Final Report Summary - GEITONIES (Generating interethnic tolerance and neighborhood integration in European urban spaces)
Project context and objectives:
In Europe multiculturalism and integration have become hotly debated issues and are often used as banners by opposing camps. Integration and even assimilation are concepts that bear increasing weight, as concerns about the cohesion of society have grown steadily due to global processes and dramatic events clothed in symbolism that stress global cultural differences. Long-established policies of dealing with immigrant populations and of 'managing' diversity and difference are currently at stake. The riots that spiralled in November 2006 across the suburbs of Paris and other cities in France gave birth to criticisms of existing integration policies and arguments about the 'failure' of the French republican model that remained blind to inequality and difference (Begag 2007). While, in the United Kingdom (UK), part of the reaction to the French experience highlighted the supremacy of the British culturally-sensitive model, it is precisely this model of multiculturalism that is currently being challenged (e.g. Goodhart 2004, Phillips 2005). The question of integration pressingly returns in the aftermath of the race disturbances in towns of the Midlands in the summer of 2001 and more bitterly, of the London transport bombings by 'home-grown' terrorists in July 2005. Just across the sea, the other European example of 'successful' multiculturalism, the Netherlands, is also being redefined (Uitermark et al. 2005).
In response to these events across Europe the ways in which difference is accommodated are being challenged and scrutinised. In general terms we are seeing what has been termed as a 'backlash against multiculturalism' (Grillo 2005) where the multicultural model has been deemed as creating a gulf of separation between communities in spatial and socio-cultural terms through its particular focus on group based rights and service provision. The discourse on the failure of multicultural models highlights the erroneous emphasis of collective groups rights over individual rights, the fact that it predetermined fixed and unchanging categories of 'difference' and emphasised identity and racial equality at the expense of other differentials such as class and inequalities (e.g. Papastergiadis 2000). These shortcomings through prioritising cultural recognition are seen as having resulted in the failure of migrant groups to meet their responsibility to integrate and indeed migrants/ethnic minority groups have been accused of segregating themselves as a means to preserve their ethnic identities and resist integration (Phillips 2006). This has been represented officially as an inherent failure of multiculturalism and has given rise to concerns centred on social identity based on the fact that ethnic or religious identities are assumed to be in opposition to a homogeneous national identity and the belief that values at odds with dominant moral values are perceived as threatening national identity (Back et al. 2002, Robinson 2005). Subsequently, these fears have resulted in a resurgence of assimilationist discourse linked with projects to promote social cohesion and integration. There appears to be a trend more generally across Europe, where a discourse of failing multiculturalism is accompanied by a de-centralised responsibility for integration, the onus falling upon the responsibility of the individual (Mitchell 2004). Some have construed this as a new technology of knowledge/power in neoliberal regimes of governance that constitutes the individual as free-thinking whereas in reality there is little choice as those who 'opt-out' are cast as individuals not willing to participate in wider social life and are liable to be excluded from citisenship rights (Mitchell 2004:645). At the same time, nevertheless, earlier and ongoing processes of settlement generate grounded experiences of everyday diversity, whereby a 'spontaneous' convivial culture may start to flourish, particularly in urban areas in Britain and elsewhere (Gilroy 2004).
Similar debates and policy shifts are taking place at present across the EU, not only in the old host countries of the North, but also in the new immigration destinations of Southern Europe and the emerging reception countries of the East. For instance, the transition of Southern Europe to a target area for immigration has taken place gradually over the past three to four decades (Fonseca et al. 2002) and to a large extent the 'new cultural encounters' (King, 2001) manifested there have been significantly marked by irregularity and stigmatisation of the migrants (Anthias and Lazaridis 1999). At the same time, the new member states of Central and Eastern Europe emerge rapidly as new migrant destinations (e.g. Salt 2006, Drbohlav, Hárs and Grabowska-Lusinska 2009). The integration debate is still limited in these countries, due to the recent character of migration and to the fact that it the numbers of migrants are still few. However, the recent developments show that policy-makers have started to identify problems that may develop if the issue is ignored (Iglicka et al. 2005).
Do the concepts of multiculturalism and integration really depict antipodal and incompatible trends? This depends of course on how they are used; in ideological discourse they tend to adopt multi-layered, versatile and ambivalent meanings. The debates surrounding the presence of Muslims in Europe, to give an example, are centred around issues of religious diversity and tolerance of difference, while ultimately they reflect global politics on terrorism (Cesari 2005, Modood et al. 2006; Levey and Modood 2009). Despite the value of these debates in political terms, they do point to a significant lack of scientific knowledge on how cultural interactions actually develop between individuals and groups of different ethnic background, in contexts were they meet on a daily basis. Also, we lack data on how these interactions are related to an interchange of skills and attitudes (acculturation) and how they bear on the social status of the individuals and groups engaged in contact.
This research aims to enhance and improve our knowledge in this respect. It introduces a new perspective to understand these old issues. It reframes the old questions of integration rigorously into a relational format, taking away the spotlight from separate categories or groups of actors and directing it to the systemic whole, the integer, concentrating on the development of interdependencies between individuals and groups in certain territorial spaces. Obviously, much of the issues with respect to integration and multiculturalism are inevitably linked to urban environments (Ray 2003). The majority of immigrants and their descendants are concentrated in European cities, with cities attracting the majority of newcomers. As urban Europe continues to be the locus of various contradictions (economic, social, spatial, cultural etc.), urban spaces first and foremost provide the arena in which interethnic relations are unfolding. The ever increasing diversity of European metropolises, the persisting social and spatial inequalities in urban areas and the shifting paradigms of welfare provision, precisely call for a redefinition of 'host-stranger' relations in cities (Alexander 2003) and set the agenda for multicultural approaches to neighbourhood governance as a means for promoting social cohesion (Allen and Cars 2001; Blockland and van Eijk 2010). Consequently, we focussed on the local level, selecting social areas (neighbourhoods) in European metropolises as the key context for our research, in which we aim to understand the representations, daily practices and interactions that affect relations between social and ethnic groups. We included the complete multi-ethnic population of these areas in our inquiry. In this way, we addressed the issue of social cohesion in urban spaces in the most direct way possible.
The association between interacting groups of people that share the same residential areas and use the same public spaces makes the notion of place central in our research. A place is a physical space where social practices and group relations materialise, where neighbourhood relations occur and give meaning and identity both to the material elements of the area and to the people that inhabit it (Tuan 1977).
To elaborate the principal research objective, further research questions central to the project relate to: the different modes of interethnic coexistence that can be identified in European urban settings; local conditions causal to the development of these different modes; the impact of levels of ethnic concentration/residential segregation on these; the impact of behavioural patterns and initiatives of residents; resources (and the effectiveness of these) on a local level to facilitate inter-group contact; daily use of space and its impact on intercultural communication; an assessment of the most effective spaces to facilitate interethnic communication and contact; the management of diversity between groups sharing a residential neighbourhood; neighbourhood factors conducive to fostering an environment of understanding and tolerance; and, finally, the role of neighbourhood stigmatisation on tolerance and xenophobia.
The perspective adopted in this project is intercultural. Departing from the cultural diversity of contemporary metropolises and assuming that culture is dynamic, the construction of spaces of intercultural dialogue is crucial to re-enforce cohesion and avoid feelings of exclusion and the outburst of conflict. Sandercock (2004) argues that the core of interculturalism implicitly refers to two rights: the right to difference and the right to the city. In this view, people have, regardless of their ethnic background, equal rights in the shared public space and an equal capacity of full participation in the public affairs of the urban living environment, not just de jure (through legal devices and access to political participation) but also de facto (in the practical situations of everyday life).
Project results:
Interethnic contact, between immigrants and natives, is embedded in unequal power relations, class positions and cultural dispositions. The study of interethnic relations may reflect the core sociological problem of 'the formation, stabilisation and change of social relations' (Esser 1986: 30). On-going immigration and increasing diversity in European societies amidst the economic and political pitfalls in the context of globalisation, restructuring and crisis, have given rise to growing concerns over interethnic relations as a prerequisite for social cohesion.
One of the key approaches that the relevant literature draws on is the so-called contact hypothesis, which assumes that as people of different backgrounds come into contact with each other, prejudice and thus social distance decreases. Two criteria are relevant here. One relates to the size of the minority group: the larger this is the larger the threat it supposedly poses, thus the more intense the competition will be. The other has to do with the wider economic conditions: when these are unfavourable resources become scarcer and therefore the perceived threat grows and competition intensifies.
Putnam (2007) has recently suggested a third possibility, which he terms constrict theory. Although he is actually concerned with the effects of diversity on societal trust and social capital, as has also been explored using GEITONIES data, his article departs from the premises set by the above approaches and thus became a reference for recent studies on interethnic relations (e.g. Vervoort et al. 2010). His key argument that diversity impacts negatively on social solidarity (in the US context), came at a time of high politicisation of immigration issues and has stimulated much debate, ranging from criticism (e.g. Giddens 2007) to replications of his study in Europe (Lancee and Dronkers 2011). Despite our objections to both the theoretical foundations and methodological tools, constrict theory suggests that diversity (read interethnic contact) may actually reduce both in-group and out-group solidarity, i.e. deteriorate not only interethnic relations but undermine social cohesion at large.
What is often downplayed in studies employing 'contact' or 'conflict' theory is a failure to account for the social context which influences individual and group relations and patterns of behaviour in various ways. On the other hand, there is also much debate on what 'context' may involve and how this relates to individual attributes and attitudes. In his study of interethnic relations among migrant workers in West German cities, Esser (1986: 36) distinguished between two categories of context. One he termed the '(territorially predetermined) macro-structure, in particular the presence of persons of the same or of the other group and the behaviour... of those in each of the groups'. This may partly echo the first of the conditions in Blalock's competition theory about the size of the minority group. Esser's second category is what he calls the primary milieu, i.e. the normative climate prevailing in the relevant reference environment. Again, partly at least, we may find parallels to what others refer to as 'third party' influences, pointing to the context of socialisation: others (the family, the ethnic community, the state, etc.) may directly or indirectly encourage or discourage contact (Kalmijn 1998). In addition, context may also refer to other characteristics of the immediate (e.g. neighbourhood) or broader (e.g. city, nation state, transnational space) social environment. Socioeconomic and socio-spatial characteristics, are often downplayed even though they often play a role (Fong and Isajiw 2000; Vervoort et al. 2010).
Moreover, Esser also reminds us of the relationship between residential segregation and ethnic segmentation, echoing the Chicago ecological school's accounts equating social to spatial distance. It is actually there whereby the study of interethnic contact may find its origins. The important issue here is the introduction of a temporal dimension in the development of interethnic relations. The following text will address, analyse and contest many of these theories and empirical results through various thematic studies and statistical analyses using data collected between 2009 and 2010 in the ambit of the GEITONIES project.
Interethnic relations
Prior to any statistical analysis of the data a preliminary descriptive analysis of the nature and frequency of interethnic relations in the 18 neighbourhoods studied across in six European cities is fundamental. This will enable us to begin reflecting on the different modes of interethnic coexistence that can be identified in European urban settings, a core research objective.
Beginning with the respondents' global social networks no significant differences were found between migrants and natives in terms of network size. As expected, relationships in the other contact fields are much more restricted regardless of background.
As far as the composition of social networks is concerned, a significant component of both migrants' and natives' global network is comprised of relatives; the social network of approximately one out of five respondents is comprised exclusively of family members. Concerning interethnic relations, significant differences can be observed between migrants and natives. While respondents overall tend to socialise mostly with people of the same origin, migrants appear to have significantly more contacts of different ethnic backgrounds than natives. Interethnic relations are more common in the 'free time' contact field, whereas relatives are overrepresented in the fields of 'help' and 'advice and confidentiality' for both natives and migrants given the nature of the contact. Taking this into account, as well as the fact that relatives form an equal share of migrants' and natives' social networks, one can conclude that relatives do not account at all for the substantial difference found between migrants and natives in terms of interethnic relations. This difference is probably explained to a large extent by the difference in size of the immigrant and native populations in the cities where research was conducted.
The findings concerning the respondents' global social network are confirmed when we narrow down our analysis to their close social circle. Again, one cannot observe significant differences in network size. In terms of the ethnic composition of close social networks, migrants were reported to have a mean of 1.19 interethnic contacts in their close social network while for natives the mean number of interethnic relations was as low as 0.12.
Interethnic relations among migrant and native residents' close social networks were found to follow very similar patterns to that of their overall social networks. Thus, we cannot argue that the higher frequency of interethnic relations among migrants is necessarily an indication of 'openness' to and intimacy with, the 'other' in general, but rather of their interaction with the host country's native population and thus a degree of assimilation in their patterns of socialisation. Following Blau's (1977) thesis mentioned previously, this type of relationships could partly be seen as a 'necessity' for immigrants, given that natives are obviously the majority (dominant) group. On the other hand, it also entails a dynamic element, since close relationships (such as those examined here) are beyond inevitable native-migrant interaction and extend to the level of intimacy, which is built over time, as immigrants become more and more settled in the host country and in the specific localities where they live (Labrianidis et al. 2011).
In order to demonstrate the relationship between the two elements of the social network measured in the survey - the global/overall network and the close network - a further analysis has been conducted specifying different types of interethnic friendships. Here, all respondents who indicated that they have a friend from a different origin in either their overall social network or among their closest friends (out of a possible eight) are included. Four subgroups are considered; respondents with no interethnic friendships, those with interethnic friendships limited to their overall social network, those with friends from another origin only among their closest friends and, lastly, those with interethnic friendships in both their overall network and closet circle of friends.
In concordance with the data presented thus far, almost 80 % of natives do not have interethnic friendships in either their close or overall social network, whilst this is true for less than 40 % of migrants. For those natives that do have interethnic friendships, the largest part is represented in the overall social network and not in the close social network. On the other hand, migrants with interethnic relations tend to have friends both among their closest circle of friends and in their broader network. At the city level for both migrants and natives the old migration cities of Vienna and Rotterdam are distinct as having the highest degree of interethnic friendships. Whilst the newer destinations have extremely low levels, this is especially true in Bilbao and Warsaw, though for example, Thessaloniki is intermediate due to the specific characteristics of mainly Soviet Greek migrants there and the cultural background and historical links that they share with the native population (Fonseca et al. 2011).
The contact continuum and modes of coexistence: the relationship between weak and strong ties
Interethnic contacts in the neighbourhood
On the contact continuum one of the weakest forms of contact maybe engaging in small talk. The informal components of the so-called 'soft-infrastructure' of any given locality are thought to include various forms of interaction that may have a bearing on the social cohesion of the area. Among migrants, regardless of network type (mixed or homogenous), in general terms there is a high degree of superficial exchange. When this type of superficial contact is considered the remainder of migrants with no form of contact with 'the other' is minimal. The picture, nevertheless, is altogether altered when the native sample is considered. Indeed, 44.8 % of natives who do not have interethnic friendships have exchanged small talk with migrants in the neighbourhood. However, this almost doubles for the smaller group of natives that have interethnic friendships. Another important point to note is the high proportion of natives with no interethnic relations who have not engaged in small talk at the local level with the other, representing 43 % of the total population. In both groups, slightly lower levels of small talk can be observed among those who only have friends in the close social network (Fonseca et al. 2011).
With regards to the causality of this relationship, Fonseca et al. (2011) analysing GEITONIES data using multilevel regression analysis found that the number of persons of natives and migrants of 'other' origin than co-ethnics in close social network is a highly significant predicator of superficial contacts (exchanging small talk) at the local level indicating that intimate relations influence daily interethnic contacts in the neighbourhood. Interestingly, however, supporting Allport's contact theory, Kohlbacher et al. (2011), testing the relationship in the opposite direction, found in their analysis of GEITONIES data that superficial contacts in the neighbourhood did not reduce anti-immigrant views or attitudes, whilst strong ties like visiting at home or intimate friendships did.
When we move along the contact continuum to a more intimate form of interaction in the local setting, visiting or welcoming neighbours at home, as one would expect the frequency of interaction decreases. Furthermore, we may reasonably assume that visiting at home may result in more affective bonds given its more intimate nature. At the local level inter-group home visits are considerably higher among the groups of natives and migrants who have interethnic friendships. Still, this difference is clearly mediated by the background of the respondents, being much more pertinent for native respondents. Welcoming or visiting migrants at home is clearly related to having interethnic relations or not for this group. Among the migrant sample the difference in home visiting among those who have and do not have interethnic relations is considerably smaller (Fonseca et al. 2011). This permits us to tentatively suggest that the neighbourhood might be a more important place for natives in stimulating or developing interethnic bonds than for migrants. Again, however, care must be taken with interpretation and the generalisation of such claims due to the very small numbers of natives with interethnic relations.
Interethnic marriage
Given the low frequency of interethnic marriage, it is unsurprising that in general terms a larger share of respondents who have interethnic friendships do not or have not had interethnic partners in the past. However, when one compares the existence or not of interethnic friendships the share of interethnic marriages is substantially higher among those who have interethnic friendships. Around one third of natives with interethnic friendship has or has had interethnic partners compared with only 6.8 % of those with in-group networks. Likewise, only 12 % of migrants without interethnic friendships have ever had an interethnic partner compared to over 40 % who have mixed friendship networks (Fonseca et al. 2011). For natives, this very intimate form of interaction appears to have particular bearing for the development of close friendships. Fonseca et al. (2011) also found a statistical relationship between interethnic marriages, as an explanatory factor influencing the probability to have intimate interethnic friendships. Moreover, 66.1 % of migrants in mixed marriages/partnerships actually developed their intimate interethnic friendships after the initiation of the partnership. A similar pattern also stands among natives: 33 % of those in a mixed marriage have intimate interethnic friendships compared to just 6.5 % among those whose partner is of the same background. In 73.1 % of the cases, intimate interethnic contacts were developed after meeting their partners (Labrianidis et al. 2011).
Work colleagues
Working in a mixed environment provides opportunities for encounter and may serve to enhance inter-group familiarity. Indeed, it is important for those who have interethnic friendships, regardless of background group. 87 % of migrants with interethnic contacts in both their close and overall social network work in a mixed environment, compared with 53 % of those with no interethnic relations. The equivalent percentages for natives are 67 and 30, respectively (Fonseca et al. 2011).
Micro-contexts and relational circumstances of interethnic intimacy
The relationship among the various dimensions of contact along the contact continuum is important to enhance our understanding of how relationships develop among individuals from different ethnic groups. It represents a key research objective of the current project, namely which factors pertaining to the behavioural patterns and initiatives of the residents (in terms of social interaction) can be pointed out as being causal to the development of modes of interethnic coexistence.
To state it more directly, one may ask to what extent does the neighbourhood matter in the composition of the social networks of the respondents? Certainly, how this differs for groups of migrants and natives both with and without interethnic relations is also a central point of interest. Due to the way the questionnaire was structured we cannot analyse the overall and close social networks together here. Thus, we will commence with the overall social network and move to the close network. With regards to the overall social network, two main tendencies can be observed. Firstly, it is apparent that the overall social network for spending free time is concentrated outside of the neighbourhood of residence for the majority of natives, (56.2 % have none or just a few of their contacts with whom they spend free time there) and the largest proportion of migrants (47.4 %) (Fonseca et al. 2011).
A similar pattern can be noted in the other contact fields of sharing confidences and advice and exchanging substantial help. However, the neighbourhood is a more important place as the locus of the overall social network for those who have completely mono-ethnic networks whether they are of migrant or of native background. This is holds true to a greater extent for migrants than for natives. Indeed, slightly over 35 % of migrants with no interethnic contacts indicated that the vast majority of their social network lives in the neighbourhood of residence compared with fewer than 20 % of those with an ethnically mixed social network. Thus, to certain extent, when the composition of the social network is in-group it is concentrated to a greater degree in the neighbourhood (Fonseca et al. 2011). Furthermore, the literature on neighbourhood effects suggests that the larger the share of contacts concentrated in the neighbourhood of any given individual the more isolated they will be from mainstream society and the more sensitive they will be to the compositional effects of the neighbourhood. Indeed, this theoretical perspective was tested and confirmed using GEITONIES data by Miltenburg and Lindo (2011) as presented later in this summary.
Briefly, it is important to note that those respondents who have interethnic relations not only have a more spatially diverse network, but also have more mixed networks in terms of gender as well as less family-centred networks.
Here we turn to the close social network. Instead of examining the share of people having interethnic contacts, we focus on the actual contacts of migrants and the micropublics (e.g. Amin 2003), where interethnic encounters, which later developed into close relationships, took place (Labrianidis et al. 2011).
Regarding the circumstances of the original encounter, out of the total interethnic contacts of migrants, the vast majority (38 %) were first met as colleagues or fellow students, proportionally more than among same-origin contacts, which are dominated by relations of kin. The picture is similar in the case of natives. Moreover, there are interesting variations by city suggesting that the micro contexts of encounter may also depend on the broader urban and national contexts. Indeed, interethnic encounter as colleagues appears to be far more important in Warsaw than in Rotterdam, where relatives are more important (owning to the higher instance of mixed marriages among migrants). Furthermore, while the share of interethnic contacts met as neighbours is slightly less than that of same-origin contacts, the neighbourhood appears to be a meeting place in Southern European cities. By contrast, while their overall shares are very low, interethnic contact is more frequent in collective organisations in northern European cities (possibly reflecting the development of civil society) (Labrianidis et al. 2011).
The neighbourhood itself emerges as the second most important 'micropublic' fostering the development of interethnic relations. It appears to be a slightly more important place of socialisation for migrants than natives (seen also for global social networks). However, in actual fact less than one out of three (29.6 %) of migrants' total interethnic contacts were developed in the neighbourhood of residence, while a similar share (30.6 %) from the total of 815 immigrants who have close interethnic friends met at least one of them in the neighbourhood (Labrianidis et al. 2011). When we consider all contacts, whether interethnic or not, of the group of migrants that have at least one interethnic contact, meeting as relatives takes joint first place and the neighbourhood slips into third place (Fonseca et al. 2011).
Interestingly the distribution of immigrants who have met at least one of their interethnic relations across the 18 neighbourhoods does not seem to follow the same pattern as the distribution of the immigrants who have developed interethnic relations in general. Paradoxically the ranking of the neighbourhood seems to be, to a certain extent, reversed. The two neighbourhoods with the highest share of interethnic relations developed in the neighbourhood, Peraia and Costa de Caparica, are among the ones with the lowest overall shares of interethnic relations. At the same time, the majority of interethnic relations in the neighbourhoods with the highest shares have not developed there but outside them. In other words, the neighbourhoods that score high in terms of their immigrants' interethnic relations do not seem to facilitate the development of interethnic relations, but simply concentrate an immigrant population which has developed many interethnic contacts (with the exception of Hoogvliet-Noord in Rotterdam). A first general conclusion that may be extracted from this observation is that the neighbourhood does not appear to be the major determinant in the development of interethnic relations for migrant residents (Labrianidis et al. 2011).
Another interesting aspect is the actual place where the initial encounter took place. The key role of the workplace as a contact point is confirmed, concerning more than a quarter of migrants' interethnic contacts compared to just 17.5 % of same-origin contacts and is even more pronounced in the countries of recent migration. Similarly, confirmed is the limited share of interethnic contacts first met in a club or association, which is more important in Rotterdam and Vienna. On the other hand, open public spaces appear to be slightly more important meeting places for interethnic relations, especially in Southern European cities (where climate and culture favour socialisation in open spaces). Finally, one can observe the importance of the private sphere of the home (the respondents' own home or the home of relatives or friends) as a meeting place in a quarter of the cases, which exceeds 30 % in Rotterdam and Vienna (to an extent reflecting the higher proportions of relatives among interethnic contacts, as well as the limited encounters in the public space), but also Thessaloniki (Labrianidis et al. 2011). For all respondents regardless of the background of friends the home is the most important place of meeting followed by the place of work/study (Fonseca et al. 2011).
Neighbourhood and compositional effects: The predictors of interethnic relations
Whilst understanding the types or different dimensions of social interactions as well as their frequency and how and where they develop is crucial, it is important to understand the predictors of interethnic interaction. A crucial objective at the core of the GEITONIES project relates to the role of the local context as well as compositional effects in the process of developing interethnic contacts, expressed in the following research question. What neighbourhood and individual factors (social network characteristics, religious affiliations, citisenship, gender, unemployment status, etc) are conducive to fostering an environment of understanding and tolerance?
In terms of individual factors a simple reading of the data uncovered the existence of particular characteristics that seem to significantly correlate with the tendency of migrants to include interethnic (and mostly native) contacts in their close social networks. These factors that emerged are directly or indirectly related to the time dimension, since - for example - more settled immigrants present higher shares of interethnic contacts, compared to recently-arrived ones, while at the same time, second generation immigrants are also more likely to develop close social networks comprising of native contacts than the first generation (90 % of the former have at least one close friend of a different ethnic background compared to 44.8 % of the latter). Also, time is relevant in respect to the migrants' life course, whereby specific circumstances (mixed marriages, birth of children) are also associated with the existence (or not) of interethnic contacts in their social networks. The proportion of migrants who have at least one intimate interethnic contact is higher (63 %) among those who do not have any children compared to parents (47.4 %). As such, the importance of time is underlined in the development of interethnic relations, as an aspect of the wider process of migrants incorporation, which is something often ignored or downplayed in the relevant literature. Further proof of the relevance of the process of migrants' incorporation for the development of interethnic relations came from our examination of factors relating to aspects of their pathways of settlement in the host society. Some of which, like legal status and language proficiency, also seem to be time-dependent, while others simply accounted for the ways individual characteristics, such as ethnicity, religion (fewer shares of migrants belonging to a religion different than the dominant religion among the native sample have interethnic relations compared to those who share the host countries religion(s) or are not religious at all) education and profession (executive and professional migrants had the highest shares of close interethnic friends) , may relate to the wider political and socioeconomic context. Safer and more secured legal status, for instance, is directly correlated with higher probabilities to develop interethnic relations. In most of the cases, of course, these statuses are likely to denote longer periods of permanent residence in the hosting country, while a similar finding stands for linguistic skills (Labrianidis et al. 2011).
Following on from the descriptive analysis at the individual level, a more sophisticated statistical analysis was conducted to explore both the neighbourhood and compositional effects conducive to the development of interethnic relations (Fonseca et al. 2011). The methodology deemed most appropriate to respond to the aforementioned research question was multilevel linear regression modelling, a type of regression analysis in which independent variables from individual and aggregate levels can be included in the same model. Understanding and tolerance were operationalised through social relations measured at both the neighbourhood level and beyond. Two analytical dimensions were taken into consideration: the interethnic dimension of daily contacts in the neighbourhood and the interethnic dimension of the respondents' close contacts. Thus two models were estimated including the following two dependent variables: the share of small talk exchanged in the neighbourhood of residence with people of a different origin (over the three months prior to the survey) and the number of close friends (out of a possible eight) of a different origin.
Two general types of individual-level independent variables were included in the models (not all of these were included in each). The first are variables related to characteristics of close contacts of the respondent: number of close contacts; share of close contacts with educational levels higher than those of the respondent; number of close persons of different origin; share of neighbours among close contacts; share of relatives among close contacts. The second are those related to the individual characteristics of the respondent: long-term residence in the neighbourhood; having children under 16; being in a mixed marriage; education level; gender; age; and religious affiliation.
At the neighbourhood level the following variables were used: share of second-generation migrants; social class; and the index of diversity. Including the diversity index at the neighbourhood level also enables us to respond to the following research question: do levels of ethnic concentration and residential segregation impact on the development of interethnic relations? Given the differences at the city level dummy variables for cities were included in the models as control variables. The main empirical results can be summarised as follows.
The first point is the differing importance of the neighbourhood level in explaining the variability in the propensity to establish interethnic relations among migrants and natives. For superficial interethnic relations, the role of the neighbourhood context is comparable for migrants and natives, though not very high, indeed, less than 10 % of the total variation of their intimate interethnic relations is attributable to differences between neighbourhoods. The analysis of close interethnic contacts resulted somewhat differently. In fact, in the reference model for migrants, 15.6 % of the total variation of their intimate interethnic relations is attributable to differences between neighbourhoods. For the native population there is no significant neighbourhood effect in explaining the variation of their close contacts with migrants. Given the low percentage of natives with no interethnic relations it is unsurprising that the variability between neighbourhoods is much lower. However, the influence of national and urban contexts in the variation of superficial interethnic contacts in the area of residence should be stressed, especially for natives (the intra class correlation coefficient for the empty model - without controlling for the city level - for natives is 40.3 % whereas for migrants it is 18.3 %). Thus, the macro-structural characteristics, relating to economic, political and ideological factors or to the migratory context, of each city have a more important effect in comparison with local factors.
Among the neighbourhood characteristics tested in the models, the share of second-generation migrants among migrants in the neighbourhood was the only characteristic at the aggregate level found to shape both superficial and intimate interethnic relations among migrants. In general, the higher the share of second-generation migrants in the neighbourhood the greater the share and intensity of interethnic relations is. Such an outcome suggests that it is the process of migrants' settlement in the neighbourhood over time that matters for the development of interethnic relations.
Among natives, neighbourhood characteristics were only significant in the model for exchanging small talk. The socioeconomic level of the neighbourhoods is significantly related with the exchange of small talk between autochthonous residents and migrants, indicating that interethnic communication is higher in the urban areas with lower socioeconomic status. This may be a result of the over-representation of minority ethnic and immigrant groups in poor neighbourhoods. This is supported further by the fact that the diversity index for each neighbourhood has a positive and statistically significant coefficient, meaning that casual social interaction at the neighbourhood level is more prevalent in those areas with larger migrant communities.
The moderate role of the neighbourhood, as a determinant of the propensity of its inhabitants to develop interethnic contacts, be it close or superficial contacts was also confirmed by the fact that individual predictors have a higher explanatory power than predictors at the neighbourhood level. Beginning with migrants, with respect to individual factors that explain variability in the propensity to engage in superficial interethnic contact in the neighbourhood, the number of interethnic relations in the close social network proves to be a highly significant predicator. This indicates that intimate relations or a positive knowledge of the 'other' influence daily interethnic contacts in the neighbourhood (Friedkin 2004; Dixon 2006). The more concentrated the social network is in the neighbourhood the lower the levels of interaction with the 'other' are. Higher levels of education correspond to lower relative frequencies of superficial interethnic contact in the neighbourhood. Gender and religion are also important predictors in the sense that men and those with a religious affiliation interact more in the public domain out with their ethnic group.
Important individual level predictors for natives exchanging small talk with migrants are the number of migrants in the individual social network, age and length of residence in the neighbourhood. Thus, those who have lived in the neighbourhood for longer periods of time interact with migrants to a greater extent. Yet, age has the opposite effect and functions as an inhibitor to daily interactions with neighbours from other origins.
Important predictors for the number of close interethnic relations among migrants include some characteristics of the close social network. First of all, the higher the number of close contacts in general a migrant has, the more likely he/she also is to develop close interethnic contacts. Moreover, migrants who tend to engage in relations with persons with a higher educational level than themselves are also more likely to develop close interethnic relations. To the contrary, the higher proportion of neighbours and relatives among close friends of the migrant, the lower number of interethnic close contacts he/she has. Selected socio-demographic characteristics also have important explanatory power, namely long-term residence in the neighbourhood, being in a mixed marriage and having children under age of 16, which presents a negative coefficient. Levels of educational achievement are also significant with better educated migrants being more likely to develop close interethnic contacts.
In general, only three predictors were found to be significant for natives: the number of close contacts, the share of relatives and the 'mixed marriage' variable. They influence the propensity of natives to develop close interethnic contacts in the same direction observed for migrants.
The role of contacts in the formation of attitudes towards immigrants
Moving to the level of more general attitudes, Kohlbacher, Reeger and Schnell (2011) conducted a specific analysis of GEITONIES data on the relationship between social interactions and attitudes towards immigrants, with the assumption that contacts have an impact on attitudes. Daily social interactions in the neighbourhood context are referred to as well as close friends as the relevant plane of reference in order to investigate how contacts in the local context and on a general level are shaping and modifying views on immigrants. Of course, contacts of different qualities, like in everyday life or in the closer circle of friends, are not the only factor influencing the way people think about immigrants and immigration. The data set we have at hand provides the opportunity to analyse the role of contacts in shaping attitudes towards immigrants in comparison with other important determinants like age, education, trust, national identity or religious affiliation.
The leading question was, whether interethnic contacts (or, in Allports terms: true acquaintances) still matter once we consider all important factors at once. Our plane of departure was the theoretical foundation of 'inter group contact theory' as proposed by Allport in 1954. The basic assumption of this theoretical argument is that interethnic contacts, under appropriate conditions, are one of the most effective ways to reduce prejudice between majority and minority group members. Thus, personal contacts are also one of the best ways to improve relations among groups that are experiencing conflict. Previous studies on anti-immigrant attitudes were mostly based on national data. However, we have the opportunity to focus on the local level. Following Allport's differentiation between the impacts of 'true acquaintance' on the one hand (close relation, friendship) and 'casual acquaintance' on the other, we further argued that it is true acquaintance which matters - especially at the local level in which attitudes are formed through daily face to face interactions. Our data provided us with the opportunity to empirically differentiate between true acquaintances and casual contacts which to date has rarely been done. We tested this assumption empirically, first with a descriptive analysis followed by the modelling of attitudes using multivariate analysis, by examining the role of interethnic contacts in addition to a set of predictors that have been identified in the literature as important. This was conducted for fifteen neighbourhoods within the cities Bilbao, Thessaloniki, Vienna, Rotterdam and Lisbon.
The dependent variable, attitudes towards immigrants, is measured using an index (mean score) constructed from two survey items. The two questions are: 'It is good for the economy that people from other countries come to live here' and 'In the future, the proportion of immigrants will become a threat to society'. The items were answered on a likert scale ranging from 'agree strongly' to 'disagree strongly'. Both survey questions have long been used to measure immigration-related attitudes (McLaren 2003, Quillian 1995, Scheepers et al. 2002, Coenders et al. 2003, Schneider 2008, Semyonov and Glikman 2009).
Our main independent variables of interest, including various types of interethnic contacts, are included as a set of dummy variables in the analysis. 'Interethnic small talks' and 'Interethnic mutual visits' indicate whether individuals experienced these types of interethnic contacts during the last three months within the neighbourhood. 'Close interethnic relations' indicate if at least one person in the closest circle of friends is of another ethnic origin All three variables serve as measures for 'true acquaintances' (in Allport's sense). Next to interethnic relations, several additional factors that have been found to affect attitudes towards immigrants are considered in the analysis. Trust is a continuous variable derived from two survey items: 'Do you think that most people try to take advantage of you if they got the chance, or would they try to be fair?' and 'Would you say that most of the time people try to be helpful or that they are mostly looking out for themselves?' Religion is a dummy variable specifying whether respondents considered themselves as belonging to a particular religion (1=Yes, 0=No). Other individual level control variables were included in the analysis: education, social class, age and gender. Socioeconomic attainment is derived from information on the labour market participation of the survey respondents. Following previous research (Phalet et al. 2007) we distinguish economically inactive or unemployed from (self-) employed persons. The latter are then subdivided into broad occupational classes by combining categories into salariat (professional and managerial occupations), intermediate (skilled non-manual and manual occupations as well as self-employed) and working class (semi- and unskilled manual occupations). Age is entered as a continuous variable while gender is a dummy variable with women being the reference category.
Finally, examining whether the total size of the migrant population and the degree of ethnic diversity within neighbourhoods matters for attitudes towards immigration, we include two contextual variables at the neighbourhood level. Firstly, we include the 'share of migrants' (expressed as percentages). Secondly, in order to estimate ethnic diversity within neighbourhoods we construct a Herfindahl index of diversity. Two additional variables enter our analysis since they have been identified as highly relevant when explaining attitudes towards immigrants: Length of residence in the survey country and identification with the country of origin.
A further major contribution of our analysis is that we explored the above expected impact separately for natives and immigrants in 15 European urban spaces of six metropolises; an approach that has not been done before, taking into account the local perspective, too. The main empirical findings can be summarised as follows:
1. Overall, we find mixed evidence of the importance of interethnic contacts in the formation of attitudes towards immigrants. We find that true acquaintances reduce prejudice for natives but not for migrants.
2. With respect to casual contacts we explored the same pattern for natives as for migrants: The presence of migrants in the neighbourhood increases anti-immigrant attitudes while at the same time the degree of diversity decreases anti-immigrant attitudes.
3. For natives, with regard to socio-demographic characteristics, contrary to what has been expected and shown in many other empirical analyses, age doesn't play a role in the formation of attitudes towards immigrants while females have a significantly higher propensity of anti-immigrant views. This has also been shown in other studies where women adopt more negative attitudes than men (e. g. Hainmueller and Hiscox 2007; Bridges and Mateut 2009; Valentova and Alieva 2010). A strong and highly significant relationship between higher levels of education and pro-immigration attitudes is observable. Natives with high levels of education are more positive in their views on immigrants. On the other hand, social class was not found to be significant this may be due to the fact that overall variations of social class positions are greater than in our rather socioeconomically homogenous case-study neighbourhoods.
4. For migrants, educational background as well as social class origin does not significantly contribute to the explanation of attitudes toward immigration across the 15 selected neighbourhoods. Although only slightly statistically significant, age turns out to be a predictor towards anti-immigration attitudes with older migrants found to be less inclined towards newcomers than their younger counterparts.
6. Besides the main set of independent variables, we found that trust, national identity and religious affiliation were all strong predictors in the formation of attitudes towards immigrants for both natives and migrants.
7. When comparing the results between natives and migrants, we first have to emphasise that the explanatory power of our empirical model was much weaker in the migrant analysis. In total, we were only able to explore 13 % of the total variance. Further research is needed to unfold the underlying dimensions of the formation of attitudes towards immigrants for the migrant population.
8. In our final empirical examination we explored whether our explanatory model accounted for cross-neighbourhood differences. Overall, after taking the set of predictors into account, only a few neighbourhoods remained significantly more or less likely to have positive or negative attitudes towards immigrants.
9. Finally, we have to stress that our study does not come without methodological caveats: The design of this study does not allow the causal direction of the contact effects to be established and the effect of having mixed friends or living in a mixed neighbourhood might be overestimated because of inverse causality.
Within the literature, there is an intense discussion on the key conditions to be fulfilled when studying interethnic contacts. However, as pointed out by Pettigrew and Tropp (2006): Allport's optimal conditions are in fact not essential for intergroup contacts to achieve positive outcomes - a finding, that is in line with our results as well.
Neighbourhood effects: Inequalities in the effect of the urban neighbourhood on residents' socioeconomic status
Thus far the specific analyses conducted using the integrated GEITONIES data set have focussed on the development of interethnic relations and the impact of individual and other neighbourhood characteristics on attitudes towards immigrants. However, given the uniqueness of the neighbourhood data collected in the GEITONIES survey, it has been possible to contribute to the body of literature on neighbourhood effects. Research has demonstrated that the neighbourhood of residence has a significant impact on life chances. However, Miltenburg and Lindo (2011) point out that more recently, some reviewers of the literature have warned against the self-evidence of assuming a uniform effect of the neighbourhood environment across all residents, as the residential area might in fact affect some people more than others (Glaster 2008, Pinkster 2007; Campbell and Lee 1990, Ellen and Turner, 1997).
The classical assumption is that neighbourhood effects transmit through contagion and socialisation models; the impact of socioeconomic characteristics of other residents in the neighbourhood is prevalent in explaining somebody's socioeconomic status. Behavioural influence in the neighbourhood, be it through socialisation, peer group activities, role models or social control, essentially transmits through local social networks (Galster, 2008: 10). Therefore, the local contacts of residents should be taken into account when estimating neighbourhood effects. In Galster's words (2008: 10), 'the intensity of exposure to such an influence would depend on the degree to which the individual's social networks were contained within the neighbourhood.' Indeed, some residents have sources of support that extend beyond the neighbourhood and they might therefore be less sensitive to neighbourhood attributes (Friedrichs and Blasius 2003, Glaster 2008).
Miltenburg and Lindo (2011), using GEITONIES data, empirically and theoretically expand on this concern. Whilst taking into account the social mechanisms through which neighbourhood effects are transmitted, the focus of the analysis is on whether the degree to which the social network of an individual resides in the neighbourhood leads to differential effects of the neighbourhood's socioeconomic status on the resident's current economic position. In other words, the main idea is that neighbourhood conditions impact differently on different members of our subpopulation of adult residents and that this difference is related to the size and quality of the social networks of each of them. Residents might have contacts and sources of support that extend beyond the neighbourhood. As a result, these residents might be less receptive to the socioeconomic composition of the neighbourhood.
By combining individual predictors and the impact of the socioeconomic composition of the neighbourhood, several hypotheses were deduced about how determinants affect residents' socioeconomic status.
1. The individual's and the parents' educational level are positively related to the resident's socioeconomic status.
2. The share of contacts within the neighbourhood is negatively related to the resident's socioeconomic status.
3. The size of an individual's network is positively related to the resident's socioeconomic status.
4. The unemployment rate and the rate of residents with low occupational attainment in the neighbourhood are negatively related to the resident's socioeconomic status.
5. The neighbourhood effects will be stronger for individuals that have solely intra-neighbourhood contacts.
In order to fully test the research hypotheses a multi-level model was required. Three levels of analysis were studied simultaneously: the individual (level one), the neighbourhood (level two) and the city (level three). The socioeconomic status of an individual, which constitutes the dependent variable, was assessed by creating the interval scale 'International socioeconomic Index of occupational status' (ISEI), which was collected in the survey. This index ranges from 16 to 90; the highest value is attributed to the highest occupational status.
In terms of neighbourhood level determinants, recent specific municipal data on socioeconomic status of neighbourhoods was not available for all cities. Therefore, the neighbourhood variables are derived by aggregation from the GEITONIES data, taking into account the stratified sample design. We weighted the rate of unemployment in the neighbourhood for the actual ratio of immigrants and natives (derived from available municipal data). To measure the rate of residents with low occupational attainment we measured the rate of residents with an ISEI-score below 30 in each neighbourhood (weighted). Both variables are measured as a level two characteristic. For the third level, the city-level, no contextual characteristics are taken into account. However, we do include this level three as fixed effects in order to control for the variance at the city- level.
In terms of individual-level determinants we include: the education of the respondent and of his or her father; the share of close friends that live in the neighbourhood; the share of overall social network living in the neighbourhood; the total number of most important people; the size of overall social network; and dummy variables for the close and overall network indicating if an individual has solely intra-neighbourhood contacts. Due to their separate measurement, two separate models, one for the most important contacts and one model for the overall social network were run. The multi-level models control for other background characteristics: gender, age and background.
In order to uncover potential differences in the neighbourhood effects, a range of explanations were tested. We found that the more contacts inside the neighbourhood an individual has (relative to the total network size), the lower the socioeconomic status of the resident (Miltenburg and Lindo 2010). We also found that the more contacts inside the neighbourhood an individual has (relative to the total network size), the lower the socioeconomic status of the resident.
For the most important contacts, the size of the network is positively related to the resident's socioeconomic status. In other words, residents with a greater number of important contacts are expected to have more access to information and resources, resulting in a higher socioeconomic status. However, this impact is not found for the overall social network in the contact fields of confidentiality and advice, spending free time and helping out. This is somewhat surprising, as it is quite often found in similar studies that the size of the network has a positive impact on somebody's socioeconomic status. How can we explain the result that the network size of the overall social network does not have any effect on the socioeconomic status? Campbell and Lee (1991: 217) found that the use of intimate name generators - as employed in our study - leads to smaller networks. Even though no numerical limit was built into the questions on the overall social network, distinguishing between the categories confidentiality and advice, spending free time and helping out is believed to result in a smaller network size. It could be the case that the different way of asking residents information on their networks is the reason we did not find a significant effect. Further research should build on these empirical suggestions.
Another important finding was that the higher the rate of residents with a low socioeconomic status in the neighbourhood, the lower the resident's socioeconomic status. This result corroborates the findings of a great deal of the previous work in this field of neighbourhood research. But in the current study, the aim was to build on these studies and assess potential differences across residents in neighbourhood effects. The most obvious finding to emerge from this study is that we found that residents that are strongly embedded in the neighbourhood are affected more strongly by the neighbourhood than those individuals who also have contacts outside the neighbourhood. This finding holds for the most important contacts and the overall network on confidentiality and advice. This was not the case for the overall social network for spending free time and helping out, possibly because of the different 'content or role' of these contacts.
In sum, the present study confirms some previous findings and contributes additional evidence that suggests that residence seems to matter for one's socioeconomic status, but the stronger the relational embeddedness in the neighbourhood the more this is the case. Even more importantly, while there is no difference between sexes, immigrants and lower-educated seem to score significantly and slightly higher in having exclusively intra-neighbourhood contacts. Consequently, these subgroups are more sensitive to neighbourhood characteristics. In other words, having your social world confined to only the neighbourhood is thus not necessarily a good thing. Dissemination activities
The project's dissemination strategy rests on five main pillars, namely dissemination of project outputs through the project website; organisation of national workshops; presentation of results at international conferences ; publication of project edited books, journal articles and other academic outputs; organisation of a final conference held in Lisbon at the end of the project.
Project website: http://geitonies.fl.ul.pt/
In Europe multiculturalism and integration have become hotly debated issues and are often used as banners by opposing camps. Integration and even assimilation are concepts that bear increasing weight, as concerns about the cohesion of society have grown steadily due to global processes and dramatic events clothed in symbolism that stress global cultural differences. Long-established policies of dealing with immigrant populations and of 'managing' diversity and difference are currently at stake. The riots that spiralled in November 2006 across the suburbs of Paris and other cities in France gave birth to criticisms of existing integration policies and arguments about the 'failure' of the French republican model that remained blind to inequality and difference (Begag 2007). While, in the United Kingdom (UK), part of the reaction to the French experience highlighted the supremacy of the British culturally-sensitive model, it is precisely this model of multiculturalism that is currently being challenged (e.g. Goodhart 2004, Phillips 2005). The question of integration pressingly returns in the aftermath of the race disturbances in towns of the Midlands in the summer of 2001 and more bitterly, of the London transport bombings by 'home-grown' terrorists in July 2005. Just across the sea, the other European example of 'successful' multiculturalism, the Netherlands, is also being redefined (Uitermark et al. 2005).
In response to these events across Europe the ways in which difference is accommodated are being challenged and scrutinised. In general terms we are seeing what has been termed as a 'backlash against multiculturalism' (Grillo 2005) where the multicultural model has been deemed as creating a gulf of separation between communities in spatial and socio-cultural terms through its particular focus on group based rights and service provision. The discourse on the failure of multicultural models highlights the erroneous emphasis of collective groups rights over individual rights, the fact that it predetermined fixed and unchanging categories of 'difference' and emphasised identity and racial equality at the expense of other differentials such as class and inequalities (e.g. Papastergiadis 2000). These shortcomings through prioritising cultural recognition are seen as having resulted in the failure of migrant groups to meet their responsibility to integrate and indeed migrants/ethnic minority groups have been accused of segregating themselves as a means to preserve their ethnic identities and resist integration (Phillips 2006). This has been represented officially as an inherent failure of multiculturalism and has given rise to concerns centred on social identity based on the fact that ethnic or religious identities are assumed to be in opposition to a homogeneous national identity and the belief that values at odds with dominant moral values are perceived as threatening national identity (Back et al. 2002, Robinson 2005). Subsequently, these fears have resulted in a resurgence of assimilationist discourse linked with projects to promote social cohesion and integration. There appears to be a trend more generally across Europe, where a discourse of failing multiculturalism is accompanied by a de-centralised responsibility for integration, the onus falling upon the responsibility of the individual (Mitchell 2004). Some have construed this as a new technology of knowledge/power in neoliberal regimes of governance that constitutes the individual as free-thinking whereas in reality there is little choice as those who 'opt-out' are cast as individuals not willing to participate in wider social life and are liable to be excluded from citisenship rights (Mitchell 2004:645). At the same time, nevertheless, earlier and ongoing processes of settlement generate grounded experiences of everyday diversity, whereby a 'spontaneous' convivial culture may start to flourish, particularly in urban areas in Britain and elsewhere (Gilroy 2004).
Similar debates and policy shifts are taking place at present across the EU, not only in the old host countries of the North, but also in the new immigration destinations of Southern Europe and the emerging reception countries of the East. For instance, the transition of Southern Europe to a target area for immigration has taken place gradually over the past three to four decades (Fonseca et al. 2002) and to a large extent the 'new cultural encounters' (King, 2001) manifested there have been significantly marked by irregularity and stigmatisation of the migrants (Anthias and Lazaridis 1999). At the same time, the new member states of Central and Eastern Europe emerge rapidly as new migrant destinations (e.g. Salt 2006, Drbohlav, Hárs and Grabowska-Lusinska 2009). The integration debate is still limited in these countries, due to the recent character of migration and to the fact that it the numbers of migrants are still few. However, the recent developments show that policy-makers have started to identify problems that may develop if the issue is ignored (Iglicka et al. 2005).
Do the concepts of multiculturalism and integration really depict antipodal and incompatible trends? This depends of course on how they are used; in ideological discourse they tend to adopt multi-layered, versatile and ambivalent meanings. The debates surrounding the presence of Muslims in Europe, to give an example, are centred around issues of religious diversity and tolerance of difference, while ultimately they reflect global politics on terrorism (Cesari 2005, Modood et al. 2006; Levey and Modood 2009). Despite the value of these debates in political terms, they do point to a significant lack of scientific knowledge on how cultural interactions actually develop between individuals and groups of different ethnic background, in contexts were they meet on a daily basis. Also, we lack data on how these interactions are related to an interchange of skills and attitudes (acculturation) and how they bear on the social status of the individuals and groups engaged in contact.
This research aims to enhance and improve our knowledge in this respect. It introduces a new perspective to understand these old issues. It reframes the old questions of integration rigorously into a relational format, taking away the spotlight from separate categories or groups of actors and directing it to the systemic whole, the integer, concentrating on the development of interdependencies between individuals and groups in certain territorial spaces. Obviously, much of the issues with respect to integration and multiculturalism are inevitably linked to urban environments (Ray 2003). The majority of immigrants and their descendants are concentrated in European cities, with cities attracting the majority of newcomers. As urban Europe continues to be the locus of various contradictions (economic, social, spatial, cultural etc.), urban spaces first and foremost provide the arena in which interethnic relations are unfolding. The ever increasing diversity of European metropolises, the persisting social and spatial inequalities in urban areas and the shifting paradigms of welfare provision, precisely call for a redefinition of 'host-stranger' relations in cities (Alexander 2003) and set the agenda for multicultural approaches to neighbourhood governance as a means for promoting social cohesion (Allen and Cars 2001; Blockland and van Eijk 2010). Consequently, we focussed on the local level, selecting social areas (neighbourhoods) in European metropolises as the key context for our research, in which we aim to understand the representations, daily practices and interactions that affect relations between social and ethnic groups. We included the complete multi-ethnic population of these areas in our inquiry. In this way, we addressed the issue of social cohesion in urban spaces in the most direct way possible.
The association between interacting groups of people that share the same residential areas and use the same public spaces makes the notion of place central in our research. A place is a physical space where social practices and group relations materialise, where neighbourhood relations occur and give meaning and identity both to the material elements of the area and to the people that inhabit it (Tuan 1977).
To elaborate the principal research objective, further research questions central to the project relate to: the different modes of interethnic coexistence that can be identified in European urban settings; local conditions causal to the development of these different modes; the impact of levels of ethnic concentration/residential segregation on these; the impact of behavioural patterns and initiatives of residents; resources (and the effectiveness of these) on a local level to facilitate inter-group contact; daily use of space and its impact on intercultural communication; an assessment of the most effective spaces to facilitate interethnic communication and contact; the management of diversity between groups sharing a residential neighbourhood; neighbourhood factors conducive to fostering an environment of understanding and tolerance; and, finally, the role of neighbourhood stigmatisation on tolerance and xenophobia.
The perspective adopted in this project is intercultural. Departing from the cultural diversity of contemporary metropolises and assuming that culture is dynamic, the construction of spaces of intercultural dialogue is crucial to re-enforce cohesion and avoid feelings of exclusion and the outburst of conflict. Sandercock (2004) argues that the core of interculturalism implicitly refers to two rights: the right to difference and the right to the city. In this view, people have, regardless of their ethnic background, equal rights in the shared public space and an equal capacity of full participation in the public affairs of the urban living environment, not just de jure (through legal devices and access to political participation) but also de facto (in the practical situations of everyday life).
Project results:
Interethnic contact, between immigrants and natives, is embedded in unequal power relations, class positions and cultural dispositions. The study of interethnic relations may reflect the core sociological problem of 'the formation, stabilisation and change of social relations' (Esser 1986: 30). On-going immigration and increasing diversity in European societies amidst the economic and political pitfalls in the context of globalisation, restructuring and crisis, have given rise to growing concerns over interethnic relations as a prerequisite for social cohesion.
One of the key approaches that the relevant literature draws on is the so-called contact hypothesis, which assumes that as people of different backgrounds come into contact with each other, prejudice and thus social distance decreases. Two criteria are relevant here. One relates to the size of the minority group: the larger this is the larger the threat it supposedly poses, thus the more intense the competition will be. The other has to do with the wider economic conditions: when these are unfavourable resources become scarcer and therefore the perceived threat grows and competition intensifies.
Putnam (2007) has recently suggested a third possibility, which he terms constrict theory. Although he is actually concerned with the effects of diversity on societal trust and social capital, as has also been explored using GEITONIES data, his article departs from the premises set by the above approaches and thus became a reference for recent studies on interethnic relations (e.g. Vervoort et al. 2010). His key argument that diversity impacts negatively on social solidarity (in the US context), came at a time of high politicisation of immigration issues and has stimulated much debate, ranging from criticism (e.g. Giddens 2007) to replications of his study in Europe (Lancee and Dronkers 2011). Despite our objections to both the theoretical foundations and methodological tools, constrict theory suggests that diversity (read interethnic contact) may actually reduce both in-group and out-group solidarity, i.e. deteriorate not only interethnic relations but undermine social cohesion at large.
What is often downplayed in studies employing 'contact' or 'conflict' theory is a failure to account for the social context which influences individual and group relations and patterns of behaviour in various ways. On the other hand, there is also much debate on what 'context' may involve and how this relates to individual attributes and attitudes. In his study of interethnic relations among migrant workers in West German cities, Esser (1986: 36) distinguished between two categories of context. One he termed the '(territorially predetermined) macro-structure, in particular the presence of persons of the same or of the other group and the behaviour... of those in each of the groups'. This may partly echo the first of the conditions in Blalock's competition theory about the size of the minority group. Esser's second category is what he calls the primary milieu, i.e. the normative climate prevailing in the relevant reference environment. Again, partly at least, we may find parallels to what others refer to as 'third party' influences, pointing to the context of socialisation: others (the family, the ethnic community, the state, etc.) may directly or indirectly encourage or discourage contact (Kalmijn 1998). In addition, context may also refer to other characteristics of the immediate (e.g. neighbourhood) or broader (e.g. city, nation state, transnational space) social environment. Socioeconomic and socio-spatial characteristics, are often downplayed even though they often play a role (Fong and Isajiw 2000; Vervoort et al. 2010).
Moreover, Esser also reminds us of the relationship between residential segregation and ethnic segmentation, echoing the Chicago ecological school's accounts equating social to spatial distance. It is actually there whereby the study of interethnic contact may find its origins. The important issue here is the introduction of a temporal dimension in the development of interethnic relations. The following text will address, analyse and contest many of these theories and empirical results through various thematic studies and statistical analyses using data collected between 2009 and 2010 in the ambit of the GEITONIES project.
Interethnic relations
Prior to any statistical analysis of the data a preliminary descriptive analysis of the nature and frequency of interethnic relations in the 18 neighbourhoods studied across in six European cities is fundamental. This will enable us to begin reflecting on the different modes of interethnic coexistence that can be identified in European urban settings, a core research objective.
Beginning with the respondents' global social networks no significant differences were found between migrants and natives in terms of network size. As expected, relationships in the other contact fields are much more restricted regardless of background.
As far as the composition of social networks is concerned, a significant component of both migrants' and natives' global network is comprised of relatives; the social network of approximately one out of five respondents is comprised exclusively of family members. Concerning interethnic relations, significant differences can be observed between migrants and natives. While respondents overall tend to socialise mostly with people of the same origin, migrants appear to have significantly more contacts of different ethnic backgrounds than natives. Interethnic relations are more common in the 'free time' contact field, whereas relatives are overrepresented in the fields of 'help' and 'advice and confidentiality' for both natives and migrants given the nature of the contact. Taking this into account, as well as the fact that relatives form an equal share of migrants' and natives' social networks, one can conclude that relatives do not account at all for the substantial difference found between migrants and natives in terms of interethnic relations. This difference is probably explained to a large extent by the difference in size of the immigrant and native populations in the cities where research was conducted.
The findings concerning the respondents' global social network are confirmed when we narrow down our analysis to their close social circle. Again, one cannot observe significant differences in network size. In terms of the ethnic composition of close social networks, migrants were reported to have a mean of 1.19 interethnic contacts in their close social network while for natives the mean number of interethnic relations was as low as 0.12.
Interethnic relations among migrant and native residents' close social networks were found to follow very similar patterns to that of their overall social networks. Thus, we cannot argue that the higher frequency of interethnic relations among migrants is necessarily an indication of 'openness' to and intimacy with, the 'other' in general, but rather of their interaction with the host country's native population and thus a degree of assimilation in their patterns of socialisation. Following Blau's (1977) thesis mentioned previously, this type of relationships could partly be seen as a 'necessity' for immigrants, given that natives are obviously the majority (dominant) group. On the other hand, it also entails a dynamic element, since close relationships (such as those examined here) are beyond inevitable native-migrant interaction and extend to the level of intimacy, which is built over time, as immigrants become more and more settled in the host country and in the specific localities where they live (Labrianidis et al. 2011).
In order to demonstrate the relationship between the two elements of the social network measured in the survey - the global/overall network and the close network - a further analysis has been conducted specifying different types of interethnic friendships. Here, all respondents who indicated that they have a friend from a different origin in either their overall social network or among their closest friends (out of a possible eight) are included. Four subgroups are considered; respondents with no interethnic friendships, those with interethnic friendships limited to their overall social network, those with friends from another origin only among their closest friends and, lastly, those with interethnic friendships in both their overall network and closet circle of friends.
In concordance with the data presented thus far, almost 80 % of natives do not have interethnic friendships in either their close or overall social network, whilst this is true for less than 40 % of migrants. For those natives that do have interethnic friendships, the largest part is represented in the overall social network and not in the close social network. On the other hand, migrants with interethnic relations tend to have friends both among their closest circle of friends and in their broader network. At the city level for both migrants and natives the old migration cities of Vienna and Rotterdam are distinct as having the highest degree of interethnic friendships. Whilst the newer destinations have extremely low levels, this is especially true in Bilbao and Warsaw, though for example, Thessaloniki is intermediate due to the specific characteristics of mainly Soviet Greek migrants there and the cultural background and historical links that they share with the native population (Fonseca et al. 2011).
The contact continuum and modes of coexistence: the relationship between weak and strong ties
Interethnic contacts in the neighbourhood
On the contact continuum one of the weakest forms of contact maybe engaging in small talk. The informal components of the so-called 'soft-infrastructure' of any given locality are thought to include various forms of interaction that may have a bearing on the social cohesion of the area. Among migrants, regardless of network type (mixed or homogenous), in general terms there is a high degree of superficial exchange. When this type of superficial contact is considered the remainder of migrants with no form of contact with 'the other' is minimal. The picture, nevertheless, is altogether altered when the native sample is considered. Indeed, 44.8 % of natives who do not have interethnic friendships have exchanged small talk with migrants in the neighbourhood. However, this almost doubles for the smaller group of natives that have interethnic friendships. Another important point to note is the high proportion of natives with no interethnic relations who have not engaged in small talk at the local level with the other, representing 43 % of the total population. In both groups, slightly lower levels of small talk can be observed among those who only have friends in the close social network (Fonseca et al. 2011).
With regards to the causality of this relationship, Fonseca et al. (2011) analysing GEITONIES data using multilevel regression analysis found that the number of persons of natives and migrants of 'other' origin than co-ethnics in close social network is a highly significant predicator of superficial contacts (exchanging small talk) at the local level indicating that intimate relations influence daily interethnic contacts in the neighbourhood. Interestingly, however, supporting Allport's contact theory, Kohlbacher et al. (2011), testing the relationship in the opposite direction, found in their analysis of GEITONIES data that superficial contacts in the neighbourhood did not reduce anti-immigrant views or attitudes, whilst strong ties like visiting at home or intimate friendships did.
When we move along the contact continuum to a more intimate form of interaction in the local setting, visiting or welcoming neighbours at home, as one would expect the frequency of interaction decreases. Furthermore, we may reasonably assume that visiting at home may result in more affective bonds given its more intimate nature. At the local level inter-group home visits are considerably higher among the groups of natives and migrants who have interethnic friendships. Still, this difference is clearly mediated by the background of the respondents, being much more pertinent for native respondents. Welcoming or visiting migrants at home is clearly related to having interethnic relations or not for this group. Among the migrant sample the difference in home visiting among those who have and do not have interethnic relations is considerably smaller (Fonseca et al. 2011). This permits us to tentatively suggest that the neighbourhood might be a more important place for natives in stimulating or developing interethnic bonds than for migrants. Again, however, care must be taken with interpretation and the generalisation of such claims due to the very small numbers of natives with interethnic relations.
Interethnic marriage
Given the low frequency of interethnic marriage, it is unsurprising that in general terms a larger share of respondents who have interethnic friendships do not or have not had interethnic partners in the past. However, when one compares the existence or not of interethnic friendships the share of interethnic marriages is substantially higher among those who have interethnic friendships. Around one third of natives with interethnic friendship has or has had interethnic partners compared with only 6.8 % of those with in-group networks. Likewise, only 12 % of migrants without interethnic friendships have ever had an interethnic partner compared to over 40 % who have mixed friendship networks (Fonseca et al. 2011). For natives, this very intimate form of interaction appears to have particular bearing for the development of close friendships. Fonseca et al. (2011) also found a statistical relationship between interethnic marriages, as an explanatory factor influencing the probability to have intimate interethnic friendships. Moreover, 66.1 % of migrants in mixed marriages/partnerships actually developed their intimate interethnic friendships after the initiation of the partnership. A similar pattern also stands among natives: 33 % of those in a mixed marriage have intimate interethnic friendships compared to just 6.5 % among those whose partner is of the same background. In 73.1 % of the cases, intimate interethnic contacts were developed after meeting their partners (Labrianidis et al. 2011).
Work colleagues
Working in a mixed environment provides opportunities for encounter and may serve to enhance inter-group familiarity. Indeed, it is important for those who have interethnic friendships, regardless of background group. 87 % of migrants with interethnic contacts in both their close and overall social network work in a mixed environment, compared with 53 % of those with no interethnic relations. The equivalent percentages for natives are 67 and 30, respectively (Fonseca et al. 2011).
Micro-contexts and relational circumstances of interethnic intimacy
The relationship among the various dimensions of contact along the contact continuum is important to enhance our understanding of how relationships develop among individuals from different ethnic groups. It represents a key research objective of the current project, namely which factors pertaining to the behavioural patterns and initiatives of the residents (in terms of social interaction) can be pointed out as being causal to the development of modes of interethnic coexistence.
To state it more directly, one may ask to what extent does the neighbourhood matter in the composition of the social networks of the respondents? Certainly, how this differs for groups of migrants and natives both with and without interethnic relations is also a central point of interest. Due to the way the questionnaire was structured we cannot analyse the overall and close social networks together here. Thus, we will commence with the overall social network and move to the close network. With regards to the overall social network, two main tendencies can be observed. Firstly, it is apparent that the overall social network for spending free time is concentrated outside of the neighbourhood of residence for the majority of natives, (56.2 % have none or just a few of their contacts with whom they spend free time there) and the largest proportion of migrants (47.4 %) (Fonseca et al. 2011).
A similar pattern can be noted in the other contact fields of sharing confidences and advice and exchanging substantial help. However, the neighbourhood is a more important place as the locus of the overall social network for those who have completely mono-ethnic networks whether they are of migrant or of native background. This is holds true to a greater extent for migrants than for natives. Indeed, slightly over 35 % of migrants with no interethnic contacts indicated that the vast majority of their social network lives in the neighbourhood of residence compared with fewer than 20 % of those with an ethnically mixed social network. Thus, to certain extent, when the composition of the social network is in-group it is concentrated to a greater degree in the neighbourhood (Fonseca et al. 2011). Furthermore, the literature on neighbourhood effects suggests that the larger the share of contacts concentrated in the neighbourhood of any given individual the more isolated they will be from mainstream society and the more sensitive they will be to the compositional effects of the neighbourhood. Indeed, this theoretical perspective was tested and confirmed using GEITONIES data by Miltenburg and Lindo (2011) as presented later in this summary.
Briefly, it is important to note that those respondents who have interethnic relations not only have a more spatially diverse network, but also have more mixed networks in terms of gender as well as less family-centred networks.
Here we turn to the close social network. Instead of examining the share of people having interethnic contacts, we focus on the actual contacts of migrants and the micropublics (e.g. Amin 2003), where interethnic encounters, which later developed into close relationships, took place (Labrianidis et al. 2011).
Regarding the circumstances of the original encounter, out of the total interethnic contacts of migrants, the vast majority (38 %) were first met as colleagues or fellow students, proportionally more than among same-origin contacts, which are dominated by relations of kin. The picture is similar in the case of natives. Moreover, there are interesting variations by city suggesting that the micro contexts of encounter may also depend on the broader urban and national contexts. Indeed, interethnic encounter as colleagues appears to be far more important in Warsaw than in Rotterdam, where relatives are more important (owning to the higher instance of mixed marriages among migrants). Furthermore, while the share of interethnic contacts met as neighbours is slightly less than that of same-origin contacts, the neighbourhood appears to be a meeting place in Southern European cities. By contrast, while their overall shares are very low, interethnic contact is more frequent in collective organisations in northern European cities (possibly reflecting the development of civil society) (Labrianidis et al. 2011).
The neighbourhood itself emerges as the second most important 'micropublic' fostering the development of interethnic relations. It appears to be a slightly more important place of socialisation for migrants than natives (seen also for global social networks). However, in actual fact less than one out of three (29.6 %) of migrants' total interethnic contacts were developed in the neighbourhood of residence, while a similar share (30.6 %) from the total of 815 immigrants who have close interethnic friends met at least one of them in the neighbourhood (Labrianidis et al. 2011). When we consider all contacts, whether interethnic or not, of the group of migrants that have at least one interethnic contact, meeting as relatives takes joint first place and the neighbourhood slips into third place (Fonseca et al. 2011).
Interestingly the distribution of immigrants who have met at least one of their interethnic relations across the 18 neighbourhoods does not seem to follow the same pattern as the distribution of the immigrants who have developed interethnic relations in general. Paradoxically the ranking of the neighbourhood seems to be, to a certain extent, reversed. The two neighbourhoods with the highest share of interethnic relations developed in the neighbourhood, Peraia and Costa de Caparica, are among the ones with the lowest overall shares of interethnic relations. At the same time, the majority of interethnic relations in the neighbourhoods with the highest shares have not developed there but outside them. In other words, the neighbourhoods that score high in terms of their immigrants' interethnic relations do not seem to facilitate the development of interethnic relations, but simply concentrate an immigrant population which has developed many interethnic contacts (with the exception of Hoogvliet-Noord in Rotterdam). A first general conclusion that may be extracted from this observation is that the neighbourhood does not appear to be the major determinant in the development of interethnic relations for migrant residents (Labrianidis et al. 2011).
Another interesting aspect is the actual place where the initial encounter took place. The key role of the workplace as a contact point is confirmed, concerning more than a quarter of migrants' interethnic contacts compared to just 17.5 % of same-origin contacts and is even more pronounced in the countries of recent migration. Similarly, confirmed is the limited share of interethnic contacts first met in a club or association, which is more important in Rotterdam and Vienna. On the other hand, open public spaces appear to be slightly more important meeting places for interethnic relations, especially in Southern European cities (where climate and culture favour socialisation in open spaces). Finally, one can observe the importance of the private sphere of the home (the respondents' own home or the home of relatives or friends) as a meeting place in a quarter of the cases, which exceeds 30 % in Rotterdam and Vienna (to an extent reflecting the higher proportions of relatives among interethnic contacts, as well as the limited encounters in the public space), but also Thessaloniki (Labrianidis et al. 2011). For all respondents regardless of the background of friends the home is the most important place of meeting followed by the place of work/study (Fonseca et al. 2011).
Neighbourhood and compositional effects: The predictors of interethnic relations
Whilst understanding the types or different dimensions of social interactions as well as their frequency and how and where they develop is crucial, it is important to understand the predictors of interethnic interaction. A crucial objective at the core of the GEITONIES project relates to the role of the local context as well as compositional effects in the process of developing interethnic contacts, expressed in the following research question. What neighbourhood and individual factors (social network characteristics, religious affiliations, citisenship, gender, unemployment status, etc) are conducive to fostering an environment of understanding and tolerance?
In terms of individual factors a simple reading of the data uncovered the existence of particular characteristics that seem to significantly correlate with the tendency of migrants to include interethnic (and mostly native) contacts in their close social networks. These factors that emerged are directly or indirectly related to the time dimension, since - for example - more settled immigrants present higher shares of interethnic contacts, compared to recently-arrived ones, while at the same time, second generation immigrants are also more likely to develop close social networks comprising of native contacts than the first generation (90 % of the former have at least one close friend of a different ethnic background compared to 44.8 % of the latter). Also, time is relevant in respect to the migrants' life course, whereby specific circumstances (mixed marriages, birth of children) are also associated with the existence (or not) of interethnic contacts in their social networks. The proportion of migrants who have at least one intimate interethnic contact is higher (63 %) among those who do not have any children compared to parents (47.4 %). As such, the importance of time is underlined in the development of interethnic relations, as an aspect of the wider process of migrants incorporation, which is something often ignored or downplayed in the relevant literature. Further proof of the relevance of the process of migrants' incorporation for the development of interethnic relations came from our examination of factors relating to aspects of their pathways of settlement in the host society. Some of which, like legal status and language proficiency, also seem to be time-dependent, while others simply accounted for the ways individual characteristics, such as ethnicity, religion (fewer shares of migrants belonging to a religion different than the dominant religion among the native sample have interethnic relations compared to those who share the host countries religion(s) or are not religious at all) education and profession (executive and professional migrants had the highest shares of close interethnic friends) , may relate to the wider political and socioeconomic context. Safer and more secured legal status, for instance, is directly correlated with higher probabilities to develop interethnic relations. In most of the cases, of course, these statuses are likely to denote longer periods of permanent residence in the hosting country, while a similar finding stands for linguistic skills (Labrianidis et al. 2011).
Following on from the descriptive analysis at the individual level, a more sophisticated statistical analysis was conducted to explore both the neighbourhood and compositional effects conducive to the development of interethnic relations (Fonseca et al. 2011). The methodology deemed most appropriate to respond to the aforementioned research question was multilevel linear regression modelling, a type of regression analysis in which independent variables from individual and aggregate levels can be included in the same model. Understanding and tolerance were operationalised through social relations measured at both the neighbourhood level and beyond. Two analytical dimensions were taken into consideration: the interethnic dimension of daily contacts in the neighbourhood and the interethnic dimension of the respondents' close contacts. Thus two models were estimated including the following two dependent variables: the share of small talk exchanged in the neighbourhood of residence with people of a different origin (over the three months prior to the survey) and the number of close friends (out of a possible eight) of a different origin.
Two general types of individual-level independent variables were included in the models (not all of these were included in each). The first are variables related to characteristics of close contacts of the respondent: number of close contacts; share of close contacts with educational levels higher than those of the respondent; number of close persons of different origin; share of neighbours among close contacts; share of relatives among close contacts. The second are those related to the individual characteristics of the respondent: long-term residence in the neighbourhood; having children under 16; being in a mixed marriage; education level; gender; age; and religious affiliation.
At the neighbourhood level the following variables were used: share of second-generation migrants; social class; and the index of diversity. Including the diversity index at the neighbourhood level also enables us to respond to the following research question: do levels of ethnic concentration and residential segregation impact on the development of interethnic relations? Given the differences at the city level dummy variables for cities were included in the models as control variables. The main empirical results can be summarised as follows.
The first point is the differing importance of the neighbourhood level in explaining the variability in the propensity to establish interethnic relations among migrants and natives. For superficial interethnic relations, the role of the neighbourhood context is comparable for migrants and natives, though not very high, indeed, less than 10 % of the total variation of their intimate interethnic relations is attributable to differences between neighbourhoods. The analysis of close interethnic contacts resulted somewhat differently. In fact, in the reference model for migrants, 15.6 % of the total variation of their intimate interethnic relations is attributable to differences between neighbourhoods. For the native population there is no significant neighbourhood effect in explaining the variation of their close contacts with migrants. Given the low percentage of natives with no interethnic relations it is unsurprising that the variability between neighbourhoods is much lower. However, the influence of national and urban contexts in the variation of superficial interethnic contacts in the area of residence should be stressed, especially for natives (the intra class correlation coefficient for the empty model - without controlling for the city level - for natives is 40.3 % whereas for migrants it is 18.3 %). Thus, the macro-structural characteristics, relating to economic, political and ideological factors or to the migratory context, of each city have a more important effect in comparison with local factors.
Among the neighbourhood characteristics tested in the models, the share of second-generation migrants among migrants in the neighbourhood was the only characteristic at the aggregate level found to shape both superficial and intimate interethnic relations among migrants. In general, the higher the share of second-generation migrants in the neighbourhood the greater the share and intensity of interethnic relations is. Such an outcome suggests that it is the process of migrants' settlement in the neighbourhood over time that matters for the development of interethnic relations.
Among natives, neighbourhood characteristics were only significant in the model for exchanging small talk. The socioeconomic level of the neighbourhoods is significantly related with the exchange of small talk between autochthonous residents and migrants, indicating that interethnic communication is higher in the urban areas with lower socioeconomic status. This may be a result of the over-representation of minority ethnic and immigrant groups in poor neighbourhoods. This is supported further by the fact that the diversity index for each neighbourhood has a positive and statistically significant coefficient, meaning that casual social interaction at the neighbourhood level is more prevalent in those areas with larger migrant communities.
The moderate role of the neighbourhood, as a determinant of the propensity of its inhabitants to develop interethnic contacts, be it close or superficial contacts was also confirmed by the fact that individual predictors have a higher explanatory power than predictors at the neighbourhood level. Beginning with migrants, with respect to individual factors that explain variability in the propensity to engage in superficial interethnic contact in the neighbourhood, the number of interethnic relations in the close social network proves to be a highly significant predicator. This indicates that intimate relations or a positive knowledge of the 'other' influence daily interethnic contacts in the neighbourhood (Friedkin 2004; Dixon 2006). The more concentrated the social network is in the neighbourhood the lower the levels of interaction with the 'other' are. Higher levels of education correspond to lower relative frequencies of superficial interethnic contact in the neighbourhood. Gender and religion are also important predictors in the sense that men and those with a religious affiliation interact more in the public domain out with their ethnic group.
Important individual level predictors for natives exchanging small talk with migrants are the number of migrants in the individual social network, age and length of residence in the neighbourhood. Thus, those who have lived in the neighbourhood for longer periods of time interact with migrants to a greater extent. Yet, age has the opposite effect and functions as an inhibitor to daily interactions with neighbours from other origins.
Important predictors for the number of close interethnic relations among migrants include some characteristics of the close social network. First of all, the higher the number of close contacts in general a migrant has, the more likely he/she also is to develop close interethnic contacts. Moreover, migrants who tend to engage in relations with persons with a higher educational level than themselves are also more likely to develop close interethnic relations. To the contrary, the higher proportion of neighbours and relatives among close friends of the migrant, the lower number of interethnic close contacts he/she has. Selected socio-demographic characteristics also have important explanatory power, namely long-term residence in the neighbourhood, being in a mixed marriage and having children under age of 16, which presents a negative coefficient. Levels of educational achievement are also significant with better educated migrants being more likely to develop close interethnic contacts.
In general, only three predictors were found to be significant for natives: the number of close contacts, the share of relatives and the 'mixed marriage' variable. They influence the propensity of natives to develop close interethnic contacts in the same direction observed for migrants.
The role of contacts in the formation of attitudes towards immigrants
Moving to the level of more general attitudes, Kohlbacher, Reeger and Schnell (2011) conducted a specific analysis of GEITONIES data on the relationship between social interactions and attitudes towards immigrants, with the assumption that contacts have an impact on attitudes. Daily social interactions in the neighbourhood context are referred to as well as close friends as the relevant plane of reference in order to investigate how contacts in the local context and on a general level are shaping and modifying views on immigrants. Of course, contacts of different qualities, like in everyday life or in the closer circle of friends, are not the only factor influencing the way people think about immigrants and immigration. The data set we have at hand provides the opportunity to analyse the role of contacts in shaping attitudes towards immigrants in comparison with other important determinants like age, education, trust, national identity or religious affiliation.
The leading question was, whether interethnic contacts (or, in Allports terms: true acquaintances) still matter once we consider all important factors at once. Our plane of departure was the theoretical foundation of 'inter group contact theory' as proposed by Allport in 1954. The basic assumption of this theoretical argument is that interethnic contacts, under appropriate conditions, are one of the most effective ways to reduce prejudice between majority and minority group members. Thus, personal contacts are also one of the best ways to improve relations among groups that are experiencing conflict. Previous studies on anti-immigrant attitudes were mostly based on national data. However, we have the opportunity to focus on the local level. Following Allport's differentiation between the impacts of 'true acquaintance' on the one hand (close relation, friendship) and 'casual acquaintance' on the other, we further argued that it is true acquaintance which matters - especially at the local level in which attitudes are formed through daily face to face interactions. Our data provided us with the opportunity to empirically differentiate between true acquaintances and casual contacts which to date has rarely been done. We tested this assumption empirically, first with a descriptive analysis followed by the modelling of attitudes using multivariate analysis, by examining the role of interethnic contacts in addition to a set of predictors that have been identified in the literature as important. This was conducted for fifteen neighbourhoods within the cities Bilbao, Thessaloniki, Vienna, Rotterdam and Lisbon.
The dependent variable, attitudes towards immigrants, is measured using an index (mean score) constructed from two survey items. The two questions are: 'It is good for the economy that people from other countries come to live here' and 'In the future, the proportion of immigrants will become a threat to society'. The items were answered on a likert scale ranging from 'agree strongly' to 'disagree strongly'. Both survey questions have long been used to measure immigration-related attitudes (McLaren 2003, Quillian 1995, Scheepers et al. 2002, Coenders et al. 2003, Schneider 2008, Semyonov and Glikman 2009).
Our main independent variables of interest, including various types of interethnic contacts, are included as a set of dummy variables in the analysis. 'Interethnic small talks' and 'Interethnic mutual visits' indicate whether individuals experienced these types of interethnic contacts during the last three months within the neighbourhood. 'Close interethnic relations' indicate if at least one person in the closest circle of friends is of another ethnic origin All three variables serve as measures for 'true acquaintances' (in Allport's sense). Next to interethnic relations, several additional factors that have been found to affect attitudes towards immigrants are considered in the analysis. Trust is a continuous variable derived from two survey items: 'Do you think that most people try to take advantage of you if they got the chance, or would they try to be fair?' and 'Would you say that most of the time people try to be helpful or that they are mostly looking out for themselves?' Religion is a dummy variable specifying whether respondents considered themselves as belonging to a particular religion (1=Yes, 0=No). Other individual level control variables were included in the analysis: education, social class, age and gender. Socioeconomic attainment is derived from information on the labour market participation of the survey respondents. Following previous research (Phalet et al. 2007) we distinguish economically inactive or unemployed from (self-) employed persons. The latter are then subdivided into broad occupational classes by combining categories into salariat (professional and managerial occupations), intermediate (skilled non-manual and manual occupations as well as self-employed) and working class (semi- and unskilled manual occupations). Age is entered as a continuous variable while gender is a dummy variable with women being the reference category.
Finally, examining whether the total size of the migrant population and the degree of ethnic diversity within neighbourhoods matters for attitudes towards immigration, we include two contextual variables at the neighbourhood level. Firstly, we include the 'share of migrants' (expressed as percentages). Secondly, in order to estimate ethnic diversity within neighbourhoods we construct a Herfindahl index of diversity. Two additional variables enter our analysis since they have been identified as highly relevant when explaining attitudes towards immigrants: Length of residence in the survey country and identification with the country of origin.
A further major contribution of our analysis is that we explored the above expected impact separately for natives and immigrants in 15 European urban spaces of six metropolises; an approach that has not been done before, taking into account the local perspective, too. The main empirical findings can be summarised as follows:
1. Overall, we find mixed evidence of the importance of interethnic contacts in the formation of attitudes towards immigrants. We find that true acquaintances reduce prejudice for natives but not for migrants.
2. With respect to casual contacts we explored the same pattern for natives as for migrants: The presence of migrants in the neighbourhood increases anti-immigrant attitudes while at the same time the degree of diversity decreases anti-immigrant attitudes.
3. For natives, with regard to socio-demographic characteristics, contrary to what has been expected and shown in many other empirical analyses, age doesn't play a role in the formation of attitudes towards immigrants while females have a significantly higher propensity of anti-immigrant views. This has also been shown in other studies where women adopt more negative attitudes than men (e. g. Hainmueller and Hiscox 2007; Bridges and Mateut 2009; Valentova and Alieva 2010). A strong and highly significant relationship between higher levels of education and pro-immigration attitudes is observable. Natives with high levels of education are more positive in their views on immigrants. On the other hand, social class was not found to be significant this may be due to the fact that overall variations of social class positions are greater than in our rather socioeconomically homogenous case-study neighbourhoods.
4. For migrants, educational background as well as social class origin does not significantly contribute to the explanation of attitudes toward immigration across the 15 selected neighbourhoods. Although only slightly statistically significant, age turns out to be a predictor towards anti-immigration attitudes with older migrants found to be less inclined towards newcomers than their younger counterparts.
6. Besides the main set of independent variables, we found that trust, national identity and religious affiliation were all strong predictors in the formation of attitudes towards immigrants for both natives and migrants.
7. When comparing the results between natives and migrants, we first have to emphasise that the explanatory power of our empirical model was much weaker in the migrant analysis. In total, we were only able to explore 13 % of the total variance. Further research is needed to unfold the underlying dimensions of the formation of attitudes towards immigrants for the migrant population.
8. In our final empirical examination we explored whether our explanatory model accounted for cross-neighbourhood differences. Overall, after taking the set of predictors into account, only a few neighbourhoods remained significantly more or less likely to have positive or negative attitudes towards immigrants.
9. Finally, we have to stress that our study does not come without methodological caveats: The design of this study does not allow the causal direction of the contact effects to be established and the effect of having mixed friends or living in a mixed neighbourhood might be overestimated because of inverse causality.
Within the literature, there is an intense discussion on the key conditions to be fulfilled when studying interethnic contacts. However, as pointed out by Pettigrew and Tropp (2006): Allport's optimal conditions are in fact not essential for intergroup contacts to achieve positive outcomes - a finding, that is in line with our results as well.
Neighbourhood effects: Inequalities in the effect of the urban neighbourhood on residents' socioeconomic status
Thus far the specific analyses conducted using the integrated GEITONIES data set have focussed on the development of interethnic relations and the impact of individual and other neighbourhood characteristics on attitudes towards immigrants. However, given the uniqueness of the neighbourhood data collected in the GEITONIES survey, it has been possible to contribute to the body of literature on neighbourhood effects. Research has demonstrated that the neighbourhood of residence has a significant impact on life chances. However, Miltenburg and Lindo (2011) point out that more recently, some reviewers of the literature have warned against the self-evidence of assuming a uniform effect of the neighbourhood environment across all residents, as the residential area might in fact affect some people more than others (Glaster 2008, Pinkster 2007; Campbell and Lee 1990, Ellen and Turner, 1997).
The classical assumption is that neighbourhood effects transmit through contagion and socialisation models; the impact of socioeconomic characteristics of other residents in the neighbourhood is prevalent in explaining somebody's socioeconomic status. Behavioural influence in the neighbourhood, be it through socialisation, peer group activities, role models or social control, essentially transmits through local social networks (Galster, 2008: 10). Therefore, the local contacts of residents should be taken into account when estimating neighbourhood effects. In Galster's words (2008: 10), 'the intensity of exposure to such an influence would depend on the degree to which the individual's social networks were contained within the neighbourhood.' Indeed, some residents have sources of support that extend beyond the neighbourhood and they might therefore be less sensitive to neighbourhood attributes (Friedrichs and Blasius 2003, Glaster 2008).
Miltenburg and Lindo (2011), using GEITONIES data, empirically and theoretically expand on this concern. Whilst taking into account the social mechanisms through which neighbourhood effects are transmitted, the focus of the analysis is on whether the degree to which the social network of an individual resides in the neighbourhood leads to differential effects of the neighbourhood's socioeconomic status on the resident's current economic position. In other words, the main idea is that neighbourhood conditions impact differently on different members of our subpopulation of adult residents and that this difference is related to the size and quality of the social networks of each of them. Residents might have contacts and sources of support that extend beyond the neighbourhood. As a result, these residents might be less receptive to the socioeconomic composition of the neighbourhood.
By combining individual predictors and the impact of the socioeconomic composition of the neighbourhood, several hypotheses were deduced about how determinants affect residents' socioeconomic status.
1. The individual's and the parents' educational level are positively related to the resident's socioeconomic status.
2. The share of contacts within the neighbourhood is negatively related to the resident's socioeconomic status.
3. The size of an individual's network is positively related to the resident's socioeconomic status.
4. The unemployment rate and the rate of residents with low occupational attainment in the neighbourhood are negatively related to the resident's socioeconomic status.
5. The neighbourhood effects will be stronger for individuals that have solely intra-neighbourhood contacts.
In order to fully test the research hypotheses a multi-level model was required. Three levels of analysis were studied simultaneously: the individual (level one), the neighbourhood (level two) and the city (level three). The socioeconomic status of an individual, which constitutes the dependent variable, was assessed by creating the interval scale 'International socioeconomic Index of occupational status' (ISEI), which was collected in the survey. This index ranges from 16 to 90; the highest value is attributed to the highest occupational status.
In terms of neighbourhood level determinants, recent specific municipal data on socioeconomic status of neighbourhoods was not available for all cities. Therefore, the neighbourhood variables are derived by aggregation from the GEITONIES data, taking into account the stratified sample design. We weighted the rate of unemployment in the neighbourhood for the actual ratio of immigrants and natives (derived from available municipal data). To measure the rate of residents with low occupational attainment we measured the rate of residents with an ISEI-score below 30 in each neighbourhood (weighted). Both variables are measured as a level two characteristic. For the third level, the city-level, no contextual characteristics are taken into account. However, we do include this level three as fixed effects in order to control for the variance at the city- level.
In terms of individual-level determinants we include: the education of the respondent and of his or her father; the share of close friends that live in the neighbourhood; the share of overall social network living in the neighbourhood; the total number of most important people; the size of overall social network; and dummy variables for the close and overall network indicating if an individual has solely intra-neighbourhood contacts. Due to their separate measurement, two separate models, one for the most important contacts and one model for the overall social network were run. The multi-level models control for other background characteristics: gender, age and background.
In order to uncover potential differences in the neighbourhood effects, a range of explanations were tested. We found that the more contacts inside the neighbourhood an individual has (relative to the total network size), the lower the socioeconomic status of the resident (Miltenburg and Lindo 2010). We also found that the more contacts inside the neighbourhood an individual has (relative to the total network size), the lower the socioeconomic status of the resident.
For the most important contacts, the size of the network is positively related to the resident's socioeconomic status. In other words, residents with a greater number of important contacts are expected to have more access to information and resources, resulting in a higher socioeconomic status. However, this impact is not found for the overall social network in the contact fields of confidentiality and advice, spending free time and helping out. This is somewhat surprising, as it is quite often found in similar studies that the size of the network has a positive impact on somebody's socioeconomic status. How can we explain the result that the network size of the overall social network does not have any effect on the socioeconomic status? Campbell and Lee (1991: 217) found that the use of intimate name generators - as employed in our study - leads to smaller networks. Even though no numerical limit was built into the questions on the overall social network, distinguishing between the categories confidentiality and advice, spending free time and helping out is believed to result in a smaller network size. It could be the case that the different way of asking residents information on their networks is the reason we did not find a significant effect. Further research should build on these empirical suggestions.
Another important finding was that the higher the rate of residents with a low socioeconomic status in the neighbourhood, the lower the resident's socioeconomic status. This result corroborates the findings of a great deal of the previous work in this field of neighbourhood research. But in the current study, the aim was to build on these studies and assess potential differences across residents in neighbourhood effects. The most obvious finding to emerge from this study is that we found that residents that are strongly embedded in the neighbourhood are affected more strongly by the neighbourhood than those individuals who also have contacts outside the neighbourhood. This finding holds for the most important contacts and the overall network on confidentiality and advice. This was not the case for the overall social network for spending free time and helping out, possibly because of the different 'content or role' of these contacts.
In sum, the present study confirms some previous findings and contributes additional evidence that suggests that residence seems to matter for one's socioeconomic status, but the stronger the relational embeddedness in the neighbourhood the more this is the case. Even more importantly, while there is no difference between sexes, immigrants and lower-educated seem to score significantly and slightly higher in having exclusively intra-neighbourhood contacts. Consequently, these subgroups are more sensitive to neighbourhood characteristics. In other words, having your social world confined to only the neighbourhood is thus not necessarily a good thing. Dissemination activities
The project's dissemination strategy rests on five main pillars, namely dissemination of project outputs through the project website; organisation of national workshops; presentation of results at international conferences ; publication of project edited books, journal articles and other academic outputs; organisation of a final conference held in Lisbon at the end of the project.
Project website: http://geitonies.fl.ul.pt/