Sample

The ESPAD target population is defined as students who turn 16 in the calendar year of the survey and are present in the classroom on the day of the survey. Students who were enrolled in regular, vocational, general or academic studies were included, excluding those who were enrolled in either special schools or special classes for students with learning disorders or severe physical disabilities. Table 2 shows the main sample characteristics. The methods are largely comparable in all countries, although there are characteristics, for example sample type, mode of administration or time of data collection, that may differ between countries.

In each participating country, a cluster sampling design was used to sample the target population, except in the Faroes, Iceland, Liechtenstein, Malta, Monaco and Montenegro, where all 1999-born target students were included. Data were collected by self-administered questionnaires. All countries used a paper-and-pencil questionnaire, except for Austria, Latvia, Liechtenstein and the Netherlands, where students answered a web-based questionnaire. Based on a methodological study in Latvia, only small differences in students’ responses to online and traditional paper-andpencil questionnaires were found (Trapencieris, 2013), and comparability was considered satisfactory. The students answered the questionnaires anonymously in the classroom, with teachers or research assistants functioning as survey leaders. The questionnaires were provided by school staff (18 countries), teachers (13 countries) or research assistants (four countries). In the majority of countries, data collection took place between February and May 2015, except for Belgium (Flanders), where data were collected 6 months earlier (autumn 2014), and Georgia and the Netherlands, where data were collected 6 months later (autumn 2015). In most countries, class was the last unit in a multistage stratified sampling process.

All samples were nationally representative, except for Belgium (only the Dutch-speaking part, Flanders), Cyprus (only government-controlled areas) and Moldova (Transnistria region not included). Sample sizes varied between 316 in Liechtenstein and 11 822 in Poland. In 2015, data on 96 046 students were collected in 35 countries covering 2.9 % of the population of adolescents born in 1999. The school participation rate (share of selected schools taking part in the survey) ranged from 21 % to 100 % and the class participation rate (share of selected classes participating) varied between 17 % and 100 %. The proportion of students of selected classes that were present on the day of the survey and answered the questionnaire was high (80-84 %). At the time of data collection, students were on average 15.8 years old, with country means varying between 15.7 and 16.4 years. The coverage of students was very high, with 30 countries reaching 90 % of the target population or more. Lower rates were reported in Denmark (78 %) and Georgia (73 %). Data were weighted in 11 countries to account for the cluster sampling design and/or to adjust the sample to the sociodemographic composition of the target population.

Measures

The questionnaire covers young people’s awareness of and experience with different licit and illicit substances, internet, gaming and gambling with money. The questions are designed to collect information on the use of psychoactive substances and the use of the internet for various activities in the lifetime, the last 12 months, the last 30 days or the last week previous to the survey, and consumption patterns such as frequency or quantity (e.g. volume, hours).

Availability of substances

The perceived availability of substances is a proxy measure for how easy or difficult it is for students to get a particular substance (cigarettes, alcohol or illicit drugs). Students were asked how easy they estimate it would be to get hold of particular substances within 24 hours if they wanted to. The response categories were ‘impossible’, ‘very difficult’, ‘fairly difficult’, ‘fairly easy’, ‘very easy’ and ‘don’t know’. The proportions of students in each country answering ‘fairly easy’ or ‘very easy’ were merged to indicate easy availability. Availability of each type of different alcoholic beverage (beer, wine and spirits) was evaluated separately. If considered relevant, countries included optional beverages such as cider or alcopops in the questionnaire.

Age of first substance use

Students were asked how old they were when they used a particular substance for the first time, started to use it on a daily basis (cigarettes) or experienced excessive use (alcohol intoxication). The response categories ranged from ‘9 years old or less’ to ‘16 years or older’, in increments of 1 year, and ‘never’. An age of initiation of 13 years or younger was taken as an indicator of early onset.

Cigarette use

Students were asked on how many occasions they had ever smoked cigarettes, with the response categories being ‘0’, ‘1-2’, ‘3-5’, ‘6-9’,’10-19’, ‘20-39’ and ‘40 or more’. Quantity of cigarette use in the last 30 days was also collected. The response categories ‘not at all’, ‘less than 1 cigarette per week’, ‘less than 1 cigarette per day’, ‘1-5 cigarettes per day’, ‘6-10 cigarettes per day’, ‘11-20 cigarettes per day’ and ‘more than 20 cigarettes per day’. Lifetime prevalence (any use) and prevalence of daily use (at least 1-5 cigarettes per day) were calculated. Daily use of cigarettes was considered as having smoked a minimum of one cigarette each day.

Alcohol use

Students were asked on how many occasions they had consumed alcoholic beverages and had been intoxicated in their lifetime and during the last 30 days. The response categories ‘0’, ‘1-2’, ‘3-5’, ‘6-9’, ‘10-19’, ‘20-39’ and ‘40 or more’. The average number of occasions was calculated as the average based on the mean value of each response category, for example 29.5 for the category ‘20-39’. For the category ‘40 or more’, the value 41 was used. Prevalence of any use (lifetime, last-30-day) and prevalence of experiencing any intoxication were also calculated (≥ 1-2 times). Moreover, heavy episodic drinking is defined as drinking a minimum of five alcoholic beverages on one occasion at least once in the last 30 days, which corresponds to a cut-off of approximately 9 centilitres of pure alcohol. The volume of alcohol intake was calculated as the total volume of pure ethanol summed across different alcoholic beverages (beer, wine, spirits, alcopops and cider).

Illicit drug use

To measure lifetime experience with illicit drugs, students were asked on how many occasions they had tried different drugs in their lifetime, with the response categories being ‘0’, ‘1-2’, ‘3-5’, ‘6-9’,’10-19’, ‘20-39’ and ‘40 or more’. Frequency of use was asked for cannabis (marijuana or hashish), ecstasy, amphetamine, methamphetamine, cocaine, crack, LSD or other hallucinogens, heroin and GHB (gammahydroxybutyrate). The average number of occasions using cannabis was calculated as the average based on the mean value of each response category, for example 29.5 for the category ‘20-39’. For ‘40 or more’ the value 41 was used.

Inhalant use

Students were asked how often they had used inhalants in their life, with the response categories being ‘0’, ‘1-2’, ‘3-5’, ‘6-9’,’10-19’, ‘20-39’ and ‘40 or more’. Prevalence of any use of inhalants was based on intake on at least one occasion.

New psychoactive substance use

New psychoactive substances (NPS) were defined as ‘substances that imitate the effects of illicit drugs such as cannabis or ecstasy and are sometimes called “legal highs”, “ethnobotanicals” or “research chemicals” and can come in different forms (herbal mixtures, powders, crystals or tablets)’. Students were asked how often they had used NPS in their life, with the response categories being ‘0’, ‘1-2’, ‘3-5’, ‘6-9’,’10-19’, ‘20-39’ and ‘40 or more’. Prevalence of any use of NPS was based on intake on at least one occasion.

Pharmaceutical use

To measure lifetime experience of use of pharmaceuticals, students were asked on how many occasions they had used tranquillisers or sedatives without a doctor’s prescription, anabolic steroids or painkillers in order to get high, with the response categories being ‘0’, ‘1-2’, ‘3-5’, ‘6-9’,’10-19’, ‘20-39’ and ‘40 or more’. Prevalence of any use was based on intake on at least one occasion.

Conditional probabilities of substance use

Conditional prevalence rates were calculated as the prevalence of use of one substance conditional on the use of another substance. This analysis is based on substance users across all countries. It is neither assumed that the use of a particular substance has occurred before the use of another substance nor assumed that the use of a substance is caused by the use of another substance.

Internet use, gaming and gambling

To assess patterns of internet use, including online gaming and gambling (online and offline), students were asked on how many of the last 7 days had they used the internet, and how many hours had they spent on the internet on an average day on which they had used the internet. This information was asked for various online activities such as social media (communicating with others), searching for information, streaming or downloading music, buying/ selling, gaming (which is defined for the purpose of this study as playing games) and gambling for money. Based on that, the average number of days using the internet and the prevalence of using the internet at least four times for each of these activities in the last 7 days (also referred to as regular use) was calculated. Gambling for money was further assessed by asking students about the frequency of particular gambling activities in the last 12 months (playing slot machines, cards or dice, lotteries or betting on sports or animals). The response categories were ‘not gambled’, ‘monthly or less’, ‘2-4 times a month’, ‘2-3 times a week’, ‘4-5 times a week’ and ‘6 or more times a week’. Prevalence rates were calculated for last-12-month (at least once) and frequent (2-4 times a month or more) gambling.

Table 2. Sampling characteristics of ESPAD 2015

Download XLS  
Country Geographic coverage Data collection mode Sample type Sampling unit(s)  Data weighted Student representa-tiveness (%) a Class participation rate (%) b Students’ presence rate (%) c Mean age d n
Albania National Pen and paper Stratified random School/class No 95 100 94 15.9 2,553
Austria National Web survey Proportionate random School/class Yes 90 17 e 90 15.9 3,684
Belgium f Flanders g Pen and paper h Stratified random School/class Yes 94 56 i 94 15.8 1,771
Bulgaria National Pen and paper Simple random Class No 99 98 84 16.0 2,922
Croatia National Pen and paper Stratified simple random School/class No 94 98 89 15.7 2,558
Cyprus National j Pen and paper Stratified random Class No >90 85 n.a. 15.8 2,098
Czech Republic National Pen and paper Stratified random School/class Yes >95 96 i 83 16.0 2,738
Denmark National Pen and paper Stratified simple random School/class No 78 k 26 i 88 15.8 1,670
Estonia National Pen and paper Stratified random School/class No 97 l 90 83 15.7 2,452
Faroes National Pen and paper Total No sample No 88 100 92 15.7 511
Finland National m Pen and paper Stratified random School/class No 93 85 89 15.8 4,049
FYR Macedonia w National Pen and paper Systematic random Class No  92 q 98 88 15.8 2,428
France National n Pen and paper Stratified random School/class Yes 94 93 87 15.9 2,714
Georgia o National Pen and paper Proportionate simple random School/class No 73 98 86 16.4 1,966
Greece National Pen and paper Stratified random Class Yes 91 95 92 15.8 3,202
Hungary National Pen and paper Stratified random Class Yes 97 93 85 15.7 2,735
Iceland National Pen and paper Total No sample No 96 79 86 15.8 2,663
Ireland National Pen and paper Stratified systematic random  School/class No 98 18 e 86 15.9 1,470
Italy National Pen and paper Stratified proportionate random Class No 99 85 88 15.7 4,059
Latvia National Web survey Stratified random cluster sampling Class Yes 95 p 42 85 15.9 1,119
Liechtenstein National Web survey Total No sample No ~99 100 93 15.7 316
Lithuania National Pen and paper Stratified random School/class No 85 99 88 15.7 2,573
Malta National Pen and paper Total No sample No 93 98 83 15.7 3,326
Moldova National r Pen and paper Simple random Class No 90 100 87 15.9 2,586
Monaco National Pen and paper Total No sample No ~99 100 91 15.8 397
Montenegro National Pen and paper Proportionate simple random Student No 94 100 87 15.9 3,844
Netherlands o National Web survey Stratified simple random School/class Yes 94 43 i 93 15.9 1,684
Norway National Pen and paper Stratified random School/class Yes 98 s 53 90 15.8 2,584
Poland National Pen and paper Stratified random School/class Yes 95 94 83 16.0 11,822
Portugal National t Pen and paper Stratified systematic random Class No 86 q 96 93 15.9 3,456
Romania National Pen and paper Systematic random School/class No 91 s 100 84 15.9 3,500
Slovakia National Pen and paper Stratified proportional random School/class No 98 100 89 15.8 2,208
Slovenia National Pen and paper Stratified random Class No 94 99 88 15.8 3,484
Sweden National Pen and paper Simple random School/class No 95 83 86 15.7 2,554
Ukraine National u Pen and paper Stratified systematic random School/class Yes 92 98 80 16.0 2,350
AVERAGE or SUM           93 87 v 88 15.8 96,046

a Proportion of ESPAD target students covered by the sampling frame.

b Proportion of selected classes participating in the survey.

c Proportion of students of participating classes answering the questionnaire.

d Based on the data collection period.

e Estimated from the maximum number of classes that could participate.

f Data collected in previous autumn instead of spring.

g Geographic population coverage 61 %: only Flanders and Dutch-speaking schools in the Brussels Capital region are covered by the sampling frame.

h A few classes in the ESPAD sample answered the online version.

i School participation rate (class participant rate unknown).

j Geographic population coverage approx. 80 %: only government-controlled areas are covered by the sampling frame.

k Boarding schools not included.

l Vocational schools not included (less than 2 % of students born in 1999).

m Geographic population coverage 99 %: the Åland Islands are not covered by the sampling frame.

n Geographic population coverage 96.5 %: DOM-TOM territories (overseas departments and territories such as French Guiana, Réunion and those in the Caribbean) are not covered by the sampling frame.

o Data collected in autumn instead of spring.

p Vocational schools not included (1.7 % of students born in 1999).

q Private schools not included.

r Geographic population coverage 85 %: the Transnistria region is not covered by the sampling frame.

s Estimations by principal investigator.

t Geographic population coverage 95 %: the islands of the Azores and Madeira are not covered by the sampling frame.

u Geographic population coverage 95 %: AR Crimea is not covered by the sampling frame.

v Only countries with class participation rates excluding Belgium (Flanders), the Czech Republic, Denmark and the Netherlands.

w Official name former Yugoslav Republic of Macedonia.

Data processing and data quality

Data were centrally cleaned in two steps. First, all cases with missing information on gender were excluded from the database. The other major reason for exclusion was poor data quality. All cases with responses to less than half of the core items were discarded, as were all cases where the respondent appeared to have followed patterns involving repetitive marking of extreme values. Across all ESPAD countries, an average of 1.8 % (0.0-7.6 %) of cases were excluded because of poor data quality or missing information on gender.

Second, logical substitution of missing values was performed in a rather conservative way. In cases where students had indicated that they had never used a specific substance and subsequently did not respond to further questions about such use, any missing values were substituted with no use for that particular substance. However, no substitutions were made if any contradictory indications of use were at hand. For the seven substance use variables where substitutions were performed, the average reduction of the non-response rate was rather small, ranging from 0.1 % to 0.5 %. The single highest country-specific reduction was found in Norway, where the non-response rate for lifetime inhalant use was reduced by 2.7 percentage points. Norway, the former Yugoslav Republic of Macedonia and Latvia were the countries where the logical substitution of missing values had the biggest impact. However, the reductions in nonresponses had only minor effects on the final prevalence estimates.

A few countries experienced modest methodological problems but, with the exception of Latvia, not of such a magnitude as to seriously threaten the comparability of the results. Compared to the ESPAD average, higher rates of inconsistencies indicate a somewhat lower data quality in Albania, Bulgaria and Cyprus. Low school/class participation rates in Austria (17 %), Denmark (26 %), Ireland (18 %), Latvia (42 %) and the Netherlands (43 %) resulted in turn in relatively small net sample sizes. In Austria (4.2 %), Cyprus (3.8 %), Latvia (7.6 %) and Norway (4.2 %), a relatively high proportion of cases had to be discarded in the central datacleaning process. Due to sampling of only one school grade or not including boarding schools, the coverage of the target student population in Denmark (78 %), the former Yugoslav Republic of Macedonia (79 %) and Georgia (73 %) was below average. Finally, a relatively high proportion of parents refused permission for their child to participate in the survey in Portugal and Romania (6.9 % each).

Due to the uncertainty of data-collection procedures, Latvia is excluded from the standard reporting and the calculation of the ESPAD average. In all tables, Latvia is reported separately to illustrate difficulties in comparability. More details on the ESPAD methodology are available online (http://www.espad.org).

Analysis

Prevalence estimates and means were calculated for each participating country, taking weights into account where necessary. In all tables, totals and gender-specific estimates for boys and girls are presented by country. Gender differences reported in Figures 1-9 were tested using either simple linear regression for quasi-continuous frequency measures or logistic regression for prevalence, with gender as predictor. Conditional probabilities expressing the use of one substance given the use of another substance were calculated for cigarettes, alcohol, cannabis, ecstasy, amphetamine, methamphetamine, cocaine, crack, LSD or other hallucinogens, heroin, GHB, inhalants, NPS, tranquillisers or sedatives, painkillers and anabolic steroids. The ESPAD average is based on 35 countries assigning equal weight to each country. Latvia was excluded from the calculation of the ESPAD average due to validity concerns. All percentages in the report were calculated on the basis of valid responses and are shown for totals, boys and girls. With the exception of frequency of alcohol use (Figures 2a, 2b), alcohol intake (Figures 3a, 3b), preference of alcoholic beverages (Figures 4a, 4b) and frequency of cannabis use (Figures 7a, 7b), where the estimates are based on consumers of a particular substance, all estimates are based on the total sample and represent population estimates.