We are testing a new beta website for gov.scot go to new site

Scottish Environmental Attitudes and Behaviours Survey 2008 - Technical Report

Listen

6 DATA ANALYSIS APPROACH

6.1 The data analysis in the study was driven by two main aims. First, to describe the prevalence of various attitudes and behaviours in relation to the environment across Scotland and within particular sub-groups. This chapter details the reporting conventions used in the main report for this.

6.2 Secondly, to analyse what drives environmental attitudes and behaviours. This was undertaken primarily by creating a typology of environmental engagement and analysing whether this typology was significant in determining different behaviours. This chapter details our approach to the creation of the typology and how it was used in the modelling of behaviour.

Reporting conventions in main report

6.3 Where percentages do not sum to 100% this may be due to computer rounding, multiple answers, or the exclusion of don't know or not applicable responses.

6.4 In some instances, reported figures combine two or more response categories, for example, combining "every day", "two or three times a week", and "once a week" to present the figures for "at least once a week". The figure given for the combined category has been rounded after the addition and therefore may not exactly sum to the figures given for the individual response categories. For example, 18.3% of respondents said that the environment was the most important issue facing the world, and a further 16.4% said that it was an important issue. While these figures are reported separately in the main report as 18% and 16%, the combined total of any mention of the environment as an important issue facing the world is reported in the text as 35% (18.3% plus 16.4% rounded to the nearest percentage point).

6.5 As noted previously, while no established theoretical framework exists to generate confidence intervals for quota sample estimates, experimental studies and comparisons have shown the robustness of such estimates. Therefore, throughout the report, differences between sub-groups are commented upon only where these are statistically significant. Significance levels were calculated at the p<0.05 level - i.e. where we can be 95% confident that such a difference has not occurred by chance. Design effects were not included in these calculations.

Methods for examining whether environmental attitudes affect behaviour

Creating the environmental engagement typology

6.6 A typology of environment engagement was created to provide a clearer analysis of the types of people who may be more open to messages aimed at changing behaviour, and to facilitate understanding of the link between attitudes and behaviours.

6.7 The segmentation approach used to develop the typology, aimed to divide the public into different groups based on their attitudes towards the environment. The groupings were determined using responses to four questions on attitudes to climate change and the environment, namely:

  • the importance of the environment versus other issues in Scotland, and globally;
  • views on the immediacy of the threat of climate change; and
  • reported levels of knowledge about climate change.

6.8 The typology segmented people into a hierarchy of five groupings:

  • Deep Greens (14%): These are people who: said unprompted that the environment was an important issue in Scotland or the most important issue in the world; believed that climate change is an immediate and urgent problem; and said they know a great deal or a fair amount about climate change. These people are the most engaged with the issues and are likely to be the most proactive in terms of adopting new or alternative behaviours.
  • Light Greens (14%): People who believe that climate change is an immediate and urgent problem and an important issue globally, but who do not necessary feel well informed about climate change or think that it is an important issue in Scotland. This group could be referred to as "aspiring greens". They may be interested in adopting new behaviours but tend to be more passive than those who are highly engaged.
  • Shallow Greens (30%): These are people who said that climate change is an immediate and urgent problem, but not one of the most important issues globally or in Scotland. These people accept that climate change is an issue, but may not be convinced of the need to take more than minimal action at present.
  • Distanced (30%): This group believe that climate change is more of a problem for the future or hold no views on climate change. It is unlikely that this group will readily accept the need for anything more than minor or relatively easy changes to their lifestyle.
  • Disengaged (14%): These are people who are not convinced that climate change is happening or, that if it is happening, believe it is not a problem. This group are likely to be the most resistant to messages about changing their behaviour.

6.9 While the rationale behind the typology approach is broadly similar to that undertaken by DEFRA15, the approach to the creation of the typology groupings differed from the DEFRA study for both practical and, to a lesser extent, theoretical reasons. The DEFRA segmentation approach was based on analysis of a batch of attitudinal questions. While some of these questions were also asked in SEABS'08, some were not. Of the 17 questions that were used in DEFRA's segmentation, 5 were included unchanged in SEABS'08, while a further one was included asking about Scotland rather than Britain. This precluded the possibility of replicating exactly the segmentation approach followed by DEFRA. Additionally, the attitudinal questions common to both surveys and that had been used in the DEFRA segmentation approach were asked in the CASI section of the SEABS'08 interview. As around 10% of respondents declined to answer this section of the questionnaire, a typology based on these questions would only provide data for a proportion of respondents, and therefore reduce the potential sample size for analysis.

6.10 The SEABS'08 typology was constructed to comprise attitudinal questions only, with the intention that it could then be used to explore whether groups with different attitudes tend to display differing patterns of behaviour. The DEFRA segmentation, in contrast, is less clearly focused on attitudes. For example, it incorporates the statement, "I would only travel by bus if I had no other choice". Arguably this - and some of the other statements in the DEFRA model - contain both an attitudinal and a behavioural element. As such, it does not offer a typology that is purely reflecting attitudes, the rationale behind the typology used for the present study.

6.11 An alternative approach to developing the typology would have been to use the attitudinal statements in the CASI section of the questionnaire. Factor analysis of the responses to these suggested that they cluster around different topics, such as transport, recycling and waste. Eight attitudinal statements, detailed in Figure 4.5 of the main report, and that correlated most closely to a single factor, could be taken to purely represent attitudes on environmental engagement. These were:

  • Tackling climate change shouldn't come at the expense of the Scottish economy.
  • The environment is a low priority for me compared with a lot of other things in my life.
  • I don't believe my behaviour and everyday lifestyle contribute to climate change.
  • The so-called environmental crisis facing humanity has been greatly exaggerated.
  • It's not worth Scotland trying to combat climate change because other countries will just cancel out what we do.
  • It's not worth me doing things to help the environment if others don't do the same.
  • The effects of climate change are too far in the future to really worry me.
  • Climate change will only have an impact on other countries, there is no need for me to worry.

6.12 The typology was compared against the combined responses to these eight statements to assess its robustness. Answers to each statement were given a score from 1 for agree strongly (a negative response in terms of environmental attitudes) to 5 for disagree strongly (a positive response). These scores were then added to create an overall standardised score. There was a high correlation between these standardised scores and the typology (see Table 6.1).

Table 6.1: Environmental engagement typology by banded standardised score from eight attitudinal statements.

Standardised score from attitudinal statements

Deep Greens

Light Greens

Shallow Greens

Distant Greens

Disengaged

Total

%

%

%

%

%

%

0-15

0

1

1

7

12

4

16-20

3

5

9

29

37

17

21-25

12

21

26

34

34

27

26-30

39

45

43

24

11

33

31-35

47

28

21

5

6

19

100

100

100

100

100

100

6.13 It is worth emphasising that the boundaries between typology groups, both in the DEFRA'07 survey and in the SEABS'08 typologies, are not totally distinctive. Attitudes, and therefore data on attitudes, tend to be less exact and bounded than behaviours and behavioural data. While it is unlikely that a respondent would give a different response to the number of children in the household if asked at two different occasions, it is highly possible that responses to attitudinal questions may vary day to day.

6.14 As noted previously, one of the objectives of SEABS'08 was to allow detection of trends over time. The use of the four questions to construct the typology has an additional potential advantage over using the attitudinal statements with regard to this objective, in that it would be easier and most cost efficient to replicate in other surveys where it may be of benefit to analyse the data by the typology of environmental engagement.

Use of regression analysis to examine significance of typology to behaviour

6.15 Logistic regression was used to model the likelihood of undertaking various behaviours - for example, recycling bottles - and to examine, in particular, whether the typology groups differed in their behaviour after the other socio-economic factors had been controlled for 16. As environmental engagement is closely linked to educational attainment and social class, by including in the regression models these variables, it is possible to separate the effect of each. This helps overcome the risk that the effect of one variable (social class for example) is confused with the effect of another (environmental engagement for example). In other words, the use of regression modelling helps examine whether positive environmental attitudes have a measurable effect on behaviour.

6.16 The following variables were routinely included in each of the regression models: age, gender, social class, economic status, educational attainment, presence of children in the household, urban/rural indicator, tenure, and property type.

6.17 Table 6.2 shows example output from the logistic regression model of whether people ever 17 use kerbside bottle recycling facilities when present. The first column indicates the different predictor factors included in the model. These include 'binary' variables such as sex (either/or variables), continuous variables such as income (variables that are measured numerically), and categorical factors such as tenure (variables including a number of different categories). Some continuous variables, age for example, have been grouped into bands and treated as categorical variables. This is to ensure that any non-linear relationships in these variables are reflected in the regression models.

6.18 A value of less than 0.05 in the fourth column suggests that this factor is significant. So, as the figure for flat (vs. house) is less than 0.05, it follows that, after controlling for the effect of all other factors in the model, the likelihood of those living in flats using kerbside bottle recycling facilities is different from those living in houses.

6.19 The second column, headed 'Beta' indicates the direction of the effect. A negative value indicates that those in the first category, for example those living in flat, are less likely to recycle bottles using kerbside facilities than those living in houses.

Table 6.2: Example output from the logistic regression model: Whether use kerbside bottle recycling facilities where present - ever versus never. (N = 1939)

Weighted logistic regression

Allowing for survey design18

Beta

S.E. of B

Sig.19

S.E. of B

Sig.

Design factor

Educational qualifications (vrs none)

0.40

0.43

Degree

-0.33

0.19

0.09

0.20

0.10

1.04

HNC/ HND

0.05

0.24

0.83

0.24

0.83

1.00

Higher

-0.06

0.21

0.79

0.23

0.81

1.08

O', Standard

0.03

0.16

0.86

0.17

0.86

1.06

Economic status (against working)

0.61

0.60

Retired

-0.17

0.23

0.47

0.22

0.45

0.94

Inactive

0.15

0.17

0.38

0.18

0.41

1.08

Studying or training

0.05

0.28

0.86

0.32

0.88

1.08

Flat (against House)

-0.43

0.13

0.00

0.17

0.01

1.34

Age (against 16-24)

0.39

0.41

25-34

-0.18

0.22

0.41

0.23

0.44

1.06

35-54

-0.13

0.21

0.55

0.23

0.58

1.09

55+

0.26

0.28

0.34

0.28

0.35

0.99

Vegetarian (against non vegetarian)

-0.31

0.25

0.21

0.26

0.22

1.04

White (vrs non-white)

-0.06

0.37

0.87

0.38

0.87

1.04

Male (vrs female)

-0.08

0.11

0.47

0.10

0.41

0.90

Rural urban indicator (against remote rural)

0.00

0.00

Large Urban

0.32

0.31

0.31

0.44

0.47

1.36

Other Urban

0.96

0.32

0.00

0.46

0.04

1.41

Accessible Small Towns

1.28

0.37

0.00

0.51

0.01

1.34

Accessible Rural

1.15

0.36

0.00

0.49

0.02

1.33

Remote Small towns

1.12

0.43

0.01

0.60

0.06

1.35

Car available (vrs not)

0.33

0.14

0.02

0.15

0.02

1.06

Tenure (against Owner-occupier)

0.36

0.47

Social Rented

-0.17

0.15

0.26

0.16

0.29

1.08

Private rented

-0.36

0.22

0.10

0.25

0.14

1.08

Other

-0.24

0.32

0.45

0.32

0.44

1.03

Typology (vrs Deep Greens)

0.62

0.59

Light Green

-0.08

0.22

0.71

0.23

0.72

1.05

Shallow Green

0.07

0.19

0.71

0.20

0.72

1.04

Distant Green

-0.01

0.19

0.95

0.20

0.96

1.07

Disengaged

-0.22

0.22

0.32

0.21

0.30

0.96

Children in household (vrs not)

-0.09

0.12

0.43

0.12

0.43

1.05

Income

0.01

0.01

0.37

0.01

0.18

1.03

Social Group (against Es)

0.01

0.02

A

1.30

0.50

0.01

0.51

0.01

1.00

B

0.95

0.26

0.00

0.27

0.00

1.03

C1

0.62

0.21

0.00

0.23

0.01

1.08

C2

0.55

0.21

0.01

0.23

0.01

1.09

D

0.42

0.21

0.04

0.22

0.05

1.08

6.20 With categorical factors, such as social group, logistic regression models compare different categories against a reference category. In the model presented above, Social Group E has been set as the reference category. Overall, all other Social Groups are more likely than those in Social Group E to recycle bottles using kerbside facilities.

6.21 Table 6.2 also suggests that the typology is not significant in determining whether people use kerbside bottle recycling facilities once the other factors have been controlled for 20. So, while there may be significant differences in the proportion of, for example, Deep Greens and the Disengaged who use kerbside bottle recycling facilities, it is not environmental attitudes but rather other factors (social class, whether household has a car, dwelling type, and social group) that is driving behaviour.

6.22 In contrast, the typology of environmental attitudes is a significant factor in determining use of non-kerbside bottle recycling facilities (see Table 6.3), whereas property type and Social Group are not. Therefore, as noted in the main report, it is possible to conclude that while environmental engagement influences use of non-kerbside facilities, it does not (measurably) drive use of kerbside facilities.

Table 6.3: Example output from the logistic regression model: Whether use non-kerbside bottle recycling facilities when kerbside facilities are not present - ever versus never (N=840).

Weighted logistic regression

Allowing for survey design21

Beta

S.E. of B

Sig.22

S.E. of B

Sig.

Design factor

Educational qualifications (vrs none)

0.01

0.01

Degree

0.95

0.35

0.01

0.38

0.01

1.04

HNC/ HND

-0.07

0.39

0.86

0.45

0.88

1.12

Higher

0.30

0.34

0.38

0.36

0.40

0.97

O', Standard

-0.24

0.25

0.34

0.25

0.34

0.99

Economic status (against working)

0.73

0.83

Retired

-0.12

0.38

0.75

0.37

0.74

1.00

Inactive

0.01

0.28

0.98

0.28

0.98

0.99

Studying or training

0.51

0.48

0.28

0.57

0.37

1.00

Flat (against House)

0.05

0.23

0.82

0.28

0.85

1.22

Age (against 16-24)

0.56

0.60

25-34

0.36

0.34

0.30

0.35

0.31

1.03

35-54

0.44

0.31

0.15

0.31

0.16

0.95

55+

0.01

0.43

0.98

0.43

0.98

0.96

Vegetarian (against non vegetarian)

0.55

0.35

0.12

0.31

0.08

0.92

White (vrs non-white)

-0.10

0.56

0.86

0.73

0.89

1.08

Male (vrs female)

-0.04

0.19

0.84

0.19

0.85

1.06

Rural urban indicator (against remote rural)

0.00

0.00

Large Urban

-1.67

0.33

0.00

0.38

0.00

1.12

Other Urban

0.14

0.34

0.69

0.36

0.71

1.05

Accessible Small Towns

-0.59

0.41

0.15

0.37

0.11

0.98

Accessible Rural

0.13

0.41

0.74

0.45

0.77

1.09

Remote Small towns

-0.20

0.49

0.68

0.36

0.57

0.78

Car available (vrs not)

0.41

0.22

0.06

0.21

0.05

0.98

Tenure (against Owner-occupier)

0.44

0.55

Social Rented

-0.09

0.25

0.72

0.27

0.74

1.10

Private rented

0.28

0.34

0.41

0.44

0.52

1.19

Other

0.76

0.61

0.22

0.62

0.23

0.99

Typology (vrs Deep Greens)

0.01

0.01

Light Green

-0.76

0.39

0.05

0.38

0.04

0.92

Shallow Green

-0.97

0.35

0.01

0.34

0.00

0.92

Distant Green

-1.31

0.36

0.00

0.37

0.00

0.98

Disengaged

-0.94

0.40

0.02

0.36

0.01

0.87

Children in household (vrs not)

-0.44

0.22

0.05

0.24

0.07

1.07

Income

0.02

0.02

0.33

0.02

0.33

1.02

Social Group (against Es)

0.35

0.24

A

0.09

0.81

0.91

0.90

0.92

1.14

B

0.37

0.41

0.37

0.41

0.37

1.02

C1

0.68

0.35

0.05

0.31

0.03

0.93

C2

0.23

0.34

0.49

0.34

0.49

1.04

D

0.48

0.33

0.15

0.32

0.13

0.97