Scottish Government

ANNEX I: ADDITIONAL MULTIVARIATE ANALYSES UNDERTAKEN

1) Analysis of the relationship between the number of positive and negative influences on mental wellbeing mentioned by respondents, and their mental health.

I.1 Previous research has shown that the absence of positive influences on mental wellbeing, rather than the presence of negative influences, is crucial in predicting suicidal behaviour ²⁵. Accordingly, analysis was undertaken to explore whether there was a relationship between the number of positive and negative influences mentioned by respondents, and their mental health as measured by the GHQ12 and WEMWBS.

I.2 For the purpose of the analysis two new continuous variables were created: one representing the total positive number of influences mentioned by each respondent ( QB1), and the other representing the total number of negative influences mentioned ( QB2). Zero-order correlational analysis was then undertaken to explore i) the relationship between the number of positive influences that respondents mention and their mental health (as measured by both the GHQ12 and WEMWBS), and ii) the relationship between the number of negative influences that respondents mention and their mental health. In addition, regression analysis was used to explore simultaneously the relationship between the number of positive and negative mentions and wellbeing - thus providing a measure of the relative importance of the reporting of positive and negative influences. The analysis was repeated for those who have experience of mental health problems (both proxy and personal) and those who do not.

I.3 On the whole, the analysis revealed that neither the number nor the type of positive and negative influences mentioned by respondents were found to be correlated with their mental health rating as measured by the GHQ12 and WEMWBS.

2) Segmentation analysis of attitudes to mental health

I.4 Segmentation analysis was undertaken on the attitudinal data to explore further the links between attitudes and socio-demographic and behavioural variables. The purpose of this analysis was to identify whether the nine attitudinal statements could be grouped into a smaller number of themes or 'factors' depending on the responses they elicit and, if so, to what extent respondents fell into distinct, like minded groups based on these factors.

I.5 The analysis involved a two stage process. First, factor analysis was used to identify related statements. Factor analysis is a statistical data reduction tool that takes a set of data and simplifies it into a smaller number of underlying themes or 'factors' by grouping together questions that are answered in similar ways. The analysis produced a four factor solution as follows:

Table I.1: Factor analysis 4 factor solution

I.6 Cluster analysis was used to identify to what extent respondents fell into like-minded groups on the basis of their attitudes in relation to each factor. A range of possible cluster solutions were identified, with the strongest of these comprising four clusters. None of the 4 clusters were particularly distinct, either in attitudinal or socio-demographic terms. However, the analysis did yield some interesting findings, several of which reinforced bivariate analyses presented in Chapter 7. Most notably it showed that:

• attitudes towards mental-ill health are characterised by a high levels of ambivalence with individuals tending to hold a mixture of positive and negative views.
• Men living in the most deprived areas of the country tend to hold the most negative attitudes overall
• People with a long term illness disability or infirmity are among those least likely to have concerns about disclosing a mental health problem to others - this may be because they are used to living with illness and to dealing with the attitudes of others towards their illness
• The most highly educated groups of respondents, tend to hold the most liberal attitudes towards people with a mental health problem but at the same time are the group most likely to agree that 'If I were suffering from a mental health problem, I wouldn't want people knowing about it'.

3) Relationship between ability to correctly diagnose the symptoms of specific mental health problems, and attitudes to mental health problems.

I.7 An analysis was undertaken to explore whether respondents' ability to correctly diagnose the symptoms described in the mental health scenarios varied in line with their attitudes to mental ill health, including their willingness to interact with people with mental health problems. For the purposes of this analysis, a new binary variable was created which reflected whether or not respondents correctly identified the mental health problem depicted in the scenario with which they were presented. This variable was then cross tabulated against responses to each of the attitudinal statements in the survey ( QC1) and with the 'willingness to interact' statements from the scenarios section ( QE5). In the event, there were no significant differences in the attitudes of those who were able to diagnose the symptoms correctly and those who were not.