Notwithstanding these health implications, for most teenagers it holds that adolescence is a phase of experimentation in the first place. A striking observation is that age specific rates of alcohol and drug use, and related to this conformity to peer pressure and fear of peer rejection, peak in adolescence and drop sharply when entering adulthood [8-10]. Nevertheless, given the possible health consequences, the (ab)use of alcohol and cannabis remains a serious public health concern and considerable and continuing efforts need to be conducted to develop effective interventions in this domain. Therefore, insight in risk factors that constitute teenage substance use is imperative.

Several studies have shown that in adulthood socio-economic differences in alcohol and drug use are relatively consistent [11-13]. People from higher socio-economic strata tend to drink more often but in smaller quantities, while their counterparts in lower socio-economic strata tend to consume alcohol less often but in larger quantities and in a more problematic way. While the existence of class-based differences in alcohol and cannabis use among adults is well established, socio-economic differences in adolescents' substance use are far less investigated, and consistency in present evidence is lacking. According to some research reports, the association between socio-economic status (SES) and substance use in adolescence is similar to patterns found in other life stages, where a lower SES is associated with increased incidence rates of alcohol and cannabis consumption [13-16]. On the other hand, several studies couldn't corroborate these findings, reporting an insignificant, diminished or reversed relationship between a teenagers socio-economic position and his/her alcohol and cannabis use [17-20]. As a possible explanation for these inconsistencies, West and others [21,22] referred to the occurrence of "a process of equalization", in which a transition is taking place from health inequality in childhood to relative equality during adolescence. According to these authors, this "process of equalization" is rooted in the defining characteristics of adolescence as a transitional period (i.e. the growing importance of peers, school environment and youth culture), by which the influence of familial background gets sharply curtailed, resulting in a homogenizing effect [21,22].

However, it can be questioned whether this equalization process is a real-existing phenomenon or rather should be interpreted as an artifact, reflecting the way in which an adolescents' socio-economic position is traditionally measured, that is, using information about the parents' socio-economic status (i.e. parental educational level, parental occupational level and family income) [23-25]. As youngsters strive to obtain more autonomy from their parents and develop their own identity, their social position gets increasingly determined by their own choices and life course plans [24,26-28]. Consequently, the use of merely parental SES markers as indicators of social status during adolescence may not be sufficient [27,29,30]. Following Bourdieu and Passeron [31] and others [24,26] we state that it may be more appropriate to use information about adolescents' educational level, as an indicator of their current individual social position, since this determines to a large extent ones future social class group.

According to Bourdieu and Passeron [31] the educational system plays a fundamental role in the reproduction of social inequalities. This reproduction works through a combination of selection and socialization processes. Selection refers to the differential validation of cultural capital in schools: students that possess the "right" (i.e. dominant middle class) cultural capital have greater chances for academic success, while other students flounder at lower levels of education. That way, scholars from lower social classes tend to concentrate in the lower status education types [32-35]. However, schools also socialize students into particular cultures: higher status education types socialize students towards the dominant middle class cultures, while lower status education types socialize towards lower class cultures [31,34,35]. Hence, the existing social order is maintained and social inequalities are even reinforced. Considering people's health behavior, the awareness of such a process of reproduction calls the question whether social gradients in alcohol and cannabis use merely are the result of differences in students' familial background or whether schools reinforce these inequalities.

In Flanders, the educational level of a student can easily be assessed by the type of education the student follows. The Flemish secondary school system is highly tracked and mainly consists of three different types of education that can be ranked in difficulty level from vocational, technical, to general secondary education. General education is a type of education that provides students with a firm theoretic knowledge foundation for going into higher education. Technical education is both practice- and theory-orientated, so that students can either enter the job market directly or continue their studies in higher education. Vocational education is a mere practical type of education and prepares students to enter the job market directly.

Unlike the existing literature on socio-economic differences, this research was conducted within a multicultural environment, i.e. Belgium's capital region. The particular nature and assembly of this urban region shapes a unique context for this study, since it additionally allows us to examine whether the impact of an adolescents' socio-economic position, as determined by both familial background and education type, on his/her consumption pattern, interacts with ones' cultural origin. Brussels is characterized by a large degree of ethnic diversity, clearly reflected in the composition of the students' population in Dutch-speaking secondary education. Hence, migrant students make up about 60% of the secondary school student body. Furthermore, previously conducted analyses [see [36]] illustrated that 75% of this migrant group are Muslims. Since Muslims are overrepresented within the migrant population and due to the fact that Islamic cultures religiously and often legally prohibit the use of alcohol and cannabis [37,38], one can expect that SES differences in substance use are much more clear for native than for migrant students. As such, their religion may act as a buffer for the effect of education type on risk behaviors such as drinking alcohol or using cannabis.

A question that arises immediately when interpreting these findings is which processes within these schools can explain the link between tracking and substance use. Here, research has shown that the specific subcultures that arise in lower educational strata are characterized by a culture of demotion, feelings of held back with bleak future perspectives, feelings of futility, frustration and strain, or eroded self-esteem [42-45]. Furthermore, teachers in lower tracks have much less expectations from their students (i.e. Pygmalion effect) and also leadership styles clearly differ between education types [46]. All these experiences may push these students towards delinquency or substance use, not only to achieve alternative sources of status in the peer group, but also to handle the strain caused by their negative experiences in these lower educational tracks.

The multicultural character of our study sample additionally allowed us to explore whether the effects of education type differed according to students' ethnic or cultural background. As we expected, effects of education type were observed in the sample of native students, while these effects were absent or negligible in the sample of migrant students. We explained this finding by referring to the buffering effect of religion in the relationship between education type and substance use. Students from Islamic background make out the majority in the group of students from another ethnic background, and these students not only profess their religion in a much more conscious way, they are also the product of families and cultures that put more emphasis on religious values such as for instance abstinence than their native Belgian counterparts [47].

A somewhat remarkable finding of our study was the positive association between parents' educational level and alcohol use. Students from families where parents are highly educated used more alcohol in the last month, independent of the students' educational level. Such a finding was not found for cannabis use. A possible explanation may be that students from high SES backgrounds simply have a more attractive financial background, allowing these youngsters to buy alcohol-related products [20,48-50]. For cannabis on the other hand, the financial strength of students' background may be of minor importance. For instance, cannabis is often consumed and shared in larger groups of teens, and especially the amount of cannabis needed to get high is much cheaper in price than the amount of alcohol needed to get drunk. Furthermore, cannabis is legally prohibited, and is dependent on having the 'right' connections or resources to buy these products, and these connections may be much less dependent on the familial SES background of these students. Other explanations are however possible and further research is necessarily to explain these results.