OUP user menu


Sylvie Ledoux , Patrick Miller , Marie Choquet , Martin Plant
DOI: http://dx.doi.org/10.1093/alcalc/37.1.52 52-60 First published online: 1 January 2002


— This paper compares samples of 15–16-year-olds from the UK and France on their usage of alcohol, tobacco and illicit drugs and also seeks to describe the associations between alcohol and other drug use with ‘family variables’ within the two countries. Compared to UK adolescents, French adolescents showed a slightly higher rate of cigarette smoking, were almost identical on cannabis use, rather lower on the use of other illicit drugs and very considerably lower on alcohol use. Family variables were related to substance use. In the two countries, children from non-intact families, those who were not satisfied with their relationships with their father or mother and those who were less closely monitored, were more likely to be heavy substance users than other students. Logistic regressions showed that parental knowledge of the whereabouts of their offspring on Saturday evenings was the strongest factor, in both countries, that family structure is frequently still significant in the UK, and that paternal relationships are highly significant among French students. Differences in national drinking culture, urbanization and parental practices are discussed in an attempt to interpret some of these findings.


Divorce has dramatically increased since the 1960s and negative long-term effects of parental divorce during childhood, adolescence or adulthood have frequently been investigated. Findings suggest that adolescents from divorced families experience poorer mental health than those from intact families (de Goede and Spruijt, 1996) and report more alcohol and other drug use and antisocial behaviours (Neher and Short, 1998). Adolescents from non-intact families tend both to drink more (Foxcroft and Lowe, 1991) or at an earlier age (Isohanni et al., 1994) and to be more frequent users of illicit drugs (Plant and Plant, 1992; Denton and Kampfe, 1994; Hoffman, 1995; Miller, 1997). Adolescents from single families were more likely to smoke (Miller, 1997) and/or to consume alcohol (Griffin et al., 2000) and scored lowest on the different aspects of psychological well-being such as mental health, psychological stress or suicidal thoughts (Spruijt and de Goede, 1997). Adolescents who have lived in a stepfamily during childhood were more likely to use tobacco, alcohol, and illicit drugs at age 18 years (Nicholson et al., 1999). Smoking prevalence rates were high among teenagers from both reconstituted and lone-parent families (Glendinning et al., 1997). Results appear conflicting regarding the impact of family structure on adolescent substance use, depending on how the concept is operationalized, but overall, Amato and Keith (1991) confirmed, in their meta-analysis, that parental divorce lowers the well-being of children, even if the effect is generally weak.

Other family variables have also been shown to be related to, or to predict, substance use. Parental use of tobacco or alcohol is a precursor of the onset of smoking or drinking among their offspring (see review, Denton and Kempfe, 1994). Low parental support or monitoring is associated with high rates of adolescent substance use (Steinberg et al., 1994; Chilcoat and Anthony, 1996; Cohen and Rice, 1997; Glendinning et al., 1997; Piko, 2000). Closeness or a positive relationship with parents reduces the risk of adolescent substance use (Kandel et al., 1978). Both Hoffman (1995) and Sokolkatz et al. (1997) pointed out the over-riding importance of family dynamics such as parent–child relationships and the degree of parental control.

Furthermore, the pattern of parental monitoring and attachment is likely to be influenced by family structure (Hoffman, 1995). In non-standard families, the relationships with the absent parent (more often the father) might be poor or non-existent, and single parent families may be less supportive than two-parent families. Thus the effects of family structure could be mediated by factors such as parental support or bonding.

It appears that no previous study has compared possible different family effects on teenage drinking and other drug use in two different countries. However, Foxcroft and Lowe (1991) suggested that there are possible cultural variations in the pattern of parental monitoring between countries. British families appear to exert less control than French ones (Hibell et al., 2001), consequently we could hypothesize that the relationships between parental monitoring and substance use are not exactly the same in the two countries. On the other hand, several studies (e.g. Miller, 1997) carried out in the UK have shown an effect of the family structure on adolescent use, but this result was not found in surveys conducted in France (Choquet and Ledoux, 1994).2

The second European School Survey Project on Alcohol and other Drugs (ESPAD) provided the participating research workers with opportunities to do cross-national studies not only on prevalence rates but also on risk factors associated with the use of alcohol, tobacco, cannabis and other illicit drugs (Hibell et al., 2001). The current study compares 15–16-year-olds from the UK and France on their usage of various psychoactive substances and also seeks to describe the associations between substance use and family variables within the two countries. Specifically, the aims of the current study were: (1) to compare the UK and French prevalence rates for the use of alcohol, tobacco, cannabis and other illicit drugs; (2) to examine and compare the associations of substance use with family structure, maternal and paternal relationships and parental monitoring within the two countries.


Sampling procedure in the UK

The population sampled consisted of students born in 1983 throughout the UK. The UK was divided into England, Scotland, Wales and the province of Northern Ireland and a separate sample was drawn from the appropriate school classes within each country. It was expected that ≥90% of the target population would be still in school at the time of the survey. Sufficient funds existed for a sample of ~225 classes in all, one from each school selected. The total sample was stratified by country so that comparisons between the four countries could be attempted. Lists were available detailing the schools in each country along with the numbers of students in each. The list for each country (including the independent schools for that country) was sampled with probability proportional to size of school, to yield self-weighting samples within each country. Results covering the UK as a whole were obtained by applying a weight to the responses of all the students within each country, this weight being the inverse of the school sampling rate within that country.

Local organizers in each selected school furnished lists of all classes in their school containing students born in 1983. One of these classes was then selected at random. The local organizers did the fieldwork during the period March to June 1999. The standardized ESPAD questionnaire (see below) was administered to the class, excluding students whose parents refused consent, under ‘examination’ conditions. Students were assured they had been randomly selected and that their replies would be anonymous. Questionnaires from students not born in 1983 were later discarded.

A total of 223 out of 302 schools approached (74%) co-operated with the survey. Comparisons were made between the participating and refusing schools on type of school and area in which the school was situated. There appeared to be no discernible differences.

The resulting 223 classes contained a total of 5192 students. The parents of 58 students (1.1%) refused to allow their child to take part. On the day of testing 419 (8.1%) were absent ill, 166 (3.2%) were absent with permission, and 105 (2.0%) were absent without permission. This left 4444 (85.6%) whose questionnaires were returned. Of these, 2774 were born in 1983. However, a further 29 frivolously completed questionnaires were discarded together with 104 incomplete ones; therefore 4.7% (133/2774) questionnaires were excluded. This left a total sample of 2641 students aged 15 or 16 years, 1280 of whom were boys and 1361 girls.

Sampling procedure in France

The ESPAD study formed a part of a larger survey of French adolescents. A national representative sample was drawn using a two-stage sampling procedure. First, 300 schools were drawn from computerized lists of schools updated in November 1998 (Ministry of Education — DPD) and then two classes were selected from each. At the first stage the sample was stratified on four variables. These were: (1) type of school: junior high school/high school/vocational school; (2) sector of school: public/private; in France, 20% of students are enrolled in private schools; this percentage varies from one region to another (it is higher in Catholic regions) and according to the type of the school (more private schools in primary education and academic schools than in vocational schools); (3) type of area: urban/rural; (4) educational characteristics of the school: ZEP/no ZEP. In France, the Ministry of Education have defined ‘Priority Zones of schooling’ (ZEP) according to several characteristics (e.g. low socio-economic status, high unemployment, high percentage of immigrant population, high percentage of drop-outs from school). The schools located in these areas (10%) receive more grants and have smaller class sizes.

Sampling was proportional to size of school. In each school, two grades/classes were drawn. 93.8% (563/600) of these classes participated. Non-participation was due to refusal of the headmaster (10 schools), students not being in school, or to a poorly applied data collection procedure. All students attending these classes were included and filled out the questionnaire during a course period. Data were collected between 22 March and 7 May. Very few students (1.0%) refused to participate; 8.9% of the students were absent. Parental permission for their children to participate was refused in only 1.2% of cases.

This procedure yielded 12 113 students in the larger survey, representing 89.1% of those in the participating classes and 83.6% (12 113/14 494) from the total number approached. Each questionnaire was then scrutinized by research assistants who excluded questionnaires: (1) whose responses were obviously frivolous or seriously inconsistent; (2) whose responses were systematic (e.g. responses ‘4’ whatever the items); (3) when more than 50% of the questionnaire was not filled out, particularly when it concerned questions about substance use.

This resulted in the exclusion of 243 questionnaires, leaving 11 870 (note that about 60% of the rejected questionnaires, i.e. ~150, would have been born in 1983). Of these, those born in 1983 constituted 1174 boys and 1110 girls from 226 schools and 321 classes.

The ESPAD questionnaire

The questionnaire used was derived from the original 1995 version (Hibell et al., 1997), modified by agreement between the researchers participating in the 1999 exercise (Hibell et al., 2001). This instrument contained a core section common to France and the UK and to all the ESPAD countries, extensively covering use of alcohol, tobacco and illicit drugs together with some demographic variables.

The independent (predictor) variables studied in the present paper were gender, family structure, relationships to the parents and parental control.

Family structure.

This was coded into four categories: intact (both natural parents were present), restructured (one natural parent and a step-parent were present), single parent (one natural parent and no step-parent), other (neither natural parent present).

Relationships to the parents.

Two variables were included based on the responses to the questions: ‘How satisfied are you usually with your relationship to your mother?’ and ‘How satisfied are you usually with your relationship to your father?’ Both questions were scored on 5-point Likert scales.

Parental control.

This was measured by the question: ‘Do your parents know where you spend Saturday evenings?’ The coding was ‘always’, ‘quite often’, ‘sometimes’ and ‘not usually’.

The dependent variables, most of which were originally measured on 7-point scales, were all dichotomized as follows: cigarette smoking in the past 30 days: at least one per day versus the rest; alcohol consumption in the past 30 days: ≥6 times versus the rest; alcohol consumption in the past year: ≥20 times versus the rest; ≥5 drinks in a row in the past 30 days: yes versus no; drunk in the past year: ≥3 times versus the rest; cannabis use ever: ≥3 times versus the rest; cannabis use in the past year: ≥3 times versus the rest; cannabis use in the past 30 days: yes versus no; volatile substance abuse ever: yes versus no; use ever of any illicit drug except cannabis: yes versus no.

Statistical analyses

In both countries, the statistical procedures used attempted to allow for the cluster sampling designs. In the UK, this was achieved using the PcCarp package from the University of Iowa (Fuller et al., 1989). In France, the SAS package procedure Proc Genmod was used. However, as the cluster effects appear to be small, between-country comparisons of proportions of subjects using substances etc. are made without correction. Using conventional methods, all the significant findings reported would still hold, were the sample sizes to be halved. Comparisons on the predictor variables within each country were achieved using logistic regressions corrected for sample design.


The French sample contained 1174 boys and 1110 girls with usable data. In the UK sample there were 1280 boys and 1361 girls.

In both countries, there were several indications that the reliability and validity of the responses were acceptable. For instance, in the UK, inconsistent responses to two versions of the same question were never higher than 5.1%, and a fictitious drug, ‘relevin’, was supposedly used by only 0.1%, whereas, in France, between 3% (tobacco, cannabis) and 6% (tranquillizers or sedatives) of the students gave inconsistent responses and 0.6% reported ‘relevin’ use.

Tables 1 and 2 set out the gender distributions for the two countries on the independent and the dependent variables. In both countries, the large majority of subjects lived in intact families, with very few living in families in which neither natural parent was present. The level of satisfaction of subject with parent relationships appeared to be lower for the girls in both countries. Furthermore, girls in both countries were more likely than boys to say that their parents knew where they were on Saturday evenings. UK girls seemed less likely to live in intact families than French ones, and were more satisfied with the maternal relationship. There were striking differences between France and the UK for both boys and girls on the parental monitoring variable. French parents were much more likely than UK ones to always know where their teenage children were on Saturday nights. From Table 2, it is apparent that the French students were slightly more likely than the UK ones to smoke cigarettes, the UK students were much heavier consumers of alcohol than the French, the two countries were about equal in cannabis use and the UK students were slightly more likely to have used volatile substances or illicit drugs other than cannabis.

View this table:
Table 1.

Distributions of family structure, relationships to parents and parental awareness of the whereabouts of the subject on Saturday nights within France and the UK (%)

View this table:
Table 2.

Prevalence of substance use (%), gender differences within France and the UK and between-country differences

Tables 3–6 show the proportions of the samples who had used alcohol, tobacco or illicit drugs within the categories of the independent variables (e.g. in Table 3, 26.5% of French students living in intact families smoked one or more cigarettes per day in the past 30 days). In both countries, the pattern of use was clear. Students living in intact families were less likely than other students to use any substances, licit or illicit. There seemed to be little to choose between the other types of family. Tables 4 and 5 suggest that the relationships of the students to their parents were important in much the same way in both countries, with those who were dissatisfied being more likely to use substances. Finally from Table 6, in both countries, it appeared that there was a strong relationship between parental awareness/monitoring and the use of all the substances: the lower the parental awareness/monitoring, the more teenagers drank alcohol, smoked tobacco or used cannabis or other illicit drugs.

View this table:
Table 3.

Cigarette, alcohol and illicit drug use according to family structure within France and the UK (%)

View this table:
Table 4.

Cigarette, alcohol and illicit drug use and the maternal relationship within France and the UK (%)

View this table:
Table 5.

Cigarette, alcohol and illicit drug use and the paternal relationship within France and the UK (%)

View this table:
Table 6.

Cigarette, alcohol and illicit drug use within France and the UK according to parental awareness of the whereabouts of the subject on Saturday evenings (%)

Table 7 sets out the significant findings when logistic regressions were run on the dependent variables entering all the predictor variables together. Because of very small numbers, those living in ‘other’ families are omitted from the calculations. The main findings were as follows. In the presence of the other variables, parental monitoring still had highly significant effects in both countries for all substances. This was not always so for the other four variables. Significant gender effects were much less common in the UK than in France. In particular, French boys were much heavier consumers of alcohol than French girls, whereas there was no significant difference between boys and girls in the UK.

View this table:
Table 7.

Odds ratios obtained when family structure, gender, maternal relationships, paternal relationships and parental awareness of the whereabouts of the subject on Saturday nights are used together to predict substance use

Family structure was frequently still significant in the UK, where students from non-intact families were more likely to be alcohol and cannabis users. Family structure had less impact among French teenagers. There were few significant effects for maternal relationships in either country. Paternal relationships were highly significant among French students for illicit drug use, tobacco and drunkenness, but there were no such effects among UK teenagers.

In summary, in the UK, alcohol and cannabis use were related to family structure and to parental awareness or monitoring. Alcohol use among French teenagers was associated with gender and parental awareness/monitoring and cannabis was related to paternal (but not maternal) relationships and parental awareness.


The methodology of school surveys is now well established and validated (Johnston et al., 1999; Hibell et al., 1997, 2001), and self-report questionnaires are considered to be probably the best instruments for investigating psychoactive substance use among teenagers (Carroll, 1995). In both countries, over 80% of the eligible sample of students participated in the study (in France 89.1% and in the UK 85.6%), the missing data on lifetime questions were extremely low (in France <2.5%, in UK <1.0%), as were the percentages of inconsistent response patterns. Reliability and validity indicators suggested that the data were a good reflection of student behaviours regarding alcohol and other drug use and the two samples described above may reasonably be considered to have been representative of 15–16-year-olds.

The findings have shown that the UK and France had different patterns of substance use: more French students smoked tobacco, very many more UK students drank alcohol and about equal proportions in both countries had used cannabis, even if the two countries differed on the frequency of cannabis use during the last 12 months or the last 30 days. Finally, the UK teenagers had a greater tendency to have used illicit drugs other than cannabis.

Regarding alcohol use, the UK teenagers of both sexes were far more likely than their French counterparts to have consumed alcohol six or more times in the last 30 days, to have had alcohol more than 20 times in the past year, to have had five or more drinks in a row more than twice in the past 30 days and to have been intoxicated three or more times in the past year.

Clearly there are important differences in the drinking cultures of the two countries. A distinction has often been drawn between the northern regions of Europe (‘dry’ areas) and the southern regions (‘wet’ areas) (Plant and Cameron, 2000). In summary, compared to dry areas, wet areas are perceived to have a greater overall alcohol consumption, with more integration of alcohol into daily life, fewer government restrictions and fewer perceived alcohol-related problems (Allamani et al., 1999, p. 59). ‘In the wine cultures around the Mediterranean, men often drink together without drunkenness, whereas in the beer- or spirits- drinking cultures to the north, they drink to get drunk.’ ‘Associated with this difference … in the south men are responsible for social order … by contrast in the north it is women who are expected to assure moderation and propriety’ (Heath, 1995, p. 337). While these distinctions may have lessened in recent years, the UK drinking culture is clearly inclined towards northern Europe and that of France towards southern Europe. However, wine consumption has decreased in France among adolescents during the last decades, and beer and spirit consumption has increased, even though adolescent alcohol use remains lower in France than in UK. Moreover, gender differences were still pronounced in France with drinking behaviours much more frequent among males than among females, whereas there were slight differences between boys and girls in the UK. However, given that, to some extent, differences in cultural context may still prevail, the findings of the present study might not be unexpected. In particular, these cultural considerations might help cause the differences in alcohol consumption but would probably have little effect on other illicit drug use where the differences were small or non-existent.

Turning to cannabis use, the results failed to show differences between the two countries. In fact, the prevalence rates of cannabis use have dramatically increased in France for several years, whereas in the UK there has been a slight decrease (Plant and Miller, 2000). In 1993, ~16% of French students aged 16 years had experimented with cannabis use (Choquet and Ledoux, 1994). In the UK, in 1995, the figure was 40.6%. In a few years, the French level has risen to that of the top European countries such as the UK, and in 1999, 35% of the students in both countries used cannabis at least once (Hibell et al., 2001).

In the two countries, three of the distributions of the family variables described above were similar. Most of the boys and girls sampled in both countries lived in intact families and there were very few in families with neither natural parent present. Similar proportions of students in both countries were satisfied with their relationships to their parents, with the girls tending to be less satisfied than the boys. However, French parents were significantly more likely than UK ones to know where their children were on Saturday nights.

In turn, this could perhaps be due to the rather different levels of urbanization between the two countries. Although the total population levels are similar (~58 million in both countries in 1995), the population density in 1995 in France was 105 persons per square kilometre as against 239 persons per square kilometre in the UK. France had fewer cities of more than 200 000 inhabitants and was 27.2% rural, compared to 10.5% for the UK (UNESCO, 1995). Perhaps, in a more rural environment, it might be easier for parents to know their children's whereabouts.

Overall, the analyses confirmed that adolescents who lived in non-intact families were more likely to use alcohol, tobacco or illicit drugs. Neher and Short (1998) have found that children from divorced parents reported more substance use than did children of married parents, although this effect was sometimes quite weak (Amato and Keith, 1991). The associations seemed strongest among girls in the UK. These findings also highlight the effects of family relationships; as shown by Anderson and Henry (1994): ‘family bonding and parental support were negatively related to the adolescent substance use’. But obviously, as ESPAD was a cross-sectional survey, it was not possible to determine the direction of causation. Bad relationships with parents could cause or contribute to adolescent drug use, but the reverse is also true; adolescent drug use could reinforce conflicts and worsen the family climate. On the other hand, the family structure could affect attachment to parents, and relationships with parents (particularly with the father) could change after parental separation.

Finally, after logistic regressions, only parental awareness, i.e. parental knowledge of children's whereabouts on Saturday nights, remained strongly associated with alcohol, tobacco, and illicit drug use for both genders. In other words, once it is taken into account, the other family variables showed greatly reduced significance. We could suggest that parental awareness was related to social class. However, in the two countries, we failed to find significant and consistent tendencies between father's level of schooling or standard of living and parental monitoring (results not shown). Our findings are thus consistent with those of previous studies in that perceived authoritative parenting by students was associated with lower substance use (Richardson et al., 1993; Steinberg et al., 1994; Cohen and Rice, 1997; Reifman et al., 1998). This implies that children from non-intact families do not have a higher risk of experimenting with psychoactive substances, if the separated parents continue to exercise loving care and control over them. Cohen et al. (1994) suggested that ‘a lack of warm and positive relations with parents, poor parental discipline and monitoring are correlated with an adolescent's association with peers who use substance’, which in turn leads adolescents to experiment with licit or illicit drugs. In summary, parental divorce has little impact on adolescent drug use when relationships with parents can be maintained. However, in a longitudinal study, Hope et al. (1998) have shown that the effect of parental divorce appears to strengthen in adulthood.

There are some slight differences between the two countries. In France, students who have poor relationships with their parents are more likely to be engaged in substance use — paternal relationship is particularly related to illicit drug use — whereas in the UK, family structure seems to be more important. Perhaps the impact of parental separation or divorce may be attenuated in France by the maintenance of good relationships with parents. In the UK, where the majority of separations occur in families of lower socio-economic status, the social consequences may be more pronounced, and relationships with the absent parent (father in particular) may be almost non-existent. However, this is a hypothesis that would need to be tested in a further study.


The UK study was mainly supported by the Alcohol Education and Research Council, the Department of Health and Social Services, Belfast and the Health Education Authority. Additional assistance was provided by Allied Domecq plc, the North British Distillery Company Limited, the PF Charitable Trust, the Sir James Miller (Edinburgh) Trust and the Drapers Fund. The French study was funded by Observatoire Français des Drogues et des Toxicomanies (OFDT), and carried out in collaboration with OFDT and the Ministry of Education (MENRT). Sylvie Ledoux thanks Michel Chavance for his statistical help.


  • * Author to whom correspondence should be addressed.


View Abstract