FormalPara Overview

  • Social network research is comparatively advanced in adolescence, based on school-based surveys conducted mainly since the 1960s.

  • Studies in adolescence focus primarily on health behavior (especially tobacco consumption but also alcohol consumption, nutrition, and physical activity) and to a smaller extent on psychosocial health.

  • To explain the homophily of peer groups in adolescence, two different mechanisms are assumed that can only be investigated in longitudinal studies:

    • Thesis of social influence: Friends influence the (health) behavior and attitudes of young people and adapt them.

    • Thesis of the selection mechanism: Adolescents choose their friends according to whether they show the same attitudes and (health) behavior as they do themselves.

    • Both theses were empirically proven.

  • While the importance of social networks in adolescence for health and health behavior has been demonstrated, there is still a considerable need for research on the importance of social networks in explaining health inequalities.

  • Only a few studies exist that identified the relevance of the peer group in the context of socioeconomic/educational inequalities and health.

  • There is a need for research regarding the role of social networks in explaining health inequalities (beyond tobacco consumption) as well as longitudinal research designs.

1 Introduction

People are connected, and so, their health is connected.

(Christakis & Fowler, 2008, p. 2257, The New England Journal of Medicine)

This paper discusses the importance of social networks for health and health behavior and especially health inequalities in adolescence. Adolescence is characterized by a variety of changes that occur in this phase of life. Adolescents are confronted with challenges such as developing their own personality, finding their identity, and dealing with developmental tasks typical for adolescents. A central developmental task is the detachment from the parental home and the simultaneous development of relationships with peers (Havighurst, 1974; Richter & Moor, 2015), which to a large extent takes place in the school context. For adolescents, the group of friends is a central context, since basic social rules such as mutuality, reciprocity, or intimacy can be learned in these power–equivalent relationships (Youniss & Jacqueline, 1986). Friendships develop, among other things, when certain characteristics or behavior patterns are found in the group. In the early 1970s, Kandel (1978) found out that there is a high congruence in the peer group when it comes to the consumption of marijuana. There is evidence that the peer group that dissolves is less common than the one that is newly developing. This congruence orientation was also found in other characteristics, such as other illegal drugs or the choice of political party.

Social network analysis (SNA), for example, makes it possible to investigate the connection between collective norms and individual behavior within the peer group. Questions that can be answered include whether young people are more likely to seek out friends who exhibit similar behavior, or whether young people within certain networks are “encouraged” to behave in a manner that is harmful to their health due to the (harmful) influence of their friends (Hall & Valente, 2007). The roles or positions that different persons (parents, circle of friends, siblings) take in the network and to what extent these influence the health and health behavior of adolescents can also be analyzed. These and other questions can be examined to the extent and in the level of detail desired with the help of SNA.

Of particular research interest is the importance of social networks for the (re)production of health inequalities. Numerous national and international studies show, for example, that adolescents with a low socioeconomic status (SES) indicate both poorer health and less favorable health behavior (Kuipers et al., 2016a; Inchley et al., 2020b; Moor et al., 2020; Bucksch et al., 2020; Ahluwalia et al., 2015; Reiss, 2013; Elgar et al., 2015). Although the relationship between social background and health or health behavior has often been investigated in adolescence, few studies have examined the extent to which adolescents’ social networks explain this association.

Accordingly, this chapter deals with SNA and health and health inequalities in adolescence. First, the focus will be put on the previous research work on SNA in adolescence (Sect. 1.1). Section 2 presents theoretical assumptions of the SNA (including homophily, thesis of influence, and selection) for health and health behavior. Subsequently, the relationship between health inequality and health (Sect. 3) is discussed, and the significance of the SNA for health and health behavior will be explained (Sect. 4), with a focus on tobacco use in adolescence (Sect. 4.3). The role of the social network in health inequalities will be discussed in Sect. 5, and a summary will be provided in Sect. 6, which will identify research gaps and critically discuss the results.

1.1 Social Network Research in Youth

SNA is an internationally established field of research (see Scott, 2011; Scott & Carrington, 2011). It has multidisciplinary applications, especially in sociology, but also in psychology, economics, and anthropology (Valente et al., 2004). The various disciplines and subject areas agree on one point: Networks have a lasting impact on access to and use of life opportunities in our society. This can be illustrated by a longitudinal study that observed selection mechanisms. It shows that those who perform well at school in adolescence also tend to seek out high-performing friends (Flashman, 2012).

First surveys and analyses of networks between young people in the school context were already carried out in the nineteenth century (Heidler et al., 2014). By the 1960s (Coleman, 1961), a systematic investigation in the context of school surveys was undertaken. These surveys became the leading object of investigation for SNA, especially in the USA (cf. (Marsden, 2011; Freeman, 2004). At the international level, a wide range of topics related to social network research has been analyzed in adolescence; however, in many countries, the debate, particularly regarding health and health inequalities, is still in its infancy.

Subject areas of SNA in adolescence and younger adulthood relate mainly to individual risk behavior and structurally unequally distributed risk exposures, because socially disadvantaged young people are exposed to a greater number of risks than adolescents from socially privileged families (Alvin & Deschamps, 1998; Friedman & Aral, 2001). For example, cross-sectional research on attempted suicide among adolescents shows clear evidence that the likelihood of planning suicide is associated with peer group characteristics such as an increased proportion of depressed friends (Fulginiti et al., 2016) as well as when friends report suicide attempts (Mueller & Abrutyn, 2015). The frequent contact in the network with friends at risk or depressed friends is particularly alarming in the USA context in the field of youth homelessness, which is also associated with a lack of safer sex practice (Kennedy et al., 2012; Craddock et al., 2016) or increased substance and drug consumption such as crystal meth (Martino et al., 2011; Barman-Adhikari et al., 2016) within the network as studies have shown.

SNA is used for a wide range of health-related research topics. Public health research topics include sexually transmitted risks (HIV) (Neaigus et al., 1995), physical activity (Simpkins et al., 2013; Macdonald-Wallis et al., 2012), the body mass index (Fletcher et al., 2011; Renna et al., 2008), the consumption of tobacco, alcohol, and illegal drugs (Kandel, 1978; Valente et al., 2004), and suicidal behavior and attempts (Mueller & Abrutyn, 2015; Mueller et al., 2021; Xiao & Lindsey, 2021). These topics will be discussed in more detail in the following sections of this paper.

2 Theoretical Assumptions

In this section, the theoretical background of the mechanisms of action of social networks in adolescence and their significance for health behavior will be outlined. Health behavior is embedded in a variety of social contexts. Therefore, social network analysis (SNA) assumes that the social network in which the respective person is located shapes individual behavior. In SNA, homophily is presented as a central mechanism of action (see also chapter “Network Analysis and Health Inequalities: A Methodological Introduction”). Homophily means that people prefer to surround themselves with people who are similar in certain characteristics. This may be the case in relation to demographic characteristics or also in relation to certain types of behavior (Daw et al., 2015). For this purpose, the assumptions of social influence and selection are described, which consequently lead to social networks comprising homophile group members.

2.1 Social Influence

There is consensus in research on the influence mechanism in a peer group that adolescents are more likely to start smoking if their friends are also smoking. Without using the possibilities of SNA, research could only rely on information provided by the adolescents about the smoking prevalence in their peer group. Information or characteristics of these friends and their smoking behavior were not taken into account (Hall & Valente, 2007); however, this is the big advantage of SNA. The problem is that young people tend to overestimate the prevalence of smoking among their friends. This has been shown to be particularly true for girls, former smokers who have friends who smoke, and students with lower school performance (Kuipers et al., 2016b). When applying SNA, there is no need to rely on these (often) distorted data, since the social network information is collected and data on all network members are often available. Regarding peer group influence, a distinction could be made between “endogenous effect,” “exogenous or contextual effect,” and “correlating effect” (Ali & Dwyer, 2009).

Endogenous Effect

This effect assumes that individual behavior reflects the behavior of the peer group. A person is more likely to smoke if there are many smokers in his or her peer group. If the behavior of one person in the group changes, that change can function as a multiplier effect, which can then also change the behavior of the entire peer group, whose members are in turn in other networks and can thus pass on the change (Ali & Dwyer, 2009).

Exogenous or Contextual Effect

This effect assumes that individual behavior depends on characteristics outside the peer group. For example, if many adults smoke in a collective group, this exposure may also affect adolescents. For example, parents who smoke are more likely to influence their children’s smoking behavior (Ali & Dwyer, 2009).

Correlating Effect

This effect occurs when people in a group behave similarly due to similar—out-of-focus or unobserved—characteristics. Accordingly, adolescents with similar socioeconomic status (SES) are more likely to form a group with similar social circumstances. Research has also shown that socially deprived adolescents are more likely to smoke than peers who are socially better off. Even if someone from that group were to quit smoking, for example, this would have a smaller effect, as these unobserved characteristics (of, for example, social background) still exist and lead to a higher risk of unhealthy behavior in general (Alexander et al., 2001; Ali & Dwyer, 2009).

2.2 Selection

In contrast to the thesis of social influence, there are not many different assumptions on selection that need further explanation. According to the selection assumption, adolescents decide for themselves and make a preference-based selection as to whom they want to be friend. They are more likely to choose those friends who have similar characteristics or ideas or who show similar behavior. The selection hypothesis also describes the exclusion of friends, that is, the persons that are excluded from the peer group. If friends do not approve their smoking behavior, for example, the adolescents will turn to those who share these behaviors and do not normally devalue smoking. However, there are often several characteristics and behaviors that create or maintain these networks (Simons-Morton & Farhat, 2010).

For many behaviors, especially smoking, both directions of influence and selection were examined. Both mechanisms appear to be central to the smoking behavior of adolescents and in some cases also have a simultaneous effect (Hall & Valente, 2007; Schaefer et al., 2012). Overall, the selection hypothesis is given more weight in tobacco use (Mercken et al., 2009, 2010, 2012; Seo & Huang, 2012; Littlecott et al., 2021). However, it is methodologically challenging to distinguish between these two effects in the analyses. Only longitudinal studies can examine these mechanisms separately.

3 Youth, Social Inequality, and Health

In adolescence, the health behavior or the subjective assessment of health mainly provides information about the well-being and health-related quality of life of the younger generation. Although health and health behavior in general have tended to develop positively over time, which is reflected in a higher assessment of very good health (Cavallo et al., 2015), higher fruit and vegetable consumption (Vereecken et al., 2015), increased physical activity (Kalman et al., 2015), and decreased tobacco prevalence (Kuntz et al., 2018; Inchley et al., 2020a), not all young people benefit from this development to the same extent. SES remains one of the most important determinants of adolescent health (Inchley, 2017; Viner et al., 2012; Inchley et al., 2020b). Adolescents with a low social status are more likely to have an unhealthy diet, less likely to be physically active, and more likely to be overweight or obese compared to adolescents with a higher social status (Inchley et al., 2020b; Chzhen et al., 2018; Inchley, 2017; Bucksch et al., 2020). Socially disadvantaged children and adolescents report poorer health and are more likely to show an increased risk of mental abnormalities, psychosomatic complaints, and lower life satisfaction compared to those who have a higher school education or live in a more socially privileged family (Moor et al., 2015b; Elgar et al., 2015; Kaman et al., 2020; Richter et al., 2012; Torsheim et al., 2004).

The results are not consistent with risk behavior, but they predominantly show that less educated young people smoke more frequently (Kuntz et al., 2018; Moor et al., 2014, 2019, 2020; Robert et al., 2018) and experience alcohol-related intoxication more often than socially better-off adolescents or those with a higher school education. However, the results are heterogeneous, showing that highly affluent adolescence in some countries consumes alcohol on a regular basis (Inchley, 2017; Moor et al., 2018).

4 Importance of Social Networks for Health and Health Behavior

Unhealthy and harmful behaviors, which can have a strong impact and consequences for health in adulthood, develop during adolescence (Daw et al., 2015; Valente, 2012). Social networks play a crucial role in the context of adolescent health and health behavior. Various studies show, for example, that social networks influence mental health (e.g., Baggio et al., 2017), alcohol consumption (e.g., Deutsch et al., 2014), smoking behavior (e.g., Lorant et al., 2017), nutrition, body weight, physical activity (e.g., Barclay et al., 2013; Simpkins et al., 2013), and drug use (e.g., Pearson et al., 2006) among adolescents. Therefore, the following sections will deal with the significance of social networks for the health and health behavior of adolescents and provide an overview.

4.1 Mental Health

Various studies have examined the influence of social networks on mental health of young people. Baggio et al. (2017), for example, have investigated how the mental health of adolescents aged 12–14 years is related to the structure of the network. They found that adolescents with good mental health are more likely to be friends with those who have similar mental health. Boys were more likely to be friends with boys and girls more likely to be friends with girls. These results are also consistent with findings of other studies (Schaefer et al., 2011). Pachucki et al. (2015) were able to show based on a longitudinal study that adolescents in early adolescence did not become more similar in terms of their mental health over an analyzed period of 3 months. Since that is a relatively short time span, these results should be interpreted with caution. Baggio et al. (2017) also showed that young people with poorer mental health are more likely to have fewer friends and be more isolated in the network than young people with better mental health. Another study found that the more friends an adolescent has in the network, the lower his or her risk is of developing depression. Conversely, those young people who are more isolated and have few connections in the network have an increased risk of being affected by depression (Okamoto et al., 2011).

4.2 Health Behavior

4.2.1 Physical Activity and Nutrition

Physical activity and diet are social behaviors that are often shared and influenced by others and can cause health consequences such as obesity (Cunningham et al., 2012; Shoham et al., 2012; Trogdon et al., 2008). For example, it has been shown that adolescents who are friends with each other have a similar body mass index (BMI)—in that aspect, the homophily of friendships is evident (Fletcher et al., 2011; Renna et al., 2008). In the study by Renna et al. (2008) with data from the “National Longitudinal Study of Adolescent Health” of more than 20,000 adolescents, however, the influence of friends on the BMI was only significant for girls. A systematic review could show regarding selection and isolation effects that school friends are similar in terms of body weight and BMI (Fletcher et al., 2011). In addition, the results indicate that overweight adolescents are less popular and have fewer friends than adolescents with normal weights in their age group (Fletcher et al., 2011). Girls and especially overweight adolescents are more influenced by their friends regarding their body weight (Trogdon et al., 2008). The influence of friends is shown, for example, by the fact that the risk of becoming overweight in a certain period of time increases by 57% if one of the friends also becomes overweight in the same period of time (Nam et al., 2015). However, there is limited evidence of the way this influence is manifested. On the one hand, this may be direct communication between friends, during which adolescents exchange different views and opinions and thus form common norms, while on the other hand, different behaviors of the friends, for example, diets or physical (in)activities, may have an impact on the body weight of adolescents (Cunningham et al., 2012). In addition to social influence, which can explain the similarity of friends in terms of body weight, selection processes also play a role (Nam et al., 2015; Shoham et al., 2012). This means that adolescents tend to look for friends with similar weights to themselves (Nam et al., 2015). Adolescents who are not overweight tend to make friends with those with similar weights (Nam et al., 2015). Analogous selection effects were found in a longitudinal study regarding the physical activity of about 1900 adolescents (Simpkins et al., 2013).

Overall, various studies could demonstrate that adolescents who are friends with each other or in a common peer group engage in physical activity (Simpkins et al., 2013; Macdonald-Wallis et al., 2012). A review showed that there are inconsistent results regarding the connection between physical activity and the selection of friends (Macdonald-Wallis et al., 2012). On the one hand, there are results that show that physically active adolescents tend to have more friends than less active adolescents, whereas other analyses could not prove an association (Macdonald-Wallis et al., 2012). Furthermore, gender-specific differences could be found, as boys tend to be more similar with each other in terms of physical activity than girls (Macdonald-Wallis et al., 2012). de La Haye et al. (2010) found that female friends are more similar in their screen-based activities, such as watching television or playing computer games, whereas boys are more similar in their consumption of high-calorie food, such as fast food (de La Haye et al., 2010). Barclay et al. (2013) also showed that in general a young person is more likely to eat a healthy diet and exercise regularly if his or her friends do the same. The closer the bond or friendship is between the adolescents, the higher the probability of similar behaviors. These relationships do not depend on same-sex friendships or on migration background (Barclay et al., 2013).

4.2.2 Alcohol and Illegal Drugs

Results from social network research on substance use among adolescents indicate that adolescents who are friends with each other also tend to be similar in their use of different substances (Valente et al., 2004; Kirke, 2004). If adolescents consume substances such as alcohol or are perceived as users, their friends are more likely to also use substances (Kirke, 2004). Adolescents using illegal drugs, for example, are also more likely to have friends who do the same (Valente et al., 2004). Moreover, the number of friends who use illegal drugs is positively associated with the young people’s own drug consumption (Valente et al., 2004). Various studies explain this similarity in the consumption behavior of young people based on two mechanisms—selection and social influence—which have already been described in the introduction. Kirke (2004) and Valente et al. (2004) were able to show with social network analyses using typical parameters such as centralization, density, and transitivity that both selection processes and the influence of the peer group explain the similarity in substance use among adolescents and that not only a single mechanism can be used to explain it.

The early consumption of alcohol is especially a major health problem among adolescents. Social networks also play a decisive role here, as they influence the start of alcohol consumption among adolescents. For example, the results of the study by Mundt (2011) show that young people who start drinking alcohol tend to have more friends and boyfriends who also consume alcohol. At the same time, they have closer contact with popular adolescents and also communicate with more friends and acquaintances than abstinent peers (Mundt, 2011). Knecht et al. (2011) found that—based on a longitudinal multilevel network analysis—for adolescents with an average age of 12 years, selection processes play a greater role than social influence mechanisms regarding consumption of alcohol as adolescents were more likely to look for friends who have similar consumption patterns (Knecht et al., 2011). Selection processes also play a role among older adolescents aged 16–17 years (Kiuru et al., 2010). At the same time, the influence of peers in this age group is more effective and decisive than among younger adolescents, and consumption tends to increase with age, underlining the social nature of alcohol among adolescents (Kiuru et al., 2010). Additional differences between the sexes could be demonstrated, as girls resemble their peer group more closely than boys in their drinking behavior (Kiuru et al., 2010). Deutsch et al. (2014) found in their prospective multilevel network analysis that the “closeness” of friendships between adolescents has an impact on their drinking behavior. For example, the influence on their drinking behavior among boys and girls increases when the closeness of friendships with boys decreases (Deutsch et al., 2014). The intimacy of friendships between girls does not influence their drinking behavior (Deutsch et al., 2014). A longitudinal study by Huang et al. (2014) showed the influence of social media, such as Facebook, on the drinking behavior of adolescents. Adolescents whose friends upload photos on social networks that depict them drinking or celebrating with alcohol have a higher risk of consuming alcohol themselves (Huang et al., 2014). However, it should be noted that only egocentric networks were surveyed and analyzed.

4.3 Importance of Social Networks on Tobacco Consumption

A particularly large amount of research on the health significance of social networks in adolescence focuses on tobacco consumption; therefore, this section will place emphasis on tobacco consumption among young people and outline the current state of research.

Although tobacco consumption in adolescence has declined significantly (Kuntz et al., 2018; Inchley et al., 2020b), experimentation with smoking and the initiation of consumption continue to take place primarily in adolescence. At around 13–14 years of age, adolescents turn to tobacco for the first time (Moor et al., 2016). In this context, an early entry age into substance use is associated with a problematic consumption behavior in adulthood (Kendler et al., 2013). In addition, regular tobacco consumption is also associated with (long-term) health risks such as increased morbidity and early mortality (World Health Organization, 2015). Since the majority of social network studies have been conducted on tobacco consumption in adolescence, they will be discussed below.

4.3.1 Importance of Different Network Members

It has been proven in many studies that a higher number of smokers in the peer group increases the probability that the adolescent will also smoke (Ennett et al., 2008). Alexander et al. (2001) were able to prove that the probability of smoking is doubled when at least half of the peer group smokes, or if one or two best friends smoke, and also with increasing smoking prevalence in the school attended. Less research has been done so far on other characteristics of friends or relationships with friends regarding adolescent smoking behavior. These include, for example, the number of friends and the closeness of the friendship, the quality of the friendship (reciprocity of friendship, out-of-school activities, commitment of the friendship), the status or position in the peer group (betweenness centrality—the extent to which adolescents connect different groups of friends), or the (further) behavior of friends are all related to smoking, as was investigated in a longitudinal study involving more than 6500 adolescents aged 11–17 years (Ennett et al., 2008).

Simons-Morton and Farhat (2010) were able to show in their review, which included longitudinal network studies on the importance of the group of friends in adolescent tobacco consumption, that the best friend has a greater influence on tobacco consumption than other friends. However, this influence was reduced if other friends show the opposite behavior (e.g., not smoking). Group behavior (social norms) also influences one’s own smoking behavior. The influence of the group of friends was given, but selection processes were of greater importance, since adolescents increasingly sought friends with similar behavior. Furthermore, the review showed that parents also play an important role—if they smoke, their children are likely to smoke as well.

In the long-term study Add Health, Ali and Dwyer (2009) investigated the importance of the influence thesis of different persons in the network on smoking behavior from adolescence to young adulthood on the basis of three survey waves (completed in 1994, 1996, and 2002). It was shown that even after controlling for socio-demographic and parental characteristics, there is a clear peer influence on tobacco consumption. If the smoking prevalence among classmates increases by 10%, the probability of one’s own tobacco consumption increases by 3%. If the smoking prevalence among close friends increases by 10%, the probability of smoking increases by 5%. The influence of close friends continues into adulthood (Ali & Dwyer, 2009). With the Add Health study, Daw et al. (2015) were also able to show that siblings, followed by friends and classmates, have the greatest influence on smoking. The influence was greater when a friendship was reciprocally indicated (Daw et al., 2015).

4.3.1.1 Position in the Network

Heterogeneous results can be found regarding the position of ego in the network and smoking behavior. Some studies found that adolescents in isolated positions—when they reported having few or no friends or friendships with classmates—were more likely to smoke (Seo & Huang, 2012; Ennett et al., 2008; Valente et al., 2004; Littlecott et al., 2021). There are heterogeneous results and interpretations. Seo and Huang (2012) assume in their systematic review of social network analyses of adolescent smoking behavior that social isolation can lead to adolescents using tobacco to reduce emotional stress. It is also conceivable that the association is vice versa, indicating that adolescents were excluded from a (former) group of friends because of their tobacco consumption.

In contrast, other studies conclude that smokers are more popular among their peer groups (Schaefer et al., 2012; Lakon & Valente, 2012; Moody et al., 2011). However, it depends on the peer group considered. Smokers are more popular in peer groups that include many smokers. In this context, selection process could be identified with the help of the Add Health study, which shows that smokers also befriend other smokers (Schaefer et al., 2012).

Both the social pressure from the peer group (Seo & Huang, 2012) as well as the school context could play a role. For example, it was reported that smokers are more popular in schools where tobacco prevalence rates are generally higher, while in schools with lower smoke prevalence, more popular students tend to smoke less. In some results of the Add Health study, popularity was measured by the summed-up friendship ratings of students (in-degree centrality) (Alexander et al., 2001).

5 Socioeconomic Inequalities in Substance Use: The Role of the Social Network

In the field of health and health behavior, there are currently only a few studies that use social network analysis to investigate socioeconomic inequalities among young people. For example, there is a lack of studies that look at the mental health of young people in the context of the network and socioeconomic status (SES). There has also been little research on physical activity and nutrition among young people. Therefore, this section aims to provide an overview of previous findings regarding substance use—especially tobacco use—among young people and the role of social networks in relation to SES.

5.1 Use of Alcohol and Drugs

On average, 39% of young people aged 15 years drink alcohol (measured by drinking in the last 30 days) (Inchley et al., 2020a). Boys consume alcohol more frequently than girls (European School Survey Project on Alcohol and other Drugs, 2016). Overall, there are only a few studies that use SNA to investigate socioeconomic differences among adolescents regarding alcohol and drug consumption. Pearson et al. (2006), for example, conducted a study on 13–15-year-olds and found that girls and adolescents with higher SES are more likely to be integrated and more popular in peer groups (friendship nominations received) and to nominate more friends themselves than boys or adolescents with low SES.

5.2 Tobacco Consumption

Adolescents with a lower educational level smoke more frequently than socially better-off adolescents (Kuntz et al., 2018; Moor et al., 2020; Robert et al., 2018). This is particularly true when considering the type of school attended: For example, in Germany, only 3.6% of girls aged 15 who attend a higher education at school report smoking at least once in the last 30 days (boys: 4.1%), but 9.2% of girls who attend a lower educational type of school report the same (boys: 7.6%), according to the findings from the HBSC study (“Health Behavior in School-aged Children”) from 2018. At this point, it remains unclear which factors are responsible for smoking or for education-specific inequalities that influence tobacco consumption. It is known that social contexts, such as family, school, and peer group, play a central role in smoking behavior in adolescence (Simons-Morton & Farhat, 2010; Simetin et al., 2011; Piko & Kovacs, 2010; Schaefer et al., 2012; Moor et al., 2015a), but the impact of social inequalities has been less investigated.

Among the few studies on smoking behavior in adolescence that take socioeconomic inequalities in the context of social network analysis into account are, e.g., SILNE (2013) (“Tackling Socioeconomic Inequalities in Smoking: Learning from Natural Experiments by Time Trend Analyses and Cross-National Comparisons”) and SILNE-R (2016/2017) (“Enhancing the Effectiveness of Programs and Strategies to Prevent Smoking by Adolescents”), which were conducted using SNA in six and seven countries, respectively, of the European Union (EU) investigating socioeconomic inequalities in tobacco use among 14–16-year-old school students (Lorant et al., 2015, 2016, 2017; Robert et al., 2018). The results of the first SILNE study indicate that socially disadvantaged adolescents smoke more often and that there are more smoking peers in their school network than among those with a higher social status. The study further found that the smoking behavior of friends and the homophily of the group mediated the link between SES and tobacco consumption (Lorant et al., 2017). Lorant et al. (2017) developed a conceptual model (see Fig. 1) that takes into account both smoking among friends and social homophily.

Fig. 1
figure 1

Tobacco inequalities: conceptual model. Source: Lorant et al. (2017)

Full size image

According to this model, which is based on DiMaggio and Garip (2011), two conditions must be met: Tobacco consumption must be interdependent, that is, dependent on others, and social relationships must be socially homophile. As already mentioned, adolescents are more likely to start smoking if their friends also smoke, and the same applies to non-smokers. Tobacco consumption or non-consumption also defines the group and its social cohesion. It has also been described regarding the second condition that social relationships do not arise by chance, rather groups are created or continue to exist because group members share similar characteristics, such as gender, SES, migration, and so forth (Lorant et al., 2017). Lorant et al. (2017) were also able to prove the model empirically; the connection between a low social status and smoking behavior was partially explained by (more) smoking friends (close and not so close) as well as by social homophily. However, the effect of selection or influence could not be clarified in this study, as it is not a longitudinal study.

Pearson et al. (2006) were also able to establish a link between social status and tobacco consumption. In their study, they found that girls and those with higher school SES (i.e., lower proportion of deprived students in school) are more likely to be integrated and more popular in a group and have a larger network. Smokers were more likely to be isolated or have a small network.

Moreover, regarding quitting smoking in adulthood, it is evident that friends with higher levels of education have a greater influence on each other than those with lower educational levels (Christakis & Fowler, 2008). This has also been shown in studies among adolescents, where homophily varies according to the educational level of the parents. Homophily in friendships is higher among those adolescents whose parents reported higher education compared to those who are less educated. This relationship was true for smoking as well as for other behaviors such as alcohol and television consumption and physical activity (Daw et al., 2015). Similar results were also shown by Robert et al. (2018) based on the SILNE study. Adolescents are not only homophilic about smoking behavior, but also about school performance. Students with poorer school performance are more likely to be friends with each other than those with varying school performance. The connection between school achievement and smoking behavior could be partly explained by smoking consumption and homophily of friendships as well as by school type.

Huisman and Bruggeman (2012) examined the importance of social networks on smoking in adolescents, taking into account both the type of school and parental education. The authors conducted a longitudinal study among 13–14-year-old Dutch adolescents in the 2008–2009 school year and analyzed the mediating role of the social network. For this purpose, the students in each school class were asked to name up to 15 other students with whom they are friends. The information in the network was analyzed using SIENA.Footnote 1 Since the social background of the parents often shapes the school education of the children, and school is a special place for social contacts, the question was to what extent the peer group played a mediating role between school type and smoking behavior. The results showed that the effect of the school type on smoking is mediated by the social network (smoking friends), even after adjusting their own smoking behavior (Huisman & Bruggeman, 2012). This is a very important result, as it means that not only is the adolescent’s school education itself responsible for smoking behavior, but also or even to a higher extent the (school) friends who smoke and mediate the effect between school type and smoking.

6 Discussion and Conclusions

6.1 Summary and Critical Reflection

For adolescence, there is a wide range of studies that have analyzed factors influencing health and health behavior in this stage of life. Interest in social determinants is increasing since health inequalities become apparent as early as adolescence and have a lasting negative impact on health and health-related behavior over the life course. However, less attention has been paid to the role of the social network in the (re)production of these health inequalities, although social network analysis (SNA) reveals promising approaches in this regard. One exception is the school context, which was given a strong emphasis in social network research at an exceedingly early stage, so that a relatively large number of studies can be drawn upon compared to studies examining other phases of life. Especially regarding smoking behavior and the importance of the social network, there is evidence that considers the entire class network and thus the entire network. However, fewer network studies have been conducted on (mental) health and other behaviors, although the number is increasing, e.g., on suicidal behavior (Xiao & Lindsey, 2021; Abrutyn et al., 2020).

6.1.1 Methodological Challenges

Methodologically, SNA on adolescence is a huge challenge, since entire classes must be surveyed to completely cover networks, but this is subject to data protection hurdles. Apart from this, the available studies must also be critically examined. For example, in the study by Schaefer et al. (2012), data from 1994 to 1996 were taken into account, while Alexander et al. (2001) evaluated data from 1994 to 1995, which is quite old as smoking norms changed enormously in the last 30 years. The question is whether these results are still valid nowadays. At that time, smoking prevalence was significantly higher and smoking itself was more socially accepted and less stigmatized. Whether or not smokers were isolated in these studies at the time would have a different meaning than whether they were isolated in more recently conducted studies, as social norms concerning smoking have changed. However, studies that only look at the school network and identify, for example, smokers as isolated individuals may come to distorted conclusions. It is possible that these students are part of a broad network outside school and are not isolated there (Pearson et al., 2006). In school-based network studies, therefore, a “blind spot” may arise which should be given consideration. In addition, it is often asked whether friends, family members, and classmates have certain characteristics, but less often the quality of the relationships (e.g., frequency of contact, positive/negative relationship) is analyzed. An essential question of SNA is whether only the relationship with different persons has a (different) influence on our behavior, or whether this influence also has a different meaning for different behavior patterns. Some studies have investigated this question, but the evidence is still insufficient. For example, one study showed stronger associations regarding social network on smoking compared to alcohol, television consumption, or physical activity (Daw et al., 2015). Further studies, such as qualitative studies, should be conducted to understand the mechanisms of action.

6.2 Conclusion and Research Desiderata

The current contribution was able to show that SNA still has significant research gaps in some areas. There is an increasing number of studies in the school context, which mainly examine tobacco consumption in association with the social network, but only few studies have been carried out on other health behaviors and especially on (mental) health. Another problem is that mostly only the school network is analyzed and not other networks, such as out-of-school friends, family network, and so forth, which could lead to inaccurate results. Many studies are based on cross-sectional studies that do not allow a causal statement. There is a lack of longitudinal studies that can more precisely identify the causal mechanisms (except for studies on tobacco consumption, where longitudinal studies are more common). It is also noticeable that SNA is mainly dealt quantitatively. There are hardly any qualitative or ego-centered studies on adolescence, although it is possible to investigate how and why the social network affects health and health-related behavior from that view. That study would have some advantages—for example, changes in norms over time could also be considered. For the relationship between tobacco consumption and the social network, the position in the network could reflect norm changes rather than friend relationships, since social acceptance and tobacco prevalence have changed over time. Those with higher social status, for example, distinguish themselves by largely refraining from smoking. People with lower social status continue to smoke, which is accompanied by an increasing social-normative devaluation, the stigmatization of smoking, and, thus, socially disadvantaged population groups (Bell et al., 2010; Chapman & Freeman, 2008; Reuband, 2014).

In conclusion, it should be noted that there is a significant lack of SNA studies that also take socioeconomic differences into account beyond tobacco consumption. However, the evidence is quite limited by the extent to which the social network can explain health inequalities. A few studies have shown a mediating effect (Huisman & Bruggeman, 2012; Lorant et al., 2017). A moderating effect has also been demonstrated, in the context that the influence of the social network in quitting smoking is more pronounced among friends with higher educational levels than among friends with lower educational levels (Christakis & Fowler, 2008). Further studies that can help add knowledge to the research gaps regarding other health outcomes are highly needed.

Reading Recommendations

  • Lorant, V., Rojas, V. S., Robert, P.-O., Kinnunen, J. M., Kuipers, M. A. G., Moor, I., Roscillo, G., Alves, J., Rimpela, A., Federico, B., Richter, M., Perelman, J., & Kunst, A. E. (2017). Social network and inequalities in smoking amongst school-aged adolescents in six European countries. International Journal of Public Health, 62, 53–62. Results of the SILNE study on the role of social networks in socioeconomic inequalities in tobacco consumption among adolescents in six countries.

  • Ali, M. M., & Dwyer, D. S. (2009). Estimating peer effects in adolescent smoking behaviour: a longitudinal analysis. Journal of Adolescent Health, 45(4), 402–408. Results of the longitudinal study “AddHealth” on the importance of the influence of different persons in the network on smoking behavior from adolescence to young adulthood.

  • Simons-Morton, B.G., & Farhat, T. (2010). Recent findings on peer group influences on adolescent smoking. The Journal of Primary Prevention, 31, 191–208. Review of peer group influences on adolescent smoking behavior, including longitudinal SNA studies.

  • Mercken, L., Snijders, T. A. B., Steglich, C., Vertiainen, E., & Vries, H. de. (2010). Smoking-based selection and influence in gender-segregated friendship networks: a social network analysis of adolescent smoking. Addiction, 105(7), 1280–1289. Longitudinal study investigating the mechanisms of influence and selection for tobacco consumption in adolescence.

  • Huisman, C., & Bruggeman, J. (2012). The social network, socioeconomic background, and school type of adolescent smokers. International Journal of Behavioral Development, 36(5), 329–337. A longitudinal study conducted to investigate the significance of social networks for health inequalities in adolescent smoking.

Data Sets/Overview

  • “SILNE” (Tackling socioeconomic inequalities in smoking: Learning from natural experiments by time trend analyses and cross-national comparisons)

SILNE is a project funded by the European Commission and based on school network data. It investigates smoking behavior and norms of adolescents aged about 14–16 years at family, socioeconomic, and school levels in six European countries (Belgium, Finland, Germany, Italy, the Netherlands, and Portugal).

  • “SILNE-R” (Enhancing the Effectiveness of Programs and Strategies to Prevent Smoking by Adolescents)

SILNE-R includes a quantitative repeated survey of SILNE with a focus on school tobacco control policies. Smoking innovations such as e-cigarettes and the health literacy of young people were also analyzed. In addition, qualitative focus groups with adolescents and expert interviews were surveyed, which can be linked to the quantitative findings for many questions.

  • Add Health (National Longitudinal Study of Adolescent Health)

Longitudinal study on adolescents in America in grades 7–12, among others, on the topics of substance use. The study offers many different network parameters and examines different relationships (school and family relationships).

  • VOCL’99 (Longitudinal Cohort Studies on Secondary Education—Cohort 1999)

In this Dutch longitudinal study, students aged 13 years on average were included in the study. The study examines the stability of youth relationships in the peer context in a longitudinal way.