Ethnic Diversity and European's Generalised Trust: How Inclusive Immigration Policy Can Aid a Positive Association
by Anna Zimdars and Gindo Tampubolon
Kings College London
Sociological Research Online, 17 (3) 15
Received: 1 Mar 2011 Accepted: 22 Mar 2012 Published: 31 Aug 2012
The present article contributes to the discussions of the link as between diversity and social capital levels by focusing on the process linking diversity and trust. We look at country-level policies rather than individual communities and evaluate how policies can influence the relationship between diversity and generalised trust. We analyse Eurobarometer data (2004) using multilevel analysis. We find a positive effect of diversity on trust, controlling for potentially confounding effect at the individual and country level. Furthermore, we find that the inclusiveness of national policies towards migrants matters. More inclusive policies as measured by the migrant integration policy index (Mipex) counterbalance potentially negative effects of increasing diversity. The findings show that while local place is crucial as the locus where diverse ethnic groups interact, countries remain important policy contexts influencing and framing interactions with immigrants. For Europe, there is nothing inevitable about a negative impact of increasing diversity on building cohesive, trusting societies. We conclude by locating our findings within a wider critical literature.
Keywords: Ethnic Diversity; Generalised Trust; Integration Policy; Mipex; Multi-Level Modelling
Introduction1.1 Immigration and demographic change are increasing the ethnic diversity of societies in advanced industrial countries (Salt, 2005, p.26). The implications of this change are debated in the media, by policy makers, and academics. One prominent concern in these immigration debates is that increasing diversity might affect societies' social capital. Specifically, there is concern that levels of generalised trust, that is how people feel they can trust and relate to strangers in their communities, might be lower in more ethnically heterogeneous communities. In such diverse communities, a 'hunkering down' might occur whereby individuals withdraw more into their personal spheres and participate less in community-level activities (Putnam 2007). Researchers show that the more ethnically diverse local places are, the less volume of social capital, or at least some forms of social capital, circulate there (for the Netherlands see Tolsma et al., 2009; for the UK see Letki, 2008; for U.S. see Alesina and La Ferrara, 2002; Hero, 2003; Putnam, 2007; for some comparative work see Delhey and Newton, 2005).
1.2 However, such 'apocalyptic' scenarios of less cohesive, trusting societies has not been without critics. Such criticism has engaged with the empirical findings and methodologies linking higher levels of diversity with lower levels of trust as well as questioning the interpretation of these findings (e.g. Hallberg and Lund 2005; Portes and Vickstrom, 2001). Empirical research has not unreservedly confirmed Putnam's hunkering down description of the U.S. or at least not without further qualifications (Fieldhouse and Cutts 2010). For example, levels of inequality, a measure correlated with diversity, has been suggested as a major alternative explanation for lower levels of social capital in diverse communities (Cheong et al 2007, p. 37; Fieldhouse and Cutts 2000).
1.3 Theoretically, it is not clear that we would expect ethnic diversity to lower levels of trust or other measures of social capital. This is because different aspects of social capital are affected differently by the diversification of communities as now illustrated in the discussion of bonding and bridging social capital. Bonding social capital is an intra-community social capital. An example would be an immigrant diaspora community meeting up to maintain their language and cultural rituals. Bonding social capital would not be expected to improve generalised interpersonal trust as it does not create interactions with people unlike oneself, and is not always socially desirable. Extremist groups like 'jihad' groups or criminal groups like the mafia have high levels of bonding, or within group social capital without promoting inclusivity and trust beyond the group. The second type of social capital, bridging social capital, brings together people across cleavages such as race, class or religion. The title of Putnam's influential book 'Bowling Alone' (2000) is illustrative (further illustration of bonding and bridging social capital can be found here: http://www.hks.harvard.edu/saguaro/glossary.htm): Bowling groups are much more likely to bring together individuals across race and occupational status and can thus, through contact, foster trust in people who are not like oneself, bowling groups 'bridge' social distances. These groups do this because they are built around interests, past-times and values rather than being based around ascriptive characteristics such as gender or ethnicity.
1.4 Bonding and bridging social capital might be differently affected by immigration processes. For example, bonding social capital might decrease in more diverse communities as arithmetic dictates a decrease in opportunities to socialising with social similars for members of the original majority group when communities diversify. Some countries also have policies to discourage continued bonding social capital in immigrant communities but to increase bridging social capital activities instead. For example, the UK recently stopped supporting language support for immigrants' mother tongues (Cheong et al. 2007) while supporting initiatives to learn English. Indeed, immigration entails opportunities to increase the more desirable type of bridging social capital nurtured in relatively heterogeneous associations. This type of social capital is more desirable because it is from such heterogenous associations that generalised trust as opposed to personal trust stems from (Putnam 1993; Tocqueville 1863). Empirically, ethnically diverse friendship ties have been found to correlate with higher levels of generalised trust (Phan 2008).
1.5 The body of literature relating to homophily can further shed light on the possible contrary expectations of how increased ethnic diversity might impact on trust. Homophily is the idea that people have an affinity for people like themselves (e.g. McPherson et al. 2001). But in our modern complex societies, the dimensions on which people match others as being 'like' themselves is fluid and socially constructed in the context of particular societies. 'Imagined communities' in friendships and social trust can be constructed across wide-ranging dimensions like having gone to the same school or university (Sommerlad and Sanderson 1998), religion / values, gender, leisure time pursuits like bowling. occupational field, or nationality (Anderson 1991). But even with regards to nationality, different notions of what it means to be a citizen of a particular nation can mean different things in different countries: in some countries notions of values of citizenship are at the heart of the national identity whereas in other countries, the idea of ancestry dominates (Ford et al. 2011). For example, in the Phillipines, national identity is very much built on the notion of ancestry whereas in the Netherlands it is more a notion of civicness. This again points to the role of national policy contexts for influencing which dimensions are salient for considering others as being 'like oneself' and hence trustworthy. Again, nation states have some power and discretion over constructing immigrants as 'the other' or constructing them as social similar.
1.6 It thus follows that the association between ethnic diversity and generalised trust could theoretically be a positive one. At the very least it is unclear that a negative association would necessarily or inevitably have to follow from increased contact at the local level.
1.7 The present article then contributes to the debates surrounding the impact of ethnic diversity on social capital as generalised trust by looking at how different policies towards migrants influence the relationship as between diversity and social capital. The research rationale for choosing national policies as a research focus is elaborated in the next section.
The centrality of local place and the under-valued role of country2.1 The vast majority of empirical research on social trust and diversity considers the local place, as the locus for interaction, holds a crucial role in understanding the relationship as between diversity and social capital. Researchers in this tradition disaggregates analysis as close to the local level as possible (e.g. Fieldhouse and Cutts 2010; Phan 2008). These local places are assumed to be crucial as the locus where actual interaction occurs, which, in turn, has been theorised to either enhancing generalised trust through 'contact' or to decrease it through 'conflict'. Local place has the potential to generate both positive as well as negative outcomes of increased diversity (for contact theory see e.g. Brown, 1995, p.268; Laurence and Heath, 2008; Martinovic et al, 2009; for conflict theory see e.g. Alesina and La Ferrara, 2002; Costa and Kahn, 2003). The mechanism can be presented as:
diversity --> local place-based interaction --> conflict or contact --> social capital erosion or generation.
2.2 In this mechanism, the more refined or disaggregated the place as a unit of analysis is, the more credible we regard the finding about a relationship as between diversity and social capital. In this conceptualisation, larger units of analysis, such as country as opposed to neighbourhood, are considered less credible as the locus for interaction.
2.3 While agreeing that the local place is key for understanding where actual interactions take place, we do not think that the role of a country (as place) should be dismissed out of hand in understanding the processes that link diversity with the creation or erosion of social capital. Country level factors or specific climates of opinions at a certain point in time have been found to affect a range of other social phenomena ranging from the propensity to vote for right wing parties, to attitudes towards progressive taxation (e.g. Knigge 1998; Entman 1989). Moreover, understanding country level effects can provide some valuable information about the possibility of influencing this relationship through policy.
2.4 The importance of country is further illustrated through two considerations. Firstly, empirically, attitudes that one would expect to be related to actual ethnic diversity at the local level for example attitudes towards immigrants sometimes have no more than a fleeting connection with actual neighbourhood diversity. An example for this phenomenon is the attitude of rural Britons towards immigration. These rural communities hold stronger views about immigration being undesirable despite the fact that virtually no immigrants live in their countryside (Pointer, 2005). More diverse metropolitan areas in Britain tend to show a more positive view towards migrants (NCVO 2002, Letki 2008, see also Oliver and Wong 2003). This may indicate that discourses in the neighbourhood, perception, and newspaper coverage are also important in the perception of immigration in addition to actual contact. Secondly, the role of country as place should not be limited to the locus of spatial interaction. Every country has a surrounding climate of ideas about how to live, what we value, what we find admirable in ourselves and others and what we strive for (Blackburn, 2003). This climate of ideas includes attitudes towards diversity, immigration and integration. While the European nation states have been described as sandwiched between the simultaneous rise of the importance of global and local concerns (e.g. Held and McGrew, 2003; Collier, 1997) and policies might converge at the supra-national EU levels countries are far from dead or negligible in shaping the climate of ideas through national policy and politics. Countries continue to be seminal social space where ideas and evidence of diversity are discussed through the mass media and where policies regarding diversity are formulated and implemented by the state.
2.5 In Europe, the response to increasing diversity and immigration could indeed be described as approximating a natural social experiment that allows us to find out more about the impact of different policies in affecting the relationship between diversity and generalised trust. Different European countries have chosen divergent policies in response to the increasing diversity of their citizens brought about by immigration, emigration and demographic change (Meyers, 2007; Joppke, 1999). This has led to country-level differences in the extent to which the societies in each nation state confront the change towards more diversity and embrace or reject this change (Schneider, 2008). For example, some countries chose more multi-cultural policies (Kymlicka and Norman, 2000) whereas others preferred the route of assimilation policies (Koopmans et al., 2005). Ideas of whether citizenship is constructed based on ancestry or based on civic values is also relevant (Ford et al. 2011). We hypothesise in the present paper that whether contact with population groups different from one's own group has positive or negative effects is influenced by the national policy context framing these interactions: whilst we might all have a homophilic preference for social similar individuals in our individual interactions, this social similarity is in itself a social construct influenced by which cleavages are viewed as salient for similarity and dissimilarity. For example, 19th century English society constructed the Irish as socially dissimilar and undesirable foreigners, but this English-Irish cleavage is significantly less salient today (MacRaild 2000). We thus hypothesise that government policy towards immigration and the construction of the other as different or alike warranting inclusion influences whether diversity erodes or creates social capital.
This hypothesised mechanism is illustrated below:
Diversity --> --> --> social capital creation or erosion.
Different State policies toward diversity/public ideas about diversity
2.6 Let us illustrate the meaning of this by way of an example. We assume that increases in ethnic diversity leads to increases in public discussion about diversity as well as to the formulation of policy geared towards managing diversity. If state policy towards diversity was accommodative, so as to minimise the different standings of various ethnic groups in say political processes, then we would expect the impact of diversity per se on social capital to be limited or negligible. In contrast, if policies towards diversity were dividing the community into fragmented or even competing groups, we might expect diversity to have a negative impact on social capital creation. We thus derive the following research hypothesis for our subsequent empirical analysis:
Hypothesis: Accommodative or inclusive state policy towards diversity means that there is no negative association between diversity and generalised trust.
2.7 Before we delve into the analysis, however, it is prudent to add some caveats about what we cannot achieve in the present article. We have to acknowledge that our hypothesis says nothing about what predicts different state approaches towards diversity policy. Such policies are quite possibly not completely independent from either the actual level of diversity in a country, or the existing stock of social capital and generalised trust in a country. For example, a public climate characterised by high levels of generalised trust, such as in the Scandinavian countries, might ease the passage of accommodative policies. Conversely, accommodating policies which incorporate heterogeneous groups in to the citizenry could foster generalised trust through the creation of mutual respect. In any case, our hypothesis and our argument are silent about these issues. It is, however, also likely that public policy towards diversity is at least partly independent from actual diversity levels and social capital or generalised trust as such policies will be influenced by ideological, party-political and other historic factors specific to particular countries. To summarise our main caveat, we introduce a third variable into our model (public policy and public climate) that we think will influence the link as between diversity and social capital. Our intervening variable is, however, not perfectly independent. Bearing this limitation in mind, we now turn to the description of our methodology.
Methodology3.1 Our analysis then seeks to isolate the net effect of diversity on social capital taking into account the above hypothesised mediating impact of national policies. We use Eurobarometer data (2004) containing individual-level information on respondents' trust, voluntary association memberships and other factors. In total, 27,008 respondents from 27 European countries are included in our analysis. In line with previous work, we use generalised social trust as an important facet of social capital (Stolle, 2002; Yamagishi and Yamagishi, 1994). The dependent variable, generalised trust, was measured by answer to the following question: 'Generally speaking, would you say that most people can be trusted or that you can't be too careful in dealing with people?' The answer categories were: (1) 'most people can be trusted'; (2) 'don't know or it depends'; and (3) 'one cannot be too careful'. We code this dependent variable as an ordered response variable rather than a binary variable (i.e. 'can be trusted', 'don't know', 'cannot be too careful'). In doing so, we diverge from several contributions to the literature that treat trust as a binary response where the answer category don't know is inconsistently collapsed into either 'can be trusted' to 'cannot be too careful'. This difference in operationalisation may explain why the effect sizes for some relationships do not always replicate previous work on the same data set e.g. Gesthuizen et al (2009).
3.2 To tease out the effect of immigration policy, we further included as many other confounding factors at the individual and country-level as possible as detailed below. Including potential confounding factors increases our confidence that our findings are not caused by other unobserved country differences let it be differences in the composition of the characteristics of citizens, or other country-specific phenomena such as unemployment and the actual level of diversity (Hooghe et al., 2009). We now briefly discuss the country-level variables included in the study.
Country level effects: Inclusiveness of Integration Policy
3.3 The main thrust of our argument concerns the impact of policy at the country-level as affecting the relationship between diversity and social capital. This crucial policy dimension is operationalised using the Migrant Integration Policy Index (Mipex). The Mipex measure combines over 140 individual indicators in six policy areas: (1) labour market access; (2) family reunion; (3) long-term residence; (4) political participation; (5) access to nationality, and (6) anti-discrimination. Countries vary on these indices. The best practice countries in the area of anti-discrimination, for example, are Sweden, Portugal, Hungary, the UK, the Netherlands and France. They ensure the most equal access to economic, social and public life for members of their societies regardless of migration status. Legislation is used to enshrine and enforce anti-discrimination and to redress grievances in these countries. In contrast, the lowest ranking countries on the anti-discrimination dimension of the Mipex measure Estonia, the Czech Republic, Denmark and Latvia have lower law enforcement levels or limited access to redress grievances through the court system. These states can be less relied upon to combat discrimination actively and to provide equal access to employment, housing, wealth and educational opportunities (e.g. Bell et al 2006). Similar discrepancies in practice exist in other areas. Migrants in the Scandinavian countries and Luxembourg are much more able to participate in public life, for exampleby voting and standing for office, than in other countries, most notably some Eastern European countries such as Latvia and Lithuania, but also Greece where voting rights might be much more restricted. Detailed country profiles regarding their migrant integration policies are available from the Mipex website.
3.4 The ranking of countries indicates that the Mipex scores map to some extent, but not perfectly, onto other cleavages such as geographic location and length of democratic governance. Most Eastern European countries score towards the lower end of the overall Mipex score and there is a clustering of the Scandinavian countries with the most inclusive integration policies at the top of the Mipex rankings.
3.5 As previously discussed, we try to minimise unobserved heterogeneity in our analysis and to maximise confidence that we actually capture the effect of migration policies rather than some other country-level differences. We include the percentage of citizens who are at least members of one organisation (Paxton, 2002, 2007). We also wish to take into account the ethnic homogeneity of a given country as a proxy of existing levels of ethnic diversity at the country level (Putnam, 2007). We do so by including a diversity index based on the percent of people who have at least one parent born in a non-European country (Hooghe et al., 2009). A dummy was included to capture whether observations were made in a former Eastern European (USSR) country as trust levels in this particular region are lower than elsewhere in Europe (Letki and Evans, 2005). We also include unemployment at the national level as it has been found to be associated with social capital levels (Putnam, 1995).
Individual level effects
3.6 We take into account an array of information at the individual and country level that has been found in previous research to be associated with social capital (e.g. Brehm and Rahn, 1997; Whiteley, 1999; Uslaner, 2002, Gesthuizen et al. 2009). This approach is best suited to estimating the effect of the policy net of other predictors of social capital. Specifically, we control for respondents' age and age squared (Robinson and Jackson, 2001), occupation (salariat or other), education (Hooghe et al., 2009) and gender (e.g. Whiteley, 1999; Putnam 2000), as well as 'lifestyle' variables such as religiosity (Delhey and Newton, 2005; La Porta et al., 1997), urban or rural residence, divorce (Rahn et al., 2003) and loneliness. It would have been desirable to include a measure of ethnicity but, unfortunately, no suitable variable was available from the data set we are using in our analysis.
3.7 In terms of modelling individual level social capital levels we take into account organisational membership (Putnam, 2000; Zucker, 1986; Axelrod, 1984; de Tocqueville, 1835) and the quantity of memberships (Stolle, 2001; Perrin, 2005) and other forms of engagement. Furthermore, we think that quantity measures of engagement are insufficient indicators of the qualitative differences of different engagement activities. Research in other fields of social enquiry strongly suggest that it is qualitative rather than quantitative differences in an experience such as education that account for differences in outcomes (Lucas, 2001). In the social capital literature, qualitative differences in participation have also grown in salience in analysis: joining a sports club is different from joining a political party or other forms of community engagement (Paxton, 2007; Seippel, 2008). The impact is not only different for the person doing the joining but also in terms of the externalities such activities create for the community (Putnam, 2000; Blau, 1977). Moreover, following Paxton's analysis of the World Value Survey, we also model the connectedness of memberships an individual holds (Paxton, 2002, 2007). Connected membership organisations such as environmental groups, charities and political parties are those where participants enjoy membership of more than one voluntary association fewer than 9 per cent of members are only members of this organisation in our data whereas in isolated organisations such as recreational, religious, trade union and elderly activities, about one-third of members are only members of this one single organisation.
3.8 Furthermore, we wish to go beyond the qualitative distinction made by Paxton (2002) to argue for the inclusion of informal social network measures in our modelling (Pichler and Wallace, 2007, Gesthuizen et al. 2009). This proposition goes back to the origin of the idea of capital a commodity that enables the proprietor to do things they would not otherwise be able to do. Here, informal social networks based on friendship and neighbourhood are enabling, perhaps even more so than formal networks (Willmott, 1987). We thus included a measure of respondents' informal social capital.
3.9 In terms of statistical approach and the presentation of findings, we use descriptive cross-tabulation and multi-level ordered logit regression (Snijder and Bosker 1999). We used ordered logit models since the dependent variable is ordinal. The ordered logit approach was used in preference to linear modelling because ordered logit models are particularly well suited to contexts in which the dependent values covered are all positive and limited. Direct comparisons of the findings concerning ordered dependent variables that were derived from both linear regression analysis and ordered logit analysis, have found that 'results based on the ordinal logit models should be considered more reliable than those from the regression analysis' (Lu 1999, p.282).
3.10 After having discussed our data and conceptual and analytical approach, we are now turning to a description of our results. We conclude with a discussion of our findings and some tentative implications for policy makers interested in sustaining social capital as our European societies become more diverse. Results
3.11 Firstly, in Table 1, we tabulate the relationship as between individual level and country level variables and generalised trust. We do this to reassure ourselves and the readers that we see the general patterns as between individual level and country level variables and generalised trust that we would expect to see given the previous research in this area, and in particular the work of Gesthuizen et al (2009) that based its social capital analysis on the same Eurobarometer data source.
|Table 1. Frequency distribution of individual and contextual level variables|
3.12 Tying in with previous work, we can see that members of groups that historically suffer more disadvantages such as women, manual workers and the unemployed, show lower levels of trust (Whiteley, 1999). There is a positive association as between trust and education and the number of memberships (Putnam, 2000) and connected memberships but not with religiosity (Paxton, 2002). We find no effect of place of residence (urban versus rural), and, contrary to Paxton (2007), we do not find an effect of being divorced on trust. This might be due to the different experiences of divorce in Europe and the US with significantly lower well-being outcomes for divorcees in the US case (Kalmijn, 2009). We shall see, however, that being married as opposed to any other form of living arrangement, has a positive effect on trust in the multi-variate analysis.
3.13 Turning to the relationship of contextual variables at the country level, we find in line with Putnam (2007) that ethnic homogeneity has a small but significant positive effect which may appear on the surface to lend some support to the previously discussed studies that show a positive effect of ethnic homogeneity on trust. Country specific membership levels have the expected significant effect (Paxton, 2007). Countries with more citizens in organisations trust more, this holds for both connected and isolated memberships (results not shown). Furthermore, respondents who trust live in countries which, on average, have lower unemployment rates than respondents who do not trust. These findings from the cross-tabulation reassure us that the patterns we observe are largely compatible with the social capital literature. Hence, our regression analysis should be capable of adding worthwhile considerations about the impact of policy on the relationship as between diversity and trust controlling for at least some of the usual sources of variations of social capital at the individual and country level.
|Table 2. Trust, membership number and informal capital: outcomes of multilevel regression analysis using Eurobarometer 2004 (*** p < .001)|
3.14 We now turn to the multilevel regression analysis (Table 2). We first look at the individual level effects on trust. Reassuringly, the patterns here again tie in with the existing social capital literature, thus enhancing our confidence in the data: the educated and those in the salariat score higher than those with less education and lower level employment across the three measures. In contrast to some (Paxton, 2007) but not all (Hooghe et al., 2009) previous research we find a positive effect of religiosity on trust and membership numbers. Those who are married are more likely to trust perhaps because married couples have each other for support and interaction. Rural location is not related to trust. To recapitulate, the individual effects are largely in the direction we expect them to be. But, perhaps more importantly, these individual effects remain largely unchanged when national level controls are added in subsequent steps.
3.15 Turning to the country-level effects, we can see that, as expected, Eastern Europeans trust less. Unemployment at the national level has a negative relationship with trust while other country specific membership levels have a positive relationship with trust. The diversity index actually has a gross positive effect and a net neutral effect on trust (Gesthuizen et al., 2009) and, to come to our conceptually most crucial measure, this effect linking diversity and trust is largely, but not completely, explained by our Mipex indicators. We find that regardless of whether individual Mipex measures or the Mipex total is being used, Mipex policies are significantly associated with social capital: the more inclusive the migration policies, the higher the levels of trust.
Discussion and Conclusion4.1 The present article aimed to investigate the role of the national context in mediating the relationship as between diversity and trust. We argued that while disaggregating analysis to the local level was a most useful advance in investigating the contact versus conflict hypothesis of diversification, such analysis were less well placed to tell us about how the country context framing such interactions can shape the outcomes. We put forward the hypothesis that country level policies towards migrants would influence the link as between diversity and social capital outcomes such as trust. We proposed that neither positive nor negative outcomes of increased diversity are inevitable but instead dependent on different countries' approaches towards managing the increasing diversity of western societies.
4.2 To test our hypothesis empirically, we used the Eurobarometer 2004 data augmented with additional measures. One limitation of this approach and our cross-sectional modelling is that we are largely limited to describing the relationship as between diversity, social capital measures and policy intervention. In the present paper, we cannot uncover the causal processes that lead some countries to adopt policies that are accommodative of migrants thus leading to high Mipex scores and other countries that score low on these measures.
4.3 Bearing this caveat in mind, our analysis first indicated that, contrary to Putnam's finding on diversity and trust (2007), the per cent in the population with foreign born parents was positively related to trust for the European context. We found no effects of the percent of foreign born parents on membership numbers or informal networks. These findings suggest that the negative association between diversity and social capital may be a U.S. specific phenomenon. We also find that the gross association as between diversity and social capital was reversed for Europe compared with the U.S..
4.4 Turning to our findings with regards to policy intervention specifically, we find that government policy as operationalised using the Mipex indicators, can have an impact on the relationship as between diversity and trust by counteracting potentially negative effects of diversity. In other words, controlling for the Mipex measure and other country level differences, the net effect of diversity on trust is small when we take into account that countries with positive diversity policies have higher levels of trust than countries with less accommodating policies. This finding confirms our research hypothesis regarding the ability of policy to influence the link as between diversity and trust: public ideas and state policy influence the link between diversity and social capital, and there is thus no inevitability about the consequences of diversity for social capital.
4.5 Policy makers have tools at their disposal to influence whether diversity is associated with negative repercussions for generalised trust in society. Thus, European societies might be simultaneously walking towards a more trusting future and increasingly diverse societies.
4.6 Furthermore, the normative interpretation that lower levels of social capital will inevitably have undesirable effects requiring policy intervention has also been challenged. Portes and Vickstrom's review of 'diversity, social capital and cohesion' reminds us that Durkheim's work on the foundations of modern societies offers a rather different take on this question: the functioning of modern societies does not depend so much on interpersonal networks or mutual expressions of trust, but on strong institutions and an effective and complex division of labour (Portes and Vickstrom, 2011, p 476). One may be tempted to add that this observation is echoed by another founding figure of the discipline, Max Weber, who saw societies as being increasingly rooted in bureaucratic systems and norms. While this has the danger of creating an iron cage (Weber  1958) this rationalisation of the modern world also explains how people can effectively move around in large anonymous cities without personally knowing the vast majority of people they see each day. Even if generalised trust was declining, it does not seem to decline to the point of challenging the habits of modern life such as urban travel or a breakdown of institutions.
4.7 Finally, Hallberg and Lund also question the implicitly taken assumption of Putnamian hunkering-down scenarios by arguing that this position implies taking the vantage point of a homogenous community and its good, but potentially exclusionary existing social networks. The two authors thus reverse the traditional social capital question of how this public and personal good is challenged by immigration to raise the question 'how are the valuable networks that constitute social capital bad for diversity?' (Hallberg and Lund, 2005, p. 61). In other words, if we valued diversity more than social capital, the question would not be how to rescue social capital but how to facilitate diversity and have a policy programme that 'flexibly incorporates immigrants' (Portes and Vickstrom, 2011, p. 467). Thus, even if social capital was declining, it might be possible to substantiate a positive interpretation of such a decline as an alternative interpretation to the negative hunkering down.
AddendumThe 2004 Eurobarometer data set used in this study - augmented with the various country-level factors such as the Mipex index - is available from the authors for further analysis. The analysis was undertaken in R and the syntax files are also available for replication and further development.
Notes1The Eurobarometer Surveys were established in 1973 and have since enjoyed between two and five annual survey waves. Each survey consists of approximately 1000 face-to-face interviews per Member State. This makes the Eurobarometer series despite some methodological imperfections (e.g. Sinnott, 1998) an authoritative source for undertaking comparative European research on various aspects of citizens' lives. Further information on the Eurobarometer series can be found here. The data sets are downloadable here
2The data set contains 29 country level units including separate entries for Germany East and Germany West and for Northern Ireland and the rest of the UK.
ReferencesALESINA, A. and La Ferrara, E. (2002). Who trust others?, Journal of Public Economics, 85, 207 234. [doi:2727(01)00084-6]
ANDERSON, B. (1991) Imagined communities: reflections on the origin and spread of nationalism. London, New York: Verso.
AXELROD, R. (1984). The evolution of cooperation. New York: Basic Books.
BELL, M, Chopin, I, Palmer, F (2006). Developing anti-discrimination law in Europe: the 25 EU member states compared. Brussels: European Commission.
BLACKBURN, S. (2003) Ethics: A very short introduction. Oxford: Oxford University Press.
BLAU, P. (1977). Inequality and Heterogeneity: a primitive theory of social structure. New York: Free Press.
BROWN, R. (1995). Prejudice. Oxford: Blackwell.
DELHEY, J. and Newton, K. (2005). Ethnic Diversity and Social Capital in Europe: Tests of Putnam's Thesis in European Countries
DE TOCQUEVILLE, A. (1835). Democracy in America. New York:Vintage.
DE TOCQUEVILLE, A. (1863). Democracy in America, Volume II. Cambridge: Sever and Francis. Translated by Henry Reeve, Esq. and Francis Bowen.
FIELDHOUSE, E and Cutts, D. (2010). Does diversity damage social capital? A comparative study of neighbourhood diversity and social capital in the US and Britain, Canadian Journal of Political Science / Canadian Journal of Political Science/Revue canadienne de science politique, 43:289-318.
FORD, R, Tilley, J. and Heath, A (2011) Land of My Fathers? Economic Development, Ethnic Division and Ethnic National Identity in 32 Countries, Sociological Research Online, 16(4) <http://www.socresonline.org.uk/16/4/8.html >. [doi:0.5153/sro.2508]
GESTHUIZEN, M.; van der Meer, T., Scheepers, P. (2009). Ethnic Diversity and Social Capital in Europe: Tests of Putnam's Thesis in European Countries, Scandinavian Political Studies, 32 (2) 121-142.
HELD, D. and McGrew, A. (2003) The global transformations reader: an introduction to the globalization debate, Cambridge: Blackwell Publishing.
HOOGHE, M., Reeskens, T., Stolle, D. and Trappers, A. (2009). Ethnic Diversity and Generalized Trust in Europe. A cross-national multi-level study. Comparative Political Studies, 42, 198-223. [doi:4008325286]
JOPPKE, C. (1999). Immigration and the Nation-State. The United States, Germany, and Great Britain. Oxford: Oxford University Press.
KALMIJN, M. (2009). Country Differences in the Effects of Divorce on Well-Being: The Role of Norms, Support, and Selectivity, European Sociological Review, advanced access online July 23.
LA PORTA, R. and Lopez-de-Silanes, F., Shleifer A. and Vishny, R.W. (1997). Trust in Large Organizations, American Economic Review, 87(2), 333-338.
LAURENCE, J. and Heath A.F. (2008). A Predictors of Community Cohesion: Multi-level Modelling of the 2005 Citizenship Survey. London: Department for Communities & Local Government.
MACRAILD, D. 2000. The Great Famine and Beyond: Irish Migrants in Britain in the Nineteenth and Twentieth Centuries. Dublin: Irish Academic Press.
MCPHERSON, M., Smith-Lovin, L. and Cook, J.M. (2001) Birds of a Feather: Homophily in Social Networks Annual Review of Sociology , Vol. 27, pp. 415-444 .
MEYERS, E. (2007). International Immigration Policy. A theoretical and comparative analysis. Basingstoke: Palgrave.
NATIONAL COUNCIL FOR VOLUNTARY ORGANISATIONS (NCVO) (2012), Third Sector Foresight: Attitudes towards immigrants, available online, viewed on Feb 15 2012, <http://www.3s4.org.uk/drivers/attitudes-towards-immigrants#fn3>.
POINTER, G. (2005). Focus on people and Migration : The UK's major urban areas. UK: National Statistics.
PUTNAM, R.D. (2007). "E Pluribus Unum". Diversity and Community in the Twenty-First Century, Scandinavian Political Studies, 30(2), 137-174.
PUTNAM, R.D. (2000). Bowling Alone: The Collapse and Revival of American Community. New York: Simon & Schuster.
RAHN, WM, KS Yoon and K Loflin. 2003. Geographies of Trust: explaining inter-community variation in General Social Trust using hierarchical linear modelling (HLM). Working Paper.
SALT, J. (2005). Current Trends in International Migration in Europe. Strasbourg: Council of Europe Publishing.
SNIJDER T. and Bosker, R. (1999), Multilevel Analysis. London: Sage.
SOMMERLAD, H and Sanderson, P (1998) Gender, Choice and Commitment: Women Solicitors in England and Wales and the Struggle for Equal Status. Dartmouth: Ashgate.
STOLLE, D. (2002). Trusting Strangers: The concept of generalised trust in perspective. Φsterreichische Zeitschrift fόr Politikwissenschaft, 7(4), 397-412.
STOLLE, D. (2001) Clubs and Congregations: the Benefits of Joining an Association, pp. 202 244, in Trust in Society, Karen S Cook, editor. New York: Russell Sage Foundation.
USLANER, E.M. (2002). The moral Foundations of Trust. Cambridge: Cambridge University Press.
WEBER, Max.  1958. The Protestant Ethic and the Spirit of Capitalism, translated by Talcott Parsons, with a foreword by R. H. Tawney. New York: Charles Scribner's Sons.
WHITELEY, P.F. (1999). The Origins of Social Capital, pp. 25 44, in Social Capital and European Democracy, Jan van Deth, Marco Maraffi, Kenneth Newton and Paul Whiteley, editors. Oxford: Routledge.
WILLMOTT, P. (1987). Friendship networks and social support. London: Policy Studies Institute.
ZUCKER, L.G. (1986). Production of Trust: Institutional Sources of Economic Structure, 1984-1920. Research in Organizational Behavior, 8, 53-111.