Mental health and wellbeing among adolescents in Scotland: profile and trends

5 Predictors of mental health and wellbeing

Key findings

There is very little difference in the predictors of mental health and wellbeing among girls and boys.

Views on school and school-work and the number and nature of friendships are the key drivers of mental health and wellbeing among girls and boys.

• Feeling pressured by schoolwork, not liking school, frequent truanting, and exclusion from school all had a large impact on the likelihood to have a borderline or abnormal overall SDQ score.
• Number of close friends, evenings spent with friends, and the ages of friends all had an impact on overall SDQ scores.

A range of other factors are associated with mental health and wellbeing such as family affluence, parental knowledge and some leisure activities. However, these are not as closely associated with mental health as views on school and friendship patterns.

Once all other factors are included, age does not have a large impact on mental health and wellbeing.

5.1 Logistic regression is used to predict an outcome using several predictor variables. In this report, logistic regression was used to predict borderline or abnormal scores on the components of the Strengths and Difficulties questionnaire and the overall SDQ score. The data used was limited to 2013 and the models were run separately for girls and boys^[10].

5.2 The main benefit of using logistic regression in this context is to clearly distinguish the different effects of the various factors. By including in the logistic regression models, for example, the Scottish Index of Multiple Deprivation (SIMD) and perceptions of family affluence, it is possible to separate the effect of each of these. This means that any significant difference by any factor is independent of any other factors (i.e. whether area deprivation is a significant factor that is separate from family affluence).

5.3 Two sets of six logistic regression models were run separately for boys and girls: borderline or abnormal versus normal scores on the five components of SDQ and the overall SDQ score. These were run separately because of the sizeable differences in the profile of mental health and wellbeing scores for boys and girls.

5.4 The potential factors included in each of the logistic regressions were those analysed across the six themes covered earlier in this report: geography, family, friends, school, leisure activities and equalities variables. The factors included are summarised in Table 5.1.

5.5 All models were run three times: with all variables included, using forward conditional selection and backward conditional selection. Forward and backward conditional selections only retain variables in the model that have some explanatory power. Overall, almost all explanatory variables had an impact in at least one of the models. The full results of every model (with the non-significant factors included) are shown in Appendix F.

Table 5.1 - Factors included in each logistic regression

Geography	Family	Friends	School	Leisure activities	Equalities
Rurality	Family status Maternal knowledge Paternalknowledge	Ageof friends Numberof close friends Number of evening spent with friends	Attitude to school Whether pressured by school work Truanting Exclusion Expectationsafter school	Group/club membership Weekly participation in various leisure activities	Sex Age Perceivedfamily affluence Scottish Index of Multiple Deprivation (SIMD) Free SchoolMealentitlement

What the models show

Key drivers of mental health and wellbeing among girls

5.6 Figure 5.1 highlights the key drivers of borderline or abnormal mental health and wellbeing scores among girls. The detailed regression results can be found in Tables 1a and 1b in Appendix F).

Figure 5.1 Summary of key drivers of borderline or abnormal mental health among girls in 2013^[11]

5.7 Feelings towards school and schoolwork were closely linked to mental health and wellbeing among girls. Feeling pressured by schoolwork, not liking school, frequent truanting, and exclusion from school all had a large impact on likelihood to have a borderline or abnormal overall SDQ score. All of these, with the exception of exclusion from school, also had a significant impact on all five components of the SDQ score separately. Being excluded from school was linked to the conduct, hyperactivity, peer components but not to the emotion or pro-social components of SDQ.

5.8 Of all factors included in the model, being pressured by schoolwork was the strongest predictor of being borderline or abnormal on the SDQ score overall, and significant for all the components individually. It is noteworthy that not being pressured by schoolwork among girls meant that they were more likely to score as borderline or abnormal on the pro-social scale. For the other four components, the relationship is in the other direction - pressure of schoolwork leads to a higher likelihood of borderline or abnormal scores.

5.9 After feelings towards school and school work were controlled for, expectations of post-school transitions were also related to mental health and wellbeing. Those that thought that they would do an apprenticeship, go into Further Education, or start working, were more likely than those who thought they would go to university to score as borderline or abnormal on the SDQ overall. Expectations of doing an apprenticeship was also linked to borderline or abnormal scores on conduct and peer problems.

5.10 Relationships with friends were also closely linked to mental health and wellbeing among girls. Having more close friends was strongly correlated to reduced emotional problems, peer problems and borderline or abnormal SDQ score overall. Number of friends, however, was not a driver of conduct, hyper-activity or the pro-social components of SDQ among girls.

5.11 Ages of friends were linked to all components of mental health and wellbeing. Those who reported that their friends were the same age as them or that their friends were of mixed ages were more likely to have good mental health and wellbeing scores across all five components. Those who said that their friends were older than them were more likely to score as borderline or abnormal.

5.12 The relationship between evenings spent with friends and the SDQ mental health and wellbeing components was more complex. Overall, spending evenings with friends was linked to positive mental health and wellbeing. It had a positive impact on the emotion, peer, and pro-social components of SDQ. However, in terms of conduct and hyper-activity, spending evenings with friends was correlated to poorer scores.

5.13 Among girls, perceptions of family affluence were related to SDQ scores, with lower affluence being associated with a higher likelihood of borderline or abnormal scores on SDQ overall. The impact of family affluence differs across the different components of SDQ. Scores for emotion and hyperactivity were closely associated with views on family affluence, and scores for conduct and peer problems were also correlated but to a lesser extent.

5.14 The relationship between affluence and mental health and wellbeing is not straightforward. While views on family affluence were a clear driver on mental health and wellbeing, once this factor is included in the modelling, entitlement to Free School Meals and area deprivation were not driving patterns of mental health and wellbeing among girls as much as other factors.

5.15 Similarly, the relationship between age and mental health and wellbeing is not straightforward. Poor mental health and wellbeing is more prevalent among S4 than S2 girls. However, age did not come out as a significant factor in the modelling of the overall SDQ score. This suggests that age by itself is not a key driver of mental health and wellbeing. Instead, it reflects the prevalence of other drivers changing as pupils get older. Most notably, S4 girls are more likely than S2 girls to feel pressurised by schoolwork and to not like school.

5.16 A number of leisure activities were included in the logistic regression model. Generally, these were less closely associated with mental health and wellbeing among girls than their views on school and schoolwork and their relationship with friends. Hanging around the streets was linked with higher borderline or abnormal SDQ scores overall, and also closely linked with scores on the conduct component. Using social networking sites was related to an increased likelihood of borderline or abnormal scores on emotion and hyperactivity, but a decreased likelihood of such scores on the peer problems scale. Doing voluntary work was related to an increased likelihood of borderline or abnormal scores on emotion and peer problems, but a decreased likelihood of pro-social borderline or abnormal scores. Full details of these are provided in the Appendix F.

Key drivers of mental health and wellbeing among boys

5.17 Although the increase in mental health and wellbeing issues among girls over time has been much more considerable than among boys, the factors that had the strongest impact on girls also had the largest impact on boys' mental health and wellbeing, with views on school and schoolwork and friendships having the strongest impact.

Figure 5.2 Summary of key drivers of borderline or abnormal mental health and wellbeing among boys in 2013

5.18 Figure 5.2 highlights the factors that had the biggest impact on the different SDQ components of mental health and wellbeing among boys (detailed regression results can be found in Appendix F).

5.19 Feelings towards school and schoolwork were closely linked to mental health and wellbeing among boys. Like girls, feeling pressured by schoolwork, not liking school, frequent truanting, and exclusion from school all had a large impact on likelihood to be borderline or abnormal on the overall SDQ score. All of these, with the exception of truanting from school, also had a significant impact on all five components of the SDQ score separately. Truanting was linked to all components of SDQ with the exception of peer problems.

5.20 As with girls, of all factors included in the model, being pressured by schoolwork was the strongest predictor of being borderline or abnormal on the SDQ score overall among boys and significant for all the components individually. Again, like among girls, not being pressured by schoolwork among boys meant that they were more likely to score as borderline or abnormal on the pro-social scale. For the other four components, the relationship is in the other direction - pressure of schoolwork led to a higher likelihood of borderline or abnormal scores.

5.21 Among boys, expectations post-school were also related to mental health and wellbeing. Those that thought that they would do an apprenticeship, go into Further Education, or start working, were more likely than those who thought they would go to university to score as borderline or abnormal on the SDQ overall. Expectations of going into work or FE, compared to going to university were linked to borderline or abnormal scores on conduct, hyper-activity, peer problems, and the pro-social component. In contrast, expectations of undertaking an apprenticeship were related to increased borderline or abnormal scores on conduct and hyperactivity, but not in relation to peer problems or the pro-social component.

5.22 Among boys, relationships with friends were also closely linked to mental health and wellbeing. Having more close friends was strongly correlated to reduced emotional problems, peer problems and borderline or abnormal SDQ score overall. Number of friends, however, was not a driver of conduct, hyper-activity or the pro-social components of SDQ. This is the same as for girls.

5.23 Ages of friends were linked to all components of mental health and wellbeing. Those who reported that their friends were the same age as them were more likely to have good scores across all five components, and those who said that their friends were of mixed ages were more likely to have good scores across on emotion, conduct and peer problems. Those who said that their friends were older than them were more likely to score as borderline or abnormal. This is broadly the same pattern as girls.

5.24 Overall, as with girls, spending evenings with friends was linked to positive mental health and wellbeing among boys. It has a positive impact on the peer, and pro-social components of SDQ. However, in terms of conduct, spending evenings with friends was correlated to poorer scores.

5.25 Among boys, perceptions of family affluence were also related to SDQ scores, with lower affluence associated with a higher likelihood of borderline or abnormal scores on SDQ overall. Again, the impact of family affluence differs across the different components of SDQ. Scores for emotion, hyperactivity, and peer problems were correlated with views on family affluence. Among boys, scores for pro-social were inversely correlated to views on family affluence - those who say their family was well-off were more likely to score borderline or abnormal on the pro-social component.

5.26 Again, once views on family affluence were controlled for, entitlement to Free School Meals and area deprivation were not strong drivers of mental health and wellbeing among boys.

5.27 As with girls, leisure activities were included in the logistic regression model. While hanging around the streets was linked with higher borderline or abnormal SDQ score overall, doing voluntary work was not. Doing voluntary work was, however, related to conduct and peer problems and was linked to decreased likelihood of pro-social borderline or abnormal scores. This is the same pattern as was seen among girls.

5.28 However, unlike girls, using social networking sites was not linked to higher borderline or abnormal SDQ score overall among boys. Full details of these are provided in the logistic regression given in Appendix F.

5.29 Overall, the analysis suggests that while the prevalence of mental health and wellbeing problems has diverged between girls and boys in recent years, the main drivers have not. Pressure of schoolwork and the nature and number of friendships are central to mental health and wellbeing.