CHAPTER TWO: RESEARCH METHODS AND SURVEY SAMPLE
2.1 A key aim of this research was to establish the prevalence of, and current attitudes towards, drinking and driving in Scotland, and to compare these with the findings from the previous study (in 2001). Therefore, it was important that the survey methodology remained as similar as possible between these 2 studies to allow comparability. At the same time, the desire to provide forward-looking recommendations meant that some change was necessary.
2.2 As with the previous phase of this research, this study consisted of 2 elements: a quantitative survey and a qualitative element. These 2 are explored in more detail in the following sections.
Quantitative methodology
2.3 To allow comparison between this report and the previous study, the same quantitative methodology was used as that of the 2001 survey. Therefore, face-to-face interviews were conducted with a pre-selected, random sample of current drivers, that is with adults aged 17 or over in Scotland who have driven in the last year. A total of 1,034 interviews were conducted with current drivers.
2.4 As in 2001, the survey was conducted using computer-assisted personal interviewing ( CAPI) for the interviewer-administered section of the questionnaire and computer-assisted self-interviewing ( CASI) for the self-completion section. The self-completion section, which comprised the majority of the questionnaire, contained questions relating directly to respondents' own drinking and driving behaviour and to their attitudes to drinking and driving and drink-driving. Less sensitive topics, such as driving habits and opinions about penalties, were asked in the interviewer-administered sections.
2.5 To help overcome potential resistance to using CASI and to allow respondents to familiarise themselves with the CAPI machine, the respondents completed a series of CASI practice questions helped by the interviewer.
Social acceptability and under-reporting
2.6 The use of self-reported measures of behaviour can lead to problems in surveys, particularly where the behaviour being measured is seen as undesirable or socially unacceptable. The under-reporting of behaviours, such as alcohol consumption is well documented, and it is to be expected that this would be exacerbated when measured in conjunction with behaviour such as drinking and driving.
2.7 Many of the issues covered in this survey could be at risk of under-reporting, including prevalence of drinking and driving, frequency of drinking and driving and consumption levels when such behaviour occurs. It is desirable, then, to minimise the level of under-reporting. At the same time if under-reporting is detected, this provides an indicator of the strength of accepted social attitudes towards drink-driving. Qualitative research, which was undertaken in the form of focus groups and depth interviews, contributed to detecting the likelihood of under-reporting in the study as a whole.
2.8 A number of measures were used to encourage accurate reporting of drink-drive behaviour, including:
- Stressing repeatedly the absolute confidentiality of all responses
- Stressing how important it is that we obtain accurate information about this issue - i.e. appealing to respondents' 'social conscience'
- Employing a Computer Aided Self-completion Interview ( CASI) element for the majority of the questionnaire, in which the respondent confidentially completed sensitive sections of the questionnaire independent of the interviewer. Previous research has demonstrated that this increases accuracy of responses
- Structuring the questionnaire in such a way that more sensitive questions were asked later on, allowing the respondent to become comfortable with disclosing less sensitive details first
- Phrasing the questions in a way that:
- encouraged respondents to believe that drinking and driving is normal
- explicitly stated that drinking certain quantities and then driving is legal
- emphasised that all situations when respondents had anything to drink, no matter how little, before driving was of interest
Sampling
2.9 Given one of the objectives of the research was to compare changes over time, it was important to keep as many survey variables as possible, including sampling method, the same as that used in the 2001 research on this topic. As before, the method of selecting households used to identify respondents to take part in the survey was based on a multi-stage stratified design, with wards selected across local authorities and interviewing locations selected from within wards based on census output areas. These were then selected with probability proportionate to number of cars, a census variable, as a proxy for the number of drivers. As was the case previously, no separate listing of current drivers was available so a screening exercise was used to identify current drivers (i.e. those who had driven a motor vehicle on a public road in the past year) at the point when addresses were selected.
2.10 Following the selection of output areas, addresses were selected at random from within those areas, and interviewers visited the selected households firstly to identify those households containing drivers and, secondly, to attempt to conduct interviews with as many drivers in each eligible household as possible.
2.11 Calls were made at different times of day and different days of the week with at least 50% made in the evening and at the weekend to further ensure 'difficult to reach groups' were included.
2.12 In total, 1034 respondents were interviewed. The sample characteristics are described in greater detail in Chapter 3, with comparisons made with the sample from the 2001 research where appropriate.
Questionnaire
2.13 As with other aspects of the survey method it was desirable to keep the questionnaire the same as that used in the previous research in 2001 to ensure comparability of results. At the same time, there was a requirement to include some different questions, in particular to explore the extent to which drivers held different attitudes about drinking and driving.
2.14 Changes to the questionnaire were kept to a minimum in the section gathering information on behaviour such as prevalence, frequency and consumption levels, to aid comparison with the previous study. Questions were added towards the end of the questionnaire exploring drivers' attitudes; this meant that some questions were dropped from the preceding sections so that the questionnaire length remained around an average of 20 minutes overall. The questions that were dropped related to the detail of respondents' social lives, concerns about driving and beliefs about other people's drink-driving habits. In addition, the options for some multiple choice questions were condensed based on the responses from the previous research. A copy of the questionnaire is in Annex A.
Qualitative methodology
2.15 The main aim for the qualitative research was to explore attitudes to drink-driving and drink-driving behaviour in more detail than was possible in the questionnaire and specifically with an audience who admitted to driving after drinking alcohol or to drink driving. An additional aim was to provide advice on future target audiences and to assist in developing strategic insight to inform drink-drive campaign/publicity initiatives.
2.16 The 2001 study conducted 6 peer focus groups (i.e. a group of people who knew each other and shared similar demographic characteristics), 4 general population focus groups composed of people who admitted drinking after driving (though not necessarily believed they were over the limit) in the past, 3 depth interviews with people admitting to being current drink-drivers and 3 depth interviews with people who had been prosecuted for drink-driving.
2.17 Rather than repeating the research with peer groups, the 2007 qualitative research instead focussed more on particular groups and individuals that were of interest in terms of the revised objectives of the study. Groups of particular interest were identified in advance of the research taking place though existing knowledge, research and statistics. The results of the quantitative survey, which was carried out as the first phase of the research, were used to refine these groups further.
2.18 The qualitative research involved 6 focus groups with those who said they did drive after drinking any amount of alcohol, 4 depth interviews with individuals who admitted driving in the past year when they suspected they were over the drink-drive limit and 2 depth interviews with drivers who had previously been disqualified for drink-driving. Particular groups of interest to the research, including people who admitted to driving after having a drink or driving when they suspected they were over the limit in the past year; females aged 25 and above; drivers in rural areas; younger men aged 17-20; and more affluent, older men were incorporated into the qualitative research.
2.19 Further information on the structure of the focus groups is given in Chapter 3.
Topic guide for focus groups
2.20 The topic guides for the focus groups were developed following on from the quantitative research ( see Annex B). The findings from the survey were used in conjunction with a communication strategy developed by The Union advertising agency 6, which outlined the strategic and creative challenges of communicating to individuals who drink and drive. A topic guide was developed with input from The Union to cover the following areas:
- What people understand by drinking and driving
- Their attitudes to drinking and driving, and how (if at all) these have changed over time
- Motivations for drinking and driving
- The risks, consequences and deterrents to drinking and driving
- Experiences of, and attitudes towards, drinking and driving
2.21 Within the focus groups, respondents were taken through a number of projective exercises which were designed to facilitate discussion around more sensitive areas which might be perceived as socially unacceptable, such as drinking over the legal limit and driving. For example, the respondents were asked to imagine how others might behave in certain scenarios, rather than be asked directly what they would do. The various projective techniques are discussed in section 2.29.
2.22 The topic guide used in the focus groups can be found in Annex B.
Depth interviews
2.23 The purpose of the depth interviews was to explore the attitudes and behaviour of those admitting to having driven in the last year when they thought they might have been over the limit, and the attitudes and experiences of disqualified drivers. Further information is given about the respondents interviewed in this phase of the research in Chapter 3.
Topic guide for depth interviews
2.24 The topic guides for the depth interviews were developed to complement the guides for the focus groups. Two separate topic guides were used: one for drivers who had been disqualified for drink-driving and one for those who had driven while over the limit. The topic guide used in the depth interviews with those who had driven while over the limit was very similar to the focus group topic guide. This can be found in Annex C.
2.25 The topic guide for disqualified drivers did not use projective techniques (see section 2.29), but instead asked respondents to tell the story of their disqualification, exploring:
- Their attitudes to drinking and driving, and how (if at all) these have changed over time
- Motivations to drinking and driving, and how (if at all) these have changed over time
- The risks, consequences and deterrents to drinking and driving
- Experiences of, and attitudes towards, drinking and driving, and how (if at all) these have changed over time
2.26 The topic guide used in the depth interviews with disqualified drivers can be found in Annex D.
Social acceptability and under-reporting
2.27 As previously highlighted, self-reported measures of behaviour can lead to problems in surveys, particularly where the behaviour being measured is seen as undesirable or socially unacceptable. For the qualitative work, it was recognised that there was a strong likelihood that participants would talk down the significance or frequency of their drinking and driving. This means techniques designed to desensitise and normalise the behaviour and assurances of confidentiality were used. Additionally, we undertook individual depth interviews (rather than focus groups) with individuals who admitted to drinking and driving currently while over the limit. Depths rather than groups were appropriate with these individuals as experience suggests that respondents would be unwilling to discuss such issues in front of others even if they had similar experiences.
Techniques used
2.28 Various techniques were used to enable respondents to discuss freely behaviour they believe to be socially unacceptable, to provide more considered views about the consequences of being caught drink-driving and to assist in developing future communications.
2.29 The following techniques were used:
2.30 Bubble drawings: A cartoon drawing of 2 people in a situation was shown to participants, who were asked individually to fill in the thought or speech bubble for the figures shown. Two bubble drawing projectives were developed for the groups; the one for the younger male groups showed 2 men in the pub, finishing their second pint with one holding car keys. The second was used for the remaining 4 groups, and showed a couple dining in a restaurant on their second glass of wine. The respondents were told that the characters had the car outside and were asked to consider what they were thinking.
2.31 This allows respondents to discuss what others might be thinking or feeling in order to unlock and clarify thoughts which can be difficult to verbalise via a simple questioning technique. It is designed to gain the confidence of participants and explore their unconscious or repressed impressions, feelings or opinions regarding drink-driving. In this case the aim was to elicit information about the personal risk audit drivers undertake before choosing whether to drink and drive or not.
2.32 Proposition statements: 15 proposition statements were developed in conjunction with TNS System Three and The Union advertising agency ( see Annex E). The statements were designed to help identify which concepts would be most effective in future communications in terms of resonating with different groups of people who might be at risk of drink-driving. For example, 'drinking and driving is for losers' was developed to resonate with younger drivers, and 'if I've had a meal as well as some drinks over the course of the evening, I think that means I am safe to drive' was intended to explore perceptions of risk among older drivers. As the statements which elicited the strongest responses were thought most likely to change behaviour, the focus group respondents were asked to sort each statement in relation to whether they agreed or disagreed with each one.
2.33 Visualisation technique: At the end of the focus group session and individual interviews, the respondents were asked to visualise being caught drink-driving. They were asked to close their eyes and imagine what thoughts would go through their minds, what their response would be and what being caught drink-driving would mean to them.
Key Points
This research aimed to establish the prevalence of, and current attitudes towards, drinking and driving in Scotland. It aimed to compare these findings with the results of a similar study undertaken in 2001. A further aim was to advise on future drink-drive campaigns.
The research consisted of 2 elements:
- Quantitative: A survey was undertaken with 1034 adults aged 17 or over who had driven in the last year. The results are broadly representative of drivers in Scotland.
- Qualitative: The qualitative research involved 6 focus groups and 6 depth interviews. The focus groups were undertaken with drivers who admitted to currently drinking and driving; 4 interviews involved drivers who suspected they have driven while over the limit in the past 12 months and 2 interviews were with people disqualified from driving within the past 2 years.
The research was designed with an awareness that people may under-report undesirable or socially unacceptable behaviours and tried to reduce under-reporting through the methodological techniques used.