Smoking, Drinking and Drug Use among Young People in England, 2021

The survey of Smoking, Drinking and Drug Use amongst Young people in England has been running since 1982. The survey takes place in secondary schools, with pupils in curriculum years 7 to 11 (mostly aged 11 to 15). Each survey since 1998 has included a core set of questions on smoking, drinking and drug use. In 2000, the survey questions changed to focus on smoking and drinking or on drug use in alternate years, but in 2016 the survey reverted back to including both drinking/smoking and drugs focused questions in one survey. However, only about half the pupils are asked the more detailed questions on smoking and drinking, with the other half asked more detailed questions on drug use. 

The 2016, 2018 and 2021 surveys were carried out by Ipsos UK on behalf of NHS Digital. The 2020 survey did not go ahead due to the Covid pandemic and was postponed to 2021. Surveys prior to 2016 were carried out by NatCen.

In 2021, 9,289 pupils in England were surveyed between September 2021 and February 2022 across 119 schools.

Where possible, specialist interviewers visited schools to support staff administering the survey. However, due to Covid outbreaks during the fieldwork period, this was not always feasible. Further information on measures adopted in response to Covid can be found throughout the appendices report.

Further detail on the 2021 survey methodology and all changes is provided in the following sections.

The questionnaires are available on the main publication page.

A2.1 Changes to questionnaire content in 2021

The questionnaire content remained largely unchanged in 2021. Just three questions were removed, four added and two amended. It is important to note that some of the new questions were added in response to the Covid pandemic and may warrant review for the next iteration of the survey in 2023.

A2.2 Changes to questionnaire content prior to 2021

Questions on wellbeing, were included for the first time in 2018.

A large-scale review of the questionnaire took place ahead of the 2016 survey. A review which included the input from the project steering group, cognitive testing and extensive piloting led to the following changes to questionnaire content:

• Additional questions on e-cigarettes, nitrous oxide, new psychoactive substances (also known as legal highs), beliefs about drinking, whether pupils had ever got drunk, consequences of drinking and family affluence.
• Removal of questions on water-pipe tobacco, types of cigarette smoked, energy drinks, drinking in pubs/bars/clubs, feelings after taking drugs, and free school meals.
• Changes to some descriptions used in the questionnaire (e.g. descriptions of e-cigarettes and legal highs, street names for some drugs).

Where comparability in time series data was affected, this is highlighted in the data tables, and details are provided in the section on 'Coherence and comparability' within the Data Quality Statement.

The survey population comprised pupils in Years 7 to 11 in secondary schools or in an equivalent year group in middle and upper schools. At the time of sampling, most of these pupils were aged between 11 and 15, although some of those pupils in year 11 would have been 16 by the time they took part in the survey. The sample included almost all types of secondary school in both the maintained and independent sectors of education in England. Special schools and pupil referral units were excluded from the survey^₁.

The survey uses a multi-stage probability design, in which first schools and then classes are selected using random methods. In 2021, 1,023 schools in England were sampled. In each participating school, three classes were then sampled (one class from Years 7 and 8, and two classes from Years 9, 10 and 11).  All pupils in the sampled classes were eligible for the survey.  More classes were sampled from the older year groups to increase the precision of estimates (see section A3.1 on sampling classes for more information).

This sampling design follows the design adopted for the first time in 2018 of sampling three whole classes. For more information, the 2018 report can be found here. However, as in 2018, some schools only offered two classes, but it felt better to include these schools than exclude them and reduce the response rate. In 2021, three schools wanted whole year groups to participate, which was accommodated with results weighted down to reflect this.

A4.1 Contacting schools

1,537 schools were sampled – 1,023 in the main sample and 514 in the reserve sample. All schools in the main sample were sent an initial letter during the summer term of 2021, explaining that they had been selected to take part in the survey and would be contacted at the start of the autumn term.

Ipsos contacted all schools again in 2021 at the start of the autumn term. Schools were invited to take part, and if they agreed, a date and time for an interviewer to visit the school were confirmed, and classes were selected. Interviewers then rang the school a few days before the survey appointment to confirm the visit time and answer any questions the school had. In 2021 schools were also given the option to administer the survey themselves with the teacher led option, questionnaires were sent to the school in advance and the recruiters would contact schools to confirm questionnaires had been completed and to arrange collection. This option was added in response to the Covid pandemic in instances where schools had restrictions on visitors to the school.

Follow-up letters, emails and calls were made to schools that did not respond to encourage them to take part in the survey. All schools in London and a random sample of schools outside of London were offered an incentive of £200 to take part. This was to assess the impact that providing an incentive has on recruitment.

However, the response rate was initially very poor, due to the ongoing impact on schools of Covid during the 2021-22 Autumn term. As such, the decision was taken to extend the incentive to all schools from November 2021.

A4.2 Carrying out the survey - interviewer led

Interviewers from Ipsos arranged with schools a convenient time to conduct the survey. Where possible, arrangements were made for the three classes to take part in the survey on the same day, either all in the same location at the same time, or at different times throughout the day (usually in their classroom). In some schools however, the survey sessions were completed over two separate visits.

Pupils completed the questionnaire under the interviewer’s supervision. The interviewer briefly explained why the survey was being carried out, what would happen to the information given, explaining how the questionnaire should be filled in, and stressing the voluntary nature of survey participation. The questionnaires are available on the main publication page.

Interviewers alternated the drug-focused and smoking-focused questionnaires when handing them out to pupils. This meant that a pupil would not be sat next to another pupil completing the same version of the questionnaire to ensure privacy of their answers.

Questionnaires were completed in ‘exam conditions’; pupils were not allowed to discuss the questions with each other or look at others’ answers. Pupils could request and receive help if they did not understand questions. A teacher was present during the completion of questionnaires, but they were not allowed to see pupils’ questionnaires at any stage of the survey. Interviewers stressed that pupils’ answers would be completely confidential, that no names were collected on the questionnaire and that their answers would not be shared with teachers. Questionnaires were serial numbered for administrative purposes, but serial numbers were not linked to pupils’ names.

In participating schools, a staff member was also asked to complete a questionnaire about how the school taught pupils about smoking, drinking alcohol and illicit drugs (the school level questionnaire).

A4.3 Carrying out the survey - school led

In 2021 schools were offered the option to complete the survey themselves without requiring an interviewer to come into the school. This change was made in response to the COVID-19 pandemic to allow schools to take part even if there were restrictions on visitors allowed into schools.  

When a school agreed to participate, recruiters sent a pack to schools which included the questionnaires and instructions on how to administer the questionnaire. Pupils completed the questionnaire under the class teacher’s supervision in exam conditions.

Once a school had completed the questionnaires, they contacted the recruiter, who would arrange for a courier to collect the questionnaires from the schools and return them to Ipsos for scanning.  For the analysis all responses were treated the same regardless of whether the survey was led by an interviewer or the school. The data tables named 'Data quality tables - Mode of survey delivery', show the impact of survey delivery mode on key estimates for smoking, drinking and drug use.

A4.4 Mop-up sessions

Mop-up sessions were held when 4 or more pupils were missing from class. Interviewers were required to schedule the mop-up session on the day of the initial visit. In cases where the school had opted to lead the survey sessions, instructions were provided to guide them through the setting up mop-up sessions at a convenient time. The absent pupils were either interviewed  in a separate session, or with another class  taking part at a later date. Mop-up sessions were conducted with 164 pupils. Information on mop sessions was provided by teachers via completed contact sheets.

A4.5 Withdrawing pupils from the survey

Schools are required to send out letters to parents/carers of pupils in selected classes to inform them when the survey will be taking place and to provide information about the survey content. The letter provides parents with an opportunity to withdraw their child from the survey.

In accordance with MRS guidelines, pupils were provided with an opportunity to decline to take part (even though a responsible adult has given permission for their participation)

In total, 61 pupils decided to opt-out or were withdrawn out by their parents/carers. Information on opt-outs was provided by teachers via completed contact sheets.

To try and ensure an equal number of schools from each region in the final sample, selection probabilities differed between regions. The different probabilities of selection were adjusted as part of the calibration weighting (see section A8). Table 3 shows the distribution of schools sampled by region.

Table 3: Distribution of pupil population by region and percentage of schools sampled

^a Pupils in Years 7 to 11, based on DfE population estimates for 2021.

The overall probability of selection (or sampling fraction) was the product of the sampling fractions at the first and second stages, i.e.

F = f₁ x f₂

where

f₁ = probability of selecting the school.

f₂ = probability of selecting the class.

Schools were sampled with probability proportional to the number of pupils in Years 7 to 11, so that roughly equal numbers of pupils could be sampled from each selected school. Thus:

f₁=n₁ x N_i/N_r

where

n₁ = total number of schools to be selected in the region.

N_i = number of pupils in an individual school in Years 7 to 11.

N_r = total number of pupils in the region in Years 7 to 11.          

Within each school pupils in years 9, 10 and 11 had a higher probability of being selected to take part in the survey. In each school, one class was selected from Years 7 and 8 and two classes were selected from Years 9,10 and 11. Within each class all pupils were selected. Thus:

f₂  = n_c/N_c

where:

n_c is the number of classes (i.e. 1 or 2) selected in the Year Grouping (i.e. Years 7 and 8 or Years 9,10 and 11) in each school.

N_c is the total number of classes in the Year Grouping in the school.

Overall, therefore, the sampling fraction for each pupil was:

F= (n₁ x (N_i/N_r)) x (n_c/N_c)

In total, 119 schools agreed to take part in the survey out of the 1023 eligible schools, a response rate of 12%. Of the 119 schools taking part, 34 took part before schools closed for Christmas 2021 and the remaining 85 from January 2022 onwards. Analysis took place in 2016 which concluded conducting surveys after Christmas did not significantly impact the results (see Appendix B for more details).

The response from selected pupils in participating schools was 92% (the pupil response rate was calculated based on information provided by teachers via completed contact sheets), yielding a total of 9289 completed, usable questionnaires. The product of the school and pupil rates gave an overall response of 11% (See Table 4 and Figure 1).

Of the 119 schools participating in the survey, 102 responded to the teacher questionnaire, a response rate of 86%.

Table 4: School and pupil response, 1982 - 2021

Figure 1: SDD Overall response rate 1982-2021

Since the survey began in 1982 school response rates have continued to decrease. This was also worsened in 2021 as a result of the Covid pandemic, which led to continued teacher and pupil absence, and thus the pressure on school time to catch-up with missed learning. Despite the drop in school response rates, pupil response rates have remained high across the surveys’ lifetime and even increased in recent years. 

Table 5 shows the response rate by region, which varied considerably. In particular, the response rate in London and Yorkshire & the Humber was poor. 

Table 5: School response by region

In the 2021-22 academic year, the coronavirus (COVID-19) pandemic continued to have a huge impact on schools, with the omicron wave leading to teacher shortages and pupil absences, resulting in a strain on school’s time and resources. This contributed to the high number of schools refusing to take part (376). Additional reasons for non-participation included a lack of time, not being interested in the subject and involvement in too many surveys. These reasons were similar to those given in 2018.

Table 6: Reasons given by schools for not taking part in the survey

Questionnaires were either returned by the school interviewers or by courier to Ipsos’s data capture department to be scanned and processed. Checks were conducted to ensure that the scanned data were captured correctly. As part of this, all handwritten text and characters were reviewed and verified. Random batches of questionnaires were then selected and checked to ensure that the data had been verified correctly.

As with previous years, the following checks were made to the data:

•   Checking that filters were correctly followed.
•   Checking ranges on consumption and age of first use variables.
•   Checking whether answers given as ‘other’ answers could be back-coded into existing codes.
•   Resolving inconsistencies between answers.

A dataset was then created for analysis purposes.

The sample is weighted to match the school population counts so that the survey’s estimates are representative of the full school population. The school population counts were derived from the most recently completed DfE School Census.

The weighting was carried out in two stages. For this survey wave, schools were allowed to include more than one class in a school year, but the sampling was based on only including one school year.  Thus, the first stage of weighting was to produce a set of weights that adjusted the number of pupils in any school year for such school back to a nominal class size of 25. 

The second stage then used cell weighting to adjust the weights from the first stage directly to the population counts for the school year by gender separately for each region (Table 7).

Table 7: Population counts used for weighting

These weights were then rescaled to sum to the achieved sample size. Both unweighted and weighted bases are shown in the tables. The unweighted bases represent the number of pupils who responded; the weighted bases have no particular meaning per se, but are used when calculating estimates.