Answers to Exercises and questions for Discussion

Is data analysis any more than choosing the right kinds of statistics to apply to a dataset?

The answer to this question, if you follow the argument in this text, is an emphatic ‘Yes’. Data analysis is much more than just applying statistics to a dataset. Data analysis is not just about performing statistical calculations on numerical variables; it is about making sense of a dataset as a whole and thinking about a range of alternative ways of approaching its analysis, taking a well-rounded view of what all the evidence is saying. Analysis becomes a dialogue between ideas and evidence. Data analysis is the process whereby researchers take the raw data that have been entered into the data matrix and create information that can be used to address the objectives for which the research was undertaken. The processes that that analysis entails will include at least preparing and describing the data. Some ‘descriptive’ analyses may stop at that, but to create information that is useful to policy makers or to clients or will help academics to understand social phenomena, it is usually necessary to go beyond just giving an account of a dataset so that data are, in addition, interpreted, related, evaluated, explained, applied and presented. This is true whatever the kind of data, but which particular activities are involved within each of these tends to differ according to whether the data are quantitative or qualitative and whether the approach is variable-based or case-based.

Access the alcohol marketing dataset which is available at https://study.sagepub.com/kent and check out the values that have been entered for the codes for the nine variables in Figure 3.1 for the non-metric measures. From Figure 3.1, try to summarize each variable by looking down each column. Try to summarize each case by looking across the variables.

The values of the nine variables are:

Drinkstatus 1 = Yes and 0 = No. This is clearly binary and has been coded as such. However, under Measure, this has been entered as Nominal in SPSS.

Intentions Codes 1–4 from ‘Definitely not’ to ‘Definitely yes’ are ordered category. ‘I’m not sure’ (code 5) could be seen as in the middle between ‘Probably yes’ and ‘Probably not’, but here is treated as outside this scale and separate from ‘Don’t know/not stated’ (code 6). It has, accordingly, been entered as Nominal in SPSS. It is not clear what the intended difference is between ‘Not sure’ and ‘Don’t know’. Perhaps codes 5 and 6 are best treated as missing values. It could then be entered as Ordinal.

Initiation This is a continuous metric variable, so should be left as Scale in SPSS. It is continuous because it is a result of a calibration process in which the value could, in principle, be any fraction of a measure (like Adrian Mole aged 13¾). Under Values, None has been recorded.

Totalseen and Totalinvolve These are discrete metric variables because they are a result of counting and can only be whole numbers. They, too, should be left as Scale.

Likeads This is an ordered category variable so has been entered as Ordinal in SPSS. However, note that this has been reverse coded so that ‘I like alcohol adverts a lot’ has been given the lowest code and so on. This is the kind of thing researchers often do, so you need to watch out for inconsistencies like this.

Gender This has been treated as nominal with 1 = Male and 2 = Female.

Socialclass This is clearly ordered category and had been entered as Ordinal. However, ‘Don’t know/not stated’ would need to be treated as a missing value if this variable is to be used in an ordinal capacity, for example in the statistic gamma.

Religion The categories are 1 = Christian, 2 = Other religion, 3 = None and 4 = DK/unstated. This, clearly, is a nominal variable.

A matrix like this can be summarized variable by variable by looking down the columns so that, for example, we can see that 5 out of the 12 have had a proper alcoholic drink, 3 would ‘definitely not’ take an alcoholic drink in the next year, and so on. Note that all three also claim never to have had a proper alcoholic drink. The metric variables like Totalseen could be summarized by calculating an average score.

An alternative, and one that is developed in Chapter 7 of this text, is to summarize each case across the variables, so that case 1 has never had a proper alcoholic drink, has no intention of taking one in the next year, has an awareness score of 14, and so on. This case can now be compared with other cases for similarities and differences. Chapter 7 explains how this can be done in a systematic manner and with the ability of showing, for example, which combinations of characteristics may be sufficient to explain the outcome of intention to drink alcohol in the next year.

What, do you think, are the key ethical issues raised by the alcohol marketing study?

Ethics are moral principles or standards that guide the ways in which individuals treat their fellow human beings in situations where they might cause actual or potential harm whether economic, physical or mental. Ethics in social research are concerned with professional standards of conduct and with the use of techniques in ways that avoid harm to respondents, to clients or to other parties. The main ethical issues that arise in the conduct of social research concern privacy, confidentiality, deception, imposition, integrity and misrepresentation.

The fact that the alcohol marketing study involves interviewing young people aged between 12 and 14 raises ethical issues in particular of transparency and consent. Certainly the consent of parents was sought with a consent form for them to sign, while an information sheet attempted to make the objectives of the study transparent. What degree of ‘consent’ the young people thought they had in agreeing to be interviewed may be an issue. In the publications of the results, no individuals and no schools are identified. However, there may be an issue of selectivity in the choice of statistical techniques so that they can be used to show that alcohol advertising has an adverse impact on young people’s drinking behaviour.