The individual, place, and wellbeing – a network analysis

Despite continuing controversy over issues of definition and measurement as well as terminology, it is clear that wellbeing has risen up the list of public health concerns [1, 2]. Most wellbeing research focusses on personal, subjective assessments of one’s feelings (i.e. eudaimonic and hedonic wellbeing) or assessments of variables relative to day-to-day living (e.g. local economy and safety). However researchers are increasingly broadening the concept of wellbeing beyond the level of the individual to encompass aspects of the community (e.g. shared values, belonging and ownership of community processes) [3]. Personal-subjective wellbeing, along with community wellbeing [3], are increasingly acknowledged as important population outcomes with advances made in their representation within international policy agendas [4‐6]. In part this is because, as with health, there exist not only individual level determinants or correlates of wellbeing but also significant wider economic, environmental and social determinants. These wider determinants reflect systemic inequities in the living circumstances of individuals and communities that are unjust and avoidable and can often arise as unwanted consequences of well-meaning but short - sighted or poorly targeted policy interventions [7, 8]. As a result, there have been heightened calls for wellbeing in all policy approaches, as enacted in the Wellbeing of Future Generation Act 2015 in Wales [6]. Research conducted under the auspices of the Community Wellbeing Evidence Programme of the What Works Centre for Wellbeing has been part of a broader effort that has begun to identify the characteristics of UK places that may contribute to poor wellbeing or languishing or that may be related to local disparities in personal wellbeing (i.e. wellbeing inequalities) [9, 10].

Although some authors argue that individual characteristics (e.g. personal health, financial status) are more robustly associated with subjective wellbeing than community or place characteristics [11], others suggest that individual, community and place characteristics should not be considered independently from one another. This is because they likely operate in a complex system of reciprocal relationships [12] as can be shown in research using mobile data collection methods for example [13]. It is incontrovertible that one’s relative flourishing or languishing will depend, at least in part, on what resources we have available to us in our neighbourhoods and communities. The place-based resources that support high levels of wellbeing are numerous but include high quality employment or other forms of occupation, enjoyable and immersive cultural and activity-based pursuits, access to good quality food and other forms of retail, reliable social support from those around us as well as wider forms of social capital, sense of belonging to and meaningful involvement with your neighbourhood and community, access to good quality open/green spaces, housing and neighbourhood living environment [14‐23].

Subjective and community wellbeing, individual characteristics, and the living environment: a complex network approach.

Further research into the complex associations between subjective and community wellbeing, place, and individual factors is warranted. The network approach [24, 25] is a promising conceptual and statistical framework for such inquiries. This approach conceptualises psychological phenomena as complex systems, wherein aetiological factors and psychological indicators (e.g. moods, behaviours) influence one another directly in a cycle of mutual reinforcement [26]. Network studies have seen notable growth in the field of mental health research in recent years. Such models are presented graphically; variables take the form of nodes (points in space), with lines linking nodes (referred to as edges) denoting the presence, strength and direction of associations between variables. In most psychological networks, edges reflect conditional dependencies; i.e. the association between variables controlling for all other variables in the network [27]. The overall connectivity of each node can be quantified using a series of metrics known as centrality, which allow us to identify the nodes that are most important to the network as a whole. Furthermore, by focusing on direct and indirect associations, network models offer detailed and nuanced insights into the associations that connect different domains of variables. Thus, network models may help clarify the complex relationships and pathways that connect individual, place-based and community characteristics with elements of hedonic and eudaimonic wellbeing. This in turn may broaden our understanding of wellbeing beyond simple eudemonic/hedonic experiences, and help establish the idea that wellbeing is a complex system consisting of individual and community elements.

To our knowledge, only one study so far has used network analysis to investigate subjective wellbeing [28]. This study focussed entirely on the associations between different aspects of hedonic and eudaimonic wellbeing, which were assessed using individual items from the Warwick-Edinburgh Mental Well-being Scale (WEMWBS). They found that items related to positive self-perception and mood were most central across four large general population samples [28]. To date, no studies have used psychological network analysis to investigate associations between subjective wellbeing and its wider determinants or broader indicators of wellbeing (e.g. community wellbeing). One study [29] has used psychological network analysis to explore how relative disadvantage and the neighbourhood environment are related to self-reported mental distress collected using continuum measures of depression, anxiety and feelings of persecution. This study, confirmed that connections across the different variable domains of neighbourhood environment and mental distress altered with level of neighbourhood deprivation, illustrating and demonstrating the important negative psychological consequences of living in more disadvantaged neighbourhoods.

In the current exploratory study, we used network analysis to examine the associations between psychological wellbeing and both individual and place characteristics. As network analysis is a highly flexible approach, allowing for investigations at different levels of granularity [30], we examined two distinct network structures: i) a network in which psychological wellbeing was treated as a uni-dimensional construct (i.e. the composite score on the short WEMWBS [31]), and ii) a network in which wellbeing was broken down into its constituent parts (i.e. the individual items from the short WEMWBS). The overall aim of this research was to identify the individual and living environment characteristics that are most strongly associated with overall subjective wellbeing, as well as those associated with the individual wellbeing items that comprise the short WEMWBS.

Descriptive statistics for all variables included in our network analyses are presented in the online supplementary materials (Table S1).

This study set out to explore the connections between and amongst individual and neighbourhood, place-based factors that, together, go some way to represent the intricate relationships within a network of variables that contribute to subjective wellbeing as measured by the SWEMWBS. We found that this network of individual, neighbourhood and wellbeing variables was densely connected, highlighting the complex relationships that exist between individual and place characteristics in the context of subjective wellbeing.

Our findings agree with the previous literature such as Propper et al. [11] that argues that individual characteristics are the strongest predictors of subjective wellbeing. In this analysis, subjective financial difficulty and physical health had the strongest connections with overall wellbeing represented in the composite Short WEMWBS score. However, notwithstanding the importance of perceived financial difficulty and overall health, we found that wellbeing is robustly associated with place characteristics, even after controlling for established individual-level correlates. This finding alone supports the conclusion that individual factors and neighbourhood/place-based factors need to be considered together in a relational way if we are to properly understand subjective wellbeing as it is spatially distributed across nations [12] and so that we can effectively intervene to improve wellbeing at national policy levels [7].

The results reported here highlight that areas characterised by lack of accessible open space, civic disengagement, a lack of neighbourhood cohesion, and housing disrepair seem to be at particular risk of low wellbeing. As such our findings are consistent with the reviews led by Marmot [45, 46] in stressing that place matters for our health and wellbeing and that people’s wellbeing is significantly affected by systemic place-based conditions over which they have very little, if any, control. These findings are also consistent with the prior research base as reviewed and reported by the community wellbeing evidence programme of the What Works Centre for Wellbeing [17, 18, 23] and supported by a recent secondary data analysis conducted by Curtis et al. [10] as part of that programme.

Our analysis highlighted that using local open or greenspace was more strongly associated with wellbeing than any other place-based/neighbourhood factor. In the context of the recent restrictions imposed in relation to control of COVID-19, the importance of open space usage is all the more pointed. According to the ONS [47], while 1 in 8 British households has no garden, 28% of people live within 5 min walk of a park. Furthermore, it seems from the ONS analysis that people living in the most deprived communities of England are twice as likely to live within 5 min of a local park or open space than those living in less deprived areas. In the North West of England, where the data analysed in the current research was collected, 30.8% of the population is reported to live within 5 min walk of a local park, rising to 55.5% when playing fields are added in [47]. This recently published information suggests that it is not availability of local open space assets per se that is the issue in relation to wellbeing but rather it is the use of those assets that is the important determinant. Therefore, there needs to be further research exploring the barriers to greenspace usage that should include a consideration of the quality of the walking journey to and from local open spaces in disadvantaged neighbourhoods.

Nevertheless, in highlighting the importance of using open spaces, our finding agrees with the wealth of published multi-disciplinary research, supported via a widely accepted theory [19], that emphasises access to and use of greenspace as being part of the public health solution to address physical health challenges, mild to moderate mental health difficulties and low wellbeing [20‐22]. Previously, it has been difficult to judge whether the use of local open space held more promise in this regard than other available intervention options [7]. This network analysis, provides some information in this regard by finding that of the neighbourhood factors measured in this analysis, use of local open space was the one most strongly associated with subjective wellbeing. It seems therefore that the accessibility, stewardship and management of local public spaces merits attention at policy level if the ambition is to improve subjective wellbeing at scale. As Snaith [48] points out, however, different cultures are likely to have different preferences for the type of public spaces they will use. With the sample analysed here being of predominantly white British origin, we must bear in mind the potential consequences of rolling out a uniform approach to the aesthetic design of parks and open spaces.

This research not only used network analysis to explore the factors associated with overall subjective wellbeing as measured by the short WEMWBS, it also embraced the complexity of the concept itself heeding the fact that this composite score is comprised of distinct aspects or different forms of subjective wellbeing. By analysing the data at the level of the individual question, we found that different neighbourhood/ place-based factors were associated with different aspects of wellbeing. In line with the overall importance of use of local greenspace to the composite wellbeing score, we found that this variable was particularly associated with the SWEMWBS questions of ‘feeling close to others’ and ‘feeling useful’. These more specific wellbeing findings relating to use of local greenspace may suggest that these areas are places where neighbours and members of the community may meet or bump into each other enabling a feeling of closeness. That use of these spaces was also associated with feeling useful may relate to the purpose of going to the areas and what activities are pursued when there.

Other notable links between individual SWEMWBS questions and neighbourhood variables included that an increased sense of neighbourhood cohesion was directly connected to the feeling of being close to others, supporting the sense that the connections being detected with the network have face validity. The finding that civic agency seemed to be most directly linked to having a sense of optimism for the future is reminiscent of previous research that emphasises the importance of joint decision-making in communities to wellbeing [18] and also to findings indicating that deciding about neighbourhood outcomes, purposes and visions in the future is an inherently optimistic process [49].

The present study has both strengths and limitations. Strengths include a novel and sophisticated analytical approach, and a dataset with rich information on both individual and place characteristics and mental wellbeing. In terms of limitations, the overall sample, although drawn from neighbourhoods of varying levels of relative deprivation, focussed on an economically deprived area of the United Kingdom, and thus findings may not generalize to areas characterised by greater social/economic advantage or less inequality. Indeed, as socioeconomic position is associated with dis/advantages across both individual and place domains, greater inequality may serve to moderate the relationships observed in the present networks. In addition, as is the case with all cross-sectional network analyses, dynamic associations and causality cannot be established between variables. However cross-sectional networks such as these are useful as an exploratory tool and can be used to identify potential causal pathways without relying on the stringent assumptions (e.g., acyclicity) of other methods (e.g. directed acyclic graphs) [50]. Future research could seek to use intensive longitudinal data and emerging dynamic network methodologies to explore the direction and temporality of theses associations, which would further unpack these complex processes. Although we employed a widely-used, reliable and valid measure of subjective wellbeing, the SWEMWBS is largely based on hedonic and eudaimonic conceptualisations. It must be noted that wellbeing is a multifaceted phenomenon that can be assessed by measuring a wide array of subjective and objective constructs. Although we aimed to include a comprehensive list of pertinent nodes, failure to incorporate all relevant nodes may lead to spurious edges and to a misrepresentation of the network structure [26]. Finally, we chose to focus largely on composite scores for most of our place and community variables, due to the formative nature of the concepts being measured [34] and topological overlap of indicators [26]. Choosing to construct networks at this level of granularity can impact the network characteristics [30].

The rich and layered analyses presented here provide important information for policymakers to address spatial disparities in wellbeing. Using network analyses to understand the complex connections between individual, community and place-based factors that correlate with wellbeing adds significant value to existing models. Our findings highlight the challenges in considering individual and place characteristics as truly separate domains. These findings can support certain evidence-based interventions based on a more sophisticated understanding of how they can affect change to population-level subjective wellbeing.