Immigration and social interaction: Do diverse environments matter? Open Access

In the immigration debate, it is a common claim that immigration not only changes the immigrants but the receiving societies as a whole. In many ways this is obviously true: The populations of many contemporary European and North American societies have become increasingly diverse as a consequence of major immigration processes. The economies of such countries cater to the needs of diverse populations, and in increasingly diverse cities this is visible for instance in the shops and restaurants available. For other areas of social life, however, the extent of change is less obvious: This includes patterns of social interactions, the topic we are especially interested in.¹

It is well-established that social interactions are shaped by preferences and opportunities, alongside individual characteristics. Immigration and the diversity arising from it transform the opportunities to interact with people of different backgrounds. Preferences for interaction with people perceived as similar to oneself may counteract the influence of changed opportunities. Sometimes it is even assumed that diversity leads to sharper boundaries as people perceive it as threatening (for a critique of this line of argument see e.g., Portes and Vickstrom 2011; Petermann and Schönwälder 2012). Further, as DiMaggio and Garip (2012: 93–5) have recently pointed out, social networks not only reflect existing boundaries but may exacerbate existing inequalities ‘when effects of individual differences are multiplied’.

Against this background, it is highly relevant that we understand the preconditions of intergroup interaction and its development. The study presented here investigates to what extent the presence of immigrants and their descendants in urban environments has led to more diverse personal social interactions of their residents. As distinct from much of the existing literature, we are not specifically interested in the degree to which immigrants ‘integrate socially’, i.e., take up interactions with the longer-term resident population (see e.g., Haug 2010). Rather, we focus on the non-immigrant majority and their responses to a higher or lower presence of immigrants² in the residential environment. How does the socio-demographic and cultural diversity of societies affect the social interactions of individuals and groups within them? To what extent do ethnicity and national origin constitute boundaries that restrict social interactions? Are social networks more diverse in a more diverse environment?

These questions are investigated with data for German cities and neighbourhoods. Germany is one of the major European countries of immigration, but a lesser studied country with respect to this topic. While studies exist that provide selected information on some forms of interaction (Wagner et al. 2006; Petermann 2011), we examine effects of immigration-related diversity in the neighbourhood and city on neighbourhood contacts as well as on weak and strong ties. We are thus able to disentangle potentially varying effects of the presence of immigrants in a neighbourhood and city on different forms of social interaction.

Following this introduction, section 2 presents the theoretical considerations informing the research question investigated here. Section 3 provides some background information on immigration and spatial segregation in Germany. This is followed by section 4 that introduces the survey analyzed in this article. Section 5 then turns to the data presentation. We first outline the frequency of interactions of different quality and examine how far they are related to contextual opportunities, i.e., the shares of immigrants in the neighbourhood. Further, we investigate whether the specific structure of the immigrant population and its heterogeneity have an impact. In addition, we present data on the locations and social contexts relevant for the formation of inter-group ties. Finally, section 6 offers possible explanations for the diverging relationship between contextual factors and different forms of interaction.

In the public debate, in Germany and other European countries, it is frequently claimed that immigrants and natives do not interact enough, that separation remains common.³ Indeed, following a well-established sociological finding this would not be surprising. In their social lives people tend to associate with others like themselves – this ‘homophily principle’ is regarded as ‘one of the most striking and robust’ (Kossinets and Watts 2009: 405) sociological findings which, according to McPherson et al. (2001: 418), applies to the range of interactions from marriage to ‘mere contact’ (see also Lazarsfeld and Merton 1954; Marsden 1990; Petermann 2002). Ethnicity and ‘race’ are cited as key dimensions structuring interaction, alongside social status, age or education (McPherson et al. 2001: 420; Fuhse 2010: 183).

One main reason for this homogeneity trend is preference. Individuals apparently prefer the company of others who have similar backgrounds, tastes and lifestyles. To the extent that we assume migration background to be linked with symbolic boundaries (Lamont and Molnár 2002: 168), differing lifestyles, or norms, it is a likely dimension of homophily in social interactions.

But, as Scott Feld argued already in 1982, individual preferences ‘can be overemphasized’. This is because they ‘can only affect the choices of associates within limited sets of available alternatives’ (1982: 797; see also Huckfeldt 1983). Indeed, individuals are heavily constrained in their choices by the opportunities available in their environment. The neighbourhoods where children grow up are usually not inhabited by a cross-section of the population but socially segregated. At work, individuals may interact more with others on a similar level of hierarchy. Leisure-time activities are often age-specific, etc. Thus, ‘the social structuring of activities tends to bring similar people into frequent contact with one another, and thereby encourages the development of relationships among them’ (Feld 1982: 797).

Immigration is in many ways systematically linked with opportunities and restrictions affecting interaction. Immigration is typically a historically uneven process. Those growing up in Europe in the 1960s were less likely to encounter immigrants in their immediate environments than those growing up in the 1990s. Some immigrants are legally restricted in their movements, i.e., confined to asylum seeker hostels or even detainment centres and thus less likely to encounter others or to be encountered than other inhabitants of the region. In the countryside, there are still fewer immigrants than in cities – thus geography determines opportunities to form heterogeneous relationships. Educational institutions and work environments are selective. At universities for instance, where many long-lasting friendships are formed – or partners found – many students are now from other countries, but in Germany students with immigrant backgrounds are still vastly under-represented. For the immigrants, opportunity contexts marked by their under-representation should favour inter-group relations, but for the non-immigrants, opportunities for such interaction are clearly restricted. In employment, opportunities for encounters between non-immigrants and immigrants differ across sectors and status. Those of immigrant background are, for example, more numerous in manufacturing than in the banking and insurance sector (Schönwälder 2007; Tucci 2011).

City and neighbourhood are particularly important contexts for personal social interactions. While several authors have argued that, in times of easier and fast communication over great distances, locality is losing importance for personal social interactions, it is still subject to debate to what extent the new opportunities for more distant relationships are taken up and to what extent this transforms the patterns of interpersonal interactions (Mok et al. 2010). Altogether, it is hardly questioned that a significant share of such interactions still takes place within the limited space of the neighbourhood or city (Wellman 1996: 353; Fuhse 2008: 81). Mewes (2009: 41) points out that, in Germany, according to the large family survey, almost half of personal ties live within a distance of not more than 15 minutes (similar results in Petermann 2002). Everyday routines are located there; leisure time is spent not too far from people's homes, meaning that unplanned encounters and many more or less regular interactions are located within limited space. The residential environment should be considered an important opportunity context for inter-personal relationships. ‘Residential context… structures friendship choices’, Susan Welch and co-authors suggest (Welch et al. 2001: 5). Like other contexts, cities and neighbourhoods offer opportunities for inter-group encounters that are structured – e.g., by the housing market, the usage of public space and by residential segregation. We should expect to see such differences reflected in the patterns of social interaction. So, to what extent and in what spatial distribution is immigration present in Germany and its cities, the context where the study presented here was conducted?

In 2010, Germany had about 15 million residents with a migration background, which is almost 20% of the total population (Statistisches Bundesamt 2011, 2013a).⁴ Individuals ‘with a migration background’ – as defined in official statistics – include residents with foreign citizenship, the naturalized, ethnic German immigrants,⁵ and the children of these groups. Over 9 million are foreign born. About 9 million residents with a migration background are German citizens. The overwhelming majority of Germany's immigrant population has European (including Turkey) roots. About half a million have an African background and about 2.5 million originate from an Asian country. Those from Turkey and the former Soviet Union are the largest national-origin groups with about 3 million each, followed by former Yugoslavia and Poland (Statistisches Bundesamt 2013b).

Germany's immigrant population lives almost exclusively in the part of the country that was the old Federal Republic but is fairly widely distributed over that territory. Still, people with a migration background have a greater tendency than others to live in cities. More than half (54%) of those with a migration background live in the cities with at least 50,000 inhabitants (the cities studied here), while of the total population 40% live there.

And yet, immigrant residential concentration in Germany is altogether far lower not only than in the USA, but also than in the UK (for more details and a discussion of possible reasons for differences see Friedrichs and Triemer 2009; Schönwälder and Söhn 2009). Most German cities – and not only a few centres of immigration – have significant immigrant populations. However, as concentration is limited, it is unusual that a district within a city (Stadtteil) has more than 50% immigrants. In Frankfurt, a city included in our study, about 39% of the inhabitants have a migration background. Of 47 Stadtteile, six have more than 50% immigrants in their populations and the highest share is 59% (Frankfurt am Main 2011: 38). Hamburg is another city investigated in our study with 28% immigrants in its population. Here, of 100 districts only five have a majority of inhabitants with a migration background, the maximum is 70% (Statistisches Amt für Hamburg und Schleswig-Holstein 2010).⁶ Typically, the immigrant populations are mixed; rather than ‘Turkish’, ‘Italian’ or ‘Arab’ neighbourhoods, we find a diversity of origins within the immigrant populations of particular neighbourhoods. Among the 50 neighbourhoods in our sample, there are only five where one nationality accounts for more than 50% of the foreign population. These structural conditions of limited segregation could be seen as favouring interaction across ethnic boundaries as they provide more opportunities than a sharply segregated environment. Still, immigrant shares in the population vary enough to allow an analysis of the impact these differences have on interactions.

The project ‘Diversity and Contact’, based at the Max Planck Institute for the Study of Religious and Ethnic Diversity in Göttingen, provides original data for Germany. Few existing studies in this field, in particular for Germany, have a design that allows for an assessment of the effects of the residential environment on individuals.⁷ For Germany, it is extremely difficult to conduct systematic quantitative research relating to a set of smaller areas because census data do not exist for the two past decades and other statistics on that spatial level are not available. However, if we are interested in diversity as experienced in an immediate context of life, comparisons of nation states or of large and heterogeneous districts are of limited value.⁸ The study presented here compares a representative sample of urban areas with on average 7200 (median) inhabitants. As distinct from other studies, we do not focus on areas with the highest immigrant shares or with a history of conflicts but rather aim to represent ‘urban normality’. The 50 areas of investigation were thus randomly selected. We investigate in West Germany only because we wanted to include cities with a longer-term experience of immigration. Furthermore, in the eastern states the immigrant population is not only mostly recent but also very small: only 5% of Germany's population with a migration background lives there. A stratified sampling procedure (for details see Petermann et al.2012) ensured that the study includes neighbourhoods with lower and higher shares of immigrants in their populations. For lack of other data, this had to be based on the shares of foreigners (for a similar approach Damelang and Haas 2012). The respective range in the sample is from 2 to 46% foreign citizens.

The data set comprises 50 interviews in each of the 50 neighbourhoods, i.e., 2500 interviews overall. Telephone interviews were conducted from May to July 2010 by a survey company. The questionnaire includes a battery of questions on the frequency of different kinds of intergroup interactions, their social locations, and evaluation. Questions refer to those ‘who are themselves not native Germans or whose parents are not from Germany’ or, if the respondents have a migration background, to those ‘who are native Germans’.⁹ As pretests (cognitive and test-run of entire questionnaire) showed, this phrasing is generally well-understood by respondents in Germany. The questionnaire also includes questions on contact with individuals of particular national or regional origin and different legal categories. We refrained from using the common term ‘Ausländer’ (foreigners) because it is exclusive and often implicitly only refers to particular groups of immigrants. Reference to ‘ethnic minorities’ is not common in Germany. The results are overall representative for the adult population of West German cities of the relevant size (for more details see Petermann et al.2012: 22–8). The survey data is matched with statistical data for the neighbourhoods. These variables (see  Appendix) were calculated on the basis of official data provided by cities. As motivations and opportunities for contact differ between majority and minority, we present in this text an analysis for non-immigrants, i.e., for people who were born in Germany as German citizens and whose parents were born in Germany.

So, to what extent do our data confirm a view of diverse neighbourhoods as opportunities for cross-group social interactions? We begin with a discussion of the frequency of contact with people of immigrant background. To capture everyday interactions, the DivCon-survey uses a question about the frequency of talking with someone of a different background in the neighbourhood, i.e. if the respondent is a non-immigrant: ‘In your neighbourhood, how often do you talk to people who are themselves not native Germans or whose parents are not from Germany?’

As Figure 1 shows, only 16% never talk with immigrants in their neighbourhood, about 40% have infrequent contact (at least monthly and less frequently), but a larger group of 45% state daily or weekly contact with people of immigrant background. The bimodal distribution indicates that frequency of contact is not normally distributed (in a statistical sense) but splits the non-immigrant population into two groups: one with frequent and regular cross-group contact and another with very little or no contact.

The distribution of those who have frequent contact and those who do not differs considerably and significantly between neighbourhoods. In some of them more than 70% state daily or weekly contact in the neighbourhood, while in others only around 20% do this. To what extent are such different contact patterns related to the contact opportunities represented by the presence of individuals with a migration background in the neighbourhood where people live?

There is a significant moderate proportional correlation between frequency of out-group contact and the share of foreigners (tau-b=.20). In the 13 neighbourhoods with the lowest shares of foreigners (under 8%; lowest quartile) 28% have daily or weekly inter-group contact while in the 13 neighbourhoods with the highest shares of foreigners (above 22%; highest quartile) 62% have frequent cross-group contact.

In a multivariate analysis with a number of control variables, share of foreigners is an influential variable for frequency of contact to immigrants in the neighbourhood (see Table 1; see  Appendix for univariate statistics). As we can show, a positive relationship between the share of foreigners in the neighbourhood and the frequency of inter-group contact exists for the non-migrant population: opportunity goes along with more frequent interaction (see Table 1, model 1). Moreover, it is one of the strongest correlations in the estimated model. The strong positive effect of the share of foreigners is significant even when contextual and individual factors are statistically controlled. Of the contextual factors, share of foreigners exerts the strongest influence. Not surprisingly, those who are more exposed to the neighbourhood, because they spend more leisure time there or because their workplace is located there, have even more frequent cross-group contact.

Two other characteristics of the neighbourhood are also relevant for cross-group interactions: population size and density. Size, density and heterogeneity are classic indicators of urbanism (Wirth 1938). Apparently, the co-existence of large, heterogeneous populations in limited space promotes interaction between immigrants and natives.

Do we find evidence for a selection of contacts according to preferences, i.e., in this case, negative or positive attitudes to immigrants, as assumed by the homophily-thesis? First, we tested whether the combined share of Turkish, African and Asian nationals among the foreign population impacts on the inter-group contact frequency. It may be assumed that boundaries hindering interaction are harsher towards particular immigrant groups, like those who are physically distinct or come from Muslim backgrounds. However, no significant correlation was found (tau-b=−.02). The national or ethnic backgrounds of the immigrants do not seem to matter for neighbourhood contacts.

Further, we used a scale of four items measuring how far respondents are willing to take the perspective of ‘foreigners’.¹⁰ The phrasing reflects common usage of the term ‘Ausländer/foreigners’ in Germany. Empathy with foreigners should go along with more frequent contact to immigrants and higher shares of immigrants among network ties. Another variable measures in-group identification with the nation, i.e., a contrasting orientation, which should affect contact with those of other national backgrounds negatively. Our findings show the assumed correlations with social inter-group interactions. Empathy with foreigners is significantly positively related to frequency of out-group contact whereas national identification has a negative effect. As empathy with foreigners has one of the strongest effects in the regression model, we interpret this in the sense that – apart from opportunities – preferences also play a strong role for the frequency of inter-group contact.

Furthermore, the social context of the individuals, here larger household size and having a partner with migration background, apparently offers opportunities to meet immigrants and thus affects frequency of cross-group contact. Altogether, favourable opportunities account for frequent inter-group contact in neighbourhoods although national identification and low empathy with foreigners counteract it.

An exchange of a few words in the neighbourhood is of course only one possible form of social interaction and may contain superficial as well as more intimate interactions. The DivCon survey also includes data on weak and strong ties, i.e., on acquaintances and friends, as well as on the number of immigrants among these personal ties. As distinct from inter-group contact, as measured in our survey, questions on weak and strong ties did not specifically refer to the neighbourhood.¹¹

Figure 2 shows the shares of intergroup ties among the weak and strong ties of non-immigrant Germans. We found a hierarchy from acquaintances to friends – the closer the interaction, the smaller the share of those having such interactions with someone of a different background. Following the homophily thesis, this is not surprising – at least if we assume that migration background constitutes a relevant dimension of the similarity people seek in their close associates. A majority of personal networks consists predominantly of homophilous ties.

However, we also found that the boundary of ethnicity and immigrant background is frequently overcome or irrelevant. Among non-immigrants 68% have acquaintances with a migration background. And about 40% among non-immigrants have at least one friend from the other category. About 10% even state that half or more of their friends are immigrants or the children of immigrants. This refers to a network of, on average, seven friends. If we take friends and acquaintances together, 75% have immigrants among them, while 25% of the non-immigrant respondents do not have any friend or acquaintance with an immigrant background.

Thus, for inhabitants of German cities, having interactions with people of different backgrounds is the norm, not an exception. Apparently, a preference for social similarity is either not the only principle governing such interactions, or the similarity individuals seek in their friends is not always defined along the lines of ethnicity or national origin. However, a trend towards homogeneity (in all likelihood produced by preference as well as opportunity) plays a significant role at the dyad level of strong tie networks as a majority of non-immigrants do not have immigrant friends.

The general picture is also confirmed by other studies. Thus Semyonov and Glikman (2009: 700) report results of the European Social Survey according to which about half of the German population report friendship contacts with immigrants. According to ALLBUS data,¹² 59.3% of Germans report contact with ‘foreigners’ as friends or acquaintances. The different phrasing of the question (foreigners versus people of immigrant background) (see Reuband 1989) may explain why the level of interaction is lower than in the DivCon survey, but here as well, the results show a majority of Germans with intergroup ties. Rippl (2008: 502) has demonstrated that such interaction has increased over time and that private contexts have become more important than work contexts.

In what ways do the closer ties differ across neighbourhoods with smaller and larger immigrant populations? Depending on the closeness and the time invested, personal relationships are commonly divided into weak and strong ties. We also have to consider that network ties – as distinct from social contacts in the neighbourhood – develop over longer periods of time and in various opportunity contexts. The neighbourhood is only one ‘focus of activity’ (Feld 1981), and not necessarily the most important one. In order to find out what spatial and social contexts are relevant for the development of intergroup ties, we asked our respondents where and how they had first met their immigrant friends and acquaintances. We found that the neighbourhood is a relevant arena for the establishment of relations to immigrants. About half (47%) said that they had first met at least one of their cross-group friends and acquaintances in their current neighbourhood (see Figure 3). A former neighbourhood was mentioned by 30%. Friends and family – a ‘nonfocus source’ in Feld's terms (1982: 799) – are the most important intermediary of intergroup ties. This is not peculiar to intergroup ties, rather the ‘transitivity principle’ is observed for all ties (Mewes 2009: 37). The workplace and educational institutions came out as almost equally important which is not surprising given that they provide an institutional context for repeated, often daily encounters. For 14% of our respondents (and 26% of those who are in employment or education), workplace or school are located in the neighbourhood, meaning that institutional and spatial contexts of interaction overlap. Formal associations and informal groups are also mentioned frequently, but are clearly of lesser importance than the already mentioned frameworks of intergroup encounters.

Given that neighbourhoods are relevant contexts for the establishment of intergroup ties, we should expect the population composition of the residential context to impact on the networks. To what extent is this the case?

Surprisingly, multivariate analyses show that immigrant shares in the neighbourhood population have no influence on the shares of immigrants among weak and strong ties (Table 1, models 2 and 4).¹³ Other contextual factors on the neighbourhood level do not affect the shares of immigrants in weak and strong tie networks.

Is the absence of a general effect of shares of immigrants in the neighbourhood on intergroup ties due to the fact that friends and acquaintances are often found in larger spatial contexts? It seems plausible that a higher share of immigrants in a city presents more opportunities for the formation of weak and strong ties between non-immigrants and immigrants and should thus be related to larger immigrant shares in the networks of the native residents. However, again we did not find correlations between the shares of foreigners in our cities of investigation and the relevance of intergroup ties (Table 1, models 3 and 5).

Does this mean that – rather than opportunities presented by the residential and city environments – only preferences account for the share of immigrants in the weak and strong tie networks? Indeed, the two attitudes tested here can be shown to affect intergroup ties. There is a small but significant negative effect of national identification and a strong positive effect of empathy with foreigners on the share of immigrants in weak tie networks. As expected, preferences play a role in the formation of inter-group weak ties. Similarly, national identification also has a significant negative effect on share of immigrants among strong ties whereas empathy with foreigners affects it in a strongly positive way. Further, and in contrast to the neighbourhood contacts, the combined share of Turkish, African and Asian nationals among foreigners in the neighbourhood has a negative effect on the share of immigrants in strong tie networks: Non-immigrants living in neighbourhoods with an immigrant population dominated by Turkish, African and Asian nationals more often have no or just one immigrant friend.

Still, to a greater extent than by such attitudes, shares of immigrants among weak and strong ties are affected by characteristics of the individuals’ social relations: a partner with migration background, inter-group contact in the area, the size of the weak tie network, and – for strong ties – inter-group weak ties all impact on the relative relevance of intergroup ties. Apparently, the social context with the opportunities it presents is more important for the network composition than the spatial context. The latter impacts on cross-group contacts, but for the closer interactions, the mere presence of larger numbers of immigrants in neighbourhood and city environments does not seem to counterbalance boundaries existing in other spheres.

This article presented mixed evidence regarding the impact of the neighbourhood as opportunity context for encounters and ensuing closer interactions. We started from the assumption that city and neighbourhood are particularly important contexts for personal social interactions and that, accordingly, differences in the population composition of these spatial contexts should be reflected in the interactions and networks of their inhabitants. We presented evidence underlining that cross-group interactions happen frequently and involve large parts of the native population. Furthermore, we could demonstrate that the immigration-related diversity of the residential environment increases the frequency of inter-group contact in these environments. However, for the weak and strong network ties a link between increased opportunities for interaction – as represented by the population of the neighbourhood – and higher shares of immigrants in the networks of the non-immigrants could not be demonstrated. Residential context does not seem to structure friendship choices. To explain the absence of direct effects of immigration-related diversity on the share of intergroup ties among all ties we offer two hypotheses:

First, many of these ties may be formed elsewhere. The residential environment is just one potentially relevant opportunity context. Work, friends and family, associations, are others – and may turn out to be more relevant. Further, differences between neighbourhoods may be related not only to the composition of their populations but to the character of social life. A systematic comparison of the importance of different spatial, institutional and social contexts (or foci) for the formation of intergroup ties is a research lacuna.

Second, friendship ties tend to be formed early in life, roughly between school and family formation. Only for the younger cohorts this phase happened in a period when the German population was fairly heterogeneous. Older cohorts may have formed life-time friendships when there were still few immigrants in their social environment. While substantial immigration occurred already until the 1980s, immigrant figures doubled in the 1990s when the diversity of backgrounds also further increased. We hypothesize that immigrant shares of neighbourhood populations will have an indirect longer-term effect on ties as they already impact on frequency of intergroup contact, which has an effect on intergroup ties.

Altogether this article aimed to shed light on the interrelatedness of neighbourhood diversity and inter-group interaction. We could show that, in German cities, interaction between the long-term residents and those of immigrant background is frequent and common – in the neighbourhoods and in the social networks more generally. At the same time, while we should not expect the immigrant minority to be present in all or most social networks, differences between the levels of neighbourhood contact, acquaintanceship and friendship circles suggest that migration background serves as a boundary in social interactions. This boundary is, however, in many ways porous. Alongside preference, opportunity shapes interaction patterns, and immigration has clearly changed patterns of interaction.

Sören Petermann is a Research Fellow at the Max Planck Institute for the Study of Religious and Ethnic Diversity. Previously he worked as a researcher and lecturer at the Institute of Sociology and at the Collaborative Research Centre 580 ‘Social development in post-socialistic societies’ at Martin Luther University Halle-Wittenberg. He obtained a Dr. phil. in 2001 and a Dr. phil. habil. in 2012, both from Halle University.

Karen Schönwälder is Research Group Leader at the Max Planck Institute for the Study of Religious and Ethnic Diversity and extracurricular professor at the Georg August University Göttingen. Previously she held positions at the Social Science Research Center Berlin (WZB), the University of London and Marburg University.