Jump to ContentJump to Main Navigation
Growing up in Diverse SocietiesThe Integration of the Children of Immigrants in England, Germany, the Netherlands, and Sweden$

Frank Kalter, Jan O. Jonsson, Frank van Tubergen, and Anthony Heath

Print publication date: 2018

Print ISBN-13: 9780197266373

Published to University Press Scholarship Online: May 2019

DOI: 10.5871/bacad/9780197266373.001.0001

Show Summary Details
Page of
date: 02 June 2020

Making Friends across Ethnic Boundaries: Are Personal Networks of Adolescents Diverse?

Making Friends across Ethnic Boundaries: Are Personal Networks of Adolescents Diverse?

Chapter:
(p.176) 7 Making Friends across Ethnic Boundaries: Are Personal Networks of Adolescents Diverse?
Source:
Growing up in Diverse Societies
Author(s):

Frank van Tubergen

Sanne Smith

Publisher:
British Academy
DOI:10.5871/bacad/9780197266373.003.0007

Abstract and Keywords

How ethnically diverse are the social networks of adolescents in Europe? The present study finds that ethnic majority youth in England, Germany, Netherlands and Sweden have very few ties to ethnic minority peers, either in the neighbourhood or at school. The networks of ethnic minorities are more mixed, because the groups to which they belong are smaller in size. Such structural opportunities for meeting members of other groups also plays a key role in understanding why some ethnic minority groups—being smaller in size—are more ‘open’ than other groups. Next to opportunities, also in-group preferences appear important. Both majority and minority groups exhibit a preference to befriend peers from their own ethnic group. Possibly, parents play a role in overcoming these ethnic boundaries. Findings suggest that having more ethnically mixed personal networks and more-positive attitudes towards other ethnic groups is transmitted from parents to their children.

Keywords:   social networks, personal networks, adolescents, ethnicity, neighbourhoods, immigrants, friendship, interethnic ties, ethnic boundaries

7.1 Introduction

SOCIAL SCIENTISTS HAVE BECOME increasingly interested in studying the ethnic segregation of social networks. The reasons for this are twofold. First, the mixing of social networks affects people’s norms, values and lifestyle. In line with contact theory (Allport 1954), for example, it has been repeatedly shown that inter-ethnic contacts promote positive out-group attitudes and reduce intergroup prejudice (Pettigrew & Tropp 2006). If the social networks of the ethnic majority population and minority groups remain largely segregated there will be more stereotyped group beliefs as a result. It has been likewise shown that among ethnic minority groups, having more inter-ethnic ties is related to their religiosity (van Tubergen 2007), ethnic identity (Vroome et al. 2014) and language acquisition (Chiswick 2008; van Tubergen & Mentjox 2014).

Secondly, having inter-ethnic ties can be regarded as social capital and is of importance for structural outcomes as well. Minority members who have contacts with co-ethnics predominantly or even exclusively might be deprived from access to relevant information about jobs or remain unconnected to influential persons. One study, using survey data on the Netherlands, showed that ethnic minority members have less resourceful networks than majority members (van Tubergen 2014). Another study, using the German Socio-Economic Panel Study, found that the development of inter-ethnic ties is positively associated with the occupational outcomes of immigrants (Kanas et al. 2012).

Given the important consequences of ethnic boundaries in personal networks, it seems imperative to study in the first place how strongly segregated minority and majority group networks are, and to unravel their driving forces. That is the aim of this chapter.

What is needed, first, is clarification of the key concepts, namely ‘ethnic boundaries’ and ‘personal networks’. The concept of ethnic boundaries is based (p.177) on the distinction (Putnam 2000) between having ties with members who are from the same group (called bonding ties) and having ties with people from another group (i.e. bridging ties). In the present case, in- and out-group refers to ethnic groups, to intra- and inter-ethnic contacts. We say that ethnic boundaries in personal networks are stronger when people have more (ethnically) bonding ties at the expense of (ethnically) bridging ties. When ethnic boundaries are strong, the networks of ethnic majority members and minority members are more segregated.

Regarding the concept of ‘personal networks’, we use a broad definition. Personal networks refer, first, to stronger ties, the smaller circle of more intimate, emotionally close social contacts. This is also called the ‘core network’, and typically consists of relations with one’s partner, friends and relatives. Many studies have been done in Europe on the prevalence and causes of inter-ethnic marriages (Lucassen & Laarman 2009), such as in Sweden (Dribe & Lundh 2011), Germany (Kalter & Schroedter 2010), England (Muttarak & Heath 2010) and the Netherlands (Kalmijn & van Tubergen 2006). Recently, this literature has been supplemented with studies on ethnic boundaries in ‘support networks’ (de Miguel Luken & Tranmer 2010) and in ‘core discussion networks’ (van Tubergen 2015). A consistent finding is that the strong-tie networks are highly segregated.

Secondly, personal networks also include the much wider group of weaker ties, that is, social contacts and ties that are more ‘superficial’ and instrumental among people who interact less frequently. Inter-ethnic contacts in the neighbourhood, at work and at voluntary organisations are typical examples. The literature on these weaker inter-ethnic ties in European countries is rapidly growing (Martinovic et al. 2009a; 2009b; Schaeffer 2013), and the evidence suggests that weaker ties, too, are strongly segregated between minority and majority populations.

What do these findings tell us about the personal networks of youth in Europe? In Europe, the empirical study of ethnic boundaries in personal networks has largely focused on the adult population. Mostly absent from this picture, therefore, are studies on the ethnic segregation of the networks of minority and majority youth. There are a few exceptions, however. These studies focus on inter-ethnic friendships made in schools, more specifically in classrooms, which are therefore characteristic of the strong-tie network. Regarding the population of interest, youth, the study of friendships made in classrooms makes sense, as students spend most of their day at school, and many of their friends are made within that setting. The school context therefore seems to be the key setting to either promote or inhibit ethnic cleavages.

One of the first studies on inter-ethnic friendships in school was that of Baerveldt et al. (2004). Using data from the Dutch Social Behavior Study, consisting of 1,317 pupils from 20 urban high schools, aged 16–18 years, they found evidence for strong ethnic segregation within adolescent friendships in schools in the Netherlands. This research has been succeeded by other studies (p.178) on ethnic segregation in schools in the Netherlands (Vermeij et al. 2009; Stark & Flache 2012) and Germany (Windzio 2012). These all suggest strong ethnic boundaries in friendship networks at schools between minority and majority youth.

In the current chapter, we elaborate on such research. First, we come up with a descriptive portrait of ethnic segregation in weaker ties. The few studies on youth to date have exclusively studied segregation in the stronger-tie networks at school, hence little is known how much segregation there is in weaker ties. From a societal perspective, it seems relevant to know whether strong-tie and weak-tie networks are both segregated and adolescents live in completely parallel societies, or that, possibly, core social ties are segregated but there are frequent inter-ethnic contacts outside the private friendship circle. Empirically, we study the contacts youth have in their neighbourhood with members of their own ethnic group and with those of other groups. Theoretically, it is difficult beforehand to predict how strongly segregated these weaker ties in the neighbourhood are compared to the strong-tie network. On the one hand, stronger ties might be driven more by homophilous preferences, that is, the preference to establish ties with same-ethnic friends might be stronger than to have same-ethnic contacts with weaker ties, such as with neighbours. On the other hand, however, neighbourhoods can be highly ethnically segregated, which would hinder opportunities to establish inter-ethnic ties—even weaker ones. Theoretically, both forces can play a role. Hence, we examine the degree of segregation in the neighbourhood empirically, to see whether one of these forces is dominant.

The second contribution of this chapter is to provide a more reliable and more comprehensive picture of ethnic boundaries of friendships in school classrooms in Europe. More reliable, because we use representative samples (for the age group we study) rather than convenience samples, as previous work has done. Earlier work has studied a few and, possibly, ‘atypical’ schools in the Netherlands and Germany. More comprehensive, because we extend earlier work on the networks of youth in the Netherlands and Germany to Sweden and England.

The organisation of this chapter is as follows. First, we present findings on the weaker ties of youth in their neighbourhood. How many contacts do majority and minority youth have within their neighbourhood with members of their own group and those of other groups? After that, we move to the strong-tie networks of youth. Before we focus on their friendships in school, however, we describe the degree of segregation within their friendship network in general, that is, the friends they have either within their school or outside. How segregated are their friendship networks overall? After having answered this question, we continue with a description of the segregation of friendship networks in school—where clearly most friends are made. Finally, after having established the ethnic segregation in schools, we attempt to explain these patterns. We show that opportunity is a major driving force to understand friendship segregation in school, but also find evidence that ethnic homophily plays a role.

(p.179) 7.2 Inter-Ethnic Contacts in the Neighbourhood

We begin with a description of the ethnic boundaries in weaker ties, as found in the inter-ethnic ties in the neighbourhood. In the Children of Immigrants Longitudinal Survey in Four European Countries (CILS4EU), we asked respondents to indicate ‘how often they spend time with persons from a certain background in their neighbourhood’. The targeted groups differ by survey country, and in each we asked about the most prominent ethnic groups. In England, we asked the adolescents about their contacts in the neighbourhood with people who are (1) white, (2) Asian, (3) black or who have (4) ‘another background’. Adolescents had to indicate how often they spent time with people from these four ethnic groups (i.e. white, Asian, black and ‘another background’) on a five-point scale which ranges from ‘never’ to ‘every day’. In Germany, the ethnic groups were Germans, Turks, Russians, Poles, Italian and ‘other’ immigrant groups. Dutch adolescents were asked about the following four ethnic groups: Dutch, Turkish, Moroccan, Surinamese or Antillean and ‘other’ immigrant neighbours. Lastly, in Sweden, a simplified distinction was made, between contacts with ‘Swedes’ and ‘foreigners’, without further differentiating between foreign groups.

In Tables 7.1a–d, we relate the adolescents’ answers to these questions to their origin. To simplify matters, we show the percentage of adolescents who reported spending time with a particular group at least once or several times a month. This is a rather minimal definition of contact frequency, which allows us to study how people of the same or different groups have at least some contact in the neighbourhood. Two findings stand out.

First, it appears that a substantial part of the majority youth does not even have such minimal inter-ethnic contact in their neighbourhood. In England, only 19% of the majority youth spend time with Asians at least once or several times per month. Figures on contact with blacks (29%) or people of another background (18%) are comparable. The same pattern can be found in other countries. In Germany, only 20% of the majority youth have at least some contact with Turks in their neighbourhood. In the Netherlands, only 13% of majority youth spend time with Turks in their neighbourhood at least once a month. In short, few majority youth have any degree of contact with minority youth in their neighbourhood. Most contacts majority youth have in their neighbourhood are with other majority members.

Second, minority groups have more ethnically diverse contacts in their neighbourhood than majority youth do. We arrive at this conclusion when we take the opposite perspective, namely from the point of view of ethnic minorities. Although there is an overall tendency to have contact with co-ethnic members, many minority youth have at least some contact with members from other groups in their neighbourhood, particularly majority members. Take as an example the Turks in the Netherlands. Among this group, 77% spend at least some time with other Turks once or several times a month. At the same time, however, 59% of the Turkish youth also have some contact with majority Dutch in their neighbourhood, (p.180)

Table 7.1a. England (contacts in neighbourhood by ethnicity, %)

Origin group

White

Asia

Black

Other

England

84

19

29

18

Pakistan

41

80

34

29

India

57

76

34

35

MENA+

75

44

48

50

Eastern Europe

69

30

38

61

NWS Europe

79

36

30

46

Asia

51

68

31

32

Sub-Sah. Africa

64

39

48

35

Caribbean

66

32

74

37

Note: Design-weighted values. NWS Europe: Northern, Western and Southern Europe; MENA+: Middle East and North Africa plus Afghanistan.

Table 7.1b. Germany (contacts in neighbourhood by ethnicity, %)

Origin group

German

Turkish

Russian

Polish

Italian

Other

Germany

95

20

20

15

13

25

Turkey

67

87

23

17

18

52

Russia

83

35

73

21

10

30

Poland

91

33

32

64

11

22

Italy

85

57

24

22

66

37

Serbia

62

60

14

16

28

65

MENA+

68

42

41

17

13

62

Eastern Europe

85

34

38

16

11

55

NWS Europe

89

38

29

25

19

67

Asia

87

43

17

20

16

43

Sub-Sah. Africa

76

43

12

6

14

63

Note: Design-weighted values. NWS Europe: Northern, Western and Southern Europe (except Italy); MENA+: Middle East and North Africa plus Afghanistan and Pakistan.

Table 7.1c. Netherlands (contacts in neighbourhood by ethnicity, %)

Origin group

Dutch

Turkish

Moroccan

Caribbean

Other

Netherlands

92

13

10

12

24

Turkey

59

77

34

18

38

Morocco

59

47

91

46

54

Suriname

72

26

29

57

44

MENA+

68

41

40

22

78

Eastern Europe

63

8

10

13

42

NWS Europe

89

16

13

14

27

Asia

89

20

26

24

43

Sub-Sah. Africa

57

23

19

34

43

Caribbean

81

19

15

42

38

Note: Design-weighted values. NWS Europe: Northern, Western and Southern Europe; MENA+: Middle East and North Africa (except Turkey and Morocco) plus Afghanistan and Pakistan.

(p.181)

Table 7.1d. Sweden (contacts in neighbourhood by ethnicity, %)

Origin group

Swedish

Foreign

Sweden

85

35

Iraq

58

83

Finland

84

48

Turkey

49

87

Bosnia & Herzegovina

70

83

Somalia

61

97

Kosovo

76

85

MENA+

66

79

Eastern Europe

71

67

NWS Europe

82

58

Asia

75

65

Sub-Sah. Africa

62

75

Note: Design-weighted values. NWS Europe: Northern, Western and Southern Europe (except Finland); MENA+: Middle East (except Iraq, Kosovo and Turkey) and North Africa plus Afghanistan and Pakistan.

34% have contact with Moroccans, 18% with Caribbeans (i.e. Antilleans and Surinamese) and 38% with other groups. The picture we get from these figures is clear: ethnic minority youth have many more inter-ethnic ties in the neighbourhood than ethnic majority youth.

7.3 Ethnic Segregation of Friendships

What levels of ethnic segregation can we see in the stronger-tie networks of youth? Do we find similar or different degrees of segregation to the case of neighbourhood contacts? To answer this question, we use information on the ethnic origin of ‘best’ friends. In the CILS4EU, the adolescents were asked to name their best friends (either inside or outside school), up to a maximum of five. Hence, these best friends could be any type of friend: within their school, even within their class, but also friends they have at their sports club, in their neighbourhood and so forth; romantic relationships were excluded, however. Adolescents were asked to indicate the ethnic background of each of their friends.

The patterns we find with regard to friendship segregation mirror those regarding segregation in the neighbourhood. However, the key finding is that ethnic boundaries are even stronger with respect to these more intimate ties. Friendships are more ethnically segregated than are the (minimal) contacts people have in their neighbourhood. Possibly this suggests that the preference to befriend co-ethnic youth is a stronger force than patterns of neighbourhood segregation.

Majority youth have very few, if any, friends outside their own co-ethnic group. Take again the ties between majority youth and Asians in England as an example. While around 19% of the majority population in England spend at least some time (p.182) in their neighbourhood with Asians (Table 7.1a), only 4% of their friends are from this group (Table 7.2a). Friendships with blacks (4%) or other groups (3%) are equally rare. The same pattern is found in Germany, the Netherlands and Sweden as well. The friendship networks of majority youth are very ethnically homogenous.

Minority youth tend to have friends largely from their own group and/or the native majority. Furthermore, it appears that there are strong differences between the various minority groups in their degree of ethnic segregation of friendships. For example, 47% of the friends of Turkish youth in Germany are other Turkish youth. Such intra-ethnic friendships are much less common among Poles in Germany, where such ethnic bonding ties make up only 13% of all friends. Poles mainly befriend majority youth, that is, 68% of their friends have a German background. Overall, however, the pattern is clear: friendship networks are highly segregated between minority and majority youth in Europe.

Table 7.2a. England (ethnic origin of friends, %)

Origin group

White

Asia

Black

Other

Total

England

89

4

4

3

100

Pakistan

15

73

8

4

100

India

36

52

8

5

100

MENA+

53

22

7

19

100

Eastern Europe

46

6

5

43

100

NWS Europe

70

6

7

17

100

Asia

29

53

5

12

100

Sub-Sah. Africa

45

17

27

11

100

Caribbean

41

11

37

12

100

Note: Design-weighted values. Totals may not add up due to rounding. NWS Europe: Northern, Western and Southern Europe; MENA+: Middle East and North Africa plus Afghanistan.

Table 7.2b. Germany (ethnic origin of friends, %)

Origin group

German

Turkish

Russian

Polish

Italian

Other

Total

Germany

86

2

2

2

1

6

100

Turkey

27

47

4

3

2

17

100

Russia

56

6

21

3

2

11

100

Poland

68

3

5

13

1

9

100

Italy

49

13

3

2

18

15

100

Serbia

35

18

3

2

6

36

100

MENA+

36

15

13

3

3

31

100

Eastern Europe

55

8

10

3

2

21

100

NWS Europe

67

5

6

4

3

15

100

Asia

67

5

3

3

2

19

100

Sub-Sah. Africa

44

10

4

1

1

41

100

Note: Design-weighted values. Totals may not add up due to rounding.NWS Europe: Northern, Western and Southern Europe (except Italy); MENA+: Middle East and North Africa plus Afghanistan and Pakistan.

(p.183)

Table 7.2c. Netherlands (ethnic origin of friends, %)

Origin group

Dutch

Turkish

Moroccan

Surinamese

Antillean

Other

Total

Netherlands

93

1

1

1

0

5

100

Turkey

26

51

6

2

0

15

100

Morocco

23

7

49

6

2

11

100

Suriname

60

5

4

13

9

9

100

MENA+

55

11

5

2

1

26

100

Eastern Europe

71

2

1

1

1

25

100

NWS Europe

85

1

1

1

1

11

100

Asia

75

1

3

2

0

19

100

Sub-Sah. Africa

60

2

3

5

2

28

100

Caribbean

71

5

1

7

4

13

100

Note: Design-weighted values. Totals may not add up due to rounding.NWS Europe: Northern, Western and Southern Europe; MENA+: Middle East and North Africa (except Turkey and Morocco) plus Afghanistan and Pakistan.

Table 7.2d. Sweden (ethnic origin of friends, %)

Origin group

Swedish

Other

Total

Sweden

83

17

100

Iraq

29

71

100

Finland

68

32

100

Turkey

23

77

100

Bosnia & Herzegovina

32

68

100

Somalia

14

86

100

Kosovo

29

71

100

MENA+

33

67

100

Eastern Europe

48

52

100

NWS Europe

67

33

100

Asia

55

45

100

Sub-Sah. Africa

33

67

100

Note: Design-weighted values. NWS Europe: Northern, Western and Southern Europe (except Finland); MENA+: Middle East (except Iraq, Kosovo and Turkey) and North Africa plus Afghanistan and Pakistan.

7.4 Ethnic Friendship Boundaries in School

How segregated are friendship networks in school? The pattern need not be the same as with friendships in general. The reason being that friends in school are only part of an overall friendship network. Furthermore, in choosing who to select as the ‘best friends’ in class, adolescents might have used a lower threshold of nominating someone as a ‘friend’ than when they had to write down their five best friends in general. These five best friends might be their top five friends so to speak, whereas the friends they nominated in class might not necessarily be in this inner circle. The question is whether we see a different level of ethnic segregation (p.184)

Table 7.3a. England (ethnic origin friendship nominations in class, %)

Ego

Alter

1

2

3

4

5

6

7

8

9

1. England

86

1

2

1

1

4

2

3

1

2. Pakistan

28

43

7

5

1

2

7

7

2

3. India

48

8

23

1

1

3

6

9

1

4. MENA+

64

8

3

4

2

5

6

7

2

5. Eastern Europe

56

6

4

2

12

5

5

8

2

6. NWS Europe

82

1

1

1

3

4

3

4

1

7. Asia

48

3

3

6

2

4

25

6

1

8. Sub-Sah. Africa

55

6

7

2

2

3

6

15

4

9. Caribbean

60

3

4

2

1

6

3

14

8

Note: Design-weighted values. NWS Europe: Northern, Western and Southern Europe; MENA+: Middle East and North Africa plus Afghanistan.

Table 7.3b. Germany (ethnic origin friendship nominations in class, %)

Ego

Alter

1

2

3

4

5

6

7

8

9

10

11

1. Germany

81

4

2

2

1

0

2

2

3

2

1

2. Turkey

40

32

4

3

2

2

7

4

2

2

1

3. Russia

63

9

8

3

1

1

6

4

2

1

2

4. Poland

66

6

7

7

2

1

3

2

2

3

1

5. Italy

60

10

2

4

4

2

3

10

3

2

1

6. Serbia

36

19

1

2

2

16

11

6

4

3

1

7. MENA+

49

14

8

3

2

3

10

5

2

2

2

8. Eastern Europe

61

10

6

2

5

1

6

2

3

3

0

9. NWS Europe

74

5

2

2

2

1

2

3

4

2

2

10. Asia

67

7

4

5

2

1

4

5

3

1

1

11. Sub-Sah. Africa

50

13

9

5

1

1

4

2

5

2

7

Note: Design-weighted values. NWS Europe: Northern, Western and Southern Europe (except Italy); MENA+: Middle East and North Africa plus Afghanistan and Pakistan.

of their friendship network in class. The answer to this question can be found in Tables 7.3a–d.1

What do the results show?

A key overall finding is that friendship networks are less strongly segregated by ethnicity in school classes, as compared to overall friendship networks. To illustrate: in Germany, around 81% of the school friends nominated by youth with a German background are from the same background (see Table 7.3b). This is lower than the percentage of same-ethnic friends found in the five best-friends network in general, that is, 86% (Table 7.2b). This pattern can also be found among immigrant (p.185)

Table 7.3c. Netherlands (ethnic origin friendship nominations in class, %)

Ego

Alter

1

2

3

4

5

6

7

8

9

10

1. Netherlands

89

1

0

1

2

1

3

2

0

1

2. Turkey

42

23

7

5

14

1

1

5

1

1

3. Morocco

41

8

30

4

5

1

3

2

2

3

4. Suriname

70

3

4

6

4

1

3

3

1

4

5. MENA+

65

10

3

2

5

5

8

1

1

0

6. Eastern Europe

82

1

1

1

10

0

1

2

0

1

7. NWS Europe

85

0

1

1

2

1

4

4

1

1

8. Asia

75

3

2

2

2

1

7

6

0

1

9. Sub-Sah. Africa

74

3

4

5

5

1

5

1

2

1

10. Caribbean

75

7

2

5

2

1

1

5

1

2

Note: Design-weighted values. NWS Europe: Northern, Western and Southern Europe; MENA+: Middle East and North Africa (except Turkey and Morocco) plus Afghanistan and Pakistan. The pattern observed for Eastern Europe should be interpreted with caution, given the small number of cases for this group.

Table 7.3d. Sweden (ethnic origin friendship nominations in class, %)

Ego

Alter

1

2

3

4

5

6

7

8

9

10

11

12

1. Sweden

81

1

3

1

1

0

1

2

3

4

3

1

2. Iraq

32

13

1

4

2

3

2

21

7

3

6

5

3. Finland

76

1

10

0

1

0

1

2

1

5

3

1

4. Turkey

35

5

0

20

2

2

2

19

4

5

4

2

5. Bosnia & Herzegovina

40

5

2

2

11

4

3

14

13

3

2

2

6. Somalia

15

8

0

5

4

28

1

21

3

1

6

8

7. Kosovo

41

3

4

2

4

0

24

6

7

3

3

1

8. MENA+

38

9

2

6

4

3

2

20

4

4

4

4

9. Eastern Europe

59

2

2

1

3

1

2

7

14

4

3

1

10. NWS Europe

75

1

3

1

1

0

1

3

4

7

2

1

11. Asia

65

3

3

1

1

1

1

7

4

4

8

2

12 Sub-Sah. Africa

48

6

2

2

1

4

1

15

6

4

2

10

Note: Design-weighted values. NWS Europe: Northern, Western and Southern Europe (except Finland); MENA+: Middle East (except Iraq, Kosovo and Turkey) and North Africa plus Afghanistan and Pakistan.

groups. In Germany, for example, we see that 32% of the school friends of Turks are also Turks (Table 7.3b), but the five best friends in general question yields a co-ethnic friendship percentage of 47% (Table 7.2b). Apparently, then, the ethnic boundaries in friendship networks in school are less strong than the ethnic boundaries in overall friendship networks. This might suggest that these overall friendship networks are more driven by ethnic homophily and selected from a large pool of potential candidates, whereas networks in class are more often ‘forced’ choices and constrained by the opportunities within the class setting.

(p.186) This is not to say that there is no ethnic segregation in school classes. On the contrary, we do see that majority youth in England, Germany, the Netherlands and Sweden have very few friendships in class with immigrant youth. And we also observe that immigrant adolescents in these countries generally tend to befriend peers of their own group, combined with friendships with majority youth. Besides these general patterns, we also see interesting ‘group-specific’ deviations. Most importantly, there are strong differences across immigrant groups in how many ethnically bonding and bridging ties they have. Some groups, like the Turks in Germany and the Netherlands for example, have higher levels of co-ethnic friendships than other groups. The ethnic minority groups that show the highest group closure in friendships in class are Pakistani in England (43% of their friends in class are of the same ethnicity), Turks in Germany (32%), Moroccans in the Netherlands (30%) and Somalians in Sweden (35%). Other groups have very few co-ethnic friends in class. Notable examples are Surinamese in the Netherlands (6% co-ethnic) and Italians in Germany (5%). In the Appendix (see Tables A7.1 to A7.4, Model 1), we also see that majority youth have significantly more same-ethnic friends than minority youth.

7.5 Opportunity and Group Size

How can we explain these patterns of ethnic segregation of friendship networks in schools? Why do majority youth in Europe have fewer friends of immigrant origin than vice versa? Why are some minority groups in England, Germany, the Netherlands and Sweden more ethnically ‘closed’ than other groups? In short, what are the conditions that promote or inhibit inter-ethnic friendships in class?

In the literature, various theoretical mechanisms are suggested. One of these was developed by the sociologist Peter Blau in the context of research on intermarriage and intergroup relations more generally (Blau et al. 1982; Blau, 1994). Blau argued that the opportunity structure is a major determinant for intergroup contacts as its shapes meeting chances. Members of numerically large groups, he argued, logically have fewer possibilities for meeting members of smaller groups than vice versa. Those who belong to smaller groups have ample contact with out-group members, and such structural opportunities therefore shape the frequency of bonding and bridging ties. Because of imbalances in the population size of groups in society, members of smaller groups are ‘forced’ to develop cross-ethnic ties, whereas those belonging to larger groups develop ties within their own group.

In the literature on intermarriage, the consequences of group size on endogamy and exogamy has been well established. For example, it is well known that larger immigrant groups (either measured at the national or regional level) show higher levels of endogamy than smaller immigrant groups (Kalmijn 1998).

Do structural opportunities also affect the frequency of intra- and inter-ethnic friendships in class? To answer this question, we measure group size at the level of the school classroom, which is the most direct opportunity structure for the (p.187) formation of ties within class. We then relate the proportion of same-ethnic peers in class and the proportion of friends in class who are from same ethnicity as the respondent. The results show that the imbalances in ethnic group size in class are a critical condition for promoting and inhibiting inter-ethnic friendship. Group size explains the asymmetry in inter-ethnic friendship ties between majority and minority members, and to a large degree also explains why some minority groups are more closed than others. To see why, we need to consider Figure 7.1, which shows the strong impact of group size in class on friendship segregation in Germany. The patterns are similar for the other countries (see Appendix, Tables A7.1–A7.4).

The figure clearly shows that the majority group in Germany are more ethnically closed than the immigrant groups, largely because many majority youth go to schools with a high proportion of majority youth, and hence find themselves surrounded by such youth. This, of course, largely reflects the population distribution at the national level, as there are far more German majority youth than minority youth. In other words, German majority youth have very few opportunities to develop ties with minorities in class, simply because they are not there. This suggests the key role of meeting opportunities at school. Though theoretically trivial and self-evident, the societal importance of the role of opportunities for the establishment of inter-ethnic ties can hardly be overstated. Opportunities for inter-ethnic friendship in class create inter-ethnic friendships and thus weaken ethnic boundaries, but apparently at present these conditions seem largely absent for majority members in Germany, England, Sweden and the Netherlands.

For minority children, however, there are more opportunities to befriend peers from another ethnicity, simply because there are usually not so many co-ethnic peers in class. As minority groups are generally smaller in size, they are ‘forced’ to develop ties outside their own group—with majority members and with youth from ethnic minority origins other than their own. These inequalities in meeting opportunities also largely explain the differences in ethnic boundaries across immigrant groups. Minority groups that are larger in size tend to be more ethnically ‘closed’ in class than minority groups that are smaller. For example, adolescents with a Turkish background in Germany more often have friends in class of Turkish origin than Poles befriend Poles, and this is, at least partly, because the Turks are a bigger group, and hence Turkish youth are more often able to sit with co-ethnics than their Polish peers. In summary: the larger the size of the own ethnic group in class, the more strongly the boundaries of that ethnic group in class.

7.6 Does Ethnic Homophily Exist?

Imbalances in group size at the school level create differential opportunities and hence is a powerful determinant of ethnic segregation in class. One might legitimately ask whether, over and above these direct inequalities in opportunities in class, there is a tendency to develop friendships in class with same-ethnic peers as (p.188)

Making Friends across Ethnic Boundaries: Are Personal Networks of Adolescents Diverse?

Figure 7.1 Relationship between proportion of same-ethnic peers in class and proportion of friends in class from same origin group as respondent, Germany

Note: Dots represent the average for the respective group in a certain class and dot size the class weight.

(p.189)

Table 7.4. Ethnic homophily of friendships in class, by ethnicity (OR)

M

SD

M

SD

England

Germany

England

1.837

1.445

Germany

1.544

2.017

Pakistan

2.988

6.299

Turkey

4.215

13.348

India

1.464

1.910

Russia

1.117

2.024

MENA+

2.815

4.636

Poland

1.734

2.786

Eastern Europe

7.890

16.705

Italy

0.581

1.110

NWS Europe

1.144

3.696

Serbia

6.823

7.540

Asia

2.365

2.969

MENA+

1.184

1.977

Sub-Sah. Africa

2.456

4.540

Eastern Europe

0.511

1.115

Caribbean

0.704

1.685

NWS Europe

0.782

1.786

Asia

0.600

1.463

Sub-Sah. Africa

18.785

22.009

Netherlands

Sweden

Netherlands

1.438

0.729

Sweden

1.687

1.260

Turkey

22.784

43.251

Iraq

2.008

2.866

Morocco

3.605

4.832

Finland

2.022

3.651

Suriname

0.910

2.513

Turkey

12.104

58.324

MENA+

1.115

4.256

Bosnia & Herz.

1.612

1.883

NWS Europe

0.820

2.640

Somalia

2.946

2.834

Asia

2.527

3.292

Kosovo

9.804

14.243

Sub-Sah. Africa

0.449

0.516

MENA+

5.594

15.575

Caribbean

0.603

1.999

Eastern Europe

3.404

7.695

NWS Europe

1.104

2.029

Asia

4.483

7.140

Sub-Sah. Africa

1.756

2.368

Note: Design-weighted values. M = Mean, SD = Standard Deviation. NWS Europe: Northern, Western and Southern Europe; MENA+: Middle East and North Africa plus Afghanistan and Pakistan.

opposed to peers from other groups. Such a tendency is labelled ethnic homophily, the preference to befriend co-ethnic peers above and beyond what is expected by random choice. To detect whether such a homophilous preference for same ethnicity exists, researchers study revealed preferences as they become evident from behavioural choices. In the present context, such revealed preferences are studied by taking the opportunity structure of the class into account, and then examining deviations in friendship patterns from friendship choices made randomly.

An often-used measure to study such ethnic homophily is based on the concept of density, which is the percentage of realised friendships from all possible friendship nominations in class (Baerveldt et al. 2004). The same-ethnic and inter-ethnic density are thus the percentages of same-ethnic and inter-ethnic friends in class of all possible such friends in that environment. The difference between these two intra-ethnic and inter-ethnic densities reflects the fact that adolescents more often choose same-ethnic friends over inter-ethnic friends; hence, this is often used as a way to capture ethnic homophily.

(p.190) In our case, however, the density measure is less useful, because students can only choose a maximum of five friends, which means that smaller (ethnic) groups can quickly reach the maximum density level, whereas for larger (majority) groups the maximum density is often impossible to realise. Therefore, the density measure is sensitive to (unequal) group sizes.

As an alternative, we use the odds ratio (OR), which is margin-free and which was introduced as a measure of ethnic homophily in adolescent friendship networks in US schools by James Moody (Moody 2001). The measure can be interpreted as the OR of a friendship between members of a same-ethnicity dyad relative to friendship in a cross-ethnicity dyad. When OR = 1, then the odds of a same-ethnicity friendship equals the odds of a cross-ethnicity friendship (Moody 2001) and there is no ethnic homophily. When the OR goes up, ethnic homophily tendencies are stronger. Table 7.4 presents the ORs for the four survey countries and the standard deviations.

The key insight of our analyses is that almost all ethnic groups have an ethnic homophily tendency, that is, most youth prefer to befriend peers from their own ethnic group. This pattern holds true for majority groups in all four countries, and for most minority groups too. In each country, the majority group has a clear tendency to befriend majority peers as opposed to minority peers (i.e. average OR > 1). The homophily parameters of the majority youth are strikingly similar across countries, that is, around 1.4 to 1.8. This suggests that—besides the structural opportunities which clearly promote intragroup ties for majority members—there seems to be a more or less stable tendency across countries to develop ties with co-ethnic majority youth.

What about the minority groups? In England, adolescents from Pakistan (OR= 3), and especially Eastern Europe (OR = 7.9) have high homophily figures on average. They stand in sharp contrast to the adolescents from Caribbean backgrounds, who more often have inter-ethnic friendships as opposed to intra-ethnic friendships, controlling for opportunity (OR = 0.7). Overall, it seems that in England the majority population, as well as virtually all minority groups, prefer to have ties with co-ethnics on average.

In Sweden, the pattern seems very similar. We see that the majority group, as well as other groups, have high homophily tendencies. In particular, the friendship choices of Turks tend to occur within their own group—again when controlling for opportunity. Turkish adolescents in Sweden have a very high homophily tendency (OR = 12.1). In the Netherlands, this in-group preference of Turks is even stronger (OR = 22.8). In Germany, their homophily rate is also high (4.2). Other groups that stand out as having strong co-ethnic preferences for friendship are Serbs in Germany (6.8) and Moroccans in the Netherlands (3.6). On the other hand, we find groups with much lower preferences for in-group ties, such as the Surinamese in the Netherlands, who do not show any evidence for homophily (OR = 0.9).

Another important consistent finding is that the standard deviations are relatively large. In fact, the distribution of Moody’s OR is wide and relatively skewed. This indicates that homophily varies considerably across school classes: (p.191) many classes are well integrated, whereas others are extremely segregated. For example, the OR is equal or smaller than 1 in 36% of the English classrooms for native students, but equal or larger than 2 in 30% of classes.

Our observations show that inequalities in ethnic group size, and hence in the opportunity structure in class, cannot completely account for the formation of inter-ethnic friendships in this environment. Apparently, above and beyond the opportunities provided in the class context, youth from almost all groups tend to prefer co-ethnic friendships, which results in more overall ethnic segregation than if such friendship choices were made regardless of the ethnic origin of peers. Overall ethnic segregation is therefore the outcome of both inequalities in group size (opportunities) and ethnic homophily (preferences). Moreover, we have seen that not all groups are alike in their ethnic preference: some groups show high levels of ethnic homophily, whereas others do not. Hence, to further understand the underlying causes of ethnic segregation, we need to explain the existence as well as the varying degree of ethnic homophily.

7.7 What Explains Ethnic Homophily?

In the literature, several explanations have been offered to explain ethnic homophily, why this happens and why some groups have more co-ethnic preferences than other groups. It would be too much to discuss each explanation here in detail, however. Elsewhere, we have reported the findings on the role of the school (Smith et al. 2016) and the importance of the neighbourhood in which adolescents live (Kruse et al. 2016). In this chapter, we address two other explanations in more detail, namely the by-product hypothesis and the possible role of parents.

7.7.1 By-product

A first possible explanation for ethnic homophily is that ethnic preferences are in fact a by-product of friendship preferences on dimensions associated with ethnicity. It could be that the strong ethnic homophily among Turkish adolescents in Germany, Sweden and the Netherlands, as suggested by studying revealed preferences, has nothing to do with Turkish peers preferring to befriend peers with the same background. It is possible that such ethnic homophily tendencies occur when Turkish peers have something in common other than their ethnic origin, which makes them likely to become friends. For example, it might be that Turkish adolescents hold more conservative attitudes towards divorce, abortion and homosexuality than other groups. If Turkish youth prefer to befriend peers in class who have conservative attitudes similar to their own, they might be attracted to such peers not because of their ethnicity but because they share the same perspectives on social issues. Thus, if conservative attitudes overlap with ethnicity, this could explain why studying revealed preferences suggest that Turks prefer Turks: in fact it is a by-product of peers preferring to (p.192) befriend like-minded peers. It is thus not ethnicity that matters but characteristics associated with ethnicity.

The by-product hypothesis therefore hinges upon two assumptions. First, there should be homophily in friendship choices based on non-ethnic traits, such as the preference to befriend peers with similar values, lifestyle, religion or socioeconomic background. In other words, friendship choices in class should be driven by other things than ethnicity, such as having the same religion. Secondly, there should be a sufficiently strong overlap, consolidation (Blau 1977; Blau et al. 1982), between ethnicity and the other dimensions that drive friendship (e.g. religion). Under these conditions, it is possible to explain (some part of) the ethnic homophily tendencies that we observe among youth. In literature on ethnic intermarriage and core discussion networks, evidence has been found for the by-product hypothesis (van Tubergen 2015).

Is the by-product hypothesis also supported among adolescents’ ethnic friendship choices in class? Our analysis of the CILS4EU data provides the following picture (Smith et al. 2014). First, we find that adolescents indeed select their friends in class on other criteria than ethnicity. In general, we observe that youth prefer to nominate peers in class as friends when they are similar to themselves. Thus, when an adolescent smokes, he/she is more likely to nominate as friends others who also smoke. Such patterns of homophily appear strong in our study, and they are not restricted to a few dimensions, such as smoking. We find that friendship choices in class are more likely when peers are similar in terms of their substance use (e.g. smoking, drinking), lifestyle (e.g. going out), delinquent activities (e.g. stealing from a shop), attitudes (e.g. about abortion, divorce) and religion. Again, all these tendencies are found when controlling for the opportunity structure in class, and hence reflect (revealed) preference for homophily. This supports the first ingredient of the by-product hypothesis, namely that homophily in friendship choices occurs along other dimensions besides ethnicity (i.e. non-ethnic homophily).

However, these patterns do not explain ethnic homophily. The reason for this is that the overlap between ethnicity and these other dimensions is not as strong within classes, where friends are made, as outside school. To illustrate: Turks in Germany do have more conservative attitudes on average than the majority group with respect to abortion and divorce. However, these are aggregated differences at the population level. When studied at the level of school classes, the differences between Turks and Germans are not so strong to make a difference. Within a school class peers from the same ethnic group are not so similar with respect to their attitudes, substance use, lifestyle, delinquent activities and religion.

The main conclusion, therefore, is that peers who are similar to each other in terms of non-ethnic attributes are more likely to be friends (i.e. non-ethnic homophily), that the non-ethnic attributes are not strongly correlated with ethnicity in class (i.e. no consolidation) and thus that the by-product hypothesis cannot explain ethnic homophily among adolescents in our countries.

(p.193) 7.7.2 Parents

If ethnic homophily is not a by-product of homophily on correlated dimensions, where does it come from? And why does it differ among ethnic minority groups? Using the CILS4EU, we investigated the role of parents (Smith et al. 2015). In the literature on intermarriage, parents are often identified as ‘third-party’ actors who interfere with the marital choices of their children (Kalmijn 1998). Generally speaking, parents often socialise their children so that they develop a stronger preference for co-ethnic spouses, and also disapprove of and possibly forbid interethnic marriages.

Do parents also affect the friendship choices of their children? Do parents disapprove of inter-ethnic friendships? Although the friendships we study here pertain to the class setting, thus largely outside parents’ monitoring, parents may nevertheless affect friendship choices. One possible mechanism is that parents transmit their own ethnic in- and out-group attitudes and practices to their children, which subsequently affects the latter’s ethnic homophily.

Indeed, our analysis provides evidence to suggest that when parents have more inter-ethnic friends themselves, their children show stronger revealed preferences for inter-ethnic friendships in class. Moreover, we find that when parents are less concerned with ‘maintaining ethnic ingroup traditions and customs important’ in society, the ethnic homophily of their children decreases. We find both patterns when controlling for co-ethnic group size in class, hence they cannot be accounted for by unequal opportunities and thus reflect ethnic preferences (Smith et al. 2015). Moreover, both relationships are partly mediated by intergroup attitudes of the children—which we were able to directly measure with the CILS4EU.

The general picture that emerges, thus suggests that when parents are more positive towards other ethnic groups and/or have more inter-ethnic ties, their children copy these values and practices, which results in lower levels of ethnic homophily.

7.8 Conclusions

Ethnic majority and minority youth in England, Germany, the Netherlands and Sweden largely live in parallel worlds. Majority adolescents in particular appear to have very few social ties to their ethnic minority peers. Within the neighbourhood context, they have barely any contact with minorities, and their friendship network almost exclusively consists of majority peers. When we study such friendship networks within the classroom, there is more ethnic mixing, but majority youth still overwhelmingly befriend their co-ethnic peers.

The strong levels of ethnic segregation among majority youth are driven by both opportunities and preference. Being a much larger group than the ethnic minority groups, many majority youth are surrounded by co-ethnic peers in class. These structural forces inhibit the formation of cross-ethnic ties among majority (p.194) youth. However, when taking these structural forces into account, it also appears that they prefer to make ties within their own group. Thus, beyond ethnic majority youth being a bigger group, ethnic homophily is another social force which promotes group closure among them. Evidence suggests that such boundaries are not a by-product of anything else that majority peers might have in common. Hence, it seems that ethnicity acts as a strong social category in shaping friendships in class.

Ethnic minority youth live in more mixed worlds than their majority peers. They tend to cluster together in their own co-ethnic neighbourhoods and have co-ethnic friends, but are not so exclusively co-ethnic as their majority peers. Being smaller in group size, ethnic minority youth live a different reality: they meet majority youth in their neighbourhood and in school more often, and hence their networks are more diverse. These structural forces are not alike for all ethnic minority groups, however, as some groups are larger than others. Those minority children who attend schools with more co-ethnics likewise develop more coethnic friendships at school. Furthermore, among minorities we also find strong preferences to befriend co-ethnics. This means that ethnic group closure is supported by both majority and minority youth.

Parents play a role in overcoming these ethnic boundaries. We find that when parents are more supportive of other ethnic groups, and when they themselves have more inter-ethnic friends, their children internalise their values and are more likely to nominate more inter-ethnic friends in class. It is possible that this intergenerational transmission of inter-ethnic values and practices has positive externalities. Peers affect each other’s values and practices as well, and hence such inter-ethnic attitudes and behaviour might diffuse through peer networks, thereby affecting many more people. Furthermore, internalised values and practices can affect the behaviour of the child outside the class, through, for example, the choice to live in more ethnically mixed neighbourhoods.

References

Bibliography references:

Allport, G.W. (1954), The Nature of Prejudice (Cambridge MA, Addison-Wesley Publishing Company).

Baerveldt, C., van Duijn, M.A.J., Vermeij, L. & van Hemert, D.A. (2004), ‘Ethnic Boundaries and Personal Choice: Assessing the Influence of Individual Inclinations to Choose Intra-Ethnic Relationships on Pupils’ Networks’, Social Networks, 26(1): 55–74.

Blau, P.M. (1977), Inequality and Heterogeneity (New York, Free Press).

Blau, P.M. (1994), Structural Contexts of Opportunities (Chicago IL, University of Chicago Press).

Blau, P.M., Blum, T.C. & Schwartz, J.E. (1982), ‘Heterogeneity and Intermarriage’, American Sociological Review, 47: 45–62.

Chiswick, B.R. (2008), ‘The Economics of Language: An Introduction and Overview’, Discussion Paper IZA DP No. 3568, Forschungsinstitut zur Zukunft der Arbeit Institute for the Study of Labor.

(p.195) De Miguel Luken, V. & Tranmer, M. (2010), ‘Personal Support Networks of Immigrants to Spain: A Multilevel Analysis’, Social Networks, 32(4): 253–62.

Dribe, M. & Lundh, C. (2011), ‘Cultural Dissimilarity and Intermarriage: A Longitudinal Study of Immigrants in Sweden 1990–2005’, International Migration Review, 45(2): 297–324.

Kalmijn, M. (1998), ‘Intermarriage and Homogamy: Causes, Patterns, Trends’, Annual Review of Sociology, 24: 395–421.

Kalmijn, M. & van Tubergen, F. (2006), ‘Ethnic Intermarriage in the Netherlands: Confirmations and Refutations of Accepted Insights’, European Journal of Population, 22: 371–97.

Kalter, F. & Schroedter, J.H. (2010), ‘Transnational Marriage among Former Labour Migrants in Germany’, Zeitschrift Für Familienforschung–Journal of Family Research, 22(1): 11–36.

Kanas, A., Chiswick, B.R., Lippe, T. & Tubergen, F. (2012), ‘Social Contacts and the Economic Performance of Immigrants: A Panel Study of Immigrants in Germany’, International Migration Review, 46(3): 680–709.

Kruse, H. & Jacob, K. (2014), Sociometric Fieldwork Report: Children of Immigrants Longitudinal Survey in 4 European Countries: Wave 1–2010/2011, V1.1.0 (Mannheim, Mannheim University).

Kruse, H., Smith, S., van Tubergen, F. & Maas, I. (2016), ‘From Neighbours to School Friends? How Adolescents’ Place of Residence Relates to Same-Ethnic School Friendships’, Social Networks, 44: 130–42.

Lucassen, L. & Laarman, C. (2009), ‘Immigration, Intermarriage and the Changing Face of Europe in the Post War Period’, History of the Family, 14(1): 52–68.

Martinovic, B., van Tubergen, F. & Maas, I. (2009a), ‘Changes in Immigrants’ Social Integration during the Stay in the Host Country: The Case of Non-Western Immigrants in the Netherlands’, Social Science Research, 38(4): 870–82.

Martinovic, B., van Tubergen, F. & Maas, I. (2009b), ‘Dynamics of Interethnic Contact: A Panel Study of Immigrants in the Netherlands’, European Sociological Review, 25(3): 303–18.

Moody, J. (2001), ‘Race, School Integration, and Friendship Segregation in America’, American Journal of Sociology, 107(3): 679–716.

Muttarak, R. & Heath, A. (2010), ‘Who Intermarries in Britain? Explaining Ethnic Diversity in Intermarriage Patterns’, British Journal of Sociology, 61(2): 275–305.

Pettigrew, T.F. & Tropp, L.R. (2006), ‘A Meta-Analytic Test of Intergroup Contact Theory’, Journal of Personality and Social Psychology, 90(5): 751–83.

Putnam, R. (2000), Bowling Alone: The Collapse and Revival of the American Community (New York, Simon & Schuster).

Schaeffer, M. (2013), ‘Inter-Ethnic Neighbourhood Acquaintances of Migrants and Natives in Germany: On the Brokering Roles of Inter-Ethnic Partners and Children’, Journal of Ethnic and Migration Studies, 39(8): 1219–40.

Smith, S., Maas, I. & van Tubergen, F. (2014), ‘Ethnic Ingroup Friendships in Schools: Testing the By-Product Hypothesis in England, Germany, the Netherlands and Sweden’, Social Networks, 39: 33–45.

Smith, S., Maas, I. & van Tubergen, F. (2015), ‘Parental Influence on Friendships between Native and Immigrant Adolescents’, Journal of Research on Adolescence, 25(3): 580–91.

Smith, S., van Tubergen, F., Maas, I. & McFarland, D. A. (2016), ‘Ethnic Composition and Friendship Segregation: Differential Effects for Adolescent Natives and Immigrants’, American Journal of Sociology, 121(4): 1223–72.

(p.196) Stark, T.H. & Flache, A. (2012), ‘The Double Edge of Common Interest Ethnic Segregation as an Unintended Byproduct of Opinion Homophily’, Sociology of Education, 85(2): 179–99.

Van Tubergen, F. (2007), ‘Religious Affiliation and Participation among Immigrants in a Secular Society: A Study of Immigrants in the Netherlands’, Journal of Ethnic and Migration Studies, 33: 747–65.

Van Tubergen, F. (2014), ‘Size and Socio-Economic Resources of Core Discussion Networks in the Netherlands: Differences by National-Origin Group and Immigrant Generation’, Ethnic and Racial Studies, 37(6): 1020–42.

Van Tubergen, F. (2015), ‘Ethnic Boundaries in Core Discussion Networks: A Multilevel Social Network Study of Turks and Moroccans in the Netherlands’, Journal of Ethnic and Migration Studies, 41(1): 101–16.

Van Tubergen, F. & Mentjox, T. (2014), ‘Minority Language Proficiency of Adolescent Immigrant Children in England, Germany, the Netherlands, and Sweden’, KZfSS Kölner Zeitschrift Für Soziologie Und Sozialpsychologie, 66(1): 241–62.

Vermeij, L., van Duijn, M.A.J. & Baerveldt, C. (2009), ‘Ethnic Segregation in Context: Social Discrimination among Native Dutch Pupils and their Ethnic Minority Classmates’, Social Networks, 31(4): 230–9.

Vroome, T., Verkuyten, M. & Martinovic, B. (2014), ‘Host National Identification of Immigrants in the Netherlands’, International Migration Review, 48(1): 76–102.

Windzio, M. (2012), ‘Integration of Immigrant Children into Inter-Ethnic Friendship Networks: The Role of “Intergenerational Openness”’, Sociology, 46(2): 258–71.

Table A7.1. Multivariate analysis of N same-ethnic friends (Poisson regression): England

Model 1

Model 1

Origin groups (ref.: majority)

   North/West/South Europe

−2.672***

−1.741***

(0.321)

(0.343)

   Eastern Europe

−2.122***

−1.222**

(0.454)

(0.433)

   Caribbean

−2.234***

−1.321***

(0.310)

(0.336)

   Middle East & North Africa

−2.736***

−1.762***

(0.431)

(0.458)

   Pakistan

−0.738***

−0.199

(0.174)

(0.124)

   Sub-Saharan Africa

−1.683***

−0.804***

(0.210)

(0.243)

   Asia

−1.217***

−0.367

(0.260)

(0.219)

   India

−1.105***

−0.367

(0.314)

(0.272)

   Other

−1.767***

−0.875**

(0.297)

(0.297)

(p.197) Generational status (ref.: majority and 2nd generation)

   Born abroad

−0.011

0.076

(0.164)

(0.138)

   Child of intermarriage

0.001

0.051

(0.119)

(0.104)

   Child of transnational marriage

−1.065***

−0.892***

(0.251)

(0.249)

Gender (ref.: male)

   Female

−0.139***

−0.107*

(0.053)

(0.052)

% same-ethnic students in class

0.013***

(0.002)

Intercept

0.997***

−0.109

(0.039)

(0.150)

No. of observations

3,240

3,240

Pseudo R2

0.159

0.207

Note: Design weighted, accounting for clustering; standard errors in parentheses;

(*) p < 0.05,

(**) p < 0.01,

(***) p < 0.001.

Excluded data: classes with invalid sociometric data (Kruse & Jacob 2014), students in small classes (fewer than ten).

Table A7.2. Multivariate analysis of N same-ethnic friends (Poisson regression): Germany

Model 1

Model 2

Origin groups (ref.: majority)

   North/West/South Europe

−2.588***

−1.452**

(0.450)

(0.465)

   Italy

−2.692***

−1.551**

(0.516)

(0.503)

   Eastern Europe

−3.642***

−2.488***

(0.313)

(0.329)

   Poland

−2.270***

−1.144***

(0.316)

(0.327)

   Russia

−2.406***

−1.325***

(0.321)

(0.313)

   Serbia

−1.672***

−0.571*

(0.285)

(0.278)

   Middle East & North Africa

−1.973***

−0.899***

(0.185)

(0.188)

   Turkey

−0.938***

−0.043

(0.078)

(0.087)

(p.198)    Sub-Saharan Africa

−2.222**

−1.053

(0.709)

(0.699)

   Asia

−4.271***

−3.101***

(0.552)

(0.539)

   Other

−2.401**

−1.299

(0.797)

(0.775)

Generational status (ref.: majority and 2nd generation)

   Born abroad

0.156

0.174

(0.145)

(0.131)

   Child of intermarriage

−0.103

−0.101

(0.156)

(0.136)

   Child of transnational marriage

−0.998***

−0.915***

(0.245)

(0.222)

Gender (ref.: male)

   Female

−0.059

−0.051

(0.036)

(0.033)

% same-ethnic students in class

0.015***

(0.001)

Intercept

1.177***

−0.049

(0.037)

(0.081)

No. of observations

3,957

3,957

Pseudo R2

0.211

0.267

Note: Design weighted, accounting for clustering; standard errors in parentheses;

(*) p < 0.05,

(**) p < 0.01,

(***) p < 0.001.

Excluded data: classes with invalid sociometric data (Kruse & Jacob 2014), students in small classes (fewer than ten).

Table A7.3. Multivariate analysis of N same-ethnic friends (Poisson regression): The Netherlands

Model 1

Model 2

Origin groups (ref.: majority)

   North/West/South Europe

−2.391***

−1.411***

(0.430)

(0.396)

   Eastern Europe

−18.072***

−15.389***

(0.401)

(0.400)

   Caribbean

−2.973***

−1.949***

(0.541)

(0.536)

   Suriname

−2.547***

−1.556***

(0.307)

(0.317)

   Middle East & North Africa

−2.520***

−1.482**

(0.505)

(0.492)

(p.199)    Morocco

−1.013***

−0.179

(0.148)

(0.154)

   Turkey

−1.410***

−0.460*

(0.180)

(0.180)

   Sub-Saharan Africa

−3.763***

−2.725***

(0.519)

(0.528)

   Asia

−2.158***

−1.159

(0.543)

(0.538)

   Other

−5.838***

−4.803***

(1.076)

(1.065)

Generational status (ref.: majority and 2nd generation)

   Born abroad

−0.429

−0.424

(0.308)

(0.293)

   Child of intermarriage

−0.496

−0.431

(0.321)

(0.315)

   Child of transnational marriage

−1.222**

−1.171**

(0.372)

(0.360)

Gender (ref.: male)

   Female

0.002

0.043

(0.040)

(0.031)

% same-ethnic students in class

0.012***

(0.001)

Intercept

1.145***

0.046

(0.033)

(0.129)

No. of observations

3,907

3,907

Pseudo R2

0.201

0.226

Note: Design weighted, accounting for clustering; standard errors in parentheses;

(*) p < 0.05,

(**) p < 0.01,

(***) p < 0.001.

Excluded data: classes with invalid sociometric data (Kruse & Jacob 2014), students in small classes (fewer than ten).

Table A7.4. Multivariate analysis of N same-ethnic friends (Poisson regression): Sweden

Model 1

Model 2

Origin groups (ref.: majority)

   North/West/South Europe

−1.950***

−0.990**

(0.308)

(0.310)

   Finland

−1.669**

−0.734

(0.591)

(0.565)

   Eastern Europe

−1.606***

−0.747***

(0.257)

(0.220)

   Bosnia & Herzegovina

−1.950***

−0.972***

(0.241)

(0.243)

    (p.200) Kosovo & Albania

−1.264***

−0.314

(0.285)

(0.284)

   Middle East & North Africa

−1.437***

−0.574***

(0.150)

(0.161)

   Iraq

−1.844***

−0.904***

(0.188)

(0.188)

   Turkey

−1.452***

−0.501*

(0.243)

(0.240)

   Sub-Saharan Africa

−2.048***

−1.067***

(0.251)

(0.246)

   Somalia

−1.098***

−0.282

(0.189)

(0.164)

   Asia

−2.179***

−1.187***

(0.284)

(0.274)

   Other

−1.839***

−0.843*

(0.349)

(0.341)

Generational status (ref.: majority and 2nd generation)

   Born abroad

0.017

0.023

(0.127)

(0.118)

   Child of intermarriage

−0.385

−0.344

(0.315)

(0.302)

   Child of transnational marriage

−0.974***

−0.915***

(0.285)

(0.273)

Gender (ref.: male)

   Female

−0.142***

−0.157***

(0.040)

(0.039)

% same-ethnic students in class

0.014***

(0.001)

Intercept

1.085***

0.020

(0.029)

(0.086)

No. of observations

3,904

3,904

R2 (Model 1: pseudo)

0.209

0.245

Note: Design weighted, accounting for clustering; standard errors in parentheses;

(*) p < 0.05,

(**) p < 0.01,

(***) p < 0.001.

Excluded data: classes with invalid sociometric data (Kruse & Jacob 2014), students in small classes (fewer than ten).

Notes:

(1) We excluded data of students in classes with invalid sociometric data (Kruse & Jacob 2014) and students in small classes (fewer than ten students).