results 3: Smoking, alcohol, and cannabis use do not show the expected association with socio-economic position, but they are inversely correlated with the quality of relations with adults.

  • Implications: Relations with adults and peers should be addressed with systemic intervention programmes when implementing health promotion policies.


This study is carried out within the frame of the international HBSC study. HBSC is an international study carried out in collaboration with WHO/EURO. The International Coordinator of the 2001/02 survey is Candace Currie (University of Edinburgh); Data Bank Manager: Oddrun Samdal (Univeristy of Bergen). For details, see http://www.hbsc.org. This study was funded by the Italian Ministry of Education and Research (Cofin 2003), University of Turin (‘quota ex 60%’, 2003) and Piedmont Region (‘Convenzione HBSC’, 2004).


View Abstract