Little is known about the relationship between socio-economic status and adolescent drinking behavior. While several studies have investigated the relationship between SES and various measures of drinking behavior in youth, findings have been very heterogeneous. The HBSC study provides a unique opportunity to allow for between-country comparisons, using the same data collection protocols.
Overall, we found only very limited evidence for a close consistent relationship between parental SES and episodes of drunkenness among 11 to 15 year olds. Our findings therefore underline previous studies which have identified weak or even no associations between parental SES and alcohol use in adolescence [20, 16–18]. Some differences are noted, though, with regard to gender. For girls, almost no significant effects for family affluence and only few effects for parental occupation were observed. For boys, on the other hand, there was at least some evidence for socio-economic differences. Adjusted for parental occupation, it was found that boys from low and/or medium affluent families faced a lower risk of frequent drunkenness than boys from highly affluent families. While such effects occurred in one third of the countries, it should be noted that in the large majority of countries, trends for low affluent boys to be less at risk than high affluent boys became apparent, even if these did – sometimes barely – miss levels of significance. Thus, for boys at least, it might be argued that the data suggest an adolescent-emergent role of SES in some countries that only become manifest in adulthood.
Kuntsche et al. have pointed out that while for adults problem drinking seems to be more common in less affluent groups, this direction might be reversed for adolescents, where accessibility of financial resources is more limited [26]. If one reasonably assumes that availability of pocket money is at least to some degree related to parents' affluence, the results of two Finnish studies, which found a clear relationship between adolescents' own financial resources, i.e. amount of pocket money and drunkenness support this view [27, 28].
As for geographical differences, it might seem difficult to identify clear patterns at a first glance. A closer look, however, reveals that it was particularly in countries with either an overall lower income level and a strong "alcohol culture" (according to overall consumption rates [29]), such as Eastern European countries like Croatia, Estonia, Hungary, Latvia, Russia or with particularly high alcohol prices such as in Sweden that effects of family affluence on drunkenness in boys became apparent. Countries with no SES effects seem to be predominantly those with either a high overall income level such as Switzerland and/or very strong "alcohol cultures" such as Denmark, Finland and France or particularly low alcohol prices, as in the case of Spain.
Regarding parental occupation there were also more effects on drunkenness for boys than for girls. Contrary to FAS, however, these effects indicated a higher risk in adolescents with low or medium occupational status parents. Thus, while for FAS the risk of repeated drunkenness decreased with family affluence, for parental occupation, the risk tended to increase with decreasing status, which is in line with some findings from other studies [14, 15]. This was even true for countries that showed significant effects on both indicators. Given the positive relationship between parental occupation and family affluence this result might appear somewhat surprising. However, it should be acknowledged that the correlation between these two indicators is rather low. Therefore, these findings suggest that the different dimensions of socio-economic circumstances also have different effects on drunkenness in adolescence.
A possible explanation for these differences might be found when the specific consequences deriving from the two SES dimensions are considered. Parental occupation to some extent reflects parental education, and parents from a higher educational background might be somewhat more likely to influence substance use in adolescence by their parenting strategies and the values and norms they transmit to their offspring. Similarly, differential processes of modeling behavior might have an impact – at least as far as fathers are concerned. European comparison studies among adults have thus shown that heavy drinking episodes are more common among lower educational groups [30], particularly among men with lower educational levels, while for women no such differences appeared [31]. In an international comparative study Bloomfield et al. also reported no significant social inequalities in terms of educational level for binge drinking in women, but relatively consistent, although often not significant, gradients for males, with the lower and middle educational groups being more likely to be heavy episodic drinkers than the higher educated [32].
Family affluence, on the other hand, appears to be more strongly related to income or spending patterns, therefore indicating availability of resources to indulge in the relatively costly consumption of alcohol in the first place. This might also explain at least part of the gender differences found in the relationship between FAS and drinking behaviour. As the overall rate of repeated drunkenness is much higher in boys than in girls, the relative relevance of financial resources, such as pocket money to buy alcohol, might also be higher among boys, thus increasing the role of family affluence for this group.
In addition, there might be other factors during adolescence, which may have a greater impact on drinking behavior than parental SES. For instance, the specific character of adolescence as a stage of experimenting with 'new' behaviors associated with an adult status might also create generalized developmental stage-related demands and temptations that tend to transcend boundaries created by the family's socio-economic background. More important, yet, might be the consideration that while health behavior in childhood is strongly determined by parents, with the onset of adolescence the influence of peers and youth culture increases in relation to more traditional family norms and values, and this shift in influences is closely related to the emergence of adolescent risk behavior [33–35]. This might especially apply to those behaviors which do not commence until adolescence such as alcohol and tobacco consumption.
Study Limitations and Strengths
Several methodological aspects restrict the explanatory power of these findings. A point which thus might be considered problematic is the usage of self-report data on parental occupation. Even though several studies indicate that the classifiable answers of adolescents (even at the age of 11 to 13) can be considered as good proxy reports of parental occupation [36–38], it remains problematic that a relatively large number of adolescents (up to 15%) do not seem to be willing or able to report their parents' occupation at all. If missing responses are unequally distributed among socio-economic groups, which might be expected, this may effect findings. Therefore, the results on occupational differences must be interpreted cautiously.
Similarly, assessing adolescent drinking behavior by means of self-report might be considered problematic. Self-reports of alcohol consumption and other substance use patterns, can not only incur general problems of social desirability, but might also result in different response tendencies according to the reference group or gender. However, it has been repeatedly shown that self-reports can usually claim a rather high degree of validity [39, 40]. Moreover, in order to underline the importance of giving honest responses, the students in the present study were assured of confidentiality and that neither their parents nor teachers would be informed about the individual results. Another source of self-reporting bias would emerge if under-reporting was associated with parental SES. Unfortunately, not much is known about these mechanisms in adolescence, but if underreporting was unequally distributed across SES, then this may have biased the results reported here. Nevertheless, while the usage of a dichotomous classification of drunkenness might be considered a crude type of assessment, it is probably less vulnerable to such reporting errors.
Even though using the highest occupational status of the parents has the advantage of reducing the numbers of missing, it does not facilitate an assessment of gender influences in the association between the prevalence of repeated drunkenness and parental occupation. Therefore, the analysis was re-run using fathers' and mothers' occupational status separately (not shown). Overall, the results support the application of the head of the household measure (i.e. the highest occupational status of each couple) as among girls – in those few countries where an effect was found – both mothers' as well as fathers' occupational status were significantly related to drunkenness. Among boys, a rather inconsistent picture emerged. The occupational status of the father was associated with drunkenness in only three countries while the status of the mother was related to drunkenness in five countries. In addition, in five countries that showed significant associations with the head of the household measure, no effect for fathers' or mothers' occupation was found.
Another issue that should be considered is that by not adjusting the significance level for multiple testing the findings might have capitalized by chance. In fact we decided against such adjustment based on the rationale that the purpose of research here is more of an exploratory kind, so that the increased likelihood of a Type II error that results from reducing type I error by Bonferroni-adjustment might be considered just as committing a type I error. Also, one could argue that in fact the different samples we looked at came from different and separate country studies (even if they were based on the same study protocol) and there was no cross-sample testing but only separate tests within the different samples/studies are reported.
However, if such an adjustment was made this would indeed leave fewer significant findings, considering the number of countries involved, and result in more non-significant trends. Nevertheless, the results would not fundamentally alter the main findings of the study, i.e. that compared to adulthood parental SES is of minor importance in the prediction of episodes of drunkenness in adolescence, and that only in some countries and only for boys a pattern of socio-economic differences in drunkenness emerges.