Problematising concepts and methods for civil society research in superdiverse neighbourhoods

This article analyses challenges for civil society research in superdiverse areas and proposes ways to overcome them. Key components of previous studies are problematised, such as the lack of attention to demographic complexity, the focus on formally registered organisations at the expense of informal ‘below the radar’ initiatives, the over-reliance on analyses using administrative data and building on dichotomous categorisations of social capital. The article calls for scholars to develop methodologies and theory that enable research across the full range of civil society activity. We argue for a holistic approach to researching civil society through comparative and mixed-methods designs that facilitate research about the nature of civil society action, its forms, patterns and experiences. The concept of ‘superdiversity’ is useful to reflect evolving demographic complexity, given age, gender, nationality, religion and immigration status, and divergent experiences of rights and the labour market.


Introduction
This article draws attention to the key role of methodology in view of the 'surprising lack of attention' to methodological debates in civil society research (Dean and Hogg, 2022: 1). We argue that civil society research needs to engage with methodological challenges, and the concept of 'superdiversity' given increasing demographic complexity, and propose ways to overcome these challenges. The past two decades have seen an increasing complexity associated with changes in migration patterns to Europe. 'Superdiversity' refers to the product of interactions and processes that contributes to population diversity and that supersedes previous experiences (Meissner and Vertovec, 2015). Major cities have experienced intense diversification processes and unprecedented levels of mobility with new arrivals, building upon long-established migrant populations and autochthonous residents (herein defined as dominant and majority populations claiming indigeneity). Much scholarship has focused on interactions between residents in superdiverse neighbourhoods (Wessendorf, 2014) but less attention has been paid to the nature of civil society and not-for-profit activities undertaken for the benefit or interests of groups (Jacobsson and Korolczuk, 2017). Further, civil society research has relied heavily on administrative or survey data, which tend to use ethno-national categorisations. Administrative data are collected and processed by charity regulators for the purpose of monitoring (McDonnell and Rutherford, 2022), whereas survey data collect information on self-reported membership and volunteering, and are vulnerable to non-response issues (Qvist, 2022), especially in superdiverse areas (Méndez and Font, 2013).
We problematise key components of previous studies, including the focus on formally registered civil society organisations (CSOs) and ethno-national groups, and argue for a holistic and local neighbourhood perspective. We outline problems associated with the dichotomous use of social capital, as a case of either bonding or bridging, within or between imagined minority/majority groups, which eschew considerations of population complexity. We also raise concerns about the overreliance on either quantitative analysis of administrative or survey data, or qualitative case studies about particular communities and activities.
Herein, we question whether ethno-national categorisations reflect contemporary demographic realities and argue that at best they offer a partial picture of the landscape of civil society in superdiverse areas. We begin by making the case for reframing civil society within conditions of superdiversity and conclude with a call for scholars to develop concepts, theory and methodologies to account for the complex nature of civil society action under conditions of diversification. We propose a holistic approach, looking across the full range of civil society activity, based on comparative and mixed methodologies.

The need for reframing and reconceptualising 'diversity'
Over the past two decades, much of Europe has experienced a transformation in migration flows and diversification processes, which has driven intensified local population diversity (Vertovec, 2007). Although the analytical ability of superdiversity, as a concept, to explain new social configurations has been questioned (Back, 2015), the term is widely used to describe evolving population complexity in urban contexts (Vertovec, 2019). Scholars tend to view superdiversity as diversity 'on the move' (Meissner and Vertovec, 2015: 550). Five superdiversification processes have been identified and underpin its emergence and evolution (Phillimore et al, 2021). The first is super-mobility and unprecedented levels of mobility at local, national and international scales. Populations are much less fixed to a place than previously observed (Robinson, 2010). The second is the speed of change wherein rapid and sometimes unexpected changes in the arrival and destinations of migrants are observed, for example following the 2015 refugee emergency. The third is the scale of change, which is uneven in terms of shifting proportions of newcomers across European countries and cities. The fourth is spread and refers to changes in migration flows and to the tendency for, for example, rural places to also change in response to increased diversity. The final process refers to demographic complexity. Previous scholarship on diversity has tended to focus on large ethno-national groups of post-colonial or economic migrants and thereby imagined within-group homogeneity. Superdiversity acknowledges the arrival of millions of migrants from wide-ranging countries, often without prior links to the countries of settlement, observed with earlier labour migrations. Many do not arrive at established communities but rather at localities where populations have become increasingly diversified and mobile (Pemberton and Phillimore, 2018). Significantly, the variation within and between ethno-national groups, along the lines of immigration status and rights, gender and age profiles, and patterns of spatial distribution, adds additional demographic complexity and is a core dimension of superdiversity. Meissner and Vertovec (2015) use the United Nations (UN) Global Migration Database, World Bank and UN Population Division data to illustrate the outcomes of superdiversification processes between 1960 and 2000, showing extraordinary changes pre-dating the 2015 refugee emergency.
The diversity of populations varies across countries, cities and neighbourhoods. Superdiverse neighbourhoods have attracted attention as places of long-term and permanent settlement, where a common neighbourhood identity may build around location and diversity instead of, or as well as, an ethno-national identity (Pemberton and Phillimore, 2018). Some urban neighbourhoods have been described as 'escalator areas' that are constantly evolving, as new arrivals find their feet in a new country before moving on (Travers et al, 2007). For example, Kensington, in Liverpool, England, saw a 20 per cent increase in the proportion of migrant residents between 2001 and 2011, with a shift from individuals originating in the Caribbean and South Asia to include those from Central and Eastern Europe, China and the Middle East, two-fifths of whom had been resident in the UK less than five years and planned to leave the neighbourhood (Pemberton and Phillimore, 2018). Since the emergence of superdiversity has brought new challenges for welfare providers seeking to meet the evolving needs of local people (Schierup et al, 2006), local civil society has increasingly been recognised as filling a gap in state service provision (Phillimore et al, 2021;Elgenius, 2017a). Civil society has enabled local participation and representation in service design in circumstances of austerity and welfare chauvinism. Yet, little is known about the nature of civil society in superdiverse neighbourhoods and how it responds to local needs (Warner and Sullivan, 2017). In the next section we highlight some of the challenges associated with existing categorisations of civil society action.

Problematising categorisations
Civil society has played a key role in filling gaps in welfare provision, especially in the face of austerity and the outsourcing of welfare services (Dahlberg, 2005;Warner and Sullivan, 2017). Policy makers increasingly recognise civil society's role in helping to mitigate marginalisation and disadvantage, acknowledging its potential to mobilise social networks and resources (McKague et al, 2015), aid communication with local institutions (Letki and Mierina, 2015) and contribute to various domains of integration (Elgenius, 2017a;2017b). Civil society has evolved to serve particular interest groups, such as migrants and refugees, and to facilitate excluded groups' access to mainstream services (Mayblin, 2017). The positive association between participating in organisations and economic and social benefits is well documented (Behtoui, 2007;Kahanec and Mendola, 2007;Wilson et al, 2020).
Previous scholarship has mainly focused on formal civil society organisations, relying heavily on the analysis of administrative data, with existing categorisations excluding unregistered initiatives falling 'below the radar' (McCabe and Phillimore, 2017). This is regrettable since unregistered civil society activity is understood to be an important mechanism of action (Soteri-Proctor and Alcock, 2012; Mayblin, 2017). The scale of unregistered actions is unknown but thought to exceed that of formal organisations. Such a lack of knowledge is problematic, since one of the key questions that has vexed scholars has been the precise relationship between the density of civil society activity and deprivation, with socio-economically deprived areas having low levels of formal activity and labelled 'charity deserts' (Mohan, 2012; or resource poor (NOA, 2016;Brå, 2017). Furthermore, research focusing on participation in registered activity finds that minority ethnic populations are less active than majority populations (Musick et al, 2000;Amnå, 2007;Gijsberts et al, 2012;MUCF, 2017). Putnam (2000; claimed that higher levels of diversity were associated with lower levels of trust in and membership of associations. Yet, Putnam also found a strong relationship between diversity and participation rates in protest marches and social reform groups, which might suggest that the nature of civil society actions may differ depending on levels of diversity. Moreover, such claims have been made in the absence of a full picture of civil society, including formal and informal activities. The nature of civil society action may differ with the extent of diversity because of a preponderance of unregistered activity rather than an absence of activity. The lines between formal and informal action may appear fuzzy but informal activity is, as a rule, organised by and for local people (Mayblin, 2017). Given that recent arrivals may be unfamiliar with the complex processes of registering organisations, they may struggle to register new civil society activities. Piper (2018) suggests that racism and discrimination impact funding opportunities, which favour mainstream organisations, with smaller groups run by diverse residents being denied the opportunity to formalise.
Intensive ethnographic work has been undertaken to capture different forms of civil society action at the neighbourhood level and has identified wide-ranging informal action that exceeds the levels of formal activity and is largely enacted by groups described as being from a minority ethnic background (Soteri-Proctor and Alcock, 2012), suggesting that a focus on registered actions overlooks a key component of activity. The bifurcation of formal and informal civil society organisation, and of so-called 'minority' and 'majority' ethnic action, may undermine attempts to grasp the range and complexity of civil society action. Without awareness of all actions, scholars generate an inaccurate picture of the civil society landscape, its actors and activities. We argue that to undertake research on civil society in superdiverse areas and to move our understanding of the relationship between diversity and civil society forward, new concepts and tools are needed.

Problematising concepts: the dichotomous categorisation of social capital
Social capital theory offers a popular framework for analysing the nature of social resources created through interactions with others. Much attention is paid to exploring the ways in which different types of networks yield different kinds of capital, which is seen as the bedrock of civil society action (Putnam, 2000). In the context of ethnically diverse places, a distinction is generally made between bonding networks and resources generated within an ethno-national group and bridging networks, which enable the development of social capital and connections between different 'ethnic' and 'mainstream' groups (Tselios et al, 2015;Weisinger and Salipante, 2005). Examples of bridging connections, frequently used as indicators of integration, include marriage between people from minority ethnic communities and the 'white majority' and inter-ethnic friendships (Kalmijn, 2010;Carol, 2013;Muttarak, 2014;Lichter et al, 2015;Wang and Morav, 2021). Social bridges tend to be valued by politicians and policy makers, while social bonds have been problematised in diverse areas and viewed as reinforcing segregated communities (Casey, 2016). We argue that the tendency to categorise social connections using binary minority-majority conceptualisations overlooks population complexity.
Thus, social capital approaches under-theorise the intermixtures and range of interactions between populations and activities in superdiverse areas. Bonding connections between 'people like me' have become associated with undermining social capital and cohesion and with helping people to get by but not to get on (de Souza Briggs, 2003). In turn, bridging connections between ethnic and autochthonous communities have been argued to yield more valuable social capital, promoting regional economic growth (Knudsen et al, 2000) and individual social mobility (Lancee, 2012). Such claims are generally founded on research using administrative (register) and survey data.
The binary language of minority/majority, immigrant/host is problematic in superdiverse areas, which may include long-established 'minority' groups into their fifth generation, residing alongside autochthonous populations and newcomers from across the globe. Such complexity leads us to question: Who is a minority? Grzymala-Kazlowska and Phillimore (2018), reflecting on social capital, highlight the importance of the social and local context in which interactions take place. They show that newcomers develop social relations with locals who may be longestablished migrants, minority groups or of mixed ethnicity, rather than solely the majority 'White' population (who may be a numeric minority). The terms 'minority ethnic' and 'majority' can make little sense in superdiverse contexts. There has been a tendency to classify civil society activities run by minority groups as bonding capital or ethnic civic community action, regardless of their sociodemographic characteristics (Fennema and Tillie, 1999;Morales and Giugni, 2011). Quantitative researchers interested in civil society often make a distinction between three types of organisations: ethnic, pan-ethnic (or cross-ethnic or pan-immigrant) and mainstream (Jacobs et al, 2004;Tillie, 2004;Pilati and Morales, 2016). Sometimes, poor data availability means these categorisations are simplified into ethnic versus mainstream organisations (for example, Vermeulen, 2005;Strömblad and Adman, 2010). As mentioned earlier, social connections that involve autochthonous actors tend to be viewed as the most beneficial (for example, Lancee, 2012). Yet in superdiverse neighbourhoods, individuals may view 'people like me' as those of the same class or length of residence (Pemberton and Phillimore, 2018). Thus, bonds can be built around a range of shared identities and may constitute bridging connections between different 'minority' groups. Without an obvious 'majority' in superdiverse areas, identifying differences between bonding and bridging activities is difficult. The two may not be mutually exclusive and may even be constitutive of each other. Bonding social capital can evolve into bridging capital for 'minority' groups as affiliations and identifications emerge based on nonethnic identifiers (Weisinger and Salipante, 2005).
Distinguishing between bonding and bridging capital may generate misleading findings about the nature of formally registered activities and how they evolve. For instance, in our fieldwork in superdiverse areas, a 'Somali football club' would be categorised as a bonding organisation by researchers adopting a social capital lens, based on its name (Somali), and classified according to a single purpose (sport). 2 Categorisations and classifications such as these rely on limited information and keyword searches, that is, the original beneficiary ('Somali') and the type of 'sports' charity ('football'). Yet our ethnographic examination revealed that the organisation addresses a wide range of needs, including recreational activities and homework support for local children from different backgrounds, and facilitates collaborations with other local organisations and state actors (Elgenius, 2020). Thus, classifying civil society endeavours requires detailed information not available in administrative data and formal records.
Additionally, quantitative research pays little attention to informal action because of a lack of data. Surveys may ask questions about 'informal sociability' or 'informal help provision' (Letki, 2008) but these are often worded differently in various surveys, making them challenging to compare over time and place. Omitting informal action is also problematic from a social capital perspective as it remains invisible in administrative data. For example, a study on access to healthcare resources in a superdiverse area found an unregistered lunch club that used a church hall and recruited volunteers from multiple backgrounds, many of whom volunteered to overcome their own sense of isolation (Grzymala-Kazlowska, 2020). With the arrival of asylum seekers to the neighbourhood, the club invited the newcomers to join. The club learned of the legal, structural and social difficulties they experienced and their activities evolved into informal mentoring, signposting for mental health issues and gardening. Organisers with detailed knowledge of the local civil society landscape facilitated connections to other local civil society organisations. Over time, boundaries blurred between volunteers and 'clients', actors and users, and between types of activities generating social capital. What started as an informal faith-based activity evolved from bonding relations to bridging capital within a network of civil society action. The nature of such social capital is invisible in registers and defies binary categorisations and formal classifications.
Thus, the dominant quantitative scholarship, originating from the social capital tradition, may have oversimplified 'minority' actions and social relationships. There are three main reasons for such oversimplification. First, the focus within the social capital tradition on the binary distinction between bridging and bonding may have encouraged researchers to translate such theoretical conceptualisation into binary empirical categories. Second, the policy debate around integration and multiculturalism in many Western countries has used language proposing minority ethnic (or immigrant) integration into the majority (autochthonous) population. Third, quantitative research is constrained by the data available in administrative registers, which by definition do not include unregistered organisations. The complexity of social connections and capital needs to be understood in light of the increasing diversification of populations, and actions within neighbourhoods. A more nuanced picture of diversity would help the identification of different forms of connection and actions, supporting an understanding of the ways in which activities evolve over time.

Problematising methods
Before we set out to consider methodologies and materials that could help to better identify and explain civil society action in superdiverse areas, we proceed to consider some limitations associated with quantitative and qualitative methodologies.

Quantitative studies, social capital and the density of civil society
As noted earlier, quantitative approaches to researching civil society tend to analyse administrative (register) or survey data, operationalising and measuring types of social capital and/or the density of formal organisations. Studies inspired by social capital theory use evidence from surveys on individuals' self-reported levels of participation in associations (Musick et al, 2000;Gijsberts et al, 2012), whereas scholarship focused on civil society 'density' explores the number and types of organisations using administrative data (Sampson, 2012;Kim, 2015;Mohan, 2015). While both traditions provide useful findings about broad patterns, neither can capture the range of civil society action in superdiverse neighbourhoods.
Most analyses find that minority ethnic populations are less active than majority populations (Musick et al, 2000) and identify a negative association between diversity and membership of civil society organisations (Putnam, 2007). However, quantitative analysts rely on broad ethnic categories. For example, Putnam (2007) uses four ethnic groupings when calculating his index of diversity; more fine-grained categories are difficult to construct, especially when examining diversity at the neighbourhood level. Individuals are therefore often classified into the 'majority' or broad 'minority ethnic' group(s), with measures capturing the proportion of the non-majority population commonly used as a proxy for diversity (Laurence et al, 2019). Categorising 'minority ethnic' groups at a national level is problematic for analyses of neighbourhoodlevel social capital, and measures capturing a few racialised minority groups are also insufficient to understand the diversity in local neighbourhoods can mean different things in different places.
Research about the relationship between area diversity and social capital in different countries has pointed in different directions, with the effect of diversity found to differ between minority and majority groups and to be generally small in magnitude (Fieldhouse and Cutts, 2010;Kesler and Bloemraad, 2010;Heath and Demireva, 2014). Once socioeconomic deprivation has been accounted for at the neighbourhood level, the negative relationship between diversity and social capital largely disappears (Laurence and Heath, 2008;Letki, 2008;Laurence, 2011;Sturgis et al, 2011), raising questions about the meaningfulness of such categorisations. Civil society in superdiverse areas therefore remains a 'black box' in quantitative studies, with little known about its nature, role and activities or the wider local and demographic context. In sum, quantitative studies offer insights about the broad patterns of area diversity (ethnic composition) and social capital through questioning attitudes and trust, or measuring membership or the density of organisations, which we turn to next.
The density and types of civil society organisation have also been analysed in relation to local disadvantage and the density is found to be lower in deprived areas (Mohan and Rolls, 2006;Clifford, 2018). However, the prevalence of formal organisations may be indicative of community disadvantage (Williams, 2003) rather than levels of civic participation, perhaps reflecting that certain types of activity are not routinely recorded. For example, Vermeulen et al (2012) show that the density and nature of organisations vary according to population diversity. Diverse neighbourhoods have a higher density of foundations, associated with bonding networks, and a lower density of leisure associations and bridging networks. Sampson (2012) offers a detailed picture of the relationship between the density of organisations and diversity, using an adapted neighbourhood survey that covered a range of activity, from community newspapers to after-school recreational activities. This extended range of neighbourhood activity is closer to a mapping approach where researchers identify actual activities and organisations in the neighbourhood. Sampson's findings were compared to both the administrative records of organisations and surveydata about the membership of organisations and they only modestly correlated with these. Thus, Sampson's extended measure tapped into domains not covered by the administrative records or survey data. Whereas the data from administrative registers or surveys provide information about the general structure of the organisational base locally, Sampson's extended measure shows how neighbourhoods self-organise for civil society action, providing details about specific services, programmes and purposes.
In other words, measures that include an extended range of neighbourhood activity predict in Sampson's terms 'collective efficacy', and residents' ability to create a safe environment, more adequately than the density of organisations per se. We take from Sampson's findings, the moderate correlation between administrative records of organisational density or survey-derived data about the membership of organisations and civil society action and argue that a local neighbourhood perspective and additional mapping techniques are required to understand how collective civic action is distributed.
Quantitative analyses can build on widely used categorisations of populations or classifications of action to find broad patterns of civil society associations in contexts of diversity and deprivation. However, quantitative approaches cannot adequately account for the complexity of local diversity, civil society activity and relationships between different forms of social capital or deprivation. Examining individual membership of formal organisations and ethnic diversity through survey data, or measuring the density of organisations in deprived areas based on administrative data, is problematic because it oversimplifies the nature of action and diversity. Using membership as a proxy for levels of civic action also overlooks the importance of informal volunteering. Letki (2008) demonstrates that there is no correlation between formal organisational membership and informal sociability at the neighbourhood level, which suggests that the association between formal and informal civil society activity should not be taken for granted. Administrative data record the address of registered organisations but not where the organisation is active, so measures of density do not equate to the locations of actions.
In sum, the focus on 'formal organisations' cannot be used as a measure of all civil society action. Likewise, the abstraction into 'minority groups' does not account for complex demographic realities. Data are therefore needed to gauge the range of activity, and purpose of formal as well as informal civil society, and to examine the relationship between formal and informal actions. In terms of categorising different types of social capital, research must move beyond binaries and reflect the ever-changing nature of populations in superdiverse neighbourhoods. Without such measures it is not possible to draw reliable conclusions about the relationship between diversity and levels of civil society action.

Qualitative studies and methodological nationalism
Qualitative studies have often moved away from fixed connections to community, ethnicity, identity and place and sought to highlight the heterogeneity of populations (Faist, 2010). However, capturing the variation of civil society activity in superdiverse areas, across time, space and population, is time-consuming. The difficulties associated with purposive sampling have sometimes led to a focus on communities of identity or communities of interest. A body of work has examined the civil society organisation of, for example, the Hindu, Jewish, Kurdish, Polish and Sikh diasporas (Safron, 2005;Cohen, 2008;Demir, 2012;Hawley, 2013;Garapich, 2016;Elgenius, 2017a;2017b;Leiding, 2020). Another body of work has explored activity in relation to refugees (Phillimore and Goodson, 2010;Benson, 2017). Qualitative studies have problematised claim-making, contestation, identities and practices around key themes, including the ethnic framing of heterogenous populations (Brubaker, 2005) and racialised logics (Elgenius and Garner, 2021). Yet, we raise two concerns that such approaches may unwittingly re-enforce groupism, albeit in a different way to quantitative approaches, and that they tend to be disconnected from quantitative civil society research.
There has been growing criticism of methodological nationalism (Wimmer and Glick-Schiller, 2002) directed towards studies that utilise ethno-national markers to categorise 'groups', assuming that individuals sharing ethno-national origins also share identities and interests. Such assumptions run the risk of essentialising cultures, reifying imagined characteristics and obscuring the importance of intersectionality, power and structure. Sampling in superdiverse populations constitutes a particular challenge in heterogenous, mobile and transnational populations (Engbersen et al, 2013;Elgenius, 2017a;Goodson and Grzymala-Kazlowska, 2017). Case studies also limit the focus to particular types of activities or mobilisations, or to certain categories of populations based on nationality, refugee status, whether disabled or non-disabled and sexual orientation. Whereas a case study may highlight the importance of immersion and reflexivity (Fitton, 2022), the focus on particular 'groups' or nature and performance of activity may potentially emphasise differences rather than looking across populations to explore similarities, and within populations identifying differences relating to contexts, conditions or other factors. Overall, material about the range of action, and how it is organised, is needed to generate a more authentic picture of civil society (Amelina and Barglowski, 2018).
Moreover, qualitative studies tend to be disconnected from quantitative research, even if triangulation would help to inform analyses. Combining and comparing data on formal and informal activity and on bonding and bridging connections and their intermixtures would offer a more holistic picture of civil society. Qualitative investigations contribute a detailed account of actual, rather than recorded activity, and would identify gaps in existing knowledge and explain patterns observed in quantitative analyses. Soteri-Proctor and Alcock (2012) offer insight into the added value of qualitative work. They used a street-mapping approach to examine the relationship between formal and informal activity in two superdiverse neighbourhoods. They attempted to record all ongoing civil society activity and identified 58 informal self-organised activities on 11 streets, none of which were listed in registers. Most initiatives were organised by local people, with some actions run by and for specific ethno-national groups and others crossing multiple boundaries. These findings reinforce the importance of treating claims about the relationship between diversity and the density of civil society activity based on an analysis of administrative data with caution. More research is needed to map activity at the street level and to identify the disparities between ongoing and local activity and administrative data. Such studies could explore and explain the heterogeneity of civil society actions, ensuring that 'ethnicity and nation' are not privileged (Amelina and Barglowski, 2018).

Towards a holistic approach of civil society, locality and place
A holistic understanding of civil society ought to account for patterns as well as classifications and move away from ethno-national lenses in favour of examining the nature of actions enacted to address the needs of both complex superdiverse as well as fixed populations. In so doing, we argue for adopting a local neighbourhood perspective. A key contribution by scholars of superdiversity has been the shift from methodological nationalism to methodological neighbourhoodism (Berg, 2019), with the attention moving from social relations between co-ethnic groups to social relations within places (Beck, 2007;Pemberton et al, 2019;Phillimore et al, 2021). An emphasis on place enables us to increase our understanding of civil society and local actors investigating social relations and connections, enabling the full gamut of civil society actions to be identified without resorting to binaries. A more holistic approach offers much potential for understanding activity in places where populations are fluid. The importance of networks and partnerships within and between civil society, state and private actors has long been acknowledged (Johnston and Blenkinsopp, 2017;Warner and Sullivan, 2017) and we suggest it is important to examine what local actors do, how they do it and how activities connect. Such research would be particularly useful to understand social welfare challenges associated with deprivation, regeneration and integration, which are often measured in relation to place. It would bring insight into the ongoing and unresolved debate about the association between diversity and the density of action.
A decisive shift is therefore needed towards comparative mixed-methods approaches that capture the entirety of civil actions. We suggest beginning with conventional analyses of administrative data to identify the type of activity officially recorded locally. Ethnographic research, including the micro-mapping of activities and interviews with civil society actors, would be undertaken in parallel. Ethnographic accounts enable a rich and contemporary picture of actual actions and their functions. By mapping activity, actions and actors, and by identifying stakeholders and the types of work they do, we can explore key questions about the relationship between different kinds of social capital, actions and diversity. Such questioning allows for an examination of different types of social connection, of how they shape and are shaped by civil society, and of the types of actions local people engage in, perhaps identifying new categories that could be included in future surveys.
Comparison has been described as one of the 'most profound and generative perspectival processes on which much of our reasoning rests' (Seawright and Gerring, 2008: 297). May (2011) identifies three purposes of comparison that are relevant for our argument: highlighting the particularity of a case; identifying convergences and deviations; and revealing causal generalisations or supporting theory development. All three are relevant to the study of civil society in superdiverse neighbourhoods. The primary purpose of comparison is the identification of parallels and differences in specific phenomena -for civil society in superdiverse areas this includes understanding the ways in which formal and informal actions interact. Through comparing quantitative and qualitative datasets we might find that some organisations are no longer active, or have changed focus and target groups. Comparison enables an understanding and explanation of the relationship between civil society and superdiversity, perhaps allowing scholars to identify different types of activities in different localities, with potential for generalisation. Given that superdiverse neighbourhoods have different demographic compositions and different levels in the availability of resources, it is important that comparisons take place at multiple scales. These range from the local to the larger scale by comparing civil society activity across countries to examine how different welfare regimes and legislative contexts shape civil society (Phillimore et al, 2021). Additionally, researchers need to reconsider the overreliance on high-level administrative data as particular types of civil society activity and neighbourhoods may suffer from underreporting. Country-level comparisons based on high-level administrative data, such as Salamon and Anheier's (1998) study, that explain the emergence of different forms of civil society through social origins theory will contribute towards making grand claims about the nature and origin of different forms of civil society cross-nationally or within different welfare regimes. Yet, the limitations of such data need to be carefully considered.
In sum, by comparing civil society across neighbourhoods, cities and countries, scholars can explore the nature of opportunity structures in which civil society action takes place and account for local population composition, deprivation levels, policy frameworks and levels of investment. In so doing we are wary of introducing a new binary of being inside/outside geographical boundaries (Amin, 2004).

Concluding remarks
We have argued in this article that the lack of attention to population complexity in civil society scholarship has either undermined efforts to understand the relationship between civil society and diversity or homogenised diversity. Claims made about low levels of civil society action and a lack of bridging activity in diverse areas have fed into a policy discourse problematising diversity as being associated with selfsegregation, lack of proclivity towards volunteering and low levels of trust (Casey, 2016). We argue that a shift is needed from methodological nationalism to a focus on local neighbourhoods, and from a focus on groups to social relations within a place, as we remain open to the possibility that informal activity may be as important as formal activity. Unregistered activities remain under the radar of official analyses, with little evidence available about the nature or efficacy of actions. Ignoring informal activity is particularly risky in superdiverse areas given the policy rhetoric around the non-participation of diverse populations and newer residents being less familiar with regulatory frameworks and so are more likely to operate below the radar (Piper, 2018).
To undertake research on civil society in superdiverse areas, scholars need to continue to develop concepts that shape our research questions and allow us to explore the full range of civil society action. Thus, a nuanced understanding of different forms of social connection and social capital is needed, which eschews the bonding/bridging, and minority/majority dichotomies. We need to analyse social capital generated by formal and informal initiatives too, and how activities evolve over time. Similarly, and in view of the complexities we have espoused, we must be wary about making claims regarding the relationship between diversity and the density of civil society action until we can account for all kinds of activity, whether based on communities of interest, identity or place, and understand how and why they differ. Unless we have a more complete picture of civil society activity and how it functions, we will not be able to appreciate its nature, role and efficacy.
To address the challenges associated with connecting civil society and diversity, we have argued for a neighbourhood approach that builds a holistic picture based on mixed and comparative research methods. Approaches are needed that seek to capture the full range of activity, changes over time and interconnections between actions, and enable us to identify mechanisms that underpin their functioning and influence on local and national opportunity structures. Combining and comparing qualitative and quantitative data enables scholars to generate a detailed picture of civil society. Without such a picture, it is difficult to imagine how evidence-based policy making will be possible.
Notes 1 Corresponding author. 2 Some institutions tasked with collecting official statistics on civil society organisations use the International Classification of Non-profit Organisations (ICNPO). One of the major categories is 'law, advocacy and politics', and, in turn, one of its subcategories is 'ethnic associations' or 'organisations that promote the interests of, or provide services to, members belonging to a specific ethnic heritage'.

Funding and disclosure
This research was supported by the Swedish Research Council for Health, Working Life and Welfare (FORTE) under grant number 2018-00181 and the title 'The role of civil society for employability and labour market participation in diverse areas: a comparative study of Sweden and the United Kingdom'. Principal investigator: Elgenius; co-investigators: Phillimore, Borkowska and Kawalerowicz. This research was funded in whole or in part by Forte . For the purpose of Open Access, the author has applied a CC BY public copyright licence to any Author Accepted Manuscript (AAM) version arising from this submission.