This chapter begins by outlining the conceptual motivation behind multilevel analyses and by identifying a core set of research questions that this approach addresses. It then introduces the idea of multilevel structures and discusses simple and complex multilevel models. It emphasizes that the key strength of multilevel models lies in modeling heterogeneity at different levels and shows how multilevel models can be extended to additional contextual levels (e.g., neighborhoods nested within regions). The estimation procedures underlying such models are discussed, showing how a multilevel framework can provide a general, unified approach to data analysis and how this can be achieved by extensions to the basic hierarchical structure of individuals nested within contexts. The chapter concludes with a discussion of issues that researchers should be aware of when applying multilevel methods.

Background. Well-being is an important determinant of health and social outcomes. Measures of positive mental health states are needed for population-based research. The 12-item General Health Questionnaire (GHQ-12) has been widely used in many settings and languages, and includes positively and negatively worded items. Our aim was to test the hypothesis that the GHQ-12 assesses both positive and negative mental health and that these domains are independent of one another.Method. Exploratory (EFA) and confirmatory (CFA) factor analyses were conducted using data from the British Household Panel Survey (BHPS) and the Health Survey for England (HSE). Regression models were used to assess whether associations with individual and household characteristics varied across positive and negative mental health dimensions. We also explored higher-level variance in these measures, between electoral wards.Results. We found a consistent, replicable factor structure in both datasets. EFA results indicated a two-factor solution, and CFA demonstrated that this was superior to a one-factor model. These factors correspond to ‘symptoms of mental disorder’ and ‘positive mental health’. Further analyses demonstrated independence of these factors in associations with age, gender, employment status, poor housing and household composition. Statistically significant ward-level variance was found for symptoms of mental disorder but not positive mental health.Conclusions. The GHQ-12 measures both positive and negative aspects of mental health, and although correlated, these dimensions have some independence. The GHQ-12 could be used to measure positive mental health in population-based research.

No abstract

It is still not known whether the places that people live affect their mental health. The principal aim of this 1991 study was to quantify simultaneously variance in the prevalence of the most common mental disorders, anxiety and depression, in Britain at the individual, household, and electoral ward levels. Data from a cross-sectional, nationally representative survey of 8,979 adults aged 16-74 years living in private households nested within 642 electoral wards in England, Wales, and Scotland were analyzed by using multilevel logistic and linear regression. Common mental disorders were assessed by using the General Health Questionnaire. Less than 1% of the total variance in General Health Questionnaire scores occurred at the ward level. This variance was further reduced and was no longer statistically significant after adjustment for characteristics of persons. By contrast, the proportion of total variance at the household level (14.4%, 95% confidence interval: 11.4, 17.5 in the null model) (p < 0.001) was statistically significant and remained so after adjustment for individual- and household-level exposures. While these findings suggest that future interventions should target persons and households rather than places, further research is first required to establish whether other (particularly smaller) areas lead to similar results.

Recent debates have suggested that increasing social diversity within Western economies is associated with adverse social consequences such as loss of community and decline of civic society, including an erosion of collective efficacy (ie shared expectations of and mutual engagement by residents in social control). In the UK and US, these debates have been given impetus by concerns about the effects of growing ethnic heterogeneity on community life. Here there is an assumption that heterogeneity undermines social cohesion and makes the established population less willing to share resources, trust fellow citizens, so that it eventually`hunkers down' and withdraws from collective life. To date there are few studies that have examined this in detail across England at the small-area level. The research presented here explores this terrain by exploiting information from the British Crime Survey on two recognised dimensions of collective efficacy: namely, social cohesion and trust, and informal social control. Multivariate, multilevel models were used to determine the importance of individual and area characteristics in the possible explanation of these outcomes, and particular attention was paid to the relative importance of neighbourhood disadvantage over and above neighbourhood diversity. Results suggest that both diversity and disadvantage are statistically associated with reduced levels of social cohesion and trust, and informal social control, but greater substantive importance is attached to neighbourhood disadvantage.