How Can Peer Group Influence the Behavior of Adolescents: Explanatory Model

The current work aims to study both the peer group and family influence on adolescent behaviour. In order to achieve the aforementioned objective, an explanatory model based on the Structural Equations Modelling (SEM)was proposed. The sample used was the group of adolescents that participated in the Portuguese survey of the European study Health Behaviour in School-aged Children (HBSC). The Portuguese survey included students from grades 6, 8 and 10 within the public education system, with an average age of 14 years old (SD=1.89). The total sample of the HBSC study carried out in 2006 was 4,877; however with the use of the SEM, 1,238 participants were lost out of the total sample. The results show that peers have a direct influence in adolescents’ risk behaviours. The relationship with parents did not demonstrate the expected mediation effect, with the exception of the following elements: relation between type of friends and risk behaviour; and communication with parent and lesser involvement in violence behaviours and increased well-being. The negative influence of the peer group is more connected to the involvement in risk behaviours, whilst the positive influence is more connected with protective behaviours.


Introduction
There are a variety of negative health indicators that show a peek during adolescence, namely homicide rates, non-intentional injuries, driving under alcohol effect or infection by sexually-transmitted diseases (Mulye, Park, Neson, Irwin & Brindis, 2009). Experimenting substances also occurs usually during adolescents, a time of development in which tolerance is lower and the risk of dependency increases (Glaser, Shelton & Bree, 2010). Peers and family have a key role in promoting health during adolescence, as well as, the perception that youngsters have of their quality of life and subjective well-being. Health does not depend solely on the delivery of health care during illness; on the contrary, influence of different settings may be crucial (Gaspar & Matos, 2008).
Behavioural problems that occur during infancy and adolescence (particularly external problems, such as substance use and violence behaviours) may continue throughout adulthood, associated to social non-adaptation, substance abuse and conflicts (Bongers, Koot, Van der Ende & Verhulst, 2008). The peer group may on one hand, serve as a model and influence behaviours and attitudes, whilst on the other hand, it may provide easy access, encouragement and an appropriate social setting for consumption (Glaser, Shelton & Bree, 2010). Social Learning Theory suggests that it is not necessary for adolescents to observe a given behaviour and adopt it; it is sufficient to perceive that the peer group accepts it, in order to be able to opt for similar behaviours (Petraitis, Flay & Miller, 1995).
Peers may strongly determine preference in the way of dressing, speaking, using illicit substances, sexual behaviour, adopting and accepting violence, adopting criminal and anti-social behaviours and in many other areas of the adolescent's life (Padilla, Walker & Bean, 2009;Tomé, Matos & Diniz, 2008). An example of this is that the main motives for alcohol consumption given by adolescents are related to social events, which usually take place in the company of friends, namely: drinking makes holidays more fun, it facilitates approaching others, it helps relaxing or facilitates sharing experiences and feelings (Kuntsche, Knibbe, Gmel & Engels, 2005). Also, mimicking risk behaviours may be greater when consumption begins in the context of a social event (Larsen, Engels, Souren, Granic & Overbeek, 2010).
On the other hand, having friends allows to share experiences and feelings and to learn how to solve conflicts. Not having friends, on the other hand, leads to social isolation and limited social contacts, as there are fewer opportunities to develop new relations and social interactional skills. Friendship is also positively associated to psychological well-being (Ueno, 2004), whilst a conflicting relation with peers is negatively associated with health (Laftman & Östberg, 2006). Stronger friendships may provide adolescents with an appropriate environment to development in a healthy way and to achieve good academic results. Adolescents with reciprocal friendships mention high levels of feelings of belonging in school; at the same time, reciprocity and feelings of belonging have positive effects in academic results (Vaquera & Kao, 2008).
School is a setting where interpersonal relations are promoted, which are important for youngsters' personal and social development (Ruini et al., 2009); it is responsible for the transmission of behavioural norms and standards and it represents an essential role in the adolescent's socialisation process. The school is able to gather different peer communities and to promote self-esteem and a harmonious development between adolescents, which makes it a privileged space for meetings and interactions (Baptista, Tomé, Matos, Gaspar & Cruz, 2008). Adolescents spend a great part of their time at school, which also makes it a privileged context for involvement in or protection from risk behaviours (Piko & Kovács, 2010). Camacho, Tomé, Matos, Gamito and Diniz (2010) confirmed that adolescents who like school were those that more often were part of a peer group without involvement in risk behaviours; whilst those that mentioned they did not have any friends reported that they liked school less.
Despite the positive influence of the peer group during adolescence, the higher the adolescent's autonomy from the peer group, the higher his/her resilience against its influence. This resilience seems to increase with age, which may mean that it is associated with youngsters' maturity; and girls emerge in several studies as more resilient than boys (Sumter, Bokhorst, Steinberg & Westenberg, 2009). Another factor that may be found in the influence of the peer group is the type of friendship, which adolescents maintain with their peer group: if friends are close they have a greater influence on the other's behaviours (Glaser, Shelton & Bree, 2010). When the friendship is perceived as reciprocal and of quality, is exerts greater influence (Mercken, Snijders, Steglich, Vartiainen & Vries, 2010). Another factor, which has been identified as a possible factor of decreasing peer influence is assertive refusal. Adolescents that are able to maintain an assertive refusal are less susceptible to the group's influence (Glaser, Shelton & Bree, 2010). These are only some variables identified as possible factors decreasing peer influence.
The relation with parents may be a mitigating factor of the negative influence by peers. Communicating family rules and parental style have been inversely associated to substance, alcohol and tobacco consumption during adolescence. This influence is essential for adolescents' development up to adulthood. Communication between parents and adolescents emerges as a protective factor for alcohol, tobacco and substance use (Newman, Harrison & Dashiff, 2008). Sen (2010) observed that family meals could lead to creating a closer relation between parents and adolescents, by strengthening a positive relationship and avoiding certain risk behaviours, such as substance use amongst girls and alcohol consumption, physical violence and robberies, amongst boys. These differences between genders may be due to a greater importance that girls attribute to family activities, but they do not reveal that boys are indifferent to them, only that the relation between genders may differ. Huebner and Howell (2003) verified that parental monitoring and communication with parents protected adolescents of both genders from being involved in risk behaviours.
Parental monitoring can be defined as parents' knowledge about their children's activities, who they hang out with and what they do. It has been associated to protection of various risk behaviours throughout adolescence, such as substance use or sexual behaviours. It may vary according to age, gender or ethnicity and it generally decreases with age (Westling, Andrews, Hampson & Peterson, 2008). Tobler and Komro (2010) confirmed with a sample made up of 2,621 adolescents in the 6 th and 8 th grades that the role of parents in the prevention of substance use during adolescence is essential; and that communication and parental monitoring were the factors, which most contributed to those results.
The greater the parental monitorings, the lower the adolescents' involvement in risk behaviour. Li, Stanton and Feigelman (2000) confirmed in a longitudinal study with afro-American children and adolescents, that parental monitoring was a very important factor in reducing risk behaviour. Parental monitoring emerged as inversely correlated with risk behaviours. The correlation persisted throughout age, suggesting that its protective effect is persistent in the long-term. Other studies found associations between parental monitoring and substance use and other behavioural problems amongst youth (Simons-Morton & Chen, 2005). Rai and collaborators (2003) also found a positive influence associated with parental monitoring, namely protection against substance use and sexual behaviours, but not condom use. On the other hand, the peer group was found to influence all risk behaviours assessed by the authors. The youngsters that had the perception of the involvement of peers in certain behaviours were more involved in similar behaviours; the same was found for those that presented a problematic relationship with their parents.
Literature seems to suggest that the peer group has an important role throughout adolescence; nevertheless it may influence negatively adolescents' risk behaviours, by enhancing their involvement in such actions. On the other hand, parents have a protective role in the same behaviours, generally associated with good communication and parental monitoring.
Taking into account the aforementioned findings, the aim of this study will be to analyse how peer influence is associated with: 1) risk behaviour, 2) violence, 3) health behaviour, 4) well-being and 5) feelings about school; and whether that influence may be moderated by adolescents' relations with parents.

Sample
The sample used was the group of adolescents that participated in the 2006 Portuguese survey (continental Portugal) of the European study Health Behaviour in School-aged Children (HBSC) (http://www.hbsc.org).
The HBSC Study started in 1982 with a team of researchers from Finland, Norway and England and from 1985/86 it has been carried out every four years. Throughout the years, the study has grown in importance and currently there are 44 participating countries from Europe and North America, in collaboration with the World Health Organisation (Currie et al., 2004;Currie, Samdal, Boyce, & Smith, 2001). Portugal has participated in the survey since 1996, when an initial pilot study adapted to the Portuguese population was carried out, in accordance with the international protocol (www.hbsc.org.com). The study aims to understand further adolescent behaviour in relation to health and to understand health and well-being in the social context, by gathering data that enable national and international comparisons.
The Portuguese study in 2006 included students from grades 6, 8 and 10 within the public education system, with an average age of 14 years (SD=1.89). The total sample is 4,877 students from 257 school classes in 125 Portuguese schools selected randomly. In the present study, due to the statistical analisis used, 1,238 students were left out, which accounts for a final total sample of 3,639 adolescents. It is representative of the abovementioned school years and stratified by Administrative Education Regions. Students were distributed in the following manner: 50,4% were girls; 31,7% were in grade 6, 35.,7% in grade 8 and 32,6% in grade 10.

Procedure
The analysis unit used was the class. In each school, classes were randomly selected with the aim to find the needed number in each class, which was proportional to the number provided by the Ministry of Education. Teachers administered the questionnaire in the classroom. Students' participation was voluntary. Schools and participating classes were selected randomly, out of a list provided by the Ministry for Education. The questionnaires were sent to teachers with the procedure, where it was explicitly asked that the students' participation was to be anonymous and voluntary. The students that did not wish to complete the questionnaire could leave the classroom.

Measures and Variables
The HBSC Survey 2006 was used for the data gathering, in accordance with the respective protocol. Amongst other, this questionnaire provides demographic data, well-being indicators (life quality related with health, happiness and life satisfaction) and indicators on peer relations Currie et al., 2004;Currie, Samdal, Boyce & Smith, 2001). In this study, variables associated with adolescents' relations with peer groups, their relations with parents, school environment and behaviours of risk, violence and well-being were used.

Statistical Analysis
With the aim to analyse the proposed explanation model, the SEM was used, as a means to assess the quality of mediation of a group of variables. The statistical software EQS, Structural Modelling Software, version 6.1 was used.
Before completely testing the model, it was necessary to test the model partially, through a confirmatory factorial analysis (CFA). Three mediation models were tested: the independent mediation model, which tested the relation of the mediation quality between the independent variables, defined by peer influence, through friends with risk behaviours, made up of five indicators about type of friends, friends with protective behaviours; three indicators about friends' protective behaviours, communication with friends comprised of three indicators and friendship quality, comprised of ten indicators about friendship quality. The mediating mediation model was comprised of the variables related with adolescents' relation with parents through communication with the father and the mother, as well as, parental monitoring, comprised of five indicators associated with what parents know about their children. Finally, the dependent mediation model, tested the quality of mediation between the dependent variables, namely low involvement in risk behaviours, comprised of three indicators associated with tobacco, alcohol and drug use; low involvement in violent behaviours, comprised of three indicators associated with provocation, carrying a weapon and involvement in fighting; health, comprised of eleven indicators related with physical and psychological symptoms; well-being, comprised of twelve indicators associated with happiness, life satisfaction and adolescents' quality of life; and feelings about school, comprised of one indicator (Bentler, 1995).
After the mediation models' analysis some indicators with saturations lower than .40 were eliminated. In total eight indicators were eliminated. Covariances between error measures were also introduced, in a total of eight covariances, two in the independent mediation model and six in the mediating mediation model.

Analysis of the Mediation Model
In the realisation of the model, 1,238 participants were eliminated from the sample. The results obtained in relation to the adequacy of the explanation model proposed showed that it presented lower levels of adequacy (see Table 1, step 1). However, the analysis of the results obtained in the Langrange Multiplier test (LM test), a test that assesses the need to add new parameters to the model (Bentler, 1995), showed that the introduction of some connections between factors, would decrease significantly the value of the qui-square, amongst other, the parental monitoring factor, the communication with parents factor and the factor on less involvement in violence behaviours. Amongst the factor on less involvement in risk behaviours and the factor friends with risk behaviours. Amongst the factor low involvement in violence behaviours, the factor on friends with protective behaviours and the factor on less involvement in risk behaviours. Amongst the factor on health and the factor on friendship quality, the communication with friends factor, less involvement in risk behaviours and the factor on well-being, and amongst the factor on feeling about school and less involvement in risk behaviour. A decision was made to add them and to re-evaluate the model (see Table 1, step 2). After the introduction of these parameters, the results show better levels of adequacy of the model. Finally, the results obtained in the Wald test, which show the non significant m odel parameters (Bentler, 1995), were analysed. These showed the existence of some non significant relations, which were eliminated, namely between communication with parents and friends with protective behaviours; between parental monitoring, friends with risk behaviours and communication with friends; between lower involvement in risk behaviours and communication with parents; between lower involvement in risk behaviours and parental monitoring; and between health and parental monitoring. The results obtained after the elimination of these parameters are shown in Table 1, step 3. Table 1 shows that the procedures undertaken have improved the adjustment index of the structural model. It can be seen that the adjustment index are acceptable and that the model shows a good adequacy.

<Table 1>
The standardised solution obtained (Figure 1) allows to confirm that the factors with a stronger impact in relation to lesser involvement in risk behaviours are having friends with risk behaviours (β=-.83). The negative indicator shows that the lower the number of friends with risk behaviours, the lower the involvement in risk behaviours. For a lower involvement in violence behaviours, the factors with a greater impact are a lesser involvement in risk behaviours (β=.58), followed by friends with protective behaviours (β=.35) and communication with parents (β=-.21). Following this, the lower the involvement in risk behaviours and the greater number of friends with Having a higher number of friends with more risk behaviours also emerges as a factor with a high impact in involvement in risk behaviours, which is in line with several studies that have identified peers as the variable with the greatest influence in the involvement in such behaviours (Glaser, Shelton & Bree, 2010;Sieving, Perry & Williams, 2000). On the other hand, the factor with a greatest impact in low involvement in violence behaviours is a low involvement in risk behaviours and having a higher number of friends involved in protective behaviours. This means that, having friends with little risk behaviours and having friends with protective behaviours, prevent violence and risk behaviours.
A lower involvement in risk behaviours is yet identified as the variable, which positively influences adolescents' health, together with feelings of well-being; and the variable with influences positively, adolescents' feelings about school. Taking into account that adolescents seem to choose to be less involved in risk behaviours when they have friends that are not involved in risk behaviours; although peers' influence is indirectly related, it is very important for adolescents' health and well-being, as suggested in other studies with similar results (Tomé, Matos & Diniz, 2008;Laftman & Ostberg, 2006;Ueno, 2004).
Parents' role seems to be more important for the variables associated with adolescents' well-being and health (Camacho et al., 2010;Newman, Harrison & Dashiff, 2008), considering that communication with parents emerges as the factor with a stronger impact in adolescents' well-being, which has a stronger impact in health. This presupposes that a better communication with parents means stronger feelings of well-being and, as a result, healthier adolescents.
Parental monitoring does not emerge as a significant variable neither in the mediation of risk and violent behaviours or the promotion of well-being, health or feelings about school. It is reinforced by communication with parents and it is dependent upon a lower involvement in violence behaviours. That dependency shows an unexpected result. Contrary to what is documented in literature (Camacho et al., 2010;Newman, Harrison & Dashiff, 2008), it seems that parental monitoring is greater when adolescents' involvement in violence behaviours is lower and not the opposite. But we cannot ignore their essential role in parents' relationship with adolescents (Huebner & Howell, 2003) and take into account the fact that there may be differences associated with gender, age and other variables (Westling et al., 2008), which may influence the results found and limit the Step 3