1. Including that reported in previous years Scottish Social Attitudes Surveys
2. The precise figure given for response rates depends on whether dwelling units whose eligibility to participate was unknown are included or excluded from the calculation. Dwelling units are coded as 'unknown eligibility' where the interviewer is unable to establish whether the property is occupied and residential. The higher response rate excludes dwelling units of unknown eligibility from the calculation, while the lower rate includes them. As some of the dwelling units whose eligibility was unknown are likely to be eligible and some ineligible, the true response rate probably lies somewhere between the two figures. For further details on response rate calculations, see the technical report.
3. British Social Attitudes survey, cited in Bromley et al, 2001.
4. In 2006, this figure stood at 19% (British Social Attitudes survey).
5. Prior to 2004, this question asked about the Scottish Parliament. However, in response to criticisms that the two questions were not comparing alike institutions, a split sample experiment was undertaken in which half the sample were asked about the Scottish Parliament and half about the Scottish Executive. The difference in wording made a negligible difference to the results (see Bromley and Given, 2005, for full results), so from 2005 onwards the question was changed to ask about the Scottish Executive.
6. The question about trust to act in Scotland's best interests used a four-point 'temporal' scale ranging from 'just about always' to 'almost never', while the 'fair decisions' question used a five-point scale based on 'amount' of trust, with options ranging from 'a great deal' to 'not at all'.
7. The main multivariate analysis technique used to inform this report was logistic regression. Logistic regression is a statistical technique used to summarise the relationship between a 'dependent' variable (in this case, trust in the Scottish Executive) and one or more 'independent 'explanatory' variables (for example, age, sex, education etc). It is particularly useful when explanatory variables are related to each other (for example, education and area deprivation), since it takes the relationships between these in to account when determining which are statistically significant. Full results of this analysis are reported in a separate technical report (Cleghorn, Ormston & Sharp, 2007).
8. Cell percentages imply that the percentages shown in each cell 'stand alone'. It is not meaningful to total such percentages across a row or down a column. For example, the figure in the first cell on the top row represents the per cent of those with no qualifications who trust the Scottish Executive to act in Scotland's best interests 'just about always' or 'most of the time'.
9. NB the question on party political identification is not funded as part of the Office of Chief Researcher core module, but is part of the background classification data collected for SSA.
10. NB although bivariate (cross-tabulation) analysis suggests that Liberal Democrat identifiers are more likely than those who support the SNP to trust the Scottish Executive, multivariate analysis suggests this may not be significant after other factors are controlled for. Labour supporters, on the other hand, are significantly more likely than SNP identifiers to express high levels of trust, even after controlling for other factors. See Cleghorn, Ormston & Sharp, 2007, for details.
11. Multivariate analysis suggests that differences in the levels of trust in the Scottish Executive to act in Scotland's best interests expressed by those with the highest and lowest levels of interest in politics are not significant after other factors are controlled for. See Cleghorn, Ormston & Sharp, 2007, for details.
12. Where the relationship between different explanatory factors is controlled for. See footnote 7 and the SSA Core module technical report (Cleghorn, Ormston & Sharp, 2007).
13. Again, see footnote 7 and the SSA Core module technical report (Cleghorn, Ormston & Sharp, 2007).
14. Although a similar pattern can be seen with respect to trust to make fair decisions, the difference is smaller and was not significant in multivariate analysis after other factors were taken into account.
15. Again, although a similar pattern can be seen with respect to trust to make fair decisions, urbanity/rurality was not significant after other factors were controlled for in multivariate analysis.
16. NB sex was significantly associated with trust to make fair decisions in the initial multivariate model, which included demographic factors (like sex, age, occupational sector and education) only. However, it was not significant once other factors, like party identification and constitutional preference, were also taken into account. This suggests that differences in trust between men and women may in part reflect gender differences in political attitudes more generally.
17. As measured by a question which asks how much respondents have seen or heard about the work of the Scottish Executive over the last 12 months.
18. The quiz is presented as a series of statements, covering both responsibilities (who makes decisions about health spending and levels of unemployment benefits in Scotland) and the mechanics of government in Scotland (how many MSPs there are and whether or not the Scottish Executive is just another name for the Scottish Parliament). Respondents are asked to say whether they think each statement is true or false.
19. With the exception of 2002.
20. Where half the sample were asked about the Scottish Executive and half about the Scottish Parliament.
21. The question wording also changed from asking people to think prospectively about the influence the parliament would have in 1999 ('When the new Parliament starts work, which of the following do you think will have most influence over the way Scotland is run?'), to its current wording from 2000 onwards ('Which of the following do you think has most influence over the way Scotland is run?').
22. Questions about trust and influence both now ask about the Scottish Executive rather than the Scottish Parliament. However, prior to 2004, this question asked about the Scottish Parliament. In response to criticisms that the two questions were not comparing alike institutions, a split sample experiment was undertaken in which half the sample were asked about the Scottish Parliament and half about the Scottish Executive. The experiment concluded that in practice it makes little difference which institution is asked about - people's response were broadly the same.
23. Again, see footnote 7 and the SSA Core module technical report (Cleghorn, Ormston & Sharp, 2007). Unless otherwise stated, only those variables found to be significant in this multivariate analysis are discussed in the text.
24. For further details, see http://www.scottishexecutive.gov.uk/Topics/Statistics/SIMD/Overview. The areas SSA respondents lived in were matched to SIMD quintiles, with 1 being the least deprived and 5 being the most deprived.
25. ' SE' is used as an abbreviation for 'Scottish Executive' in this and the following tables.