ethnic diversity and solidarity - ferry koster · 2018-08-16 · british library cataloguing in...

Ethnic Diversity and Solidarity

Ethnic Diversity and Solidarity:

A Study of their Complex Relationship

Edited by

Paul de Beer and Ferry Koster

Ethnic Diversity and Solidarity: A Study of their Complex Relationship Edited by Paul de Beer and Ferry Koster This book first published 2017 Cambridge Scholars Publishing Lady Stephenson Library, Newcastle upon Tyne, NE6 2PA, UK British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Copyright © 2017 by Paul de Beer, Ferry Koster and contributors All rights for this book reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the copyright owner. ISBN (10): 1-4438-8171-6 ISBN (13): 978-1-4438-8171-5

CONTENTS Acknowledgements ................................................................................... vii Introduction ................................................................................................. 1 Paul de Beer Chapter One ............................................................................................... 13 Towards an Interdisciplinary Theory of Solidarity Paul de Beer, Maarten Berg, Laurens Buijs, Ferry Koster and Dorota Lepianka Chapter Two .............................................................................................. 43 Why Would Diversity Harm Solidarity? Paul de Beer Chapter Three ............................................................................................ 55 Heterogeneity and Human Sociality Ferry Koster Chapter Four .............................................................................................. 95 Diversity, Similarity, and Solidarity: Results of a Vignette Study Maarten Berg and Ferry Koster Chapter Five ............................................................................................ 127 More or Less Strangers: Diversity in the Dutch Media Depiction of Ethnic Minorities and its Implications for Interethnic Solidarity Dorota Lepianka Chapter Six .............................................................................................. 155 Solidarity in a Multicultural Neighbourhood: Results of a Field Experiment Paul de Beer and Maarten Berg

Contents

vi

Chapter Seven .......................................................................................... 189 Conclusion Paul de Beer Affiliation of the Authors ........................................................................ 199

ACKNOWLEDGEMENTS This book is the result of a research project that was conducted by the Amsterdam Institute for Advanced Labour Studies (AIAS) (www.uva-aias.net) of the University of Amsterdam and was made possible by a grant of Instituut Gak (www.instituutgak.nl). We thank Aafke Komter, Ruud Koopmans and Wim van Oorschot for their very helpful comments and suggestions. A very special thanks to Merle Zwiers for her invaluable assistance during the research project.

INTRODUCTION

PAUL DE BEER In the past decade, both policy makers and academic scholars have called attention to the possible detrimental effects of ethnic heterogeneity on socially desirable outcomes, such as public goods provision, economic growth, trust, social capital and solidarity (e.g., Alesina and LaFerrara 2005; Leigh 2006; Putnam 2007; Habyarimana et al. 2007; Hooghe et al. 2009). In their view, in ethnically heterogeneous neighbourhoods, cities, regions or countries, citizens find it more difficult to identify with each other, trust each other less and are more reluctant to co-operate. This would be related to the perceived social distance between people who belong to different ethnic groups (cf. van Kempen and Bolt 2009). Some scholars claim that ethnic diversity might also hamper trust and co-operation between people within the same ethnic group (Putnam 2007). The consequences of ethnic heterogeneity might thus be broad and far-reaching. Ethnically diverse neighbourhoods may be characterized by less community activities and be plagued by vandalism and crime (Wilson 1987; Ellen and Turner 1997; Morenoff et al. 2001; Haynie et al. 2006), ethnically diverse cities may lack effective governance and may have less public provisions such as parks and sports facilities , and ethnically diverse countries may offer its citizens less protection by social services and income transfers and may turn out to be less prosperous in the long run.

As a consequence of immigration and, in addition, fertility rates which are often higher among ethnic minority groups, ethnic diversity is growing in many regions and countries (Zorlu and Hartog 2002). If the worries expressed above are justified, this may have serious detrimental effects. Ultimately, increasing ethnic diversity might even result in a higher incidence of riots, civil war and ethnic cleansing. Both from a scientific and from a societal point of view it is, therefore, of great importance to know more about the societal consequences of ethnic diversity. Up to date, the results of empirical studies are, at best, mixed. Some studies confirm the expectation that ethnic diversity erodes trust, social capital and public goods provisions (cf. Kleinhans et al. 2007), while other studies find no

Introduction

2

relationship at all between ethnic diversity and various outcomes at the community level, and a few studies even find positive effects of ethnic diversity (cf. Portes and Zhou 1993; Portes 1995).

This volume adds to the existing knowledge of the effects of ethnic diversity in three important ways.

Firstly, of the many societal outcomes of ethnic diversity that have been studied, this study focuses on just one, namely social solidarity, rather than social capital, trust or public goods provision. By solidarity we mean any act that purposefully benefits another person at a cost for the agent, without any guarantee of an equivalent return. While social capital, trust or co-operation does not impose any cost on the actor, solidarity does. Even though this cost may be repaid in the future, there is no guarantee that this will happen. Thus, solidarity is a stronger indicator of pro-social behaviour than most other indicators. Consequently, an empirical study that focuses on the intention to act solidary or on solidary behaviour itself offers a more robust test of the effects of ethnic diversity than most previous studies.

Secondly, in most previous studies ethnic diversity or heterogeneity is measured by a single, one-dimensional indicator, such as the fractionalization index, which measures the probability that two persons, meeting each other at random, belong to different ethnic groups. In this volume, we take a broader perspective on diversity by using various ways to measure ethnic diversity. Thus, we do not assume a priori that there is one best way to measure diversity or that there is a simple linear relationship between ethnic diversity and solidarity. To illustrate, we do not assume a priori that there is a dichotomy between natives (autochthones) and non-natives or immigrants (allochthones), but we also examine the possible tensions between various ethnic groups and consider the possibility that religious differences or linguistic differences matter more than ethnic differences.

Thirdly, this study applies various research methods, including a field experiment in a multicultural neighbourhood, a content analysis of mass media, a vignette study of a nation-wide survey and a statistical analysis of a large cross-country survey. The use of different research methods allows us to cross-examine the outcomes of the various studies that this volume reports on. Thus, the conclusions we draw do not depend on one particular research method and one particular pool of data, but are corroborated by various methods.

Ethnic Diversity and Solidarity: A Study of their Complex Relationship

3

In this introductory chapter, we will take a closer look at the two central concepts of this volume, viz. solidarity and diversity.

What is solidarity?

As mentioned above, we define a solidary act as any act that purposefully benefits another person at a cost for the agent, without any guarantee of an equivalent return. This definition of solidarity includes a broad range of pro-social behaviours, such as alms-giving to a beggar, helping your neighbour, raising children, voluntary community work, donating money to a charity organization, paying an insurance premium and paying taxes. These examples show that solidarity can be both informal and formal, either voluntary or mandatory, and one-sided as well as two-sided or reciprocal (cf. De Beer and Koster 2009, chapter 2).

Informal solidarity, which one might also call ‘warm’ solidarity, springs from a direct involvement and sympathy with concrete other persons, for whom one has ‘warm feelings’. The most intense forms of informal solidarity are usually found within the family: between husband and wife, and between parents and children. This solidarity can be so strong that one is prepared to sacrifice one’s life to save another, e.g., a parent that risks her life in trying to save her drowning child. Less intense forms of informal solidarity are often found between good friends, neighbours or colleagues. The compassion expressed by giving alms to a beggar is also an example of informal solidarity.

Formal or ‘cold’ solidarity originates from commitment to or sympathy with anonymous others, whom one does not know personally, but to whom one is nevertheless connected through a formal bond. Some examples are the solidarity with poor fellow citizens, which are supported by a social benefit system to which everybody contributes, solidarity with the unlucky person whose house burns down and who is compensated by the insurance company to which one regularly pays an insurance premium, and the solidarity with poor people in developing countries to which your government pays development aid, which is financed from taxes.

Whereas informal solidarity is expressed by the direct relation between persons, the formal solidarity between persons who do not know each other is always mediated by an institution such as the state or an insurance company.

Introduction

4

A solidarity act can be voluntary as well as compulsory. Showing solidarity voluntarily is a choice; compulsory solidarity is enforced, usually by the state. Compulsory solidarity does not rule out that one is actually prepared to act out of solidarity, but this is, contrary to voluntary solidarity, not a prerequisite.

There is often a close relationship between the (in)voluntariness of solidarity and the organization of solidarity. However, not all informal solidarity is voluntary and not all formal solidarity is compulsory. Contributing a premium to a fire and theft insurance is an example of formal voluntary solidarity, and the solidarity of parents with their children is informal but compulsory, since they are legally obliged to take care of their children.

In case of two-sided solidarity the agent expects, on balance, to benefit just as much from others as s/he contributes to others. It should be stressed that this only refers to the expected, ex ante, balance between contributions and receipts. The actual, ex post, contributions and receipts will, in general, differ. Indeed, this is the feature which distinguishes two-sided solidarity from an ordinary market exchange, which one undertakes knowing that the benefits outweigh the costs. In case of two-sided solidarity one does not know beforehand whether one will end up being a net-contributor or a net-receiver.

In case of one-sided solidarity a person expects beforehand, ex ante, that his/her contributions and receipts will not balance. This is clearly the case if one expects nothing in return for one’s contribution. In giving alms to a beggar or in donating money for the victims of an earthquake in a faraway country, one cannot reasonably expect to get anything in return. In this sense, one-sided solidarity is always unselfish, although one may of course feel satisfied or get a ‘warm glow’ from showing solidarity. The taxes levied on rich persons for financing social assistance, from which they will probably never benefit themselves, is also a form of one-sided solidarity.

Our definition restricts solidarity acts to relations between persons. The persons towards whom one acts solidary may be either concrete persons or anonymous representatives of a group (e.g., the victims of a natural disaster). In the last case, one could also say that solidarity is a relationship between a person and a group.

The term solidarity is sometimes also used for the contribution of an individual to a public good or an ideal. For instance, a financial


5

contribution (by means of taxes) to the construction of roads or taking part in the activities of a political party or an environmental organization can also be called solidary acts. We will, however, leave these kinds of solidary acts aside and limit ourselves to solidarity between persons.

In our study, solidarity is merely a descriptive term for a particular kind of behaviour. We do not address the normative question whether solidarity is desirable from a moral point of view. Although the term solidarity probably has a positive, desirable flavour for most readers, the relevance and valence of our research does not depend on this positive connotation.

Since our definition of solidarity refers to a particular kind of behaviour, the underlying values, motives or attitudes do not determine whether a particular act is considered to be a solidary act. Thus, a solidary act need not be motivated by altruism or benignity, but may also arise from selfish or egotistic motives, as long as the act benefits others. This does not mean that we are not interested in the motives for solidarity and the feelings or emotions that trigger solidary behaviour. On the contrary, one of the main purposes of this research is to find out which attitudes, feelings or emotions motivate solidary behaviour. However, explaining solidary attitudes or feelings is not our goal as such, but only a way to understand solidary behaviour.

What is diversity?

Diversity or heterogeneity – two terms which we will use interchangeably – can be simply defined as the negation of uniformity or homogeneity. But, while a group can be uniform or homogeneous in only one way, that is, that all group members are identical with respect to a particular characteristic, there are many ways in which a group can be diverse or heterogeneous.1 Most measures for group diversity or heterogeneity that are used in the literature start from a dichotomous relationship between group members: two arbitrary group members are either equal or different with respect to a particular characteristic (ethnicity, mother country, language, etc.). Next, an indicator is constructed to aggregate the dichotomous relationships between all possible pairs of persons. A well-known indicator is the fractionalization index or, alternatively, one minus

1 Cf. the distinction between income equality and inequality: income equality can only be measured in one way (all incomes are equal), but there are many different ways to measure income inequality, e.g., the Gini coefficient, the Theil coefficient, the coefficient of variation, the D9/D1 decile ratio, etc..

Introduction

6

the Herfindahl index, which measures the probability that two randomly selected individuals belong to a different (ethnic) group. Although this index may be a useful indicator for (ethnic) diversity in many contexts, it has two serious limitations. First, it is based on the assumption that there is no gradation in the extent of difference between two persons. Two persons are either the same or different, but not a little different or very different. As we know from studies of social distance, the extent to which people experience or perceive a distance to people from other ethnic groups, may vary considerably depending on the particular ethnic group. For example, the (perceived) distance between a native Dutch and a fellow citizen with a Surinamese background may be smaller than the distance with a Moroccan Dutch. In studying the effects of ethnic heterogeneity, it is important to take these differences into account.

Secondly, the fractionalization index is only one way to aggregate differences between individuals into an overall indicator of the heterogeneity of a group or community. It assumes, for example, that the diversity of a community consisting of many different, relatively small ethnic groups is larger than the diversity of a community that is made up of two large ethnic groups. In a literal sense, this is of course true, but when studying the societal impact of ethnic heterogeneity, two large ethnic groups that confront each other may have worse consequences for social cohesion or solidarity than when a larger number of smaller groups cohabitate. To illustrate, compare the tensions between the Flemings and the Walloons in Belgium to the more harmonious relationship between the various ethnic groups in Switzerland.

It is also important to take into account, that the ethnic diversity of a community may vary considerably, depending on the level at which we measure diversity. To illustrate, an ethnically homogeneous neighbourhood may be part of an ethnically diverse city, which is situated in a largely homogeneous region in an ethnically divided country. Depending on the level at which one measures ethnic diversity, one may find quite different societal effects.

The concept of ethnicity is, of course, also far from self-evident. It can refer to a common ancestry or tradition, to people with the same culture, religion, language, et cetera. We do not claim that ethnicity has an objective, essentialist meaning, but only that ethnicity is one of the dimensions along which people tend to categorize each other and one of the characteristics that people use to evaluate other people whom they do not know in person (Brubaker 2002). Thus, the fact that another person is


7

(perceived as) Turkish, may conjure up a particular image of that person, referring, for example, to his/her competence (‘Turkish people are diligent’) or his/her sympathy (‘Turkish people are callous’). Whether this stereotype is correct – which, as we know, is usually not the case – is not relevant for our research, but only whether people act on it. Do people show less solidarity towards people from another ethnic group because they have an unfavourable image of that group? For this reason, we also study (in chapter five) the images that are depicted of various ethnic groups in the Dutch mass media.

Outline of the book

This book is divided into two parts. The first part, consisting of chapters one and two, sketches a theoretical framework that can be used to study the relationship between diversity and solidarity. The second part, consisting of chapters three to six, reports on a number of empirical studies of diversity and its relation to solidarity. The book concludes with a chapter which draws some general conclusions.

The first chapter sketches the outlines of an interdisciplinary theoretical framework for studying solidarity. The chapter first briefly discusses the explanation of solidary behaviour offered by five scientific disciplines: sociology, anthropology, social psychology, economics and socio-biology. Next, it combines the common and complementary elements of these disciplines, to construct an interdisciplinary framework of solidary behaviour. This framework distinguishes between other-regarding and self-regarding motives for solidary behaviour on the one hand, and between particularistic and generalized kinds of solidarity on the other hand. These two dimensions give rise to four ideal-typical kinds of solidarity: empathic solidarity, bilateral solidarity, multilateral solidarity and normative solidarity. The chapter concludes that most solidary acts in real life are a mix of these four ideal-typical kinds of solidarity.

The second chapter discusses, from a theoretical point of view, which kind of effects ethnic diversity is expected to have on solidarity. Although many recent studies start from the assumption that ethnic diversity is detrimental to social goods, such as trust, social capital and solidarity, this chapter argues that, from a theoretical perspective, the relationship between ethnic diversity and solidarity is ambiguous. The chapter distinguishes between the aggregate effect of an ethnic difference between pairs of individuals, on the one hand, and the direct effect of ethnic diversity as a characteristic of a group or community, on the other hand. It concludes

Introduction

8

that solidarity is possible both between similar and between different persons and both in ethnically homogeneous and in heterogeneous groups. Very large differences or very large heterogeneity are generally detrimental to solidarity. This is simply so, because solidarity presupposes recognition of the other or some common purpose, interest or values. However, this does not imply that the more similar persons are the stronger will be their willingness to act solidary towards each other. This depends on the motives for solidary behaviour and on the possibility (and willingness) to identify with others.

Chapter three investigates the relationship between ethnic diversity and a number of solidarity intentions (the willingness to contribute to the welfare state) using international comparative data. Prior studies rely either on archival data at the national level or use census data at the neighbourhood level within a single country. Both approaches have some limitations. The first approach does not allow to investigate variation in diversity within countries and the second approach misses the possibility to investigate cross national differences. This chapter brings these two approaches closer together by constructing diversity measures based on the European Social Survey (ESS). The ESS includes individual level data that allow replicating earlier measures of ethnic, linguistic, and religious diversity for thirty European countries. Furthermore, since respondents are asked to indicate in what region they live, it is possible to construct disaggregated measures at the regional level. Our analyses show that regional diversity is more strongly related to different kinds of sociality than diversity at the national level.

Chapter four presents the outcomes of a so-called vignette analysis, based on a representative survey among the Dutch population, to analyse the preferences of the respondents for helping fellow citizens belonging to different ethnic groups. As resources are scarce and solidarity can never be completely unrestricted, we need to think about whom to direct our solidarity at. For this chapter, we conducted two vignette analyses. Respondents were asked to express their preferences for helping fellow citizens with varying characteristics. Because each vignette (a fictional person) is constructed from several characteristics, this methodology is less vulnerable to socially desirable answers. In the public solidarity study, we varied eight characteristics: the sex and age of the beneficiary, his/her ethnic background, working experience, type of welfare state entitlement, willingness to do volunteer work, familiarity in the neighbourhood and the number of under-age children. We identify the variables that foster (public) solidarity, and also study the interaction with the characteristics of the


9

respondent. In the private solidarity study (about concrete helping situations directed at neighbours) the same basic design was used. Five attributes varied between the vignettes: sex, age, ethnic background, interactions in the neighbourhood and willingness to reciprocate. We found that people systematically favour certain beneficiaries over others. In particular, natives were more willing to help people with a Surinamese than with a Moroccan background. From these findings it is likely that persons are motivated by motives such as helping the needy, rewarding deservingness and favouring similarity, including ethnic similarity. The same mechanisms seemed to underlie both public and private solidarity, although a few interesting differences emerged as well.

Since the attitudes of citizens towards people with a different ethnic background depend strongly on their image of various ethnic groups, chapter five analyses the images of ethnic minority groups as depicted in some of the most widely read and viewed Dutch news media. The investigation rests on the assumption that by voicing specific norms and values, and by presenting a particular image of social life in a society, the media contribute to creating and maintaining symbolic boundaries between groups, thus influencing the nature of inter-group relations. The analyses revolve around the question of diversity in the media presentation of ethnic minorities, focusing on the visibility of different ethnic groups in the Dutch news media (diversity in presence) as well as the content of their portrayal (diversity in presentation). The results show an imbalance in the media presence of various groups and some noteworthy discrepancies in the content of their presentation, with Turks enjoying the most favourable descriptions and Moroccans the most unfavourable ones. The analyses uncover important nuances in the media construction of ethnic minorities, showing how different aspects of negativity (e.g. hostility, illegality and/or incompetence) take precedence in the description of various ethnic groups. The results of the media analysis are juxtaposed with some factual information on the “performance” of various groups in Dutch society.

Chapter six studies the effects of ethnic diversity of a group on the solidarity between the group members by conducting a field experiment in the multicultural Dapper neighbourhood in the city of Amsterdam, the Netherlands. In the experiment, the participants played a game in groups with varying ethnic composition. The experiment allows to discriminate between Putnam’s (2007) constrict theory, which states that ethnic diversity hurts in-group solidarity, and the alternative conflict theory, which maintains that an ethnic difference between two persons results in

Introduction

10

less solidarity. We find some evidence for the latter theory, but not for Putnam’s constrict thesis. A difference in ethnic background between two players has a strong negative impact on the gift they bestow each other in the experiment. This result provides support for the conflict theory, which asserts that members of different ethnic groups discriminate against each other.

We also find an effect of the ethnic composition of a group on solidarity, but there is no straightforward linear relationship between ethnic diversity and solidary behaviour. We find no evidence that group diversity reduces in-group solidarity. On the contrary, in-group solidarity among natives is the lowest in fully native groups. Moderate ethnic diversity increases in-group solidarity, although maximum diversity reduces it again. We suggest that these results may be explained by the fact that individuals try to avoid belonging to a minority within their group.

In chapter seven we draw some general conclusions from the empirical studies in the previous chapters. We conclude that ethnicity is indeed an important factor in understanding patterns of solidarity. However, there is not a simple linear relationship between ethnic diversity and solidarity. Even though ethnic difference in itself may be a source of discrimination, one cannot conclude from this that increasing ethnic diversity will necessarily result in less solidarity. Under particular circumstances, ethnic diversity may even be beneficial for overall solidarity.

References

Alesina, A., and LaFerrara, E. (2002). Who trusts others? Journal of Public Economics, 85, 207-234.

Brubaker, R. (2002). Ethnicity without groups. Archives Européenes de Sociologie, 43, 163-189.

de Beer, P., and Koster, F. (2009). Sticking together or falling apart? Solidarity in an era of individualization and globalization. Amsterdam: Amsterdam University Press.

Ellen, I., and Turner, M. (1997). Does neighbourhood matter? Assessing recent evidence. Housing Policy Debate, 8, 833-866.

Habyarimana, J., Humphreys, M., Posner, D.N., and Weinstein, J.M. (2007). Why does ethnic diversity undermine public goods provision? American Political Science Review, 101, 709-725.

Haynie, D.L., Silver, E,. and Teasdale, B. (2006). Neighborhood characteristics, peer networks, and adolescent violence. Journal of Quantitative Criminology, 22, 147-169.


11

Hooghe, M., Reeskens, T., Stolle, D. and Trappers, A. (2009) Ethnic diversity an generalized trust in Europe. A cross-national multilevel study. Comparative Political Studies, 42: 198-223.

Kempen, R. van and Bolt, G. (2009). Social cohesion, social mix and urban policies in the Netherlands. Journal of Housing and the Built Environment, 24, 457-475.

Kleinhans, R., Priemus, H., and Engbersen, G. (2007). Understanding social capital in recently restructured urban neighbourhoods: two case studies in Rotterdam. Urban Studies, 44, 1069-1091.

Leigh, A. (2006) Trust, inequality and ethnic heterogeneity. The Economic Record, 82, 268-280.

Morenoff, J.D., Sampson, R.J., and Raudenbusch, S.W. (2001). Neighborhood inequality, collective efficacy, and the spatial dynamics of homocide. Criminology, 39, 517-560.

Portes, A., and Zhou, M. (1993). The new second generation. Segmented assimilation and its variants. Annals of the American Academy of Political and Social Science, 530, 74-96.

Portes, A. (1995). The economic sociology of immigration. Essays on networks, ethnicity, and entrepreneurship. New York: Russell Sage Foundation.

Putnam, R.D. (2007) E Pluribus Unum. Diversity and community in the twenty-first century. The 2006 Johan Skytte Prize Lecture. Scandinavian Political Studies, 30, 137-174.

Wilson, W.J. (1987). The truly disadvantaged. The inner city, the underclass and public policy. Chicago and London: The University of Chicago Press.

Zorlu, A., and Hartog, J. (2002). Migration and immigrants. The case of the Netherlands. In Rotte, R. and Stein, P. (eds.). Migration policy and the economy: International experiences. Munchen: Hans Seidel Stiftung.

CHAPTER ONE

TOWARDS AN INTERDISCIPLINARY THEORY OF SOLIDARITY

PAUL DE BEER, MAARTEN BERG, LAURENS BUIJS, FERRY KOSTER

AND DOROTA LEPIANKA

Introduction

Pro-social behaviour, or solidarity, for short, has puzzled social scientists for a long time. Acting for the benefit of another at a cost for oneself has often been considered as anomalous or aberrant behaviour, at odds with the self-interest of the acting person. For economists and evolutionary biologists, in particular, pro-social behaviour seemed to be inconsistent with their basic assumptions and was thus largely disregarded as an anomaly, which did not fit into their theories. For sociologists, social psychologists and anthropologists, pro-social behaviour has always been a more familiar phenomenon, but for them, too, it was often difficult to reconcile solidarity with the simultaneous existence of self-interested behaviour.

At first sight, both the salience and the explanation of solidarity vary widely between the various scientific disciplines. It may thus seem an impossible task to integrate insights from these various disciplines into one overarching and encompassing interdisciplinary theory. However, on further consideration one notices remarkable similarities between the various disciplinary approaches, which are hidden from sight due to the fact that different disciplines use different terms for similar concepts. In this chapter, we attempt to draw the outline of a comprehensive theory of solidarity which combines insights from sociology, anthropology, social psychology, economics and socio-biology.

Chapter One

14

We first give a brief overview of the interpretation and explanation of solidarity in each of these five disciplines. Next, we enumerate the similarities and differences between the disciplines and draw up an interdisciplinary framework by using the common and complementary elements of the various disciplines. Before starting with the overview of the five disciplines, we first explain what we mean by pro-social behaviour or solidarity.

Solidarity

Pro-social behaviour and solidarity will be used interchangeably in this volume. We restrict the definition of solidarity to a particular kind of behaviour, irrespective of the underlying attitudes, motives or preferences. Pro-social behaviour or a solidary act is defined as any act that purposefully benefits another person at a cost for the agent, without any guarantee of an equivalent return. Some examples of pro-social behaviour are alms-giving to a beggar, helping your neighbour, raising children, voluntary community work, donating money to a charity organization, paying an insurance premium1 and paying taxes. The last example shows that solidary acts, according to our definition, need not be voluntary. Moreover, solidarity includes both informal behaviour and formal behaviour, that is, solidarity behaviour that is mediated through a formal organization, such as the state or an insurance company.

This definition implies that we exclude normative interpretations of solidarity and focus only on objective interpretations of actual behaviour. Therefore, we refrain from the question whether solidarity is a good thing, worthwhile to be promoted or not.

The fact that we restrict our definition of solidarity to behaviour, does not mean that we are not interested in the motives for solidarity and the feelings or emotions triggering solidarity. On the contrary. One of the main purposes of this chapter is to find out which attitudes, feelings or emotions, according to the various disciplines, motivate solidary behaviour. However, explaining solidary attitudes or feelings is not our goal as such, but only a way to understand solidary behaviour.

1 A voluntary insurance may be considered a borderline case, since it can also be interpreted as an ordinary market transaction, in which the guarantee for the insured person of receiving a compensation in case of a damage or an injury is equivalent to the premium.

Towards an Interdisciplinary Theory of Solidarity

15

Sociology

From the very beginning, solidarity has been a central concept in sociology, since it relates to social order and collective action, generally considered as the alpha and omega of classical sociological theory (Hechter 1998). Many scholars, both classical and contemporary, placed the concept at the core of their theories and empirical research. This has led to a wide variety of definitions, operationalisations, and applications of the term.

In early sociology, the concept of solidarity came to the fore in response to the changes wrought by the development of industry, the rise of market economies and the expansion of cities. New kinds of communities emerged that were bigger, denser and more diverse than the rural communities that dominated before. Most classical theorists regarded the cohesion or solidarity of agrarian communities as unproblematic. In their view, such communities were technologically and demographically stable and were not involved in long-distance trade. Social mobility was minimal and most children were destined to follow in the (professional) footsteps of their parents (cf. Hechter 1998). Since these agrarian communities offered little scope for individual choice, the attainment of solidarity was viewed as relatively straightforward. According to Tönnies (1887), such communities (which he termed Gemeinschaften) were breeding grounds for social relations based on strong emotional, quasi-familial commitments. Émile Durkheim (1893) coined the term ‘mechanical solidarity’ for this kind of society, implying that it had a certain automatic quality. Mechanical solidarity is based on common values and internalized social norms and beliefs in a community with strong moral (often religious) norms, which tie the individuals to the society as a whole. In these communities the individual consciousness is fully integrated into the collective consciousness. This binds the individual to society and makes him or her act in accordance with the shared norms (Durkheim 1893: 60-61).

Mechanical solidarity appeared to be threatened, however, by the rise of modern industry and the expansion of market economies in the Western world, starting in the late eighteenth century. Among many other effects, this transformation increased the size and scope of social networks, thereby offering individuals more options in their daily life, ranging from the choice of a marriage partner to the choice of an occupation. This resulted in a growth of individualism and the concomitant decline of the conscience collective (Simmel 1922). It was far from evident how groups, communities, and societies could maintain solidarity in the wake of this

Chapter One

16

increasing individualism. At that time, many theorists were deeply worried: “This transformation means the doom of culture itself if none of its scattered seeds remain alive.” (Tönnies 1887)

Durkheim, the foremost early theorist of solidarity, shared Tönnies worries of the decline of ‘mechanical solidarity’, but believed it was replaced by a new form of solidarity. In his book The Division of Labour in Society (1893), he posed that industrial societies were held together by individuals’ mutual functional interdependence. This new form of solidarity was a result of the division of labour – a form of solidarity that he termed ‘organic’. By distinguishing organic from mechanic solidarity, Durkheim emphasized that a coherent social organization based on individualism and large communities was indeed possible. He acknowledged that the nature of solidarity changed, but it remained the foundation of any society.

In a modern industrial society, Durkheim (1893) said, social cohesion depends upon the division of labour. Individual specialization leads to mutual interdependence and ensures cooperation for a final end. However, Durkheim warned that an extreme division of labour would result in anomaly and endanger the priority of the whole over the individual. He believed that the intensity of social conflicts was positively correlated with the lack of perception of solidarity. This made social tensions ultimately a political problem. For Durkheim, this showed the importance of the modern democratic state. Solidarity had to be maintained by keeping the right balance between social pressures and individual liberty, constantly enhancing the perception of solidarity (Hechter 1998).

With these early attempts to define the influence of societal processes on solidarity, sociology established itself among the social sciences.

Durkheim was not very explicit about the mechanism by which the division of labour translates into social solidarity. For a self-interested individual, acknowledging that one is dependent on others does not necessarily result in solidary behaviour, since it might also elicit free-riding on the efforts of others. This would mean that people take advantage of the solidary behaviour of their fellow men without contributing to the common good themselves. However, if free-riding becomes the dominant strategy, the free riders will ultimately crowd out the co-operators and solidarity will disappear in the end. According to Durkheim (1893) civil law, in particular contract law, plays an important role in preventing free-riding behaviour, but contract law can only prevent non-compliance with


17

private, bilateral contracts. More is needed to explain why individuals contribute to the common good if this does not directly affect the benefits they reap from the common good.

In rational choice theory, it is assumed that individuals can only be made to contribute to the common good if the expected cost of free-riding is larger than the cost of contributing. Thus, there has to be a credible threat of a sanction in case of non-compliance with the social norm of helping others. This sanction can take a wide range of forms, from a simple verbal rebuke to a physical beating or ostracism. A self-interested person “will comply only if the probability of detection multiplied by the sanctions imposed given detection equals or exceeds the benefits from noncompliance” (Chai & Hechter 1998: 36, 37).

However, the monitoring and punishment of defectors are also costly. These costs will have to be added to the expected contributions to the common good, for example in the form of a tax. If these costs are very high (for example, because detecting non-compliance is difficult), it may be impossible to maintain solidarity, because many group members will leave the group (Hechter 1987, Chai & Hechter 1998).

An alternative way to deter free-riders is spontaneous punishment by other group members. However, since punishing is costly (partly because it requires some effort, partly because the defector may retaliate by hurting the punisher), a purely selfish person will not punish a defector (unless a non-punisher is also punished, but this results in an infinite regress). Thus, this explanation needs the additional assumption that people are willing to unselfishly punish a defector (so-called altruistic punishment, cf. Boyd et al. 2003), which does not fit well within the rational choice framework which starts from the assumption of self-interest (Coleman 1990: 31-32).

Anthropology

In anthropology the classical example of solidarity is the reciprocal exchange of gifts. The act of gift-giving binds groups and individuals; it creates and reinforces a network of rights and obligations that generates and sustains social cohesion.

Exchange of gifts in archaic societies is far more than an economic transaction. As noted by Lévi-Strauss, “[g]oods are not only economic commodities, but vehicles and instruments for realities of another order, such as power, influence, sympathy, status and emotion” (Lévi-Strauss

Chapter One

18

1967 [1949]: 54). The goods that are exchanged embody not only – or not even primarily – material value but also social meanings (Malinowski 1985 [1922]). These goods are by no means limited to material objects; they include borrowing things, getting help, accepting an obligation, or assuming responsibilities for another.

Gifts are common and ever present in archaic societies. They do not constitute an exception for special occasions (such as our birthday gifts), but are rather the norm. Lévi-Strauss (1969:54) talks about a “passion for the gift, accompanied by the ritual obligation on the recipient to accept and to give”.

A gift is rarely purely voluntary. All anthropologists stress the compulsory nature of reciprocity: gift-giving is generally considered to be a social obligation, just as the duty to repay the gift (Malinowski 1985 [1922]:182). Accepting a gift is obligatory, too: a gift cannot be properly refused.

The exchange of gifts is not the result of economic self-interest (Sahlins 1974:160). Most often, the exchange of gifts brings no tangible results or profits, and after the exchange, the partners are no richer than they were before. In fact, from an economic point of view, most transactions do not make any sense, as usually neither partner acquires any real material benefit. Actually, it is the exchange itself that matters and not the object of exchange!

Anthropologists distinguish between various types of gifts, primarily depending on the extent of reciprocity, although there is no agreement on the number and kind of categories.

Malinowski (1985 [1922]) distinguished seven types of gifts. At one extreme, there is the pure gift, defined as “an act, in which an individual gives an object or renders a service without expecting or getting any return”, which according to him is exceptionally rare. Mauss rejected the idea of ‘free gift’ altogether. The idea of a ‘free gift’ is to him a contradiction, a misunderstanding. Every gift triggers an obligatory counter-gift and perpetuates a (dominant) system of reciprocity.

At the other extreme is ordinary trade (e.g. barter), in which mutual advantage to the trading partners is clearly present. In between are various forms of customary gifts that are partially or conditionally returned. The value of the counter gift is (almost) never strictly equivalent to the value of


19

the original gift and the economic value of the counter gift might be even symbolic.

Sahlins (1974) makes an interesting distinction between reciprocity, i.e. “vice-versa movements between two parties”, and pooling or redistribution, i.e. “centralized movements: collection from members of a group … and redivision within this group”. An important difference is that “pooling is socially a within relation, the collective action of a group … [while] reciprocity is a between relation, the action and reaction of two parties” (Sahlins 1974:188). Within the category of reciprocal exchanges, Sahlins distinguishes between generalized reciprocity, balanced reciprocity and negative reciprocity (Sahlins 1974:191-196).

Generalized reciprocity is similar to Malinowski’s pure gift, but it seems broader and is not necessarily restricted to the closest kin. It is help given freely, altruistically, out of disinterested concern and without any (open) stipulation of return. It includes ‘sharing’, ‘hospitality’, ‘free gift’, ‘help’, ‘generosity’, but also ‘noblesse oblige’. If there is an obligation to reciprocate, it is vague, meaning that the counter-obligation “is not stipulated by time, quantity or quality: the expression of reciprocity is indefinite”. Generalized reciprocity is a “sustained one-way flow”. As stated by Sahlins, “[f]ailure to reciprocate does not cause the giver of stuff to stop giving: the goods move one way, in favour of have-not, for a very long time” (Sahlins 1974:194).

Balanced reciprocity refers to direct mutual exchange. Reciprocation always takes place within a finite and narrow period of time in the form of “the customary equivalent of the thing received” (Sahlins 1974:194). The economic and social interests of the parties are central and clearly stipulated, and the failure to reciprocate within the given time may cause a disruption of the relation between the parties involved (Sahlins 1974:195). Transactions have usually a utilitarian purpose as well, but the ‘moral’ purpose of “renunciation of hostile intent or of indifference in favour of mutuality” remains central (ibid., p. 220).

Sahlins also distinguishes negative reciprocity, which is defined as “an attempt to get something for nothing with impunity” (Sahlins 1974:195). However, this form of transaction is exceptionally rare and might be charged with negative social sanctions.

Central to anthropological theory is, that gift-exchange is not considered to perform an economic function, but is seen primarily as a social act to

Chapter One

20

acquire and maintain social status, to create and consolidate good relations or to comply with social norms.

One of the main motives for gift giving, is according to Malinowski, the “fundamental human impulse to display, to share, to bestow” (Malinowski 1985 [1922]:175). This fundamental impulse may be fed by vanity, but may also serve the purpose of acquiring and maintaining prestige and power (ibid., p. 175).

The exchange of gifts also helps in creating, sustaining and strengthening social ties. According to Douglass, this purpose lies at the heart of gift giving: “A gift that does nothing to enhance solidarity is a contradiction” (Douglass 2004: x). To Lévi-Strauss, “reciprocity is … the most immediate from of integrating the opposition between self and others” and the gift, which “makes … individuals into partners”, constitutes the symbolic agent of this integration (Lévi-Strauss 1969 [1949]:84). According to Sahlins, gift-giving “underwrites or initiates social relations” (Sahlins 1974: 186). Malinowski states that the purpose of ceremonial re-payments is “to thicken the social ties from which arise the obligations” (ibid., p.182).

Malinowski repeatedly stresses the obligatory nature of gifts. Gifts are almost always given to fulfil social obligations “and with a great deal of formal punctillo” (Malinowski 1985 [1922]:174). The obligation to (give, accept and) reciprocate is frequently enforced by social sanctions. However, as noted by Sahlin, “[t]he norms are relative and situational rather than absolute and universal. A given act… is not so much in itself good or bad – it depends on who the “Alter” is” (Sahlins 1974:199). It is important to note, that the obligatory nature of the gift-exchange rules out pure altruism. Mauss notes a peculiar integration of interest and disinterest, generosity and egoism, altruism and selfishness of the gift exchange.

Social psychology

The term ‘solidarity’ is not frequently used in the social psychological literature. The related terms in social psychology are ‘pro-social behaviour’ (any act performed with the goal of benefiting another person) and ‘altruistic behaviour’ (pro-social behaviour that involves a cost to the helper). Social psychology is not based on one theoretical perspective but combines insights from various perspectives, which often originate from other disciplines. For example, evolutionary psychology stems from


21

evolution theory in biology and will here be discussed under the heading of socio-biology.

Social exchange theory borrows from economics and rational choice theory the concept of social exchange. It stresses the role of self-interest in pro-social behaviour by assuming that people aim to maximize their own utility or happiness. The (psychological) social exchange theory is broader than its economic equivalent, in the sense that it is less focused on monetary rewards. Relationships between people are also explained in terms of costs and benefits. Helping others can be beneficial if others return the favour (now or in the future) or when a confrontation with misery is stressful for the helper (e.g. Eisenberg and Fabes, 1991). Moreover, by helping others we gain psychological rewards, such as social approval and increased feelings of self-worth. However, if the costs of helping are large (for example when it is dangerous, embarrassing or time-consuming), people are less inclined to help (e.g. Dovidio et al., 1991).

Another motive for solidary behaviour can be empathy, which social psychology borrows from psychology and socio-biology. Batson (1991) argues that people sometimes help other people purely out of the goodness of their hearts, which presupposes the ability to feel empathy. Toi and Batson (1982) manipulated the level of empathy and found that when empathy is low, there is a big difference in willingness to help, depending on whether or not helping is costly. When empathy is high, however, willingness to help is almost independent of the cost of helping.

The specific contribution of social psychology to explaining solidary behaviour is its focus on the conditions and circumstances that promote solidarity.

For example, the willingness to help others depends on someone’s mood. Isen and Levine (1972) manipulated mood (‘mood induction’) by leaving small amounts of money on a telephone cell. People who had just found the money seemed to be much more willing to help others than people who did not find any money. When people feel good, they do good. When they perform well on a test, receive a gift, have happy thoughts or listen to pleasant music, they contribute more to charity, are more willing to donate blood or to help co-workers on the job (e.g. Carlson, Charlin and Miller, 1988; Salovey, Meyer and Rosenhan, 1991). Feeling good makes us look on the bright side of life. Moreover, helping others is an excellent way of prolonging our good mood (e.g. Williamson and Clark, 1989).

Chapter One

22

But also negative emotional states might foster solidary behaviour. An example is feeling guilty (e.g. Estrada-Hollenbeck and Heatherton, 1998). Churchgoers were more likely to donate money to charity before attending confession than afterwards, presumably because confessing to a priest reduced their guilt. But also feeling sad can boost helping behaviour, as sad people are motivated to engage in activities that make them feel better (Wegener and Petty, 1994). This is sometimes called the ‘negative-state relief hypothesis’ (e.g. Cialdini et al, 1987). According to this theory, the helping behaviour does not have to be related to the source of our sadness.

Although not directly related to solidary behaviour, an important strand of social psychology that is relevant in studying solidarity is social identity theory. According to this theory, people try to build a positive self-image through social categorization (thinking in terms of groups), social identification (identifying with certain ‘in-groups’) and social comparison (differentiating their in-group positively on particular dimensions from relevant out-groups). Each individual has a repertoire of (social and personal) identities open to her/him, each identity informing the individual of who he/she is and what this identity entails. Which of these many identities is most salient for an individual at any time varies according to the social context. Where personal identity is salient, the individual will relate to others in an interpersonal manner, depending on their character traits and any personal relationship existing between the individuals. However, under certain conditions a group identity might take precedence.

What is a relevant out-group? According to the concept of distinctiveness, a relevant out-group is not too different from the corresponding in-group. The most rivalry exists between quite similar football-clubs, such as Ajax and Feyenoord, or between two Ivy-league universities, or between Protestantism and Catholicism (rather than Buddhism). Important for social identity theory is that no interaction is required to speak of ‘a group’. When a single woman is in a meeting with ten other men, her social identity (female) is salient, despite the fact that there is no (physical) contact with other women.

Self-esteem is an important motive for identification. People strive for positive differentiation of their in-group.

Self-categorization theory is an elaboration of social identity theory (e.g. Turner, Oakes, Haslam and McGarty, 1994). According to this theory, identity is a flexible and multi-dimensional concept. What part of your identity is active is dependent on the social context.


23

Economics

The mainstream economic perspective on human behaviour, neoclassical economics, starts from the assumption of the homo economicus, i.e. an individual who rationally seeks her/his self-interest. That is, he/she has only self-regarding preferences and is not interested in the well-being or ‘utility’ of other persons. Altruism, in the sense of wilfully acting for the benefit of others, is, thus, impossible. This basic assumption seems to be at odds with solidary or pro-social behaviour. Nevertheless, even purely self-interested individuals may act solidary if this is in their (long-term) self-interest.

There may be three self-interested motives for solidarity in economics.

A first motive is that a risk-averse individual2 wants to insure her/himself against the risk of a harmful future event. This self-interested individual prefers paying a premium for an insurance policy which guarantees compensation in case a harmful event occurs, even if it costs (slightly) more than the expected (probabilistic) loss. The premium paid by the insured person is used to compensate other policyholders who incur a loss. If the harmful event never occurs, the insured person will have contributed to the benefit of others without having received anything in return. In this sense, paying an insurance premium is a solidary act, since there is no guarantee that it will be refunded. However, since the insurance policy stipulates the legal right to compensation, it is usually considered to be equivalent to other market transactions in which no solidarity is involved.

A second motive for pro-social behaviour is that one expects to benefit from a future return of the beneficiary (direct reciprocity). If the beneficiary has no legal obligation to pay back the favour bestowed on her/him, the benefactor apparently has no incentive to favour the beneficiary. However, if both persons interact with each other repeatedly, pro-social behaviour may be optimal for both parties. An optimal strategy may be to help the other person as long as he/she helps you, but to retaliate in the next meeting as soon as the other party defects (the so-called tit-for-tat-strategy, cf. Axelrod 1984). Both parties have then an interest in continuing their co-operative behaviour as long as it is uncertain when the sequence of interactions will end.

2 A risk-averse person prefers a fixed amount of money to a variable amount with the same expected monetary value. Thus a sum of € 1,000 is preferred to a fifty-fifty chance of € 500 and € 1,500, respectively.

Chapter One

24

A third self-interested motive for solidary behaviour is that it elicits admiration and praise from others, which the benefactor values positively. Thus, a feeling of pride may compensate for the costs of the solidary act. However, this explanation of self-interested solidarity raises the question why people who are purely self-interested would praise the solidary behaviour of a benefactor from which they do not benefit themselves.

These three motives explain why a self-interested individual may act solidary voluntarily. However, economic theory can also explain why people may support mandatory solidarity. This can be the case if a voluntary, private insurance scheme is not feasible or is much more costly than a mandatory scheme. Various market failures may render compulsory insurance optimal from the point of view of self-interested individuals (e.g. Barr, 2004). These market failures include interdependent risks, such as inflation and unemployment, adverse selection, which may occur if there is asymmetric information about risks between the individual persons and the insurer, and economies of scale, which make a single public scheme more efficient and thus cheaper than a multitude of private schemes.

It is important to realize that even if an individual prefers a compulsory public insurance scheme, he/she still has a strong incentive to free ride and evade paying the mandatory contribution if that does not nullify her/his entitlements to a benefit. Thus, a reasonable probability of detection and a credible sanction are needed to sustain such a compulsory scheme.

The three motives that have been discussed so far exhaust the possible motives of a purely self-interested person for acting solidary. However, if one stretches the basic assumption of neoclassical theory somewhat, one may also include other-regarding preferences in the utility function of an individual. Actually, this means that insights from other disciplines, such as psychology and sociology, are incorporated into the economic framework.

Three kinds of other-regarding preferences may straightforwardly result in solidary behaviour.

First, someone may enjoy gift-giving. More formally, this means that, ceteris paribus, giving something to someone else (not necessarily an arbitrary person) raises your utility.

Second, one may care about the well-being of others. This is usually called sympathy or empathy (Smith 1759). As a consequence, your utility


25

(or well-being, happiness, etc.) increases if someone else’s utility increases. If you care about the well-being of someone else, you may be willing to act in favour of that person if it raises her/his well-being. A rational person will weigh the increase of her/his own well-being as a consequence of the higher well-being of the beneficiary against the reduction of her/his well-being due to the cost of acting for the benefit of the other.

In this case, contrary to the joy of giving, it does not matter to your well-being whether the well-being of the beneficiary is promoted by your own act or by someone else’s solidary behaviour. This implies that sympathy may give rise to the problem of free-riding. Even though you care about the well-being of another person, you would rather have someone else help that person instead of you. To sustain solidarity based on sympathy, especially if it concerns the rather weak sympathy with ‘distant’ persons, it might, once again, be necessary to oblige individuals to contribute to it by way of compulsory solidarity.

A third other-regarding motive for solidarity is commitment (Sen 1985). This means that a person feels obliged to act in the interest of someone else, even if it does not raise her/his utility or well-being. This feeling of obligation may result from the internalization of a social norm, which may have been the result of the socialization process during upbringing. Contrary to the two preceding other-regarding motives, commitment cannot be traded off against the utility of self-interested behaviour, since commitment is not directly related to your utility. The internalized social norm is actually a second (higher) order preference, which is superior to first order preferences (compare your preference for a fair treatment of your child to your preference for marmalade). Thus, the analysis of solidary behaviour because of commitment cannot be easily incorporated into the usual neoclassical economic framework.

Socio-biology

Studying solidarity using insights from socio-biology implies trying to understand the social behaviour of different species from the perspective of evolutionary theory. An application of this perspective is based on the following points of departure.

First, it means that the main mechanisms explaining social behaviour involve a reference to evolutionary processes such as natural selection, reproduction, fitness, and adaptation. One of the basic premises of

Chapter One

26

evolutionary approaches to social behaviour is that focusing on these processes helps to uncover the ultimate causes of behaviour rather than more superfluous explanations based on proximate causes (Diamond 1997; Henrich and Henrich 2007).

Secondly, these social behaviours can be found in different species. As such, to explain a certain kind of behaviour it does not matter whether it is performed by a germ, an ant or a human being (Kropotkin 1902 [2009]; Axelrod and Hamilton 1981; Ridley 1997). In that sense, this perspective places stronger emphasis on the similarity between species than on their differences. As a result, little attention is paid to the question whether certain kinds of behaviour are expressed consciously or not. Therefore, the question whether the consequences of a particular behaviour are rationally considered or consciously reflected on, has a less prominent place in this field of inquiry compared to investigations in psychology, sociology, and economics.

In socio-biology, the term solidarity is rarely used. Most research focuses on altruistic behaviour, while far less attention is paid to cooperation between individuals.

Within socio-biology, altruism has a specific meaning, namely behaviour decreasing the fitness of the individual who behaves altruistically, while increasing the fitness of another individual (e.g. Hamilton 1964; Trivers 1971; Wilson 1975; Bell 2008). This definition emphasizes costs of altruism for the giver and the benefits for the receiver. One of the central questions that socio-biology tries to answer is how such costly behaviour can have evolved through natural selection (e.g. Komter 2010). Understanding altruistic solidarity is much more difficult than cooperative solidarity, since the latter also yields gains to the individual who acts solidary. As in other scientific disciplines, this puzzle is translated into game theoretic notions, such as the prisoner’s dilemma, in which those acting selfish can take advantage of the benevolence of altruists. In the socio-biological literature several mechanisms are proposed to explain how altruism may evolve (Nowak 2006).

A central explanation of altruism is based on the idea that natural selection can favour cooperation between genetic relatives. If the giver and the receiver are genetically close to each other, organisms carrying an altruistic gene that increases the fitness of their relatives, will also foster the reproduction of that gene, since these relatives carry a copy of that gene. This mechanism is referred to as kin selection (Fisher 1930; Haldane


27

1955; Hamilton 1964; Smith 1964). It explains why there is more help between relatives than between non-relatives, thus maximizing the inclusive fitness of the organism (Hamilton 1964).

However, kin selection is not able to explain altruism within groups consisting of non-relatives. The occurrence of altruism in groups of unrelated organisms can be explained with a reference to reciprocal altruism (Trivers 1971). If the organisms interact more than once, there is the possibility that an altruistic act from one is reciprocated with an altruistic act later on by the other. Clearly, for reciprocal altruism to work, organisms should recognize the other to be able to return the favour. By following a tit-for-tat strategy, adjusting their behaviour to the behaviour of others, organisms can yield higher outcomes than following an egoistic strategy (Axelrod and Hamilton 1981).

Direct reciprocity is based on symmetric exchange relationships between individual species. In contrast, indirect reciprocity involves asymmetric relationships. This occurs when an organism helps another, without the possibility (or perhaps the need) for reciprocation. Reputation is the main mechanism explaining altruism in such situations. If someone having a good reputation is likely to receive more help from others, helping others is beneficial to oneself (Nowak and Sigmund 1998). Reputational effects involve a complex system of calculation to keep track of who helps others. This is supported by the development of language to enable information sharing and the evolution of social norms (Nowak 2006).

Another possible explanation of altruism focuses on the mechanism of group selection (Wynne-Edwards 1962; 1986). The basic idea of group selection is that groups that are better able to adapt to their environment have a greater chance of survival and that nature therefore benefits groups with certain traits, rather than individuals. If altruism strengthens the chance of survival of a group, a group with a large share of altruists among its members is expected to have a higher chance of survival than a group with a smaller share or no altruists.

The idea of group selection was accepted for quite some time until it was heavily criticized by biologists like Smith (1964) and Williams (1966), who argued that group selection is not a strong enough evolutionary force, because selection only takes place at the level of individuals or genes (Dawkins 1976). Clearly, a weak point of the theory of group selection concerns the possibility that individuals free ride on the altruism of others within the group. Whereas a group with a large share of altruists may have

Chapter One

28

a higher chance of survival, it will also benefit egoistic individuals who do not contribute to the survival of the group but take advantage of the altruists. Ultimately, if the egoists have a higher survival probability than the altruists, the proportion of egoists within the group will increase and the probability of survival of the group drops. Only recently has the possibility of group selection regained interest. The most recent development in this area is the development of multilevel selection theory, which argues that selection takes place at different levels – both the genetic level and the group level – at the same time (Wilson and Sober 1994; Wilson and Wilson 2008). This latter theoretical development also fuelled ideas about the co-evolution of genes and cultures with an emphasis on social learning and norms (Gintis 2003; Boyd and Richerdson 2009).

Socio-biological perspectives on altruistic and cooperative solidarity are inherently functional, as they explain the actions of individuals by referring to the impact on their own fitness and the fitness of others without reference to the underlying motivations. In an attempt to explore the motivational basis of altruism and to go beyond models that try to understand altruism by focusing on the benefits that it may have for individuals, de Waal (2008) suggests that empathy is among the most important impulses explaining altruistic acts. These impulses result from emotional identification and social bonding. If others experience pain, sadness or distress, the empathy mechanism leads to an emotional response and comforting attempts.

Overview: two dimensions of solidarity

Despite the widely varying premises and assumptions of the various theoretical perspectives on solidarity, there is a common tendency within most disciplines to distinguish different kinds of solidarity along two dimensions. The first dimension refers to the prime motive for solidary behaviour, in particular whether a solidary act serves the actor’s self-interest (we will call this self-regarding solidarity) or whether it is meant to further the interests of someone else (other-regarding solidarity). The second dimension refers to the relationship between the donor and the beneficiary, in particular whether there is a personal tie between the two (personal or particularistic solidarity) or the benefactor and the beneficiary do not know each other personally (communal or generalized solidarity).

We will discuss these four kinds of solidarity in relationship to the various disciplines below.


29

Self-regarding solidarity

The first common approach to solidarity in the various disciplines assumes that it is in the interest of the individual her/himself to act solidary, since this contributes, mostly indirectly, to her/his well-being, her/his survival or the transfer of her/his endowments to others. This approach does not need other-regarding preferences or motives, such as altruism, empathy, affection or charity, to explain solidary behaviour. It suffices that individuals are aware of their mutual dependence on others or that they are genetically programmed to show solidarity. This perspective is most prominent in economics and in socio-biology.

In the standard version of neoclassical economic theory the fully rational homo economicus will always pursue her/his self-interest. Consequently, solidarity is only possible if it is in the self-interest of the agent, since s/he expects at least an equivalent return, i.e. the value of the expected return must be sufficient to compensate the agent for the cost of the solidary act. This equivalence can be achieved in case of an insurance (if the agent is risk averse), in case one expects the beneficiary to return the favour equivalently in the future (direct reciprocity) and in case a solidary act increases one’s reputation which fosters the likelihood of support by others in the future (indirect reciprocity).

In the socio-biological approach solidary behaviour can only be explained if it yields some evolutionary advantage. This means that the genetic traits that enhance solidary behaviour can be successfully transferred to the next generation by increasing their likelihood of survival and their procreation. From this reasoning, pro-social behaviour towards genetically related individuals (kin selection), such as solidarity of parents towards their children and solidarity between siblings, can be easily explained. Solidarity towards non-kin is only possible if the beneficiary is likely to do something in return (direct reciprocity) or if showing solidarity enhances one’s reputation, which will trigger support by others (indirect reciprocity). These latter two motives are identical to those in neoclassical theory.

Within social psychology, the branch of evolutionary psychology is closely related to socio-biology. It, too, focuses on kin selection and reciprocity as the main motives for solidary behaviour. The social exchange theory is quite similar to the economic approach in focusing on the weighing of costs and benefits. Solidary behaviour presupposes that the (long-term) benefits outweigh the costs. These benefits can either be an equivalent return by the beneficiary (direct reciprocity), or a psychological benefit of

Chapter One

30

feeling proud for helping another or being esteemed by others, which may contribute to pleasant feelings of self-esteem and self-respect.

Although in anthropology gift-giving is never purely self-interested, reciprocity is generally required, implying that the costs of the solidary act are partly or fully repaid at a later moment (balanced reciprocity, in Sahlins’ terms). Consequently, the net cost for the benefactor is generally modest. Moreover, gift-giving can benefit the benefactor in indirect ways, for example by increasing her/his status.

Within sociology the rational choice approach comes closest to the preceding perspectives. It, too, assumes that individuals make a rational trade-off between the costs and the benefits of various options. Costly pro-social behaviour, thus, assumes that there are positive revenues. These may be due to a reciprocal service by the beneficiary (direct reciprocity) or to praise by the community for acting in conformity with a social norm. Violating a social norm of pro-social behaviour – by refusing to help someone who one should help according to the dominant norm – may be punished by various sanctions, ranging from a mere rebuke to ostracism.

In the functionalist branch of sociology, which builds on Durkheim, (organic) solidarity is explained by the mutual interdependence of citizens in a modern society. This interdependence is caused by functional (occupational) specialization resulting in a strong division of labour, rendering individuals dependent on each other.

Other-regarding solidarity

Opposite to the self-regarding kinds of pro-social behaviour, discussed above, most disciplines also distinguish a kind of solidarity based on motives which put the interests of others – or of the group as a whole – forefront.

This perspective is most prominent in sociology, which almost by definition emphasizes group mechanisms instead of individual considerations. Durkheim’s classical explanation of mechanical solidarity in traditional rural, homogeneous communities, traces solidary behaviour back to shared values and social norms which prescribe pro-social behaviour that enhances the survival of the group. If these norms of solidarity are internalized during the upbringing of children (primary socialization) and are reinforced by peers in later phases of life (secondary socialization),


31

people will generally comply with these norms, even if there is little social control and no threat of sanctioning in case of violation of norms.

In anthropology, gift-giving and repaying the gift are generally considered to be a social obligation. Solidary behaviour is thus tantamount to complying with a social norm, which may or may not be internalized. A purely altruistic gift is considered to be rare, but the obligation to repay a gift is often vague. According to Sahlins, generalized reciprocity often boils down to a “one-way flow”.

In social psychology, the innate capacity of humans for empathy may explain why they act in the interest of others and, in particular, help others who are in need. Witnessing the suffering of someone else evokes similar feelings of distress with the observer, who is thus stimulated to relieve the suffering of the other person. However, social psychology cannot explain why humans have a tendency to feel empathy with others.

Social identity theory states that people are willing to help members of their in-group, simply because they identify with this group. Even though this help may not be reciprocated, it may enhance his or her self-image and self-esteem.

Also in socio-biological theory empathy may explain pro-social behaviour towards non-related individuals who are not expected to reciprocate solidarity. An abundance of empirical evidence supports the claim that humans as well as higher animal species (such as primates) have the capacity to feel empathy – sometimes even towards animals from other species. However, from an evolutionary perspective, it is still a puzzle how to explain the survival of empathic genes. Individual selection processes seem to weed out empathic genes in the long run (although de Waal claims that they may be a by-product of empathy for an animal’s offspring). A process of group selection may better explain the origin and survival of empathy, but it is still a matter of controversy among evolutionary theorists whether group selection processes really exist, because they seem to be at odds with the dominant mechanism of individual (or gene) selection.

The assumption of homo economicus in mainstream neoclassical economic theory effectively excludes the possibility of other-regarding behaviour. One has to stretch the assumptions underlying the homo economicus or, actually, abandon the concept of this rational, selfish individual altogether to create room for an other-regarding economic subject. One option, which

Chapter One

32

does not depart too much from the standard assumptions, is to include the well-being of others in an individual’s utility function. This may be justified by the fact that people enjoy giving to others (‘warm glow’) or are directly affected by the well-being of others. Actually, this latter assumption is identical to introducing empathy in the economic model. However, the existence of this tendency cannot be explained within the economic discipline.

Particularistic versus generalized solidarity

The second dimension along which we can distinguish different mechanisms that foster solidarity refers to the relationship between the benefactor and the beneficiary. At one extreme, the target of a solidary act (the beneficiary) is a particular person (or particular persons) with whom one has a personal relationship, at the other extreme it is an anonymous person, with whom one has no personal relationship and whose identity is not relevant as such. We will call the former kind of solidarity particularistic solidarity and the latter generalized solidarity.

Particularistic solidarity is central to the explanation of solidary behaviour in social psychology and economics. In social psychology, it refers to mechanisms such as empathy or affection that can motivate a person to act in the interest of another person. In this case, it matters whom one helps or supports, since one has to experience a personal tie with the beneficiary. In economics, the personal relationship with the other person matters, too. Since the homo economicus will only support another person if he expects an equivalent return in the future, it is essential that the two persons interact with each other repeatedly and trust each other.

In case of generalized solidarity the benefactor does not necessarily know the beneficiary of a solidary act personally. It may be an anonymous person, of whom one does not know more than that s/he is in need and is a member of your own community or group, who therefore deserves support. This kind of solidary behaviour is usually strongly influenced by values, social norms, customs or social pressures.

Both sociology and anthropology emphasize these mechanisms as a cause of solidary behaviour. Even if the benefactor expects some return for his or her solidary behaviour, it need not be repaid by the same person that benefited from the solidary act. Thus, one may be willing to help an anonymous person, expecting to be helped oneself by another anonymous person in the future in case one should be in need.


33

Four ideal-types of solidary behaviour

Combining the two dimensions of solidarity that we distinguished in the previous section, other-regarding versus self-regarding solidarity and particularistic versus generalized solidarity, we can define four ideal-typical kinds of solidarity (see Figure 1).3 We have chosen names for these ideal-types which are not directly derived from one particular discipline, to emphasize that the distinction between the ideal-types does not coincide with disciplinary borders.

Figure 1. Two dimensions and four ideal-typical kinds of solidarity

Although, up till now, we treated the different kinds of solidarity in sociology, anthropology, social psychology, socio-biology and economics as mutually exclusive, polar cases, it is actually more accurate to consider them as the ends of a continuum ranging from pure altruism to pure self-interest on one dimension and from purely particularistic to purely generalized solidarity on the other dimension.

3 A similar fourfold distinction is made by van Oorschot & Komter (1998).

empathic solidarity

bilateralsolidarity

multilateral solidarity

normative solidarity

hybrid solidarity

particularistic solidarity

generalized solidarity

other-regarding solidarity

self-regarding solidarity

Chapter One

34

Empathic solidarity is a form of other-regarding solidarity between two persons that is driven by a strong sense of connectedness (affection and empathy), which may cause feelings of pain or pleasure. Altruism might be the clearest example of this. Empathic solidarity will occur if the benefactor is emotionally affected by the experience of someone else who is suffering or who needs support.

Bilateral solidarity is also a form of solidarity between two persons, which is, however, based on self-interest. Whereas empathic solidarity is usually one-sided (unilateral), bilateral solidarity is, by definition, two-sided and manifests itself as a bilateral exchange (quid pro quo), in which the (initial) benefactor expects a return from the beneficiary which is at least of equal value to the benefactor. Thus, reciprocity is essential for bilateral solidarity. Purely bilateral solidarity will only occur if the value of the expected future return is larger than the costs of the solidary act.

Both empathic solidarity and bilateral solidarity originate from the personal relationship between two individuals and are not (strongly) influenced by societal forces, although the way in which the two persons interact may of course be shaped by a particular culture or tradition.

Normative solidarity, on the contrary, is strongly embedded in the culture (norms and values) of a particular group or community. Acting solidary is dictated by specific social norms, which are usually internalized during upbringing. As a consequence, most people really believe in the norms and comply with them more or less voluntarily. The only reward they get from complying with the norm of solidarity is a feeling of pride or self-esteem (and shame or guilt in case they violate the norm).

Normative solidarity may also be the result of more personal norms or values, which may not coincide with the prevailing norms and values in a person’s community. For example, one may endorse particular values regarding a fair or just society or the desired level of income inequality, which urge one to show solidarity in accordance with these values.

Thus, a potential benefactor will act out of normative solidarity if he/she has a strong belief or conviction that acting solidary is the (morally) right thing to do. Consequently, normative solidarity predominantly goes one way: from the benefactor to a largely anonymous beneficiary (or beneficiaries).

Multilateral solidarity differs from normative solidarity in that the dominant social norm is not internalized. Individuals realize that they are


35

mutually dependent on other group members, but they are also aware of the advantage of free-riding on the solidary behaviour of others. Therefore, monitoring and sanctioning are indispensable to maintain the norm of solidarity. Fear of loss of reputation, which may result in less support by others in the future, may also be an effective way to foster norm compliance. Multilateral solidary behaviour will occur if the benefactor values the costs of non-compliance with a social norm, such as a verbal rebuke or punishment, higher than the costs of compliance, i.e. acting solidary.

Figure 1 may suggest that there exist different, clearly distinguishable kinds of solidarity, which are prevalent under different circumstances or among different individuals. For example, in market relationships bilateral solidarity would be expected to dominate, while in family relations we would primarily expect empathic solidarity. However, this is not what we claim. On the contrary, we postulate that actual solidary behaviour is mostly a mixture of these four ideal-types. Thus, a particular solidary act may be partly bilateral, partly empathic, partly normative and partly multilateral at the same time. As an example, consider the case that you witness your neighbour, carrying a heavy bag of groceries, stumble and fall on the ground. If you assist her to get up and pick up the groceries, this act of solidarity may be motivated by the fact that you feel sorry for her (empathic solidarity), by the expectation that she will help you in the future in a similar case (bilateral reciprocity), by the conviction that you should help someone who is in need (normative solidarity), and by the social pressure that you feel from passers-by (multilateral solidarity), all at the same time.

To illustrate, in Figure 2 we have ordered some examples of solidary behaviour along the axes of other-regarding versus self-regarding and particularistic versus generalized solidarity. It should be stressed that the exact position of the various solidary acts on the two dimensions is somewhat arbitrary and may vary between persons and circumstances.

For example, lending money to a friend is placed on the self-regarding side of the horizontal dimension, because it is supposed that the donor will expect her/his friend to repay the loan on a later moment when s/he is able to. However, it is also conceivable that you would not expect your friend to ever pay back the loan. In that case, the solidary act should be placed more on the other-regarding side. In both cases, it is probably essential that it is a particular friend who requests a loan, and therefore it is placed at the particularistic side of the vertical dimension in Figure 2.

Chapter One

36

Figure 2. Examples of solidary acts ordered along two dimensions

Voluntary community work is placed on the other-regarding side of Figure 2, because it is assumed that the volunteer does not expect any concrete return from her/his volunteering. Nevertheless, some people may volunteer because they actually do expect some reward, such as increasing employment opportunities. In that case, voluntary work should be shifted to the self-regarding side of the horizontal dimension. The extent of particularism may also vary in case of voluntary work. If you help elderly in a local nursing home, this is probably motivated by involvement with the people in your neighbourhood. If you join a non-governmental development organization, this is more likely motivated by general feelings of human solidarity.

In real life, the relative weight and the mixture of the four ideal-typical kinds of solidarity may vary between persons as well as between circumstances. If there are no bystanders, external social pressure cannot play a role. If you do not know the woman who stumbles, you will probably not expect any future return benefit from her. In case of solidarity with anonymous persons, apparently only the normative motives

other-regarding solidarity self-regarding solidarity

gene

raliz

ed s

olid

arity

parti

cula

ristic

sol

idar

ity

bilateral solidarity

multilateral solidarity

normative solidarity

empathic solidarity

social assistance

unemployment insurance

development aid

donation to charity org.

neighbourly help

voluntary community work

raising children

babysitting for grandchildren

alms-giving to beggar

donating a kidney

lending money to a friend

gift-giving

helping a driver with car trouble


37

play a role. But even then, particularistic motives can be important, for example if you donate money to help the victims of a natural disaster after seeing a TV report showing some individual victims.

We believe that an act of solidarity which can be characterized by only one of the ideal-types in Figure 1 is exceptional. Moreover, there are probably no persons who cannot be motivated by one of the four ideal-typical motivations in Figure 1, except in case of rare psychological disorders, which cause people to have no capacity for empathy or never experience feelings of compassion, such as extreme cases of autism or other kinds of serious psychopathology.

Thus, our synthetic proposition regarding solidary behaviour reads, that any solidary act is motivated by a combination of empathy, bilateral reciprocity, internalized norms and values, and social pressure.

General conditions for solidarity

In the previous sections we focussed on the motives that people may have to act solidary towards particular people. In doing so, we took for granted that there was some reason to act solidary in the first place. As a conclusion to this chapter, we address three more general conditions that have to be met for solidarity to occur. The first refers to the need for solidary behaviour, the second to the capacity for solidary behaviour and the third to the efficacy of solidary behaviour.

According to our definition, solidary behaviour refers to a (costly) act of a benefactor that benefits another person (the beneficiary). In most cases, the very reason to act solidary is that the beneficiary is in need.4 This need may refer to a broad range of causes, such as illness, frailty, disability, poverty, et cetera. Apart from the fact that the beneficiary is in need, the potential benefactor must also be aware of this. Often, this is unproblematic, because it can be easily seen (e.g., a blind person who has trouble crossing the street). But there are also cases in which it is not self-evident that a person is in need, for example because s/he is ashamed to show it or because the potential benefactor does not know the person(s) in need (for instance, in case of a famine in a third world country).

4 There may be some exceptions to this rule. The most important one is gift-giving. If you give a present at a birthday or a wedding, this is not because the recipient is in need, but because it contributes to continuation and strengthening of the relationship (even though the giver may not be aware of this motive).

Chapter One

38

In those cases in which the benefactor acts out of self-interest, i.e. in the case of bilateral and multilateral solidarity, s/he must face the risk of being in need at some future date her/himself, too. Thus, uncertainty and risk can also be important conditions for solidary behaviour. Again, an additional condition is that one must be aware of this risk.

The second general condition is, that the benefactor must be able to act solidary. In general, this means that s/he should have the command over resources that can be employed to support the beneficiary. Usually, the person who gives money to a poor person, has a higher income her/himself, or at least can spare some of her/his income. Apart from money, these resources can also be spare time or even a kidney.

The third general condition for solidarity is, that a solidary act is efficacious. This means, that the beneficiary is expected to really benefit from the support offered by the benefactor. Often, efficacy is enhanced by a personal relationship between the benefactor and the beneficiary, because the benefactor can quite easily determine what kind of support the beneficiary needs. Moreover, it is relatively easy to check whether the support achieves its goal. But even then, it is not always obvious what kind of support is in the interest of the beneficiary. Even though an homeless person may beg for alms, a passer-by may doubt whether this will actually help him (“probably he will by a beer from it …”) and may decide it is better not to help than to give the wrong incentive.

The efficacy of a solidary act is also affected by the provision of help by others. If other persons are willing to help someone in need, the additional contribution of your help will probably be smaller than if you are the only helper.

The condition of efficacy is often a more serious hindrance in case of generalized solidarity. Since the benefactor does not know the beneficiary personally, it is more difficult to decide what the best support is and if it is really effective. To illustrate, the many reports about the alleged ineffectiveness of developmental aid, have probably eroded the public support for transferring money to developing countries. Another example is the recurrent criticism of the alleged ineffectiveness of the welfare state, which, according to some critics, enhances rather than alleviates poverty.


39

Conclusion

In this chapter we have combined insights from various disciplinary perspectives in order to construct an interdisciplinary framework for analysing solidarity. At first sight, different disciplines, such as sociology, anthropology and economics, approach the phenomenon of solidarity from completely different angles. However, closer examination reveals that there are many parallels, similarities and overlaps, which are hidden from sight due to the use of different terms and concepts by the various disciplines. Two common dimensions, which enable to us to distinguish between different motives and mechanisms that explain solidary behaviour, have emerged from this overview. Along the first dimension, self-regarding and other-regarding motives are distinguished. At one extreme, a solidary act is motivated by the fact that it benefits the benefactor her/himself, because s/he expects to get something in return for her/his solidary act, either from the beneficiary (in the case of bilateral solidarity) or from the community at large (in the case of multilateral solidarity). At the other extreme, a solidary act has no tangible benefit for the benefactor, apart from the fact that it satisfies an internal urge to help. Along the second dimension, we distinguish between motives which are directed at concrete persons, with whom the benefactor has a personal relationship, and motives which are directed at anonymous members of one’s community. The combination of these two dimensions results in four ideal-typical kinds of solidarity: empathic solidarity, bilateral solidarity, normative solidarity and multilateral solidarity. However, we do not claim that the wide variety of solidary acts in ‘real life’ should be classified into four categories, but rather that (almost) all solidary behaviour is a mixture of these four ideal-types. Although the weight of the four motives may differ, depending on the person and on the circumstances, in most cases several motives underlie a solidary act.

References

Axelrod. R. (1984). The evolution of cooperation. New York: Basic Books.

Axelrod, R., and Hamilton, W. D. (1981). The evolution of cooperation. Science, 211, 1390-96.

Barr, N. (2004) Economics of the welfare state (4th ed.). Oxford: Oxford University Press.

Chapter One

40

Batson, C.D. (1991). The altruism question: Toward a social-psychological answer. Hillsdale, NJ, England: Lawrence Erlbaum Associates.

Bell, G. (2008). Selection: the mechanism of evolution. Oxford: Oxford University Press.

Boyd, R., and Richerson, P.J. (2005) The origin and evolution of cultures. Oxford: Oxford University Press.

Boyd, R., Gintis, H., Bowles, S., and Richersen, P.J. (2003). The evolution of altruistic punishment. Proceedings of the National Academy of Science (PNAS), 100, 3531-3535.

Carlson, M., Charlin, V., and Miller, N. (1988). Positive mood and helping behaviour: A test of six hypotheses. Journal of Personality and Social Psychology, 55, 211-229.

Chai, S.-K. & Hechter, M. (1998). A theory of the state and of social order. In P. Doreian & T. Fararo (eds.). The problem of solidarity. Theories and models (33-60). Amsterdam: Gordon and Breach Publishers.

Cialdini, R.B., Schaller, M., Houlihan, D., Arps, K., Fultz, J,, and Beaman, A.L (1987). Empathy-based helping: Is it selflessly or selfishly motivated? Journal of Personality and Social Psychology, 52, 749-758.

Coleman, J.S. (1990). Foundations of social theory. Cambridge, MA: Belknap Press of Harvard.

Dawkins, R. (1976). The selfish gene. Oxford: Oxford University Press. de Waal, F.B.M. (1996). Good natured: The origins of right and wrong in

humans and other animals. Cambridge, MA: Harvard University Press. de Waal, F.B.M. (2008). Putting the altruism back into altruism: The

evolution of empathy. Annual Review of Psychology, 59, 279-300. Diamond, J. (1997). Guns, germs, and steel. W.W. Norton. Durkheim, E. (1893). The division of labour in society (translated by

George Simpson). New York: The Free Press. Eisenberg, N., and Fabes, R.A. (1991). In M.S. Clark (Ed.), Review of

Personality and Social Psychology: Vol. 12. Prosocial Behavior. (34-61) Newbury Park, CA: Sage.

Estrada-Hollenbeck, M., and T. Heatherton. (1998). Avoiding and alleviating guilt through prosocial behavior. In J. Bybee (Ed.), Guilt and children (215-231). San Diego: Academic Press.

Fisher, R.A. (1930). The genetical theory of natural selection. Clarendon Press.


41

Gintis, H. (2003). The hitchhiker’s guide to altruism: Genes, culture, and the internalization of norms. Journal of Theoretical Biology, 220, 407–18.

Haldane, J.B.S. (1955). Population genetics. New Biology, 18, 34–51. Hechter, M. (1987). Principles of group solidarity. Berkeley: University of

California Press. Henrich, N., andHenrich, J. (2007). Why humans cooperate: A cultural

and evolutionary explanation. Oxford: Oxford University Press. Isen, A. M., and Levine, P. F. (1972). Effect of feeling good on helping:

Cookies and kindness. Journal of Personality and Social Psychology, 21, 384-388.

Komter, A. (2010). The evolutionary origins of human generosity. International Sociology, 25, 443-464.

Kropotkin, p. (2009 [1902]). Mutual aid. London: Heinemann. Lévi-Strauss, C. (1969 [1949]). The elementary structures of kinship.

London: Eyre & Spottiswoode. Malinowski, B. (1985 [1922]). Argonauts of the western Pacific: An

account of native enterprise and adventure in the archipelagoes of Melanesian New Guinea. London etc.: Routledge and Kegan Paul.

Maynard Smith, J. (1964). Group selection and kin selection. Nature, 201, 1145-1147.

Nowak, M.A. (2006). Five rules for the evolution of cooperation. Science, 314, 1560–1563.

Nowak, M.A., and Sigmund, K. (1998). The dynamics of indirect reciprocity. Journal of Theoretical Biology, 194, 561-574.

Ridley, M. (1997). The origins of virtue. Viking. Sahlins, M. (1974). Stone age economics, London: Tavistock Publications. Salovey, P., Mayer, J.D., and Rosenhan, D.L. (1991). Mood and helping:

Mood as a motivator of helping and helping as a regulator of mood. Review of Personality and Social Psychology, 12, 215, 237.

Sen, A. (1985). The standard of living. The Tanner Lectures. Cambridge: Cambridge University Press.

Simmel, G. (1955 [1922]). Conflict and the web of group affiliations (translated and edited by Kurt Wolff). Glencoe: Free Press.

Smith, A. (1975 [1759]). An inquiry into the nature and causes of the wealth of nations. London: J.M. Dent & Sons.

Toi, M., and Batson, C.D. (1982). More evidence that empathy is a source of altruistic motivation. Journal of Personality and Social Psychology, 43, 281-292.

Tönnies, F. (1988 [1887]) Community and society (Gemeinschaft und Gesellschaft). Chicago: Michigan State University Press.

Chapter One

42

Trivers, R. L. (1971). The evolution of reciprocal altruism. The Quarterly Review of Biology, 46, 35–57.

Turner, J. C., Oakes, P.J., Haslam, S.A., and McGarty, C. (1994). Self and collective: Cognition and social context. Personality and Social Psychology Bulletin, 20, 454-63.

van Oorschot, W., and A. Komter (1998). What is it that ties …? Sociale Wetenschappen 41, 5-25.

Wegener, D. T., and Petty, R. E. (1994). Mood management across affective states: The hedonic contingency hypothesis. Journal of Personality and Social Psychology, 66, 1034–1048.

Williams, G.C. (1966). Adaptation and natural selection. Princeton, N.J.: Princeton University Press.

Williamson, G. M., and Clark, M. S. (1989). The communal/exchange distinction and some implications for understanding justice in families. Social Justice Research, 3, 77-103.

Wilson, D.S., and Sober, E. (1994). Reintroducing group selection to the human behavioral sciences. Behavioral and Brain Sciences, 17, 585-608.

Wilson, D.S., and Wilson, E.O. (2008). Evolution “For the Good of the Group”. American Scientist, 96, 380-389.

Wilson, E.O. (1975). Sociobiology: The new synthesis. Cambridge MA: Harvard University Press.

Wynne-Edwards, V.C. (1962). Animal dispersion, in relation to social behaviour. Edinburgh: Oliver and Boyd.

Wynne-Edwards,V.C. (1986). Evolution through group selection. Oxford: Blackwell Scientific.

CHAPTER TWO

WHY WOULD DIVERSITY HARM SOLIDARITY?

PAUL DE BEER

Introduction

In the past decade, a rapidly growing body of academic literature has addressed the question whether ethnic diversity is detrimental to social capital, trust, public good provision or other measures of social cohesion (e.g., Alesina and LaFerrara 2005; Leigh 2006; Putnam 2007; Habyarimana et al. 2007; Hooghe et al. 2009; Gijsberts et al. 2012). Although the empirical evidence for a negative relationship between ethnic diversity and various socially desirable outcomes is still mixed, the starting point of all these studies is that such a negative relationship may be expected from a theoretical point of view. The common idea underlying these theoretical expectations is that (ethnic) similarity between people fosters recognition, communication, cooperation and trust. Consequently, the more ethnically heterogeneous a community is, the less similar its citizens are and the less they will recognize each other as equals, communicate and cooperate with each other, and trust each other.

Although the main subject of this book, social solidarity, is clearly related to the concepts of social capital, trust and social cohesion, these terms do not refer to identical phenomena and, thus, need to be studied separately. The most important difference is that the latter concepts refer primarily to attitudes (recognition, trust) or to more or less enduring relationships (social capital, social cohesion), rather than to the actual behaviour of citizens, which is a crucial element of our definition of solidarity. This difference is important both from a theoretical and from an empirical perspective. Theoretically, attitudes and behaviour may not coincide. As we have argued in the preceding chapter, solidary behaviour need not be motivated by other-regarding motives, such as empathy, but may also be based on self-regarding motives, such as the expected return from helping

Chapter Two

44

or from a gift. Even if ethnic diversity would erode other-regarding attitudes, this would not necessarily result in less solidary behaviour if self-regarding solidarity would be enhanced.

Empirically, testing the effect of ethnic diversity by focussing on solidary behaviour may yield more robust outcomes than focussing on attitudes, since the opinions and preferences that people express in surveys are generally more susceptible to social desirability. Moreover, their answers may strongly depend on the exact wording of the questionnaire. To illustrate, if the share of the respondents in different countries who confirm that they think most people can be trusted, differs, this may simply reflect a difference in social desirability of this answer in the various countries rather than point to a real difference in the way that people treat strangers.

Another point to note about most empirical studies of the effect of ethnic diversity is that they do not (explicitly) distinguish between two ways in which ethnic diversity may affect individual behaviour. The most obvious effect runs through the interaction between people who belong to different ethnic groups. The more diverse a community is, the more often will people encounter people from a different ethnic group (unless, of course, they live in segregated ethnic enclaves). If these interactions are in some sense less satisfying or positive than interactions with people from the same ethnic group, than an ethnically diverse community will, in the aggregate, have less satisfying relations than an ethnically homogeneous community. However, there may also be a direct effect of the ethnic composition of the community, independent of the dyadic relationships between individual persons. To give an example, an ethnically diverse community may be characterised by a plurality of social norms, which gives individual citizens more latitude to act in the way they prefer instead of complying with the dominant social norm. Moreover, the informal enforcement of social norms may be more difficult in ethnically diverse communities, if people are only sensitive to social pressure which originates from their own group. Hence, ethnic diversity of a community may affect the relationship between people belonging to the same ethnic group as well as the relationship between people from different ethnic groups.

In this chapter we will discuss, from a theoretical perspective, what the impact of ethnic diversity on solidary behaviour may be. We take the four ideal-typical kinds of solidarity that were introduced in the previous chapter as our starting point. We identify which (societal) conditions are conducive and which are detrimental for the various kinds of solidarity.

Why Would Diversity Harm Solidarity?

45

We start with conditions that affect the dyadic relation between two persons and analyse what role ethnic similarity or difference plays. Next, we discuss the conditions that are related to the community and examine the effect of ethnic diversity, independently of the individual characteristics of the persons involved.

Solidarity between ethnically different people

According to a folk wisdom, likeness breeds solidarity and difference hampers solidarity. Thus, the more similar people are, the more likely they are to act solidary towards each other, and the more they differ, the less solidary behaviour we will observe. However, as we will show in this section, solidarity between two persons is often based on some differences as well as similarities between the two persons, at the same time.

The verbal similarity of ‘being like someone’ and ‘to like someone’ suggests that affection or empathy is strongly promoted by the similarity of the (potential) benefactor and beneficiary. In case of solidarity between blood relatives this is indeed so, since, in the socio-biological theory of kin-selection, the genetic relatedness determines the costs that individuals are willing to make to support one another. However, as we know from field studies of empathy, the capacity to experience similar feelings and emotions as another individual is certainly not restricted to blood relatives or to subjects who are very close to each other. The Internet abounds with movie-clips of helping behaviour among animals that are not familiar with each other and sometimes even belong to different species. Everyone is familiar with the feeling of sympathy or empathy with a victim of a famine or a natural disaster which is shown on television, and which can encourage people to make a substantial donation to a charity. Although closeness and likeness may make it easier to trigger empathy, they are certainly not a precondition for empathy and, thus, for altruistic behaviour.

In many circumstances, the difference between the benefactor and the beneficiary is even the main reason to act solidary. Because of this difference the former person supports or helps the latter, for example when a parent helps her child, a young woman supports a feeble old man or a native citizen assists a foreigner who does not speak the language. Actually, empathic solidarity often occurs if two persons differ in an important respect but nevertheless are able to identify with each other. Thus, identification is much more important than similarity in understanding empathic solidarity.

Chapter Two

46

Consequently, a difference in ethnicity will only be detrimental to empathic solidarity if ethnicity is a barrier to identification with the other person. Whether this is the case probably depends strongly on the dominant values in a society and the extent to which these have been transferred during upbringing (socialization).

Purely bilateral solidarity requires that the cost to the benefactor of a solidary act is lower than the value of the expected return from the beneficiary. Consequently, two conditions have to be met for bilateral solidary behaviour to occur.

First, the benefactor must know – or at least expect – that the beneficiary will be able to repay the gift in the future. This presupposes that the beneficiary is only temporarily in need, e.g. due to illness, or that the benefactor and the beneficiary have different skills or abilities, so that they both benefit from an exchange between them. Thus, in general, the benefactor and the beneficiary should consider each other as equivalent.

A priori, it is an open question whether ethnic differences go together with differences in skill or ability. If particular skills are related to cultural background, such as craftsmanship (e.g., cookery or music), ethnic difference may even enhance the conditions for bilateral solidarity if the skills that are characteristic for different ethnic groups complement each other. However, if people from a specific ethnic group simply lack skills or education, this can hinder bilateral solidarity, since there is not much room for a mutually beneficial exchange between the groups.

Secondly, the benefactor must be confident that the beneficiary will repay the gift. Since a contractual obligation to repay a gift is excluded from our definition of solidarity (because this is tantamount to a market exchange), this requires mutual trust between the benefactor and the beneficiary. Trust between two persons can be enhanced if they interact regularly. Both have then an interest not to betray the other in order to continue their co-operation in the future. As is know from game theory and laboratory experiments, a strategy of (initially) co-operating with another until the other defects, in which case one retaliates by defecting too (tit-for-tat), is generally quite successful. Thus, the more frequently people from different ethnic groups interact, for example because they live in a mixed neighbourhood, the more likely mutual trust will develop.

Whether two persons will interact repeatedly depends primarily on the value of these interactions to both of them. Put differently, two individuals


47

may interact repeatedly either because they like each other or because they benefit from each other. The chance that they like each other may be enhanced by some similarity between them, but mutual beneficence may depend, as mentioned above, on some differentiation, such as specialization.

When people interact for the first time, the willingness to help each other may depend on the level of generalized trust that is prevalent in a community. We come back to this condition in the next section. The more two people recognize each other as equals, the more likely it is that they are willing to support each other, expecting some return benefit in the future.

So, generally speaking, a different ethnic background need not hamper bilateral solidarity as long as both persons consider each other equal (or, at least, equivalent) and trust each other. Whether this is the case, may partly depend on the prevailing norms and values in their community.

Although the motives for empathic solidarity and bilateral solidarity are clearly different, they are not mutually exclusive. On the contrary, it is quite likely that repeated interactions between two persons, who first act purely out of self-interest, breed some sympathy or affection for each other. This enhances the willingness to help each other, even if the two persons would gradually grow apart with the passing of time (for example, if one of them would become chronically ill; cf. Putnam’s (2000) famous example of two bowling partners).

Ethnic diversity of the community and solidarity

As we stated above, apart from the difference in ethnicity between two persons, the ethnical composition of the community itself may affect solidary behaviour. The reason for this is that the extent of ethnic heterogeneity influences the social context in which individuals act. An ethnically diverse community may differ from an ethnically homogeneous community with respect to the prevailing norms and values and the way that people behave in public.

Both normative and multilateral solidarity are strongly dependent on the prevailing social norms regarding solidary behaviour in a community or society. These norms may vary considerably between groups within a community or between communities. The next three conditions are crucial.

Chapter Two

48

First, it is important which persons are considered to be a member of the in-group and are, thus, eligible for assistance if in need.

Secondly, a social norm may determine under what conditions a group member deserves help. Is being in need the only condition, or is it also important whether the beneficiary is her/himself responsible for her/his distress and whether the beneficiary has assisted other group members in the past? In other words, which needy persons are considered to be deserving and which are non-deserving? These eligibility conditions may be more or less stringent.

Thirdly, since normative solidarity is directed towards anonymous persons, it is important in which situations a person is expected to offer help. This may be guided purely by considerations of efficacy – the person who is able to help another person effectively at the lowest cost – but also by other criteria, for example based on some ranking of the population (e.g., younger people are expected to assist older people, or men should assist women).

The first condition, the definition of the in-group, is of particular importance and may show considerable differences between groups. One important element is how large the group is with whom one identifies and whose members one is supposed to help. The broader the definition of the in-group, the larger the number of persons that people are willing to act solidary towards. It makes a lot of difference whether the in-group is the extended family, the neighbourhood or a village, or the nation-state. Of course, these groups need not exclude each other. It is conceivable that the solidarity within the extended family is the strongest, but that people are also expected, albeit to a lesser extent, to help neighbours, villagers or fellow countrymen.

The more similar people are, the more likely it is that they are included in the in-group. But, what characteristic is crucial for people to be regarded as similar can vary widely. Thus, an in-group can be very homogeneous with respect to one characteristic (e.g. religion), but strongly heterogeneous with respect to another characteristic (e.g. ethnicity, cf. the Jews in Israel; see also chapter 3). Consequently, although ethnic or cultural diversity may result in the partitioning of the society in more homogeneous groups, this need not be the case if there is some other common characteristic which binds these groups together. However, if ethnicity is closely related to other characteristics, such as religion, language and culture, such a partitioning becomes more likely.


49

Another important element is whether ‘strangers’ who live amidst the members of a group are recognized as group members or are treated as outsiders who do not deserve the solidarity of the group members. The treatment of newcomers or ‘strangers’ varies widely between groups and communities. Nevertheless, each in-group has some restrictions on the acceptance of newcomers, otherwise there would be no distinction between group members and outsiders. Acceptance of a newcomer will, in general, be easier, the more similar the newcomer is to the group members. Once again, this similarity may refer to one specific trait of the newcomer, for example, religion, language or skin colour.

An additional characteristic that distinguishes group members from outsiders is simply that the former have been living in this particular community for a longer time. Thus, all else remaining the same, the longer one lives in a particular place, the more one is recognized as an insider, irrespective of any other characteristic. This means that the group members can differ in many respects, except that they share the same history. But even if individual characteristics do not play an important role, it may be pretty hard for ‘immigrants’ to be recognized as members of the group (cf. the new inhabitants of a small closely-knit village community).

The second condition refers to the prevailing norms and values in a community, that prescribe under what circumstances a person in need is considered to be deserving to receive help. If deservingness depends on the responsibility of the potential beneficiary for being in need or if it depends on the (solidary) behaviour of the potential beneficiary in the past, fewer people will be entitled to help than if these criteria do not apply. These deservingness criteria are not directly related to the diversity of the group, unless the characteristics of the persons which determine the heterogeneity of the group, are an indicator for their deservingness. For example, if the members of a specific ethnic group are thought to be responsible for their neediness themselves and if self-responsibility is part of the deservingness criteria, then group diversity may hamper solidarity. Put differently, the broader the definition of deservingness, the less likely it is that group diversity hampers solidarity.

There might be a relationship with generalized trust, too. If generalized trust is strong, most people may be considered to be deserving, even if deservingness criteria are quite strict, since people tend to believe that others will only ask for help if they meet these criteria. In this case, diversity may not hamper solidarity.

Chapter Two

50

Some communities are characterized by a higher level of generalized trust than other communities. One might expect the level of generalized trust to be higher in communities with a homogeneous population than in diverse communities. Nevertheless, it is a matter of (historical) contingency to which personal characteristics (e.g. ethnicity, religion, age, gender) the homogeneity or heterogeneity of a community relates. Thus, ethnic diversity will hamper generalized trust only if ethnicity is considered to be an important individual characteristic that warrants mistrust of persons belonging to a different (ethnic) group.

The existence of particular norms regarding solidarity is not sufficient to elicit solidary behaviour. These norms must also be internalized or externally enforced. As explained in the previous chapter, the main difference between normative and multilateral solidarity is that, in the case of the former, the solidarity norm is internalized, while in the latter case it has to be externally enforced.

Internalization of a social norm presupposes a socializing process in which the norms are learned and gradually become the subject’s own. If the norm is fully internalized, i.e. if it has become part of a subject’s belief system, s/he complies with the norm out of conviction, without the need of an external pressure. According to developmental psychologists, the early phases of upbringing are crucial for the internalization of social norms (primary socialization). However, in later phases of life, these norms may be reinforced or weakened by other external influences from, for example, peers and mass media (secondary socialization). It should be noted that internalization of a norm is usually more effectively enhanced by the conduct of others which confirms the norm, than by verbally transmitting the norm (‘practice what you preach’).

According to Durkheim, in a homogeneous society, the socialization of social norms may show itself in a collective consciousness (Durkheim 1893: 61). “The solidarity that derives from similarities is at its maximum when the collective consciousness completely envelops our total consciousness, coinciding with it at every point. At that moment our individuality is zero.” (Durkheim 1893: 84)

Ethnic diversity of a community may imply a differentiation between ethnic groups with respect to the norms that are internalized. Of course, this presupposes that different ethnic groups endorse different social norms. This can result in a variety of solidary behaviours that may affect the secondary socialization process by peers, mass media, et cetera. For


51

example, if your peers’ interpretation of deservingness is stricter than yours, this may persuade you to adapt your own deservingness criteria to your peers’ standards. Thus, prevailing solidarity norms may evolve over time. If these changing norms translate into other solidary behaviour, this process may reinforce itself. However, this process can move in both directions, either increasing or decreasing solidarity. Thus, there is no general presumption that increasing ethnic heterogeneity will result in less socialization of norms of solidarity.

In case of multilateral solidarity, the norms of solidary behaviour need to be externally enforced, since people have not fully internalized the norm. The external pressure that forces people to act according to the social norm, can take various forms. It can be formal or informal, material or psychological, negative or positive.

Informal social pressure usually presupposes that others have internalized the social norm and are therefore willing to reprimand a person who violates the norm or to praise a person who complies with the norm. Thus, the larger the proportion of group-members who have internalized the norm, the larger the probability that one will experience external pressure to comply with the norm (cf. de Beer & Mosch 2007). The form of the social pressure exercised by group members can vary from a verbal rebuke to a physical beating or ostracism. The social pressure can also be positive, in case a person who helps an anonymous group member is praised by other group members. Apart from the pleasure of being praised, the status that one derives from this may also enhance the probability of being helped by other group members in the future (indirect reciprocity).

Formal social pressure is exercised by a formally appointed supervisor, who is legally entitled to inflict a punishment. In the case of formal, compulsory solidarity, this can be, for example, a tax inspector who is authorized to fine a tax evader. Formal pressure may also be positive, for example if people who act solidary (e.g., volunteers) are awarded a decoration.

In a case of multilateral solidarity in which the subject does not believe in the social norm at all, the subject will only comply with the norm if the (expected) costs of non-compliance are higher than the costs of acting solidary. However, a more common situation is probably that a person has partly internalized the norm, but is nevertheless tempted to violate the norm because of the costs of compliance. In that case, the expected

Chapter Two

52

punishment need not equal the full costs, since the subject will be willing to bear at least some of the costs of acting solidary.

The effectiveness of deterrence of norm violation depends on the extent to which individuals care about the rebukes or punishment by others. Informal deterrence usually does not take the form of physical coercion, but of a verbal rebuke or a reprimand. It is then crucial that the potential benefactor recognizes the punisher as someone about whose rebuke s/he cares because the punisher belongs to her/his own group. This requires that they are not too different and have something in common. Thus, a reproach from a by-stander from your own ethnic group, inflicted because you fail to help another in-group member who is in need, may be a much stronger incentive to act solidary, than a reproach from a by-stander from another ethnic group. Consequently, increasing ethnic diversity may reduce the effectiveness of reprimands.

This need not be the case if there is a formal supervisor who is legally entitled to apply a sanction. However, the effectiveness of a formal supervisor usually depends not only on her/his formal competence but also on her/his personal authority, which may be less if the trespasser belongs to a different ethnic group.

The importance of perceptions

It is important to recognize that many of the possible effects of ethnic diversity that have been addressed in the preceding sections depend strongly on the perceptions of people who belong to other ethnic groups. Consequently, it is not so much the actual differences between (typical members of) ethnic groups that matter for solidarity, but the perceived differences. This is self-evident with respect to generalized trust, but it also applies to seemingly objective characteristics, such as the skills and needs of people from other ethnic groups. Of course, also the perception of the members of one’s own ethnic group matters. As a consequence, if perceptions change, without any real change taking place, this may affect solidarity between groups.

There is abundant empirical evidence that the perception of the members of another (ethnic) group is often based on stereotypes which do not coincide with their real characteristics. These stereotypes may, for example, be based on a few experiences with atypical members of the group. But they are probably also strongly influenced by the mass media. In particular, regarding people from (ethnic) groups that one rarely meets


53

in person, the image that the media depict of these groups may have a strong influence on the perception of any member of these groups. Although it is notoriously difficult to establish whether the mass media affect the dominant views among the population or that they simply reflect these dominant views, it is hard to deny that the media play an important role in disseminating the prevalent views of various ethnic groups. Therefore, in chapter 5, we will pay attention to the way that the mass media portray various ethnic groups in the Netherlands.

Conclusion

From the preceding discussion we can conclude that solidarity is possible both between similar and between different persons and both in ethnically homogeneous and in heterogeneous groups. Nevertheless, very large differences or very strong heterogeneity are generally detrimental to solidarity. This is simply so, because solidarity presupposes recognition of the other or some common purpose, interest or values. If two persons or groups have nothing in common, there is simply no sense in acting solidary. However, this does not imply that the more similar persons are, the stronger will be their willingness to act solidary towards each other. It depends on the motives for solidary behaviour and on the possibility (and willingness) to identify with others.

In case of other-regarding solidarity (both empathic and normative solidarity), the identification with another person or with a group is a crucial condition for solidarity. Identification is often fostered by similarity, but this need not always be the case. Identification can also be (purposefully) stimulated or manipulated, for example by the mass media.

In case of self-regarding solidarity (bilateral solidarity or multilateral solidarity), the awareness of mutual interdependence is the crucial condition for solidarity. In general, interdependence is created by differentiation and specialization. Thus, some heterogeneity may actually breed self-regarding solidarity. However, if individuals or groups become too different, this may weaken interdependence and reduce trust and, consequently, erode self-regarding solidarity. A crucial condition is then the extent to which the differences between ethnic groups are perceived as complementary, in the sense that they can easily benefit from each other, or as an unbridgeable divide.

In the remaining part of this volume, we study the relationship between ethnic diversity and solidarity empirically in various ways. We start with

Chapter Two

54

taking a cross-country perspective by analysing the relationship between various indicators of ethnic diversity at the national and at the regional level and some measures of solidarity in a large scale survey (Chapter 3). Next, we present the outcomes of a so-called vignette analysis, based on a representative survey among the Dutch population, in which the respondents were asked to express their preferences for helping fellow citizens with varying characteristics, including their ethnicity (Chapter 4). In the following step, we focus on the role of the mass media, by analysing the images that are depicted of various ethnic minority groups in some of the most widely read and watched Dutch news media (Chapter 5). Next, we examine the solidary behaviour – or its absence – between citizens with a similar or different ethnic background at the local level, by conducting an experiment with the visitors to a renowned market in a multicultural neighbourhood, called the Dapperbuurt, in the city of Amsterdam (Chapter 6). In the concluding chapter we sum up the main results and draw some general conclusions.

References

Alesina, A., and LaFerrara, E. (2002). Who trusts others? Journal of Public Economics 85, 207-234.

de Beer, P.T., and Mosch, R.H.J. (2009). The waning and restoration of social norms. DNB Working Paper No. 131, Amsterdam: De Nederlandsche Bank.

Gijsberts, M., van der Meer, T., and Dagevos, J. (2012). ‘Hunkering down’ in multi-ethnic neighbourhoods? The effects of ethnic diversity on dimensions of social cohesion. European Sociological Review 28, 527-537.

Habyarimana, J., Humphreys, M., Posner, D.N., and Weinstein, J.M. (2007). Why does ethnic diversity undermine public goods provision? American Political Science Review 101, 709-725.

Hooghe, M., Reeskens, T., Stolle, D., and Trappers, A. (2009). Ethnic diversity an generalized trust in Europe: A cross-national multilevel study. Comparative Political Studies 42, 198-223.

Leigh, A. (2006). Trust, inequality and ethnic heterogeneity. The Economic Record 82, 268-280.

Putnam, R.D. (2000). Bowling alone. New York/London/Toronto/Sydney: Simon & Schuster.

—. (2007). E Pluribus Unum. Diversity and community in the twenty-first century. The 2006 Johan Skytte Prize Lecture. Scandinavian Political Studies 30, 137-174.

CHAPTER THREE

HETEROGENEITY AND HUMAN SOCIALITY

FERRY KOSTER

Introduction

The term social distance refers to the classification of individuals or groups based on how far they are apart from each other relationally, rather than geographically (Park 1924; Bogardus 1925; Akerlof 1997). Social distance is not a unilateral concept and may refer to any socially relevant difference between actors, such as from group membership, social class, ethnic background, and so forth. Measures of social distance may be based on liking, in-group and out-group distinctions, and the extent to which there is contact between people having different backgrounds. The present chapter focuses on the latter of these three approaches to social distance by examining how social structural diversity, indicated as the chance that two actors with a different background meet each other within a given space (Smith 1960; Blau 1977; Koster 2013), relates to solidarity. For that purpose, international comparative survey data are analysed. Given that such surveys do not include in-depth information on actual behaviour, this chapter investigates the attitudes, motivations, and preferences underlying and enabling solidarity. Therefore, the central topic closely relates to what is called human sociality governing the social relationships between individuals (e.g. Henrich, Boyd, Bowles, Camerer, Fehr and Gintis 2004). In what follows, the focus is on two kinds of sociality that differ with regard to the way in which they are organized, namely informally, based on interpersonal relationships, and formally, involving impersonal means of organization. As an indicator of informal sociality, the level of interpersonal trust is examined and formal sociality includes trust in a country’s institutions and support for the welfare state.

This chapter investigates to what extent variation in people’s level of interpersonal trust, institutional trust, and welfare state support can be

Chapter Three

56

explained by the social structure in which these individuals reside. There are several dimensions along which the heterogeneity of social structures can be classified. Three of these dimensions received considerable attention in previous research, namely those based on ethnical, linguistic, and religious distinctions between individuals (Alesina, Devleeschauwer, Easterly, Kurlat and Wacziarg 2003). Since these social structural divisions may have an impact on human sociality through distinguishable mechanisms, their effects are considered separately.

Ethnic heterogeneity

There are contrasting theoretical expectations regarding the effects of ethnic heterogeneity. In his review of the “contact hypothesis” (which states that contact between people from different ethnic groups increases mutual understanding and social trust) and “conflict theory” (which in contrast states that in-group solidarity increases at the cost of less out-group solidarity as ethnic groups compete for scarce resources), Putnam (2007) argues that both of these theoretical approaches miss an important aspect of social reality, namely that ethnic heterogeneity has a negative impact on solidarity altogether as in-group as well as out-group relationships are negatively affected by ethnic heterogeneity. The idea that ethnic heterogeneity decreases in-group as well as out-group solidarity is the core of what is termed the “constrict hypothesis” (Putnam 2007). To a certain extent, this criticism holds when focusing on interpersonal trust, as the majority of studies do in line with Putnam’s study. Nevertheless, in the case of public goods provision, meaning that everyone benefits once the public good is provided, the distinction between the conflict and the constrict hypothesis may be less relevant as they both lead to the prediction that ethnic heterogeneity decreases welfare state support. According to conflict theories, support for certain types of welfare state arrangements may be supported more strongly by one ethnic group than the other if the social risks are unevenly spread across different ethnic groups. Even though this may mean that some groups are more in favour of some of the welfare state arrangements, it also follows that other groups will be less supportive of these arrangements. The end result is that in ethnically heterogeneous social structures there is less support for the welfare state compared to ethnically homogeneous societies, which is the same prediction that follows from the constrict hypothesis. In both cases, the main mechanism underlying this expectation is that heterogeneity decreases the possibilities for providing positive and negative sanctions which are required to sustain a norm of cooperation that in turn generates

Heterogeneity and Human Sociality

57

institutional trust and support for government redistribution (Habyarimana, Humphreys, Posner and Weinstein 2007; Miguel and Gugerty 2005).

Linguistic heterogeneity

Linguistic heterogeneity is often viewed as a kind of social structural diversity that is strongly related or even similar to ethnic heterogeneity. Empirically, it is not uncommon to use measures based on the ethno-linguistic fractionalization (ELF) of countries (e.g. Easterly and Levine 1997). Although the ethnic and linguistic heterogeneity of a society may be strongly related, they do not always overlap completely. As Alesina et al. (2003) show, combining ethnic and linguistic heterogeneity into a single measure may lead to a misspecification of the actual level of diversity of a country. Their data show, for example, that most countries in Latin America and the Caribbean are much more divided in terms of groups sharing the same language than along the lines of ethnic groups, while the situation is completely the opposite in European countries. As a result, a measure consisting of ethnic and linguistic heterogeneity blends different mechanisms explaining human sociality and using distinct measures allows theorizing about their unique effects instead. In discussing the different mechanisms explaining why ethnic diversity undermines cooperation, Habyarimana et al. (2007) – drawing on Deutsch (1966) and Hardin (1995) – argue that a shared language is part of the reservoir of common cultural material. Speaking the same language accommodates interpersonal communication and cooperation that gives ethnic homogeneous groups an advantage in producing public goods. In other words, linguistic homogeneity may lead to more efficient social interactions (Anderson and Paskeviciute 2006). However, instead of interpreting a shared language as a part of ethnic heterogeneity, linguistic heterogeneity is investigated here as a specific kind of heterogeneity affecting both informal and formal human sociality through a different mechanism, namely fewer possibilities for collective action due to communication problems, instead of lack of social sanctions.

Religious heterogeneity

Religious heterogeneity refers to the presence of people from different religious denominations within a certain social setting. Although religious affiliation may be one of the dividing lines between ethnic groups, it turns out that religious heterogeneity is only weakly related to ethnic and linguistic heterogeneity (Alesina et al. 2003). Thus, there is an empirical

Chapter Three

58

reason for distinguishing their effects. In addition to that, it is also theoretically sound to separate the effects of religious heterogeneity on welfare state attitudes. Religious denominations provide the possibility for members to organize themselves, developing feelings of belonging to a specific group, and exclude those who are not a member of the group (Koster, Goudriaan and van der Schans, 2009). This means that membership of a religious group may create a strong dividing line between insiders and outsiders as it facilitates a clear difference between members and non-members. As such, religious affiliations provide a clear point of reference to distinguish between “them” and “us” which may in turn have consequences for the social behaviour of group members (Yamagishi and Mifune 2008). Furthermore, since the members interact on a regular basis, such divisions may strengthen over time, increasing interpersonal trust within religious groups while decreasing the level of interpersonal trust in outsiders (Stolle 1998). As a result, sharing the same religious background may provide the opportunity structure for interacting with similar others while decreasing social contacts with outsiders and making these differences more pronounced. This, in turn, may have negative consequences for both informal and formal human sociality.

In short, these different theoretical considerations lead to the general expectation that social structural diversity has a negative effect on human sociality through different mechanisms. Ethnic diversity is believed to hinder social sanctioning, linguistic diversity hinders efficient communication, and religious diversity can lead to a stronger division between members and non-members. Based on these mechanisms it is hypothesized that interpersonal trust, institutional trust, and support for government redistribution are negatively related to heterogeneity. These individual level attitudes may in turn have macro level consequences. Although interpersonal trust does not necessarily have a direct influence on people’s political preferences, it may have an indirect effect on citizenship behaviours, for example. (It should however be noted that the opposite can also be hypothesized: if interpersonal trust is higher, people may be less trusting towards the government.) The macro level consequences are more clearly related to both institutional trust and support for government redistribution through welfare state legitimacy and support, based on the following considerations. The production of collective goods depends on social structural conditions (e.g. Oliver 1993) and structural heterogeneity may be among the conditions restricting the mechanisms bringing about public support for such goods. In terms of theories emphasizing explanations based on social mechanisms, as for example proposed by authors like Coleman (1990), Hedström and Swedberg (1998) Gross


59

(2009) and Hedström and Ylikoski (2010), the focus is on “situational mechanisms” that connect the macro level with the micro level as social structures and environments shape actions, desires, and believes. In the analyses presented here, the macro level conditions are the level of heterogeneity at the regional level and the micro level concerns individual welfare state attitudes. Each of the three kinds of heterogeneity – ethnic, linguistic, and religious – relates to these individual outcomes through a different mechanism. Ultimately, these individual actions, desires, and believes generate intended and unintended social outcomes through so-called “transformational mechanisms” (Hedström and Swedberg 1998), which translate support into actual social policies (Brooks and Manza 2006).

Prior studies

To date, two strands of literature can be identified in which societal heterogeneity and human sociality are investigated empirically. The first branch, which may be called the Putnam strand – as one of the main references is his article “E Pluribus Unum” (2007) – focuses on the relationship between heterogeneity and interpersonal trust. Most of the studies that belong to this part of the literature consist of single country analyses of social structural heterogeneity at relatively low levels of aggregation (e.g. the neighbourhood) using census data (Leigh 2006; Putnam 2007; Tolsma, van der Meer and Gesthuizen 2008). These investigations provide mixed evidence for a negative relationship between heterogeneity and interpersonal trust. A strong point of these studies is that they rely on fine-grained indicators of heterogeneity. They do, however, not allow for a comparison of the effects of diversity across countries. Therefore, they do not provide information about how the impact of diversity differs depending on cross-national variation in institutions and economic circumstances and the consequences they have for the welfare state. This latter issue is the topic of the second branch of the heterogeneity literature, which can be referred to as the Alesina strand with “Fractionalization” by Alesina et al. (2003) as a main source of reference, consisting of international comparative studies that measure diversity at the national level using archival data and examine how it relates to social policies at the national level (Fearon 2003; Alesina and Glaeser 2004) and individual level outcomes such as social capital, citizenship behavior, and welfare state attitudes (Anderson and Paskeviciute 2006; Gesthuizen, van der Meer and Scheepers 2009; Mau and Burkhardt 2009; Burgoon, Koster and van Egmond, 2012). These

Chapter Three

60

studies show that, at the national level, heterogeneity and the welfare state are negatively related. However, when relating national level heterogeneity with individual outcomes, these studies have only found a moderate or weak association.

Each of the two approaches to heterogeneity and human sociality lead to important insights, but they also overlook some aspects that need further investigation. The Putnam studies investigate data from one country, usually focus exclusively on ethnic heterogeneity, and do not investigate attitudes towards formal institutions. The Alesina studies, in contrast, do include a wider range of countries, more dimensions of heterogeneity, and different kinds of attitudes, but do not generate very strong conclusions about the relationship between heterogeneity and human sociality. A potential problem with the latter approach is that measures of heterogeneity at the national level may simply be too abstract and general to have an impact on individual attitudes. It may be questioned whether national level indices measure the relevant parts of people’s social context and their preferences, attitudes and behaviour. The present chapter tries to bridge the gap between these two empirical approaches by investigating to what extent individual attitudes relate to regional diversity across European countries (see also Koster, 2013). Two different steps are central for achieving this goal. First, new measures of heterogeneity are constructed at lower levels of aggregation than the ones that are investigated in the Alesina studies. Clearly, these new measures are not at the same level of aggregation as the Putnam studies, but they do allow for cross-national comparisons, which is largely understudied in this part of the literature. The newly constructed measures are compared with external sources to investigate their external validity. Having new measures is not a goal in itself, but it is a stepping stone to answer a more relevant question, namely whether these disaggregated indicators of heterogeneity do explain variation in interpersonal trust, institutional trust, and support for government redistribution. The second step, therefore, involves an empirical comparison of the effect of regional and national level heterogeneity.

Data and method

Survey data

This study uses the European Social Survey (ESS) as its main source to investigate the relationship between heterogeneity and human sociality. The ESS has been held every two years since 2002 and at the time of this research the data of four rounds were available (ESS 2002; 2004; 2006;


61

2008). To include as much information as possible, these four rounds are combined and analysed together. All of the variables investigated here are asked in all rounds of the ESS as they belong to its core module. Table 1 shows which countries participated per ESS round. In total 33 countries participated in these four rounds. However, the number of times that countries participated varies since not every country participated in every round. There are 15 countries for which data are available for all four rounds, 7 countries participated three times, 7 other countries participated two times, and in 4 countries data have been gathered only once. Round 1 includes 22 countries, in Round 2 26 countries participated, Round 3 includes 23 countries, and Round 4 consists of 28 countries. To get data from a broad range of countries, the average scores are computed for each country across the four rounds of the ESS. Table 1 provides an overview of the ESS Rounds, the countries, and the number of respondents.

Table 1. Number of respondents per country and per ESS round

Round 1

Round 2

Round 3

Round 4 Total

Austria 2,257 2,256 2,405 --- 6,918 Belgium 1,899 1,778 1,798 1,760 7,235 Bulgaria --- --- 1,400 2,230 3,630 Switzerland 2,040 2,141 1,804 1,819 7,804 Cyprus --- --- 995 1,215 2,210 Czech Republic 1,360 3,026 --- 2,018 6,404

Germany 2,919 2,870 2,916 2,751 11,456 Denmark 1,506 1,487 1,505 1,610 6,108 Estonia --- 1,989 1,517 1,661 5,167 Spain 1,729 1,663 1,876 2,576 7,844 Finland 2,000 2,022 1,896 2,195 8,113 France 1,503 1,806 1,986 2,073 7,368 United Kingdom 2,052 1,897 2,394 2,352 8,695

Greece 2,566 2,406 --- 2,072 7,044 Croatia --- --- --- 1,484 1,484 Hungary 1,685 1,498 1,518 1,544 6,245 Ireland 2,046 2,286 1,800 --- 6,132 Iceland --- 579 --- --- 579 Italy 1,207 1,529 --- --- 2,736 Luxembourg 1,552 1,635 --- --- 3,187 Latvia --- --- --- 1,980 1,980

Chapter Three

62

Netherlands 2,364 1,881 1,889 1,778 7,912 Norway 2,036 1,760 1,750 1,549 7,095 Poland 2,110 1,716 1,721 1,619 7,166 Portugal 1,511 2,052 2,222 2,367 8,152 Romania --- --- --- 2,146 2,146 Sweden 1,999 1,948 1,927 1,830 7,704 Slovenia 1,519 1,442 1,476 1,286 5,723 Slovakia --- 1,512 1,766 1,810 5,088 Turkey --- 1,856 --- 2,416 4,272 Total 39,860 47,035 38,561 48,141 173,597 Source: European Social Survey

Comparing heterogeneity and fractionalization

International comparative surveys offer a unique opportunity for constructing indicators of social structures. For the present study, data from the ESS are examined for the following reasons. First, this dataset allows construct heterogeneity scores and to compare them empirically across the majority of European countries with the existing fractionalization scores. Secondly, the ESS includes a variable indicating the region where people live, thus allowing for the construction of indicators at disaggregated levels.

The variables of interest are the country of origin, which is measured with the question “In which country were you born?”, language, which is asked with the question “What language or languages do you speak most often at home?” and religion, measured with the item “Do you consider yourself as belonging to any particular religion or denomination” and if the respondent answers yes, the follow up question is asked: “Which one?”. These individual answers are aggregated to the country level to construct the national level indicators and to the NUTS level 1 level to construct the regional indicators. This procedure results in a mean value for the categories within a country or a region, which is used to construct heterogeneity scores for ethnic, linguistic and religious diversity per country and per region. These heterogeneity scores are in line with the most common measures found in the literature. Both the Putnam studies and the Alesina studies use the same logic in constructing these scores. The basic formula underlies these measures (which is referred to as Simpson’s diversity index, Blau’s index or the Herfindahl-Hirschman index).


63

where

p = proportion of individuals per category N = number of categories.

A score of 0 indicates a perfectly homogenous population and 1 a perfectly heterogeneous one (every person belongs to a different group). The score of the index indicates the probability that two randomly selected individuals from the population belong to different groups (e.g. Alesina et al. 2003).

Comparing the number of foreigners in the ESS

Even though the ESS aims at including a representative sample of citizens of each country, the question can be asked to what extent the data reflect the national population in terms of nationality or country of birth. To investigate this, data from the Eurostat database on the proportion of foreigners in a country (defined as non-nationals, meaning persons who are not citizens of their country of residence) (Eurostat 2007) are compared to the ESS data. These data are available for 25 of the 33 countries for the period between 2000 and 2008. Table 2 summarizes the outcomes. On average, the foreign population constitutes 7 per cent of the total population of the 25 countries according to the Eurostat dataset. The Eurostat data are compared to the ESS indicator asking about people’s country of origin. Overall, the number of foreign-born people in the ESS is slightly higher than the percentage foreigners reported by Eurostat (0.08 versus 0.07). As table 2 shows, in most countries these numbers are somewhat higher in the ESS than in the Eurostat database. On the other hand, the differences between this ESS indicator and the Eurostat measures are considerable for countries like Luxembourg and Latvia (both 9 per cent difference). In addition to the differences between the Eurostat and the ESS indicators, they are related to each other to investigate their similarity. Figure 1 represents the relationship between the proportion of foreigners according to Eurostat and those interviewed in the ESS. Figure 1 shows that these indicators are positively related. This comparison shows that the ESS includes a sizable number of foreigners and that the proportion of foreigners in the ESS is strongly related to that of an external source (Eurostat), but also that it should be realized that the numbers slightly underestimate the number of actual foreigners (non-nationals) and

Chapter Three

64

that this particularly holds for some of the outliers in the dataset.

Table 2. Shares of foreigners, Eurostat and ESS % Foreigners % Foreign

born Difference

Austria 0.09 0.08 0.01 Belgium 0.08 0.09 -0.01 Bulgaria 0.00 0.01 -0.01 Switzerland --- (0.19) --- Cyprus 0.11 0.07 0.04 Czech Republic 0.02 0.03 -0.01 Germany 0.09 0.08 0.01 Denmark 0.05 0.06 -0.01 Estonia 0.19 0.20 -0.01 Spain 0.05 0.07 -0.02 Finland 0.02 0.03 -0.01 France 0.06 0.09 -0.03 United Kingdom 0.05 0.10 -0.05 Greece 0.08 0.09 -0.01 Croatia --- (0.09) --- Hungary 0.01 0.02 -0.01 Ireland 0.06 0.09 -0.03 Israel --- (0.35) --- Iceland --- (0.03) --- Italy 0.03 0.02 0.01 Luxembourg 0.39 0.30 0.09 Latvia 0.23 0.14 0.09 Netherlands 0.04 0.08 -0.04 Norway --- (0.07) --- Poland 0.00 0.01 -0.01 Portugal 0.02 0.06 -0.04 Romania 0.00 0.01 -0.01 Russia --- (0.06) --- Sweden 0.05 0.11 -0.06 Slovenia 0.02 0.08 -0.06 Slovakia 0.00 0.03 -0.03 Turkey --- (0.01) --- Ukraine --- (0.12) --- Total 0.07 0.08 -0.01 Sources: Eurostat 2000-2008, European Social Survey 1-4


65

Figure 1. Comparison of percentage foreigners

Comparing heterogeneity with fractionalization

Table 3 shows the fractionalization scores of Alesina et al. (2003) together with the heterogeneity measures based on the ESS data. The average scores of the indicators that are constructed using the ESS are consistently lower than the fractionalization scores. While the ethnic fractionalization of the European countries is 0.25, their average ethnic heterogeneity is 0.15 based on the country of origin indicator of the ESS. According to the fractionalization dataset the mean level of linguistic fractionalization is also 0.25, whereas the ESS reports a mean of 0.18. The level of religious fractionalization is 0.37 compared to 0.26 in the ESS data. Again there are some noteworthy country differences, as can be read in Table 4. For a number of countries it holds that the level of fractionalization and the ESS based heterogeneity is quite similar. For some countries, the measures are exactly the same: Norway, Portugal, and Sweden have a difference of zero. There are 15 countries in the dataset for which the difference

Chapter Three

66

between the ethnic fractionalization score and the ESS heterogeneity measure based on the nationality of respondents falls within a range of a 0.10 and this number increases to 0.18 if ethnic fractionalization is compared to the country of origin measure of the ESS. On the other hand, there are countries with a difference that is higher than 0.30 (Bulgaria and Croatia, for example) and there are two countries with a difference of 0.46 (Belgium and Ukraine). Focusing on the country of origin indicator of the ESS, the differences are somewhat smaller, but it also turns out that for some countries (namely Sweden and Israel) the level of ethnic heterogeneity according to the ESS is higher than the ethnic fractionalization scores.

Comparing the two datasets with regard to linguistic diversity shows that the fractionalization scores and the ESS based indicator are closer to each other than the ethnic diversity measures. The indicators fall within a range of 0.10 for 25 of 33 the countries. The difference between the level of linguistic fractionalization and the level measured with the ESS dataset is particularly large in the Czech Republic (a difference of 0.30), Cyprus (a difference of 0.36), and the Netherlands (a difference of 0.42). The linguistic fractionalization score of Italy is lower than the ESS based measure of heterogeneity (a difference of 0.13).

With regard to the level of religious diversity, the two types of indicators show the following similarities and differences. There are 18 countries with a difference within the 0.10 range and five countries have a difference that is larger than 0.30, notably the Czech Republic (0.41), Cyprus (0.37), Spain (0.35), Ukraine (0.32), and Luxembourg (-0.34).

The relationships between the different indicators of diversity are shown in Figures 2 through 4. There is a positive relationship between the fractionalization scores and the corresponding indicators measured with the ESS. The correlation coefficients presented in Table 5 confirm this relationship. Ethnic fractionalization is correlated with the country of origin measures of the ESS (r = 0.455), the two indicators measuring linguistic diversity are positively related (r = 0.818) and the same holds for religious diversity (r = 0.695). Table 5 also shows the interrelations between the different indicators. In accordance with the correlation coefficients calculated for the whole sample of the fractionalization data, ethnic and linguistic fractionalization are positively related to each other (r = 0.770) and are less strongly related to religious fractionalization (r = 0.177 for ethnic fractionalization and r = 0.346 for linguistic fractionalization). The patterns are somewhat similar when the ESS measures

Het

erog

enei

ty a

nd H

uman

Soc

ialit

y

67

Tab

le 3

. Fra

ctio

naliz

atio

n an

d E

SS h

eter

ogen

eity

Frac

tiona

lizat

ion

Het

erog

enei

ty

E

thni

c Li

ngui

stic

R

elig

ious

O

rigi

n Li

ngui

stic

R

elig

ious

A

ustri

a 0.

11

0.15

0.

41

0.15

0.

06

0.21

Be

lgiu

m

0.56

0.

54

0.21

0.

16

0.51

0.

22

Bulg

aria

0.

40

0.30

0.

60

0.02

0.

25

0.36

Sw

itzer

land

0.

53

0.54

0.

61

0.35

0.

53

0.58

Cy

prus

0.

09

0.40

0.

40

0.12

0.

04

0.03

Cz

ech

Repu

blic

0.

32

0.32

0.

66

0.06

0.

02

0.25

G

erm

any

0.17

0.

16

0.66

0.

15

0.07

0.

57

Den

mar

k 0.

08

0.10

0.

23

0.11

0.

04

0.13

Es

toni

a 0.

51

0.49

0.

50

0.33

0.

43

0.54

Sp

ain

0.42

0.

41

0.45

0.

14

0.25

0.

10

Finl

and

0.13

0.

14

0.25

0.

04

0.12

0.

07

Fran

ce

0.10

0.

12

0.40

0.

17

0.10

0.

24

Uni

ted

Kin

gdom

0.

12

0.05

0.

69

0.18

0.

07

0.58

G

reec

e 0.

16

0.03

0.

15

0.16

0.

07

0.07

Cr

oatia

0.

37

0.08

0.

04

0.16

0.

01

0.14

H

unga

ry

0.15

0.

03

0.52

0.

04

0.01

0.

43

Irela

nd

0.12

0.

03

0.16

0.

17

0.05

0.

10

Isra

el

0.34

0.

55

0.35

0.

57

0.51

0.

37

Icel

and

0.08

0.

08

0.19

0.

06

0.01

0.

16

Italy

0.

11

0.11

0.

30

0.04

0.

24

0.03

Lu

xem

bour

g 0.

53

0.64

0.

09

0.50

0.

50

0.43

Cha

pter

Thr

ee

68 Latv

ia

0.59

0.

58

0.56

0.

25

0.42

0.

70

Net

herla

nds

0.11

0.

51

0.72

0.

16

0.09

0.

62

Nor

way

0.

06

0.07

0.

20

0.12

0.

09

0.20

Po

land

0.

12

0.05

0.

17

0.03

0.

00

0.03

Po

rtuga

l 0.

05

0.02

0.

14

0.11

0.

02

0.07

Ro

man

ia

0.31

0.

17

0.24

0.

01

0.14

0.

22

Russ

ia

0.25

0.

25

0.44

0.

11

0.10

0.

22

Swed

en

0.06

0.

20

0.23

0.

20

0.10

0.

28

Slov

enia

0.

22

0.22

0.

29

0.15

0.

04

0.13

Sl

ovak

ia

0.25

0.

26

0.57

0.

05

0.20

0.

32

Turk

ey

0.32

0.

22

0.00

0.

02

0.23

0.

03

Ukr

aine

0.

47

0.47

0.

62

0.21

0.

50

0.30

To

tal

0.25

0.

25

0.37

0.

15

0.18

0.

26

Sour

ces:

Ale

sina

, Dev

lees

chau

wer

, Eas

terly

, Kur

lat a

nd W

aczi

arg

(200

3) a

nd E

urop

ean

Soci

al S

urve

y 1-

4


69

Table 4. Differences between fractionalization and ESS heterogeneity

Ethnic/origin Linguistic Religious Austria -0.04 0.09 0.20 Belgium 0.40 0.03 -0.01 Bulgaria 0.38 0.05 0.24 Switzerland 0.18 0.01 0.03 Cyprus -0.03 0.36 0.37 Czech Republic 0.26 0.30 0.41 Germany 0.02 0.09 0.09 Denmark -0.03 0.06 0.10 Estonia 0.18 0.06 -0.04 Spain 0.28 0.16 0.35 Finland 0.09 0.02 0.18 France -0.07 0.02 0.16 United Kingdom -0.06 -0.02 0.11 Greece 0.00 -0.04 0.08 Croatia 0.21 0.07 -0.10 Hungary 0.11 0.02 0.09 Ireland -0.05 -0.02 0.06 Israel -0.23 0.04 -0.02 Iceland 0.02 0.07 0.03 Italy 0.07 -0.13 0.27 Luxembourg 0.03 0.14 -0.34 Latvia 0.34 0.16 -0.14 Netherlands -0.05 0.42 0.10 Norway -0.06 -0.02 0.00 Poland 0.09 0.05 0.14 Portugal -0.06 0.00 0.07 Romania 0.30 0.03 0.02 Russia 0.14 0.15 0.22 Sweden -0.14 0.10 -0.05 Slovenia 0.07 0.18 0.16 Slovakia 0.20 0.06 0.25 Turkey 0.30 -0.01 -0.03 Ukraine 0.26 -0.03 0.32 Total 0.09 0.07 0.10 Source: European Social Survey 1-4

Chapter Three

70

are investigated. The ethnic diversity measure based on country of birth is positively related to linguistic diversity (r = 0.567 and r = 0.652). The correlations between religious diversity on the one hand and ethnic and linguistic diversity on the other hand are different than those reported for the fractionalization measures: ethnic and religious heterogeneity measured with the ESS are positively related (r = 0.462 and r = 0.472), while linguistic and religious heterogeneity have the lowest correlation (r = 0.419).

Figure 2. Ethnic diversity


71

Figure 3. Linguistic diversity

Chapter Three

72

Figure 4. Religious diversity

Furthermore, Table 5 provides insight how the diversity scores from the different sources are related to each other. With regard to ethnic fractionalization, it is worth mentioning that this indicator of diversity is most strongly related to the linguistic heterogeneity measured in the ESS (r = 0.821). Moreover, linguistic fractionalization is also strongly related to the ESS based indicator of linguistic heterogeneity (r = 0.818). A possible interpretation of this result, in combination with the strong relation between ethnic and linguistic fractionalization and the fact that the ethnic fractionalization and ethnic heterogeneity measured with the ESS are less strongly correlated, is that despite the effort to distinguish ethnic and linguistic fractionalization, the ethnic fractionalization indicator mainly reflects the linguistic diversity within a country. However, before this conclusion can be stated succinctly additional data explorations are required, which can be the subject of future studies. The results regarding religious fractionalization are clear-cut. This variable is not related to any of the heterogeneity scores based on the ESS except for religious


73

heterogeneity (r = 0.695), suggesting that these indicators essentially measure similar aspects of the countries investigated in this sample.

Table 5. Correlation coefficients fractionalization and heterogeneity Fractionalization Heterogeneity Ethnic Linguistic Religious Origin Linguistic Fractionalization Ethnic Linguistic 0.770** Religious 0.177 0.346* Heterogeneity Origin 0.455** 0.628** 0.053 Linguistic 0.821** 0.818** 0.160 0.652** Religious 0.414* 0.506** 0.695** 0.472** 0.419* N = 33 countries ** p < 0.01; * p < 0.05 Sources: Alesina, Devleeschauwer, Easterly, Kurlat and Wacziarg (2003) and European Social Survey 1-4

The comparison of the ESS data with two external sources supports the following conclusions. First, on average, the ESS includes shares of foreigners that are quite similar to the figures provided by Eurostat. However, since for some countries the ESS shares of foreigners deviate considerably from the Eurostat numbers, it is advisable to explore whether this affects the resulting indices. In empirical analyses, it is possible to investigate how much inclusion or exclusion of these countries affects the outcomes by conducting a sensitivity analysis. Secondly, comparing different indicators measuring societal diversity, leads to the conclusion that the ESS based scores are consistently lower than the fractionalization scores, which were taken as a point of reference. With regard to this outcome, it cannot simply be stated that the ESS underestimates the level of diversity, as it is not clear what the true level of diversity is in the countries included in the analysis. It may as well be concluded that the fractionalization scores overestimate the level of diversity. There may be a number of reasons for these differences. First, the ESS may include too few respondents from different backgrounds, thus lacking the ability to capture the diversity to the fullest. Secondly, as noted above, related but different dimensions of diversity may be measured in both datasets. In particular, this remark relates to the close relationship between ethnic and linguistic diversity. And, thirdly, part of the difference can result from the difference between using a more objective source (archival data measuring fractionalization) and self-reports involving a certain level of subjectivity.

Chapter Three

74

Given the differences between the ESS indicators and the external sources, it is acknowledged that ESS data are not completely the same as the Eurostat and fractionalization data. However, a focus on the similarities between the measures, by investigating country plots and correlations, also shows that the indicators are interrelated and thus provide evidence that the measures at least reflect a relatively equal ordering of the countries in the dataset in terms of diversity.

Regional heterogeneity measures

So far, the focus was on cross national variation in diversity, a topic that has received quite some attention in prior research, whereas other studies have focused on diversity at lower levels of aggregation. A combination of these two strategies, investigating diversity at regional levels across a sample of countries, is not available yet. With the ESS data this can be investigated as respondents are asked about their region of residence. Appendix 1 shows the number of respondents and the ESS heterogeneity scores for each country disaggregated to the NUTS level 1 region. For 11 countries the country level is the same as the NUTS level. This means that the diversity measures cannot be disaggregated to a lower level based on this classification scheme. The level of intra-country diversity of the remaining 19 countries can be further examined as they fall into a number of different NUTS level 1 regions. The number of regions ranges from 2 (Bulgaria and Finland) to 16 (Germany).

The data in Appendix 1 provide some evidence that NUTS level 1 regions within countries differ with regard to their level of heterogeneity measured with the ESS indicators. Table 6 investigates these differences more closely. Table 6 reports the country means, standard deviations (based on the NUTS level 1 regions) and the range of heterogeneity (the difference between the minimum and the maximum value of the NUTS level 1 region within a country).

With regard to ethnic heterogeneity, Table 6 shows the following. Bulgaria and Finland have the lowest value on this measure with a standard deviation of 0.00 and a range of 0.01. In contrast, Belgium and the United Kingdom have a standard deviation of 0.08 and 0.10 and a range of 0.30 and 0.39, respectively. In Hungary and Poland, the level of linguistic heterogeneity varies the least across the NUTS level 1 regions of the countries in this sample (the standard deviations for both countries are 0.01 and the range is also 0.01). Quite different is the situation for Turkey and Spain where the standard deviations are 0.22 and 0.18 and the scores

Het

erog

enei

ty a

nd H

uman

Soc

ialit

y

75

Tab

le 6

. Reg

iona

l het

erog

enei

ty

E

thni

c he

tero

gene

ity

Lin

guis

tic h

eter

ogen

eity

R

elig

ious

het

erog

enei

ty

M

ean

SD

Mea

n SD

M

ean

SD

Aus

tria

0.14

0 0.

026

0.06

3 0.

025

0.20

7 0.

050

Bel

gium

0.

240

0.15

4 0.

187

0.17

7 0.

307

0.22

0 B

ulga

ria

0.02

5 0.

007

0.25

0 0.

057

0.36

0 0.

000

Cro

atia

0.

160

0.04

4 0.

007

0.00

6 0.

137

0.03

8 C

ypru

s 0.

130

---

0.04

0 --

- 0.

030

---

Cze

ch R

epub

lic

0.06

0 --

- 0.

020

---

0.25

0 --

- D

enm

ark

0.

110

---

0.04

0 --

- 0.

130

---

Esto

nia

0.32

0 --

- 0.

430

---

0.54

0 --

- Fi

nlan

d 0.

055

0.00

7 0.

130

0.02

8 0.

075

0.00

7 Fr

ance

0.

148

0.08

3 0.

101

0.06

0 0.

214

0.11

1 G

erm

any

0.13

9 0.

069

0.07

2 0.

039

0.43

4 0.

108

Gre

ece

0.10

4 0.

062

0.04

1 0.

037

0.21

6 0.

196

Icel

and

0.06

0 --

- 0.

010

---

0.16

0 --

- Ita

ly

0.04

6 0.

024

0.23

8 0.

093

0.03

6 0.

013

Latv

ia

0.24

0 --

- 0.

420

---

0.70

0 --

- Lu

xem

bour

g 0.

430

---

0.50

0 --

- 0.

430

---

Net

herla

nds

0.13

5 0.

049

0.10

8 0.

116

0.50

8 0.

181

Nor

way

0.

124

0.04

6 0.

084

0.03

2 0.

184

0.09

3 Po

land

0.

033

0.02

2 0.

005

0.00

8 0.

032

0.01

5 Po

rtuga

l 0.

110

---

0.02

0 --

- 0.

070

---

Rom

ania

0.

010

0.00

8 0.

130

0.17

7 0.

210

0.21

1

Cha

pter

Thr

ee

76 Slov

akia

0.

060

---

0.20

0 --

- 0.

320

---

Slov

enia

0.

150

---

0.04

0 --

- 0.

130

---

Spai

n 0.

133

0.05

3 0.

170

0.16

9 0.

094

0.04

4 Sw

eden

0.

180

0.06

6 0.

093

0.04

0 0.

260

0.10

0 Sw

itzer

land

0.

303

0.04

9 0.

310

0.10

8 0.

512

0.07

6 Tu

rkey

0.

045

0.10

9 0.

179

0.23

4 0.

028

0.04

9 U

nite

d K

ingd

om

0.15

2 0.

105

0.06

0 0.

059

0.53

7 0.

077

Sour

ce: E

urop

ean

Soci

al S

urve

y 1-

4


77

on linguistic heterogeneity have a range 0.62 and 0.41, respectively. Finally, religious heterogeneity also turns out to vary across the NUTS level 1 regions within countries. With respect to this kind of diversity, Bulgaria and Finland stand out with a standard deviation of 0.00 and a range of 0.00 and 0.01, respectively. On the other side there are countries like Romania and the Netherlands with a standard deviation of 0.18 and 0.16 and a range of 0.47 and 0.41, respectively.

This quick overview of how the measures of heterogeneity constructed with the ESS data vary across regions within countries provides evidence for the conclusion that in some countries the different regions are far more similar to each other than in other countries. In practice, this means that in some countries people from different regions experience the same level of diversity, whereas in other countries this depends much more on where the person lives. With respect to the impact that diversity may have on people’s behaviour, attitudes and opinions, these differences may indicate that studies relating national level diversity to individual level outcomes miss an important share of the variation in diversity across regions within countries.

Effects of national level heterogeneity versus regional level heterogeneity

In this section, three kinds of sociality are investigated that are close to the ones examined in previous studies, namely interpersonal trust, institutional trust, and support for government redistribution. The main aim of this part of the analysis is to examine whether regional heterogeneity affects human sociality using international comparative data rather than single level data. To examine whether regional level indicators of heterogeneity yield different individual outcomes than the ones at the national level the results for these indicators are compared.

Multilevel analyses

To take the nested structure of the data into account a three-level multilevel model is constructed. The three dependent variables are investigated following the same steps. First a control model (Model 1) is estimated that functions as a point of reference to compare subsequent models. Then, three models are constructed investigating the effects of ethnic, linguistic, and religious heterogeneity at the regional level (Models 2a-c). And finally, three models are estimated investigating ethnic,

Chapter Three

78

linguistic and religious heterogeneity at the national level (Models 3a-c). The deviance (the difference in -2 log likelihood) is computed using full information maximum likelihood to assess the change in the fit of the models (Snijders and Bosker 1999). The parameters in these models are estimated by the maximum likelihood method and the regression coefficients are tested by Wald tests (Goldstein 2003; Snijders 2003).

The results of the multilevel analyses are reported in Table 7, 8, and 9. Concentrating on the models with the control variables, the analyses show that the outcomes for the three kinds of sociality differ markedly. While people report higher levels of interpersonal trust and support for government redistribution in later rounds of the ESS compared to the first, their trust in institutions is lower in the later rounds of the ESS. Women report higher levels of interpersonal trust and they are more in favour of government redistribution, but gender does not explain variation in institutional trust. Interpersonal trust and support for government redistribution is negatively related to age, while institutional trust is not related to the age of respondents. Educational level is positively related to interpersonal trust and institutional trust and negatively related to support for government redistribution. Finally, while interpersonal trust and institutional trust are higher in more wealthy countries, GDP per capita is negatively related to support for government redistribution.

Turning to the models relating regional heterogeneity and national level fractionalization to interpersonal trust, Table 7 shows that none of these indicators explains variation in this kind of sociality. All forms of regional and national level diversity, except religious heterogeneity of the region, have a negative sign, but none of these effects is significantly different form zero.

Investigating institutional trust, Table 8 shows that it is positively related to ethnic heterogeneity and linguistic heterogeneity at the regional level and negatively to ethnic fractionalization and religious fractionalization of the country. Nevertheless, focusing on how much the variables contribute to the fit of the model, it is evident that including both effects at the regional level leads to a significant improvement (Deviance = 3.900; p < 0.05 for ethnic heterogeneity and Deviance = 15.800; p < 0.01 for linguistic heterogeneity), while including both heterogeneity indicators at the national level does not result in significant improvements of the model (Deviance = 3.800; n.s. for ethnic fractionalization and Deviance = 3.300; n.s. for religious fractionalization).

Het

erog

enei

ty a

nd H

uman

Soc

ialit

y

79

Tab

le 7

. Mul

tilev

el a

naly

sis o

f int

erpe

rson

al tr

ust

(1

) (2

a)

(2b)

(2

c)

(3a)

(3

b)

(3c)

R

egio

nal l

evel

Ethn

ic h

eter

ogen

eity

-0.2

95

(0.3

77)

Ling

uist

ic h

eter

ogen

eity

-0

.169

(0

.227

)

Rel

igio

us h

eter

ogen

eity

0.31

6 (0

.233

)

Cou

ntry

leve

l

Ethn

ic fr

actio

naliz

atio

n

-1

.197

(0

.726

)

Ling

uist

ic fr

actio

naliz

atio

n

-0.5

15

(0.6

92)

Rel

igio

us fr

actio

naliz

atio

n

0.

662

(0.5

86)

Con

trol

var

iabl

es

ESS

Rou

nd 2

0.

038*

* (0

.013

) 0.

038*

* (0

.013

) 0.

038*

* (0

.013

) 0.

038*

* (0

.013

) 0.

038*

* (0

.013

) 0.

038*

* (0

.013

) 0.

038*

* (0

.013

)

ESS

Rou

nd 3

0.

111*

* (0

.013

) 0.

111*

* (0

.013

) 0.

111*

* (0

.013

) 0.

112*

* (0

.013

) 0.

111*

* (0

.013

) 0.

111*

* (0

.013

) 0.

111*

* (0

.013

)

Cha

pter

Thr

ee

80 ESS

Rou

nd 4

0.

129*

* (0

.013

) 0.

129*

* (0

.013

) 0.

129*

* (0

.013

) 0.

129*

* (0

.013

) 0.

129*

* (0

.013

) 0.

129*

* (0

.013

) 0.

129*

* (0

.013

)

Gen

der (

1 =

fem

ale)

0.

071*

* (0

.009

) 0.

071*

* (0

.009

) 0.

071*

* (0

.009

) 0.

071*

* (0

.009

) 0.

071*

* (0

.009

) 0.

071*

* (0

.009

) 0.

071*

* (0

.009

)

Age

-0

.003

**

(0.0

00)

-0.0

03**

(0

.000

) -0

.003

**

(0.0

00)

-0.0

03**

(0

.000

) -0

.003

**

(0.0

00)

-0.0

03**

(0

.000

) -0

.003

**

(0.0

00)

Yea

rs o

f edu

catio

n 0.

054*

* (0

.001

) 0.

054*

* (0

.001

) 0.

054*

* (0

.001

) 0.

054*

* (0

.001

) 0.

054*

* (0

.001

) 0.

054*

* (0

.001

) 0.

054*

* (0

.001

)

GD

P pe

r cap

ita

0.60

9**

(0.1

25)

0.62

5**

(0.1

25)

0.61

3**

(0.1

23)

0.60

4**

(0.1

23)

0.59

1**

(0.1

19)

0.63

0**

(0.1

26)

0.62

8**

(0.1

22)

Inte

rcep

t 4.

826

(0.1

30)

4.82

7 (0

.129

) 4.

829

(0.1

29)

4.83

0 (0

.129

) 4.

847

(0.1

25)

4.83

2 (0

.129

) 4.

844

(0.1

28)

Var

ianc

e co

untry

leve

l 0.

465

(0.1

27)

0.46

2 (0

.126

) 0.

459

(0.1

25)

0.45

6 (0

.125

) 0.

425

(0.1

16)

0.45

7 (0

.125

) 0.

443

(0.1

21)

Var

ianc

e re

gion

al le

vel

0.05

2 (0

.009

) 0.

052

(0.0

09)

0.05

2 (0

.009

) 0.

051

(0.0

08)

0.05

2 (0

.009

) 0.

052

(0.0

09)

0.05

2 (0

.009

)

Var

ianc

e in

divi

dual

leve

l

3.13

6 (0

.011

) 3.

136

(0.0

11)

3.13

6 (0

.011

) 3.

136

(0.0

11)

3.13

6 (0

.011

) 3.

136

(0.0

11)

3.13

6 (0

.011

)

D

evia

nce

0.61

6 0.

557

1.84

4 2.

613

0.54

6 1.

248

Indi

vidu

als =

166

,458

; Reg

ions

= 1

23; C

ount

ries =

30

**

p <

0.0

1; *

p <

0.0

5

Sour

ces:

Eur

opea

n So

cial

Sur

vey

1-4,

Ale

sina

, Dev

lees

chau

wer

, Eas

terly

, Kur

lat a

nd W

aczi

arg

(200

3), a

nd E

uros

tat

Het

erog

enei

ty a

nd H

uman

Soc

ialit

y

81

Tab

le 8

. Mul

tilev

el a

naly

sis o

f ins

titut

iona

l tru

st

(1

) (2

a)

(2b)

(2

c)

(3a)

(3

b)

(3c)

R

egio

nal l

evel

Ethn

ic h

eter

ogen

eity

0.81

1*

(0.3

88)

Ling

uist

ic h

eter

ogen

eity

0.

933*

* (0

.224

)

Rel

igio

us h

eter

ogen

eity

0.31

4 (0

.241

)

Cou

ntry

leve

l

Ethn

ic fr

actio

naliz

atio

n

-1

.202

* (0

.592

)

Ling

uist

ic fr

actio

naliz

atio

n

-0

.531

(0

.570

)

Rel

igio

us fr

actio

naliz

atio

n

-0

.873

* (0

.462

)

C

ontr

ol v

aria

bles

ESS

Rou

nd 2

-0

.178

**

(0.0

14)

-0.1

78**

(0

.014

) -0

.178

**

(0.0

14)

-0.1

78**

(0

.014

) -0

.178

**

(0.0

14)

-0.1

78**

(0

.014

) -0

.178

**

(0.0

14)

ESS

Rou

nd 3

-0

.092

**

(0.0

14)

-0.0

92**

(0

.014

) -0

.092

**

(0.0

14)

-0.0

92**

(0

.014

) -0

.092

**

(0.0

14)

-0.0

92**

(0

.014

) -0

.092

**

(0.0

14)

ESS

Rou

nd 4

-0

.140

**

-0.1

40**

-0

.140

**

-0.1

40**

-0

.140

**

-0.1

40**

-0

.140

**

Cha

pter

Thr

ee

82

(0.0

14)

(0.0

14)

(0.0

14)

(0.0

14)

(0.0

14)

(0.0

14)

(0.0

14)

Gen

der (

1 =

fem

ale)

-0

.006

(0

.009

) -0

.006

(0

.009

) -0

.006

(0

.009

) -0

.006

(0

.009

) -0

.006

(0

.009

) -0

.006

(0

.009

) -0

.006

(0

.009

)

Age

0.

000

(0.0

00)

0.00

0 (0

.000

) 0.

000

(0.0

00)

0.00

0 (0

.000

) 0.

000

(0.0

00)

0.00

0 (0

.000

) 0.

000

(0.0

00)

Yea

rs o

f edu

catio

n 0.

042*

* (0

.001

) 0.

042*

* (0

.001

) 0.

042*

* (0

.001

) 0.

042*

* (0

.001

) 0.

042*

* (0

.001

) 0.

042*

* (0

.001

) 0.

042*

* (0

.001

)

GD

P pe

r cap

ita

0.52

0**

(0.1

03)

0.47

6**

(0.1

09)

0.49

6**

(0.1

08)

0.51

4**

(0.1

07)

0.50

0**

(0.0

98)

0.54

2**

(0.1

04)

0.50

0**

(0.0

97)

Inte

rcep

t 4.

998

(0.1

07)

4.84

0 (0

.131

) 4.

843

(0.1

30)

4.84

3 (0

.131

) 4.

861

(0.1

26)

4.84

6 (0

.130

) 4.

858

(0.1

29)

Var

ianc

e co

untry

leve

l 0.

302

(0.0

86)

0.33

4 (0

.093

) 0.

347

(0.0

96)

0.33

3 (0

.093

) 0.

267

(0.0

77)

0.29

5 (0

.084

) 0.

263

(0.0

76)

Var

ianc

e re

gion

al le

vel

0.06

2 (0

.010

) 0.

052

(0.0

09)

0.05

2 (0

.009

) 0.

052

(0.0

09)

0.05

2 (0

.009

) 0.

052

(0.0

09)

0.05

2 (0

.009

)

Var

ianc

e in

divi

dual

leve

l

3.11

3 (0

.012

) 3.

113

(0.0

12)

3.11

3 (0

.012

) 3.

113

(0.0

12)

3.11

3 (0

.012

) 3.

113

(0.0

12)

3.11

3 (0

.012

)

D

evia

nce

3.90

0*

15.8

00**

1.

500

3.80

0 0.

800

3.30

0 In

divi

dual

s = 1

43,4

70; R

egio

ns =

123

; Cou

ntrie

s = 3

0

** p

< 0

.01;

* p

< 0

.05

So

urce

s: E

urop

ean

Soci

al S

urve

y 1-

4, A

lesi

na, D

evle

esch

auw

er, E

aste

rly, K

urla

t and

Wac

ziar

g (2

003)

, and

Eur

osta

t

Het

erog

enei

ty a

nd H

uman

Soc

ialit

y

83

Tab

le 9

. Mul

tilev

el a

naly

sis o

f gov

ernm

ent r

edist

ribu

tion

(1)

(2a)

(2

b)

(2c)

(3

a)

(3b)

(3

c)

Reg

iona

l lev

el

Ethn

ic h

eter

ogen

eity

-0.8

25**

(0

.226

)

Ling

uist

ic h

eter

ogen

eity

-0

.219

(0

.142

)

Rel

igio

us h

eter

ogen

eity

-0.4

61**

(0

.128

)

Cou

ntry

leve

l

Ethn

ic fr

actio

naliz

atio

n

0.

002

(0.2

37)

Ling

uist

ic fr

actio

naliz

atio

n

-0

.163

(0

.215

)

Rel

igio

us fr

actio

naliz

atio

n

-0

.372

* (0

.163

)

C

ontr

ol v

aria

bles

ESS

Rou

nd 2

0.

044*

* (0

.007

) 0.

044*

* (0

.007

) 0.

044*

* (0

.007

) 0.

044*

* (0

.007

) 0.

044*

* (0

.007

) 0.

044*

* (0

.007

) 0.

044*

* (0

.007

) ES

S R

ound

3

0.08

2**

0.08

2**

0.08

2**

0.08

2**

0.08

2**

0.08

2**

0.08

2**

Cha

pter

Thr

ee

84

(0.0

07)

(0.0

07)

(0.0

07)

(0.0

07)

(0.0

07)

(0.0

07)

(0.0

07)

ESS

Rou

nd 4

0.

040*

* (0

.007

) 0.

041*

* (0

.007

) 0.

041*

* (0

.007

) 0.

041*

* (0

.007

) 0.

040*

* (0

.007

) 0.

041*

* (0

.007

) 0.

040*

* (0

.007

)

Gen

der (

1 =

fem

ale)

0.

130*

* (0

.005

) 0.

130*

* (0

.005

) 0.

130*

* (0

.005

) 0.

130*

* (0

.005

) 0.

130*

* (0

.005

) 0.

130*

* (0

.005

) 0.

130*

* (0

.005

)

Age

-0

.002

**

(0.0

00)

-0.0

02**

(0

.000

) -0

.002

**

(0.0

00)

-0.0

02**

(0

.000

) -0

.002

**

(0.0

00)

-0.0

02**

(0

.000

) -0

.002

**

(0.0

00)

Yea

rs o

f edu

catio

n -0

.028

**

(0.0

01)

-0.0

28**

(0

.001

) -0

.028

**

(0.0

01)

-0.0

28**

(0

.001

) -0

.028

**

(0.0

01)

-0.0

28**

(0

.001

) -0

.028

**

(0.0

01)

GD

P pe

r cap

ita

-0.1

49**

(0

.039

) -0

.103

**

(0.0

39)

-0.1

44**

(0

.040

) -0

.138

**

(0.0

37)

-0.1

49**

(0

.040

) -0

.143

**

(0.0

40)

-0.1

53**

(0

.036

)

Inte

rcep

t 3.

633

(0.0

40)

3.63

4 (0

.042

) 3.

627

(0.0

41)

3.62

8 (0

.038

) 3.

633

(0.0

41)

3.63

4 (0

.040

) 3.

626

(0.0

37)

Var

ianc

e co

untry

leve

l 0.

034

(0.0

12)

0.04

0 (0

.013

) 0.

037

(0.0

12)

0.02

9 (0

.010

) 0.

034

(0.0

12)

0.03

3 (0

.011

) 0.

027

(0.0

10)

Var

ianc

e re

gion

al le

vel

0.02

6 (0

.004

) 0.

022

(0.0

03)

0.02

5 (0

.004

) 0.

024

(0.0

04)

0.02

6 (0

.004

) 0.

026

(0.0

04)

0.02

6 (0

.004

)

Var

ianc

e in

divi

dual

leve

l

0.94

8 (0

.003

) 0.

948

(0.0

03)

0.94

8 (0

.003

) 0.

948

(0.0

03)

0.94

8 (0

.003

) 0.

948

(0.0

03)

0.94

8 (0

.003

)

D

evia

nce

11.9

93**

2.

297

12.1

87**

0.

000

0.56

8 4.

964*

In

divi

dual

s = 1

67,3

67; R

egio

ns =

123

; Cou

ntrie

s = 3

0

** p

< 0

.01;

* p

< 0

.05

Sour

ces:

Eur

opea

n So

cial

Sur

vey

1-4,

Ale

sina

, Dev

lees

chau

wer

, Eas

terly

, Kur

lat a

nd W

aczi

arg

(200

3), a

nd E

uros

tat


85

Table 9 shows the results for support for government redistribution. At the regional level, two of the three heterogeneity measures are significantly related to this kind of sociality, namely ethnic heterogeneity (b = -0.825; p < 0.01) and religious heterogeneity (b = -0.461; p < 0.01). In both these cases, the model fit improves significantly (Deviance = 11.993; p < 0.01 for ethnic heterogeneity and Deviance = 12.187; p < 0.01 for religious heterogeneity). Besides, support for government redistribution is significantly (at the 5 per cent significance level) negatively related to religious fractionalization (b = -0.372; p < 0.05; Deviance = 4.964; p < 0.05).

These results refute both the Putnam studies and the Alesina studies. In contrast with the expectation found across the Putnam studies, no relationship is found between heterogeneity and interpersonal trust in this international comparative study. Clearly, the outcome with regard to interpersonal trust may be attributed to the data that were used, as they can be criticized for being too crude to take the social context of individuals into account. Nevertheless, a different interpretation of this outcome is that cross-national differences in interpersonal trust should be explained by other factors than heterogeneity alone.

Although a large share of the Alesina studies do not address individual values and voting behaviour explicitly, it is a central part of the explanation for why national heterogeneity negatively affects the welfare state. Again, for this prediction, the analyses presented here offer very little support (with the relationship between religious heterogeneity and support for government redistribution being the only exception). This seems to suggest that decreasing lack of institutional trust and support for the welfare state are not the prime explanation for Alesina’s finding that national level heterogeneity undermines the welfare state. Nevertheless, at the regional level, heterogeneity is related to institutional trust and support for government redistribution. Somewhat surprisingly, institutional trust is positively related to ethnic and linguistic heterogeneity at that level. This finding is in contrast with the mechanisms most often discussed in the literature. Thus, rather than undermining trust in formal institutions, ethnic and linguistic heterogeneity may strengthen it. With regard to support for government redistribution, the analyses do support the expectation that heterogeneity – the ethnic and religious dimension – decreases this attitude of individuals. Combining these outcomes leads to the conclusion that the link between heterogeneity and human sociality is not as straightforward as is suggested by a large share of the literature. While regional heterogeneity is positively related to institutional trust, it is negatively related to support

Chapter Three

86

for government redistribution. This suggests that people are less willing to support the welfare state, but not because their trust in government institutions is in decline.

Conclusion

Both social structure and solidarity belong to the core topics of the social sciences. Societal changes such as migration and ageing of the population have spurred scientific and societal interest as they provide prime examples of how these two abstract concepts relate to each other empirically. This chapter argues that to date research efforts have focused on two distinct but related approaches, namely the more micro oriented research relating heterogeneity at the neighbourhood level (in particular the ethnical dimension) to interpersonal trust (which was referred to as the Putnam strand of research) and the macro oriented studies that relate national level heterogeneity to welfare state arrangements (termed the Alesina strand). The analyses discussed in this chapter aim at bringing these two strands closer together as their theoretical underpinnings and empirical findings are relevant for both of them. Nevertheless, given the observed gaps between the two kinds of studies it is impossible at the moment to integrate them completely. To name a few important differences: while the Putnam strand includes single country data, the Alesina studies aim at understanding cross national variation, and whereas the Putnam studies investigate interpersonal trust, the Alesina studies examine welfare state arrangements. As a result, prior studies do not allow relating the macro level to the micro level as they have not been studied in unison. This chapter is an effort in offering such an integrated investigation by using international comparative data. These data allow studying the effects of heterogeneity at multiple levels of analysis on a series of individual attitudes. A number of conclusions can be drawn from this effort.

First, as the empirical comparison with which this chapter starts shows, survey data may be used to generate useful information about the social structure of countries and regions. Such a research strategy is attractive in a number of ways. For one thing, as international surveys aim at generating international comparative data, researchers gathering the data put a lot of effort in defining and measuring the data unequivocally. Therefore, measurement errors due to differences in meaning of the variables should be reduced as much as possible. Existing datasets, for example the one presented in Alesina et al. (2003) do not have this advantage as they are compiled from different sources, which may increases


87

such errors. Furthermore, it offers a way to update the existing datasets and to examine changes in heterogeneity. While some of the data on which a number of international comparative studies are based go back to the 1960s, the data presented here are quite recent. As such, they offer a means to use more recent data. Finally, it opens the way to investigate longitudinal patterns of heterogeneity, whereas the existing studies are static. Certainly, there is a downside to the use of these survey data. For example, its potential relies on the quality of the data. If certain groups are underrepresented, this will lead to unreliable estimates of the level of heterogeneity. Furthermore, the number of countries included in the dataset investigated here is smaller than in the one used by Alesina et al. (2003). Nevertheless, there is no reason not to extend the present data with other datasets, such as the World Values Survey (WVS) and the International Social Survey Programme (ISSP), to get a wider range of countries.

With regard to the effects of heterogeneity, the outcomes are mixed. The expectation found in the Putnam studies, stating that heterogeneity decreases interpersonal trust, was neither confirmed at the national nor at the regional level. The implications of these findings need to be considered. Clearly, one may argue that the data were simply not accurate enough to find the proposed association. Even though that may be true, if we assume that there is some merit in the analyses provided here, it does shed a light on the theoretical expectations and the empirical outcomes of previous studies. To begin with, it underlines the importance of the level at which heterogeneity is assumed to affect interpersonal trust. If the Putnam studies are correct, the present analysis shows that heterogeneity affects interpersonal trust at low levels of aggregation and that the effect disappears as the analyses move to higher levels of aggregation. This may confirm the idea that the neighbourhood is indeed an important point of reference when it comes to trusting others in society. Another interpretation should also be considered. Whereas the Putnam study is based on data from the US while the current analyses are based on European data, the possibility that heterogeneity plays out differently in these two parts of the world cannot be excluded (see for example the studies by Hooghe, Reeskens, Stolle and Trappers (2009) and Gesthuizen, van der Meer and Scheepers (2009) that do investigate the Putnam thesis across European countries and arrive at different conclusions than the original study). Finally, a point worth mentioning is that interpersonal trust is positively related to GDP per capita. This finding shows that country characteristics do explain the level of trust that people have in others, but heterogeneity turns out not to be among them. In addition to that, the

Chapter Three

88

expectations based on the Alesina branch of heterogeneity studies receive some support. However, it cannot be concluded that the empirical support is conclusive. Most and for all, the original hypothesis in this part of the literature concentrates on national level heterogeneity, for which no support was found in the current analyses. When looking at the effects of regional level heterogeneity, opposite outcomes emerge: institutional trust is higher and support for redistribution is lower. Somehow, this seems to suggest that the public views the government as a reliable and legitimate actor as heterogeneity is higher, but they do not seem to demand a more extensive welfare state. This latter point means that we do not find evidence for a compensation hypothesis type of logic that was found in a study where the stock of migrants within an economic sector was taken into account (Burgoon, Koster and van Egmond 2012). The first point means that the public does expect something from the government, but for now it remains an open question what that is.

All in all, this chapter shows that heterogeneity may affect human sociality. Nevertheless, it also shows that the present research is far from conclusive and needs additional work in all kinds of directions. For starters, more theoretical progress is required to understand the effects of heterogeneity at different levels of analysis. To date, the focus has been too much on the direction of the relationship rather than its theoretical justification. In addition to that, the theoretical mechanisms need to be explicated and investigated. For a full understanding, controlled experiments will be extremely valuable. The same holds for qualitative accounts, which may range from participant observation to an in-depth content analysis to understand processes such as stereotyping in heterogeneous societies. In sum, for the moment, we have patterns and regularities, but the whole range of social scientific methods should be used to understand them.

References

Akerlof, G.A. (1997). Social distance and social decisions. Econometrica, 65, 1005-1027.

Alesina, A., Devleeschauwer, A., Easterly, W., Kurlat, S., and Wacziarg, R. (2003). Fractionalization. Journal of Economic Growth, 8, 155-194.

Alesina, A., and Glaeser, E.L. (2004). Fighting poverty in the US and Europe. A world of difference. Oxford: Oxford University Press.


89

Anderson, C.J., and Paskeviciute, A. (2006). How linguistic and ethnic heterogeneity influence the prospects for civil society: A comparative study of citizenship behavior. Journal of Politics, 68, 783-802.

Blau, P.M. (1977). Inequality and heterogeneity: A primitive theory of social structure. New York: Free Press.

Bogardus, E.S. (1925). Measuring social distances. Journal of Applied Sociology, 9, 299-308.

Burgoon, B., Koster, F., and van Egmond, M. (2012). Support for redistribution and the paradox of immigration. Journal of European Social Policy, 22, 288-304.

Brooks, C., and Manza, J. (2006). Why do welfare states persist? Journal of Politics, 64, 816-827.

Coleman, J.S. (1990). Foundations of social theory. Cambridge: The Belknap Press.

Deutsch, K.W. (1966). Nationalism and social communication: An inquiry into the foundations of nationality. Cambridge: MIT Press.

Easterly, W., and Levine, R. (1997). Africa's growth tragedy: Policies and ethnic divisions. Quarterly Journal of Economics, 112, 1203-1250.

European Social Survey (2000-2010). European Social Survey. Bergen: European Social Survey Data Archive.

Eurostat (2007). Regions in the European Union. Nomenclature of territorial units for statistics. Luxembourg: Office for Official Publications of the European Communities.

Fearon, J.D. (2003). Ethnic and cultural diversity by country. Journal of Economic Growth, 8, 195-222.

Gesthuizen, M., van der Meer, T., and Scheepers, P. (2009). Ethnic diversity and social capital in Europe: tests of Putnam's thesis in European countries. Scandinavian Political Studies, 32, 121-142.

Goldstein, H. (2003). Multilevel statistical models. London: Edward Arnold.

Gross, N. (2009). A pragmatist theory of social mechanisms. American Sociological Review, 74, 358-379.

Habyarimana, J., Humphreys, M., Posner, D.N., and Weinstein, J. (2007). Why does ethnic diversity undermine public goods provision? American Political Science Review, 101, 709-25.

Hardin, R. (1995). Self-interest, group identity. Nationalism and Rationality, 14-42.

Hedström, P., and Swedberg, R. (1998). Social mechanisms. An analytical approach to social theory . Cambridge: Cambridge University Press.

Hedström, P., and Ylikoski, P. (2010). Causal mechanisms in the social sciences. Annual Review of Sociology, 36, 49-67.

Chapter Three

90

Henrich, J., Boyd, R., Bowles, S., Camerer, E., Fehr, E., and Gintis, H. (2004). Foundations of human sociality. Oxford: Oxford University Press.

Hooghe, M., Reeskens, T., Stolle, D., and Trappers, A. (2009). Ethnic diversity and generalized trust in Europe A cross-national multilevel study. Comparative Political Studies, 42, 198-223.

Koster, F. (2013). Sociality in diverse societies. A regional analysis across European countries. Social Indicators Research, 111, 579-601.

Koster, F., Goudriaan, H., and van der Schans, C. (2009). Shame and punishment. An international comparative study on the effects of religious affiliation and religiosity on attitudes to offending. European Journal of Criminology, 6, 481-495.

Leigh, A. (2006). Trust, inequality and ethnic heterogeneity. Economic Record, 82, 268-280.

Mau, S., and Burkhardt., C.(2009). Migration and welfare state solidarity in Western Europe. Journal of European Social Policy, 19, 213-229

Miguel, E., and Gugerty, M. K. (2005). Ethnic diversity, social sanctions, and public goods in Kenya. Journal of Public Economics, 89, 2325-2368.

Park, R.E. (1924). The concept of social distance. Journal of Applied Sociology, 8, 339-344.

Putnam, R.D. (2007). E Pluribus Unum: diversity and community in the twenty-first century. Scandinavian Political Studies, 30, 137-174.

Smith, M.G. (1960). Social and cultural pluralism. Annals of the New York Academy of Sciences, 83, 763-785.

Snijders, T.A.B. (2003). Multilevel analysis. In M.S. Lewis-Beck, A.E. Bryman, and T.F. Liao (eds.), The SAGE encyclopedia of social science research methods (673-77). London: Sage.

Snijders, T.A.B., and Bosker, R. (1999). Multilevel analysis: An introduction to basic and advanced multilevel modeling. London: Sage.

Stolle, D. (1998). Bowling together, bowling alone: The development of generalized trust in voluntary associations. Political psychology, 497-525.

Tolsma, J., van der Meer, T., and Gesthuizen, M. (2009). The impact of neighborhood and municipality characteristics on social cohesion in the Netherlands. Acta Politica, 44, 286- 313.

Yamagishi, T., and Mifune, N. (2008). Does shared group membership promote altruism? Fear, greed, and reputation. Rationality and Society, 20, 5-30.


91

Appendix 1. Heterogeneity per region, NUTS level 1

Country Nuts 1 region N Origin Language Religion Austria Ostösterreich 2926 0.17 0.09 0.26 Südösterreich 1513 0.12 0.06 0.20 Westösterreich 2479 0.13 0.04 0.16 Belgium Vlaams Gewest 4451 0.11 0.10 0.17

Région de Bruxelles-Capitale/Brussels Hoofdstedelijk Gewest

466 0.41 0.39 0.56

Région Wallonne 2318 0.20 0.07 0.19

Bulgaria Severna i iztochna Bulgaria 1927 0.03 0.29 0.36

Yugozapadna i yuzhna centralna Bulgaria 1703 0.02 0.21 0.36

Croatia Sjeverozapadna Hrvatska 413 0.14 0.00 0.11

Sredisnja i Istocna (Panonska) Hrvatska 420 0.21 0.01 0.18

Jadranska Hrvatska 560 0.13 0.01 0.12 Cyprus Cyprus 2210 0.13 0.04 0.03 Czech Republic Czech Republic 6400 0.06 0.02 0.25

Denmark Denmark 6108 0.11 0.04 0.13 Estonia Estonia 5167 0.32 0.43 0.54 Finland Manner-Suomi 6259 0.05 0.11 0.07 Åland 1854 0.06 0.15 0.08 France Île de France 1109 0.31 0.20 0.46 Bassin Parisien Est 583 0.10 0.08 0.09 Bassin Parisien Ouest 714 0.10 0.12 0.12 Nord - Pas-de-Calais 546 0.09 0.07 0.20 Est 734 0.13 0.20 0.27 Ouest 1014 0.07 0.04 0.13 Sud-Ouest 925 0.11 0.05 0.18 Sud-Est 952 0.16 0.07 0.25 Méditerranée 791 0.26 0.08 0.23

Chapter Three

92

Germany Schleswig-Holstein 295 0.13 0.07 0.29 Hamburg 207 0.22 0.13 0.45 Niedersachsen 854 0.15 0.06 0.43 Bremen 82 0.13 0.07 0.28 Nordrhein-Westfalen 2028 0.20 0.11 0.58 Hessen 614 0.25 0.13 0.54 Rheinland-Pfalz 440 0.15 0.07 0.53 Baden-Württemberg 1112 0.22 0.11 0.59 Bayern 1356 0.19 0.10 0.46 Saarland 107 0.18 0.07 0.43 Berlin 698 0.15 0.10 0.58 Brandenburg 647 0.05 0.05 0.35 Mecklenburg-Vorpommern 535 0.06 0.03 0.38 Sachsen 1123 0.06 0.03 0.31 Sachsen-Anhalt 693 0.04 0.01 0.30 Thüringen 665 0.05 0.01 0.44 Greece Voreia Ellada 2400 0.13 0.08 0.07 Kentriki Ellada 1496 0.11 0.04 0.05 Attiki 2457 0.20 0.09 0.07 Nisia Aigaiou, Kriti 691 0.15 0.06 0.07 Hungary Közép-Magyarország 1511 0.07 0.00 0.44 Dunántúl 2122 0.04 0.02 0.31 Alföld és Észak 2612 0.03 0.00 0.50 Iceland Iceland 579 0.06 0.01 0.16 Ireland Ireland 6132 0.16 0.05 0.10 Italy Nord Ovest 591 0.04 0.16 0.05 Nord Est 468 0.08 0.28 0.03 Centro (IT) 521 0.06 0.12 0.03 Sud (IT) 799 0.03 0.29 0.02 Isole (IT) 357 0.02 0.34 0.05 Latvia Latvia 1980 0.24 0.42 0.70 Luxembourg Luxembourg (Grand-Duché) 3187 0.43 0.50 0.43 Netherlands Noord-Nederland 878 0.08 0.28 0.49 Oost-Nederland 1718 0.13 0.04 0.61 West-Nederland 3635 0.20 0.07 0.67 Zuid-Nederland 1681 0.13 0.04 0.26 Norway Oslo og Akershus 1530 0.21 0.14 0.30 Hedmark og Oppland 572 0.08 0.04 0.09 Sør-Østlandet 1291 0.15 0.07 0.32 Agder og Rogaland 1000 0.14 0.08 0.20 Vestlandet 1292 0.10 0.08 0.13 Trøndelag 672 0.11 0.11 0.14 Nord-Norge 738 0.08 0.07 0.11


93

Poland Centralny 1580 0.01 0.00 0.02 Poludniowy 1428 0.02 0.01 0.02 Wschodni 1393 0.02 0.02 0.06 Pólnocno-Zachodni 874 0.04 0.00 0.03 Poludniowo-Zachodni 652 0.07 0.00 0.03 Pólnocny 1239 0.04 0.00 0.03 Portugal Continente (PT) 8152 0.11 0.02 0.07 Romania Macroregiunea unu 502 0.00 0.39 0.51 Macroregiunea doi 647 0.02 0.04 0.09 Macroregiunea trei 561 0.01 0.00 0.04 Macroregiunea patru 436 0.01 0.09 0.20 Slovakia Slovakia 5061 0.06 0.20 0.32 Slovenia Slovenia 5723 0.15 0.04 0.13 Spain Noroeste 1143 0.09 0.38 0.04 Noreste 735 0.11 0.13 0.10 Comunidad de Madrid 971 0.22 0.06 0.11 Centro (ES) 998 0.07 0.03 0.05 Este 2140 0.17 0.44 0.17 Sur 1574 0.11 0.04 0.08 Canarias (ES) 283 0.16 0.11 0.11 Sweden Östra Sverige 2728 0.24 0.13 0.36 Södra Sverige 3412 0.19 0.10 0.26 Norra Sverige 1564 0.11 0.05 0.16 Switzerland Région lémanique 1465 0.39 0.39 0.54 Espace Mittelland 1642 0.24 0.49 0.50 Nordwestschweiz 1317 0.30 0.27 0.58 Zürich 1187 0.30 0.26 0.51 Ostschweiz 1437 0.31 0.24 0.57 Zentralschweiz 544 0.28 0.21 0.37 Ticino 212 0.40 0.35 0.16 Turkey Istanbul 533 0.01 0.08 0.05 Bati Marmara 88 0.00 0.02 0.02 Ege 277 0.05 0.01 0.00 Dogu Marmara 280 0.08 0.04 0.10 Bati Anadolu 214 0.01 0.04 0.01 Akdeniz 314 0.01 0.06 0.00 Orta Anadolu 67 0.00 0.06 0.00 Bati Karadeniz 80 0.02 0.00 0.02 Dogu Karadeniz 31 0.00 0.00 0.00 Kuzeydogu Anadolu 68 0.00 0.55 0.00 Ortadogu Anadolu 129 0.00 0.50 0.00 Güneydogu Anadolu 335 0.01 0.62 0.00

Chapter Three

94

United Kingdom North East (ENGLAND) 411 0.06 0.02 0.54 North West (ENGLAND) 1035 0.13 0.06 0.59 Yorkshire and The Humber 749 0.14 0.06 0.57 East Midlands (ENGLAND) 666 0.10 0.02 0.48 West Midlands (ENGLAND) 727 0.18 0.09 0.56 South West (ENGLAND) 783 0.11 0.02 0.43 Eastern 843 0.21 0.05 0.51 London 760 0.45 0.23 0.74 South East 1180 0.18 0.04 0.51 Wales 510 0.06 0.08 0.51 Scotland 752 0.11 0.04 0.52 Northern Ireland 279 0.09 0.01 0.49 Total 171,619 0.12 0.12 0.26 Sources: European Social Survey 1-4

CHAPTER FOUR

DIVERSITY, SIMILARITY, AND SOLIDARITY: RESULTS OF A VIGNETTE STUDY

MAARTEN BERG AND FERRY KOSTER

Introduction

Solidarity and sharing behaviour are necessary elements of viable, well-functioning societies. Solidarity can take different forms, and a relevant distinction is between ‘public’ (formal) solidarity and ‘private’ (informal) solidarity. The prototypical example of public solidarity is the financial support of certain societal groups through the mechanism of the welfare state. Private solidarity refers to voluntary helping behaviour directed at concrete individuals, e.g. taking out the garbage for an elderly neighbour (e.g. de Beer and Koster, 2009).

Solidarity, whether measured in terms of time, money, consumer goods, practical assistance or emotional support, has an economic aspect in the sense that it is always scarce. As all resources are limited, solidarity can never be unrestricted. This raises questions about the degree of our potential for solidary behaviour, but also about the allocation of our solidarity. In this chapter we address the latter question. We study what types of potential beneficiaries have the strongest appeal on our sense of solidarity. When forced to choose, towards whom do people feel most solidary?

To study this question, we use a vignette approach (Rossi and Nock, 1982). Respondents are asked to choose between two fictive beneficiaries who are both described on several dimensions (‘attributes’). Although reading a vignette is different from experiencing a stimulus or action in real life, the vignette approach is thought to reveal the values and perceptions of respondents. Moreover, this is done in a way that is less vulnerable to socially desirable responses than the standard survey

Chapter Four

96

approach. We will elaborate on the vignette approach, its pros and cons, and its implementation below.

The theoretical starting point of this paper is the assumption that ‘diversity’ in a society or a community affects the potential for and the direction of solidarity. Although it is not always clear and it cannot be fully determined what the relevant aspects of diversity are, common ‘distinctions’ between people that are mentioned both in the literature and in policy debates relate to the old and the young and different ethnic groups (de Beer and Berg, 2012). These attributes coincide with some of the ones used in the current chapter.

In this chapter we conceptualize diversity not as a characteristic of a group, but as a dyadic relationship between two individuals. Although solidary behaviour necessarily takes place within a broader social context, in this chapter we are interested in the attributes of potential beneficiaries that trigger solidarity and how these attributes relate to attributes of the contributor.

Hypotheses

When we think of solidarity as primarily ‘other-regarding’, we should place great importance on the attributes of the beneficiary. His/her neediness is the main reason that solidarity exists in the first place. We hypothesize that needier beneficiaries will generally appeal more strongly to our sense of solidarity (van Oorschot, 2006). This might have to do with internalized social norms to help the needy or with feelings of empathy. In study 1 of this chapter (on public solidarity), we compare different types of entitlements (retirement, unemployment, disability and welfare benefits). We assume that disabled persons are generally perceived to be the most needy and, therefore, will evoke most solidarity. We assume that people on pension benefits are perceived to be the least needy and will evoke the least solidarity1. Study 1 also compares people with varying numbers of under-aged children. We assume that people with many under-aged children are perceived to be needier than people with few under-aged children and will evoke more solidarity. In the context of study 2 (on private solidarity), we assume that elderly people are perceived as relatively needy (e.g. because of physical limitations) and will receive the most solidarity. 1 Note that in the Netherlands the term ‘pension’ refers to a supplementary occupational (second pillar) pension and not to state pension (first pillar).

Diversity, Similarity, and Solidarity: Results of a Vignette Study

97

Apart from classifying beneficiaries in terms of their neediness, we must also look at their ‘deservingness’ (e.g. van Oorschot, 2000). For example, it seems likely that people are generally more willing to help others who are not to blame for their needy situation than those who brought it onto themselves. Deservingness also depends on past contributions by the beneficiary. In study 1, we compare people with varying numbers of working years. We assume that people with a lot of working experience are perceived to be more deserving than people with lesser working experience and predict that the number of years of working experience contributes to the willingness of others to show solidarity.

Helping behaviour is arguably not only other-regarding, but also (partly) self-regarding in some (or even most) situations (see chapter 1). Helping might be functional in the sense that it encourages reciprocation, thereby strengthening the social fabric of communities. We expect people to be more willing to help people who are prepared to make some future contribution (if possible) in return (Fong, Bowles and Gintis, 2003). In study 1, we compare people who are willing to do volunteer work with people who are not. We predict most solidarity with people who are very willing to volunteer. In study 2, we compare people who are offering to reciprocate favours in the future with people who do not explicitly offer help in the future.

In this chapter we are especially interested in the role of ethnic diversity. There are several theoretical reasons to expect people from different ethnic backgrounds to be less inclined to help one another than people from the same ethnic background (see chapter 2). For example, they might have greater difficulty in identifying with one another.

Traditionally, left-wing political parties (and their supporters/voters) in the Netherlands have more favourable attitudes towards multiculturalism and (the role of) migrants than right-wing parties (Koopmans, 2013). Therefore, we expect right-wing voters to react more strongly to the ethnicity of the potential beneficiaries than left-wing voters. That is, they will be more unfavourable when the beneficiary has a different ethnic background.

The fact that some groups of voters are expected to respond differently to different ethnic groups illustrates a more general notion. Solidarity needs not be just a unilateral relationship directed at and affected by (characteristics of) the beneficiary. It might be truly relational in the sense that both the attributes of the contributor and the beneficiary matter, as

Chapter Four

98

well as the ‘distance’ between them. We hypothesize that the differences between contributor and beneficiary have an independent effect up and above the characteristics of the beneficiary per se.

In this chapter we look at both public solidarity (provided through the welfare state) and private solidarity (shown voluntarily between individuals). This distinction might coincide with different psychological mechanisms (Koster, 2009). Social norms and social pressure might be of greater importance in the context of the welfare state. On the other hand, personal contact and visibility might be more important for private solidarity than for public solidarity and may trigger empathy and direct reciprocity. Therefore, we hypothesize that some different results will emerge for the public and the private solidarity studies.

The following hypotheses are tested.

Hypothesis 1: All other things being equal, people express more solidarity towards people who are in need. More specifically, we predict most solidarity towards disabled persons and the least solidarity towards people on pension benefits. Also, we expect more solidarity towards people with relatively many under-aged children. In the context of private solidarity, we expect more solidarity towards the elderly.

Hypothesis 2: All other things being equal, people express more solidarity towards people who are deserving. More specifically, we predict a positive relationship between the number of working years of a beneficiary and the solidarity that s/he encounters.

Hypothesis 3: All other things being equal, people express more solidarity towards people who are willing to reciprocate. More specifically, we predict that people who are willing to do volunteer work will encounter more solidarity. In the context of private solidarity, we predict the same for people who offer to reciprocate a favour.

Hypothesis 4: All other things being equal, people express more solidarity towards people with the same ethnic background than towards people with a different ethnic background.

Hypothesis 5: Persons with a ‘right-wing’ political orientation will be more sensitive to ethnic differences and discriminate more strongly towards people with a different ethnic background than left-wing persons.


99

Besides these hypotheses, we explore (post hoc) the relative importance of the several attributes and discuss differences between public and private solidarity.

Methods

The vignette approach

A vignette is a short, carefully constructed description of a person, object or situation, representing a systematic combination of characteristics. McFadden et al (2005) describe vignettes as “hypothetical situations that can be used to elicit preferences, judgments or anticipated behaviour”.

In this chapter a vignette refers to a person. An example of a vignette from our study on public solidarity (as shown to our respondents) is:

• Male • 47 years old • Moroccan background • 22 years working experience • On welfare • 4 under-aged children • Living in your neighbourhood • Willing to do unpaid work (volunteering)

There are many different variations of the vignette approach (for a review: Atzmüller and Steiner, 2010). Sometimes respondents have to evaluate only one vignette, whereas in other cases multiple vignettes are used. In the current chapter, respondents deal with several pairs of vignettes. From each pair, they have to choose one.

When studying solidarity from a standard survey approach, we could ask respondents to what extent they are willing to support certain groups (e.g. elderly people, people from different ethnic backgrounds, etc.) and to formulate their answer on a verbal or numerical scale. This approach has several limitations. First, the question may seem artificial, because of a lack of contextual information. People might not know if they want to express solidarity towards ‘an elderly person’, because they do not know who this person is. Is the person a male or a female? Is s/he a nice person? Et cetera. The vignette approach is more realistic in this respect, as it presents hypothetical persons in a multidimensional manner.

Chapter Four

100

Secondly, the multidimensional approach is ideal for studying interaction effects (Wason et al., 2002). In this study, we include such interactions by examining how solidarity depends on the combination of characteristics of the hypothetical person and those of the respondents.

Thirdly, the standard survey method can be sensitive to socially desirable (or strategic) statements (e.g. de Wolf and van der Velden, 2001). For example, persons with negative stereotypes about migrants might actually express positive attitudes when they want to appear (to the experimenter or even to themselves) open-minded and tolerant. This kind of ‘impression management’ is much harder with the vignette approach, as the respondents have to deal with complex stimuli and several dimensions at the same time. They are forced to answer mainly intuitively (rather than cognitively and deliberately). This advantage of the vignette method is especially valuable when sensitive topics are addressed (such as attitudes towards immigrants). Although some impression management might still be possible, one would have to restrict it to one or two dimensions. We expected that the risk of impression management would be greatest with respect to the ethnic dimension. For that reason, we decided to decrease the saliency of this dimension (as explained below).

A last advantage of the vignette approach compared to survey studies is that the vignette method is ideally suited for assessing the relative weight of different aspects that are relevant for the decision or evaluation at hand.

A big advantage of the vignette approach over laboratory experiments is that the former method is much less time (and money) consuming. This makes it possible to (drastically) increase the number of subjects and to study several dimensions simultaneously. In other words: when we want to study multiple dimensions simultaneously, the experimental method is often simply infeasible.

Participants

The data were obtained from the LISS-panel (Longitudinal Internet Studies for the Social Sciences), an Internet panel survey that aims to be representative for the Dutch population2. The LISS panel consists of 5,000 households, comprising 8,000 individuals (although in our specific study 5,605 people participated). Some distributional characteristics are

2 See Knoef and De Vos (2009) for an elaborate discussion of the representativeness of the panel.


101

presented in Table 1. The panel is based on a true probability sample of households drawn from the population register by Statistics Netherlands. Every month, panel members complete online questionnaires of about 15 to 30 minutes in total. They are paid for each completed questionnaire. One member in the household provides the household data and updates this information at regular time intervals

Table 1: Distribution of some characteristics of 5,605 respondents and of the total Dutch population3 Variable Value N % in the

sample % in the

population Sex Male 2,594 46.3 49.5 Female 3,011 53.7 50.5 Age 18-29 years 889 15.9 18.7 30-39 years 747 13.3 15.3 40-49 years 965 17.2 19.3 50-59 years 1,102 19.7 17.7 60-69 years 1,183 21.1 15.0 70 and older 719 12.8 14.1 Ethnic background

Dutch 5,081 91.1 79.1

Other 4994 8.9 20.9 Missing 25 -- -- Researchers from different universities can incorporate their studies or questions in the LISS-panel. Moreover, they can use the data on the background characteristics of the respondents. In our case, this enabled us to study not only how vignette conditions relate to respondents’ solidarity, but also how these effects interact with the characteristics of the contributor (respondent).

Study 1: public solidarity

All participants were given ten pairs of vignettes, the first five pairs relating to public solidarity and the last five pairs to private solidarity.

3 Data are from Statistics Netherlands (CBS), 2012. 4 Of the 499 non-Dutch respondents, 60 had a Surinamese background and 33 a Moroccan background.

Chapter Four

102

Regarding public solidarity, the respondents were asked to imagine the Dutch government having to balance the budget and cut spending on social welfare. Given that context, the question was: What person should be spared most should the government decide to cut its expenditures?5

The ‘public’ vignettes consisted of eight variables:

- Sex (two categories: ‘male’ and ‘female’). - Age (all values between 18 and 90). - Ethnicity (five categories: ‘autochthone Dutch’, ‘born in Morocco’,

‘Moroccan background’, ‘born in Surinam’, ‘Surinamese background’6).

- Number of years of working experience (all values between 0 and ‘age minus 18’, with a maximum of 50).

- Type of entitlement (four categories: ‘welfare’, ‘unemployment benefit’, ‘disability benefit’, ‘pension’). People above 65 were automatically placed in the ‘pension’ category, people under 65 in one of the other three categories.7

- Number of under-aged children (all values between 0 and 6). - Distance (two categories: ‘lives in your neighbourhood’, ‘lives in

another part of the Netherlands’). - Willingness to do volunteer work (three categories: ‘is eager to do

volunteer work’, ‘might be interested to do volunteer work’, ‘prefers not to do volunteer work’).

The theoretical number of vignettes (combining all values of all factors) is enormous (even compared to the large number of respondents). Therefore, we constructed 500 vignettes independently from each other, using a random generator. In principle, the different values of the independent variable had an equal chance to be included in a specific vignette.8 There were, however, some exceptions:

5 It was a forced choice question. Maybe respondents felt like both persons (or neither person) should be spared. Still, they had to make a choice. 6 For the analyses, we recoded the five categories into three categories: Dutch, Moroccan, Surinamese. 7 This means that the different independent variables were not perfectly independent from one another. This creates some multicollinearity in the analyses, but we considered this problem to be smaller than using highly unrealistic vignettes. 8 This did not mean that the numbers had to match perfectly. We did not strive for exactly 250 male and 250 female vignettes, for example.


103

- For ethnicity, we decided to use a 70% chance for the vignette to be ‘autochthone Dutch’ and to use equal chances of 7.5% for the four other categories of this variable. We wanted respondents to face sufficient pairs of two autochthone Dutch, in order to decrease the saliency of the ethnic dimension in this study. With only 1/3 (or even 1/5) of the vignettes being Dutch, it would have been very clear to the respondents that we were most interested in the role of ethnic diversity. Still, compared to the actual ethnic distribution within the Netherlands, we actually included a lot of people from Morocco and Surinam in our vignette population.

- For the number of under-aged children, we used different distributions for different age groups (as shown in Table 2).

Table 2: Distribution of numbers of under-aged children by age group (%) Number of under-aged children Age group

0 1 2 3 4 5 6

18-19 95 5 0 0 0 0 0 20-29 50 20 20 5 5 0 0 30-39 40 15 15 10 10 5 5 40-49 30 15 15 10 10 10 10 50-59 40 15 15 10 10 5 5 60-69 80 10 5 5 0 0 0 70-79 90 5 5 0 0 0 0 80-90 100 0 0 0 0 0 0 Although we took measures (using certain distributions), to guarantee realistic vignettes (we wanted to avoid 87 year old women with 6 under-aged children), we loosely checked the vignettes after they were generated. No vignettes had to be excluded.9

For each respondent, ten different vignettes (five pairs) were randomly drawn from the population of 500. The eight attributes were placed vertically and two vignettes (a pair) were placed next to one another. For every respondent, the order of the eight attributes was identical for each

9 When asked to comment on the study, very few (still some) respondents mentioned unrealistic vignettes. However, we have to realize that atypical people exist in the real world as well.

Chapter Four

104

pair (in order to avoid confusion). The order varied between subjects. It was possible that a respondent had to choose between two pairs that were identical on one or more dimensions, although it was impossible that they had to choose between the exact same vignettes.

Study 2: private solidarity

The study on private solidarity dealt with helping behaviour in the neighbourhood. Again, there were five pairs of vignettes. The recurring question was what neighbour would be most likely to get the help from the respondent. In contrast to the study on public solidarity, in the private solidarity study five different helping situations were used for the five pairs of vignettes.

Question 1: Your neighbour rings at your bell at night and asks if you are willing to look after his/her (grand)child for about an hour. Which neighbour are you most inclined to help?

Question 2: Your neighbour rings at your bell at night and asks if you can help buying groceries for a party (as there are too many groceries for one person to carry). Which neighbour are you most inclined to help?

Question 3: Your neighbour rings at your bell at night and asks if you can help with some paper work that gives him/her some trouble. Which neighbour are you most inclined to help?

Question 4: Your neighbour rings at your bell at night and asks if you can help take out the household refuse. Which neighbour are you most inclined to help?

Question 5: Your neighbour rings at your bell at night and asks if you can help him/her to construct a small cabinet (since it is much easier for two persons to do this). Which neighbour are you most inclined to help?

The vignettes consisted of five variables (the first three were identical to the ones used in the study on public solidarity):

- Sex (two categories: ‘male’ and ‘female’). - Age (all values between 18 and 90).


105

- Ethnicity (five categories: ‘autochthone Dutch’, ‘born in Morocco’, ‘Moroccan background’, ‘born in Surinam’, ‘Surinamese background’).10

- Encounters (two categories: ‘run into him/her in the neighbourhood frequently’, ‘seldom run into him/her in the neighbourhood’).

- Reciprocity (two categories: ‘offers to return the favour in the future’, ‘says nothing about possibly returning the favour in the future’).

Although the situations differ, and hence it is possible that the effects of the vignette conditions differ across these situations, separate analyses and analyses in which the situations are combined lead to similar results (not shown here). Therefore, the results of the combined analyses are shown here.

Procedure

Data were gathered in the beginning of 2012. Participants filled out the LISS panel questionnaire, of which our study was only a part. When coming to our part of the questionnaire, all participants (after reading the instructions) first dealt with the vignette pairs on public solidarity and then turned to the vignette pairs on private solidarity. For all participants, the same order of questions (helping situations) was used for the private solidarity part.

After completing the main part of their task, respondents were asked whether they had based their choices mainly on intuition or on conscious deliberation (evaluating and weighing the different attributes in a systematic way). Subsequently, they answered the question about ethnic identity (already mentioned above).

Next, there were several questions about their attitudes regarding the Dutch welfare state (scoring from less to more favourable) and regarding solidarity towards the young, the old, and persons with different cultural backgrounds. Also some questions about media usage were included. Respondents had to indicate how often they watched the NOS journal and RTL nieuws (TV news), or read De Volkskrant and De Telegraaf (newspapers) or nu.nl (a news website).

10 The same distribution was used as in the study on public solidarity.

Chapter Four

106

Finally, they answered some questions about the study itself. Was it comprehensible, interesting, and so on? They were also able to comment on the study.

Method of analysis

In both studies, the dependent variable is dichotomous since respondents choose either vignette 1 or vignette 2. In our analyses, each choice between two vignettes was taken as a separate observation. Thus, an ‘act of solidarity’ is a dyadic relation between the respondent and a (fictive) beneficiary. Furthermore, we have observations nested within individuals, as respondents answered 5 vignettes for both kinds of solidarity. Therefore, we applied logistic multilevel regression analysis. The total number of observations is 56,050 (distributed among 5,605 persons and two studies). 16 missing observations from the public solidarity study and 91 observations from the private solidarity study could not be analysed.

Since respondents compared several pairs of vignettes, their choice may have been affected by fatigue. To make sure that this does not affect the outcomes, we tested, using interaction effects for each vignette condition, whether there were such fatigue effects (not shown here). This analysis shows that the results are unaffected and remain the same.

Our independent variables were either continuous (e.g. age) or nominal (e.g. ethnic background). The continuous variables (using the difference between vignette 1 and vignette 2) were easy to include in our multivariate regression models and to interpret in combination with our dependent variable.

The dataset includes information about the fictive person displayed on the vignette and information about the respondents. By combining these two sources of information, it is possible to calculate the similarity between the person on the vignette and the respondent. We approached this as follows:

− With regard to the effect of ethnic identity, we are interested in investigating whether similarity in ethnic background matters. To assess this, it is necessary to combine characteristics of the vignettes with the ethnic background of the respondents. For the ethnic identity dimension, we created a dummy variable that scored 0 when the persons in the vignettes both had a different ethnic background than the respondent, and scored 1 when the person on the vignette had the same ethnicity as the respondent.


107

− Age similarity was approached in the same way as ethnic identity. Nevertheless, since age was measured as a continuous variable instead of as categorical variable, we did not calculate a dummy variable. To assess age similarity (or distance in terms of years), the absolute difference between the age of the respondent and the age of the person on the vignette was calculated. To account for the fact that the effect of age may be curvilinear, also the squared root of absolute age was included in the analyses.

Finally, to investigate whether the individual characteristics matter for solidarity preferences, interaction effects were added to the analyses. For example, to see whether these preferences are affected by the political orientation of respondents, interactions were included between the answers on this variable and each vignette condition, and to investigate whether there is a gender bias in solidarity, we added an interaction term between the gender of the respondent and the gender of the person on the vignette.

Respondents had to indicate on a five-point scale whether or not they thought it was difficult to choose between the vignettes. Recoding the answers to three categories gave the following distribution: easy (60.6%), neither easy nor difficult (18.0%), and difficult (21.4%). Moreover, 76.7% of the respondents thought the questions were clear, 16.3% were ambivalent and 7.0% thought that they were not clear. Considering these findings, it is fair to say that the results below do in fact reflect the (conscious or unconscious) attitudes and preferences of the panel.

Study 1: Public solidarity

The descriptive statistics of the study on public solidarity are shown in Table 3 and based on 28,009 observations/choices.

Respondents chose more female (fictive) persons than male persons (compared to their prevalence in the vignette population): 52.1% versus 47.9%. The difference is significant: χ2(1)=98.630, p<0.001. Hence, respondents seem to show more solidarity towards women than towards men.

Chapter Four

108

Table 3: Descriptive statistics (public solidarity) Variable Value Number of

times included in a vignette

Number of times

included in the chosen vignettes

Chance of

being chosen

Sex Male 28,613 (51.1%)

13,719 (49.0%)

47.9%

Female 27,405 (48.9%)

14,290 (51.0%)

52.1%

Age 18-30 years 11,159 (19.9%)

5,025 (17.9%)

45.0%

31-45 years 12,113 (21.6%)

7,222 (25.8%)

59.6%

46-64 years 13,811 (24.7%)

8,236 (29.4%)

59.6%

65-90 years 18,935 (33.8%)

7,526 (26.9%)

39.7%

Ethnic background

Dutch 41,813 (74.6%)

21,384 (76.3%)

51.1%

Moroccan 6,182 (11.0%)

2,726 (9.7%)

44.1%

Surinamese 8,023 (14.3%)

3,899 (13.9%)

48.6%

Working years

0-10 years 25,916 (46.3%)

12,135 (43.3%)

46.8%

11-20 years 12,372 (22.1%)

6,622 (23.6%)

53.5%

21-30 years 7,003 (12.5%)

3,913 (14%)

55.9%

31-40 years 7,155 (12.8%)

3,531 (12.6%)

49.4%

41-50 years 3,572 (6.4%)

1,808 (6.5%)

50.6%


109

Type of entitlement

Welfare 11,771 (21.0%)

6,092 (21.8%)

51.8%

Unemployment 12,260 (21.9%)

6,295 (22.5%)

51.3%

Disablement 13,052 (23.3%)

8,096 (28.9%)

62.0%

Pension 18,935 (33.8%)

7,526 (26.9%)

39.7%

Number of under-aged children

0 35,706 (63.7%)

15,017 (53.6%)

42.1%

1 6,975 (12.4%)

4,038 (14.4%)

57.9%

2 5,739 (10.2%)

3,644 (13.0%)

63.5%

3 2,157 (3.9%)

1,459 (5.2%)

67.6%

4 2,666 (4.8%)

1,926 (6.9%)

72.2%

5 983 (1.8%)

692 (2.4%)

70.4%

6 1,792 (3.2%)

1,233 (4.4%)

68.8%

Distance Same neighbourhood

27,342 (48.8%)

13.675 (48.8%)

50.0%

Other part of the country

28,676 (51.2%)

14.334 (51.2%)

50.0%

Volunteer work

Very willing 18,930 (33.8%)

10,666 (38.1%)

56.3%

Somewhat willing

18,606 (33.2%)

9,654 (34.5%)

51.9%

Rather unwilling

18,482 (33.0%)

7,689 (27.5%)

41.6%

In Table 3, four age groups are compared.11 Up to the age of 64 (excluding the last category), there is a positive relationship: χ2(2)=672.266, p<0.001. Older (fictive) persons receive more solidarity, although the difference does only exist between the first (18-30) and the second (31-45) category and not between the second and the third (46-64) category. A perhaps

11 For the regression analyses below, however, we used the original age scores.

Chapter Four

110

counterintuitive finding is that the oldest category (aged 65-90) receives the least solidarity of all. We must note, however, that this age group is confounded with ‘type of entitlement’. All persons above 65 fall automatically in the ‘pension’ category. Therefore, it seems likely that the low score for persons above 65 is an ‘entitlement effect’ rather than an ‘age effect’.

There is also a difference between the three ethnic groups. Moroccan persons had a 44.1% chance of being chosen, compared to 48.6% of the Surinamese persons and 51.1% of the Dutch persons. Again, this difference was highly significant: χ2(2)=114.324, p<0.001. The same holds for the direct comparisons between the Dutch and persons with a Surinamese background and between those with a Surinamese and Moroccan ethnicity.

We categorized the years of working experience into five categories.12 We found a curvilinear relationship with solidarity increasing from category 1 (0-10 years) to category 2 (11-20 years) and category 3 (21-30 years), and then decreasing again for category 4 (31-40 years) and category 5 (41-50 years). Despite the fact that categories 4 and 5 score lower than category 3 (which is not easy to explain), the data fit a linear relationship: χ2(4)=264.473, p<0.001. This confirms hypothesis 2 (based on ‘deservingness’ considerations).

Vignettes with retired persons had the lowest probability of being chosen: 39.7% (which may in part be explained by an underrepresentation of elderly in the LISS panel). The respondents were more solidary with unemployed persons (51.3%) and people on welfare (51.8%). They were most solidary with respect to the disabled (62.0%). The difference between the four groups is highly significant: χ2(3)=1,575.069, p<0.001. This is, however, not the case for the difference between the unemployed and people on welfare. The results provide support for hypothesis 1.

For the number of under-aged children, solidarity initially increases substantially with the number of children (supporting hypothesis 1). The percentages for 0, 1, 2, 3 and 4 children are 42.1%, 57.9%, 63.5%, 67.6% and 72.2%, respectively. But after that point solidarity decreases again (refuting hypothesis 1). The percentage for five children is 70.4%, and it is 68.8% for six children. The differences between the seven groups are

12 For the regression analyses below, however, we used the original working years scores.


111

significant: χ2(6)=2,706.067, p<0.001. There is no significant difference between vignettes with three children and vignettes with either five or six children. Neither is there a significant difference between vignettes with four children and vignettes with five children. All other specific pairs are significant.

‘Distance’ had no influence at all. The expected scores are almost identical to the actual scores. Respondents in this study did not base their choice on whether or not people live in their neighbourhood.

Persons in a vignette who were willing to do volunteer work had the best chance of being chosen (56.3%). Persons who were somewhat willing scored less (51.9%), but substantially better than vignettes who preferred not to do volunteer work (41.6%). The difference between the three groups is significant: χ2(2)=852.581, p<0.001. The differences between ‘very willing’ and ‘somewhat willing’ and between ‘somewhat willing’ and ‘preferably not’ are also significant. These findings confirm hypothesis 3 (the ‘reciprocity’ hypothesis).

Because some of the independent variables are correlated, we conducted several multivariate regression analyses. This also enabled us to study the influence of respondent characteristics (to test hypothesis 4). The results are in Table 4.

In the baseline model A of Table 4, we look at the various characteristics of the persons described in the vignettes. The results more or less replicate those of Table 2. Women, persons with the same ethnic background, persons with a lot of working experience, persons who are willing to do volunteer work and persons having under-aged children can count on greater solidarity. Our respondents felt least solidary with persons with a pension benefit and most solidary with disabled people. Age and physical distance turned out not to matter for the level of public solidarity.

The result that a difference in ethnic background between the respondent and the beneficiary13 undermines solidarity14 is in line with hypothesis 4.

13 The fact that the characteristics of both the respondents and the fictive persons of the vignettes were used for this variable, is of course a bit atypical for model A. 14 An additional analysis (not in Table 2) shows that autochthones evoke more solidarity than people with a Surinamese background and the latter group scores better than people with a Moroccan background.

Cha

pter

Fou

r

112

Tabl

e 4:

Log

istic

mul

tilev

el r

egre

ssio

n an

alys

is o

f pub

lic so

lidar

ity

(A)

(B)

(C)

b ex

p(b)

si

g.

b ex

p(b)

si

g.

b ex

p(b)

si

g.

Vig

nett

e

Mal

e -0

.017

0.98

30.

002

-0.0

190.

981

0.00

4-0

.031

0.97

00.

000

Age

0.

000

1.00

00.

153

0.00

01.

000

0.27

30.

000

1.00

00.

187

Age

diff

eren

ce

0.00

01.

000

0.16

60.

000

1.00

00.

039

-0.0

020.

998

0.00

1 A

ge d

iffer

ence

^2

0.00

01.

000

0.34

90.

000

1.00

00.

627

0.00

01.

000

0.07

3 Sa

me

ethn

icity

0.

039

1.03

90.

000

0.03

61.

037

0.00

00.

047

1.04

80.

000

Wor

king

yea

rs

0.00

71.

007

0.00

00.

007

1.00

70.

000

0.00

61.

007

0.00

0 En

title

men

t W

elfa

re

0.17

61.

193

0.00

00.

185

1.20

30.

000

0.16

21.

176

0.00

0 U

nem

ploy

ed

0.18

11.

198

0.00

00.

195

1.21

50.

000

0.16

61.

181

0.00

0 D

isab

led

0.29

51.

343

0.00

00.

310

1.36

30.

000

0.28

01.

323

0.00

0 R

etire

d re

f re

f re

f re

f re

f re

f re

f re

f re

f N

umbe

r of c

hild

ren

0.05

41.

055

0.00

00.

054

1.05

50.

000

0.05

31.

054

0.00

0 D

ista

nce

0.00

71.

007

0.19

50.

003

1.00

30.

644

0.00

71.

008

0.19

3 Vo

lunt

eerin

g Ve

ry w

illin

g 0.

157

1.17

00.

000

0.15

11.

163

0.00

00.

157

1.17

00.

000

Som

ewha

t will

ing

0.12

01.

128

0.00

00.

115

1.12

20.

000

0.12

01.

127

0.00

0 U

nwill

ing

re

f re

f re

f re

f re

f re

f re

f re

f re

f

Div

ersi

ty, S

imila

rity,

and

Sol

idar

ity: R

esul

ts o

f a V

igne

tte S

tudy

113

Lef

t-ri

ght o

rien

tatio

n Le

ft-rig

ht o

rient

atio

n -0

.003

0.99

70.

648

x m

ale

0.00

41.

004

0.13

4x

age

0.00

01.

000

0.84

9x

age

diffe

renc

e 0.

000

1.00

00.

402

x ag

e di

ffere

nce^

2 0.

000

1.00

00.

350

x sa

me

ethn

icity

0.

011

1.01

10.

001

x w

orki

ng y

ears

0.

000

1.00

00.

028

x w

elfa

re

-0.0

100.

991

0.13

8x

unem

ploy

ed

-0.0

040.

996

0.56

6x

disa

bled

-0

.013

0.98

70.

047

x re

tired

re

fre

fre

fx

num

ber o

f chi

ldre

n -0

.003

0.99

70.

013

x di

stan

ce

-0.0

020.

998

0.58

2x

very

will

ing

to v

olun

teer

0.

003

1.00

30.

358

x so

mew

hat w

illin

g to

vo

lunt

eer

0.

001

1.00

10.

736

x

unw

illin

g to

vol

unte

er

ref

ref

ref

Cha

pter

Fou

r

114

Indi

vidu

al x

vig

nett

e M

an

-0.0

170.

984

0.04

3 N

umbe

r of c

hild

ren

(I)

0.00

11.

001

0.61

8 M

an x

mal

e

0.02

91.

029

0.01

2 A

ge x

age

diff

eren

ce

0.00

011.

000

0.00

1 A

ge x

age

diff

eren

ce^2

0.

000

1.00

00.

743

Num

ber o

f chi

ldre

n (I

)(V

) 0.

002

1.00

20.

240

Dut

ch

0.03

41.

035

0.00

3 D

utch

x sa

me

ethn

icity

0.

076

1.07

90.

192

Mor

occa

n

-0.0

130.

987

0.74

4 M

oroc

can

x sa

me

ethn

icity

0.

067

1.06

90.

545

Surin

ames

e

0.07

31.

076

0.17

0 Su

rinam

ese

x sa

me

ethn

icity

-0

.110

0.89

60.

144

Inte

rcep

t 0.

235

1.26

40.

000

0.23

71.

267

0.00

00.

246

1.27

90.

000

-2 lo

g lik

elih

ood

37

764

2876

4

37

733


115

Comparing the relative effects for Model A (the continuous variables were standardized to make the effects comparable), shows that entitlement and reciprocity (in terms of willingness to do volunteer work) are the most important factors in explaining public solidarity. The effects of the number of working years and the number of under-aged children are less strong. Ethnic similarity and gender have the weakest effect on public solidarity.

Model B adds the political orientation of the respondent as well as all interaction effects. This slightly changes the baseline results of model A. There are some interesting interactions. Right-wing voters have a stronger preference for people with the same ethnicity (supporting hypothesis 5). Furthermore, the tendency to favour persons with more working years is higher among right-wing voters. Right-wing respondents are less inclined to favour persons on pension benefits. Finally, right-wing voter show less solidarity towards people having under-aged children than left-wing voters.

While model A already includes the effect of similarity between the respondent and the beneficiary with regard to ethnicity and age, model C investigates the effect of the ‘similarity’ in other respects between the contributor (respondent) and the beneficiary. This leads to the following outcomes.

First, there seems to be a gender effect when it comes to public solidarity, as the interaction effect between the gender of the respondent and the gender of the fictive person is positive and significant. Older respondents assign more weight to age differences than younger respondents. While there is a direct relation between ethnic similarity and public solidarity, the results show that these preferences do not vary across the different ethnic groups (none of the interaction effects is significant): in-group preferences do not differ between Dutch, Surinamese or Moroccan respondents.

Intuitive versus deliberate decision making

We did some additional analyses concerning the role of intuitive versus deliberate decision-making. Respondents who consciously and deliberately weighted several attributes before making a decision did not react differently to the ethnic background of (fictive) persons than intuitive respondents. This distinction does not seem to play a role in explaining ethnic discrimination. However, we found some other results. People that choose consciously reacted relatively strongly to the number of under-aged children. Also, they had an even stronger preference for disabled

Chapter Four

116

persons over people on welfare and the employed compared to intuitive respondents. Their preference for persons on welfare over persons on pension benefits was relatively weak.

Study 2: Private solidarity

The study on private solidarity enabled us to test hypothesis 5 that states that different mechanisms may explain public and private solidarity. The descriptive statistics of the study on public solidarity are in Table 5 and are based on 27,934 observations/choices15.

Table 5: descriptive statistics (private solidarity) Variable Value Number

of times included

in a vignette

Number of times included

in the chosen

vignettes

Chance of being

chosen

Sex Male 29,250 (52.4%)

12,998 (46.5%)

44.4%

Female 26,618 (47.6%)

14,936 (53.5%)

56.1%

Age 18-30 years

11,047 (19.8%)

4,132 (14.8%)

37.4%

31-45 years

10,735 (19.2%)

4,019 (14.4%)

37.4%

46-64 years

15,537 (27.8%)

7,399 (26.5%)

47.6%

65-90 years

18,549 (33.2%)

12,384 (44.3%)

66.8%

Ethnic background

Dutch 37,822 (67.7%)

18,969 (67.9%)

50.2%

Moroccan 7,492 (13.4%)

3,627 (13.0%)

48.4%

Surinamese 10,554 (18.9%)

5,338 (19.1%)

50.6%

15 In Table 4, the five different helping situations are pooled. Depending on the specific helping situation, respondents might prefer person A over person B. However, we found no significant differences between the different helping situations.


117

Encounters Frequent encounters

30,049 (53.8%)

18,106 (64.8%)

60.3%

Seldom encounters

25,819 (46.2%)

9,828 (35.2%)

38.1%

Reciprocity Mentions returning the favour

25,598 (45.8%)

14,056 (50.3%)

54.9%

Does not mention returning the favour

30,270 (54.2%)

13,878 (49.7%)

45.8%

Respondents chose relatively more female (fictive) persons than male persons (compared to their prevalence in the vignette population): 56.1% versus 44.4%. The difference is significant: χ2(1)=759.796, p<0.001. Female (potential) beneficiaries seem to appeal more to our sense of solidarity than males. This effect is even stronger than in the public solidarity study16. We do not find a curvilinear relationship with the age group of ‘65 and older’ scoring worst. Instead, we found a linear relationship, with most solidarity for the eldest group: χ2(3)=3,498.902, p<0.00117. This provides support for hypothesis 1 (based on ‘neediness’ considerations).

As in the public solidarity study, there was a difference between the three ethnic groups. Moroccan vignettes had a 48.4% chance of being chosen, compared to 50.6% of the Surinamese vignettes and 50.2% of the autochthone Dutch vignettes. The difference was significant (but less so than in the public solidarity study: χ2(2)=9.327, p<0.01). Moreover, as opposed to the study on public solidarity, Surinamese persons were actually chosen more often than the Dutch, although the difference was not significant. The difference between the Dutch and the Moroccans was significant.

The ‘encounters’ variable from the private solidarity study resembles the ‘neighbourhood familiarity’ variable from the public solidarity study. While there was no effect of neighbourhood familiarity, there was a distinct effect of ‘encounters’. If the respondent encountered the (fictive)

16 We actually compared the public and private study by using only the attributes they have in common (sex, age and ethnicity). This comparison (not shown in this paper) shows that the preference for female rather than male beneficiaries is much stronger in the case of private solidarity. The difference is highly significant. 17 Again, the difference with the public solidarity study was highly significant.

Chapter Four

118

person a lot, this person had a 60.3% chance of being selected, compared to only 38.1% of the persons with whom there were few encounters. This difference is highly significant: χ2(1)=2735.130, p<0.001.

Willingness to reciprocate is also an important factor. Persons that offer to reciprocate are selected 54.9% of the times. Persons that make no reference to future reciprocation only have a 45.8% chance. This difference is highly significant: χ2(1)=455.696, p<0.001. This supports hypothesis 3 (the ‘reciprocity’ hypothesis).

All and all, we do not only find similarities, but also some differences between the public solidarity and the private solidarity study. This confirms hypothesis 5, although we need to look at the multivariate analyses in Table 6 for a better understanding.

The simplest model, model A, mimics the results discussed above. Gender, age, encounters in the neighbourhood and reciprocity all matter. Women, older persons, familiar persons and persons willing to return favours can count on more solidarity. Besides, age similarity between the respondent and the person displayed on the vignette does not affect solidarity. Again, a difference in ethnic background undermines solidarity. Assessing the relative importance of the determinants, we find that ‘neighbourhood encounters’ is the most important, followed by the age of the beneficiary, the gender of the person, the willingness to reciprocate and a difference in ethnic background.

In model B, we look again at the interactions with political orientation. Right-wing voters favour people from their own ethnic background more than left-wing voters.

In model C, the effects of similarity on solidarity are further explored. In model A, we found that similarity with regard to age does not affect private solidarity, while similarity with respect to ethnic background increases the willingness to help others. In model C, when we interact characteristics of the respondent with the characteristics of the fictive persons in the vignettes, the results are the following. Gender similarity does not increase private solidarity. Furthermore, the effect of age similarity is stronger if respondents are older. While age similarity does not matter if we take into account all age groups (as was shown in Model A), it turns out that older people find these differences more important. The interaction effect of age with age similarity shows that age similarity is more important for private solidarity if the respondents are older themselves.

Div

ersi

ty, S

imila

rity,

and

Sol

idar

ity: R

esul

ts o

f a V

igne

tte S

tudy

119

Tabl

e 6:

Log

istic

mul

tilev

el r

egre

ssio

n an

alys

is o

f pri

vate

solid

arity

(A

)

(B

)

(C

)

b

exp(

b)

sig.

b

exp(

b)

sig.

b

exp(

b)

sig.

V

igne

tte

M

ale

-0.1

280.

880

0.00

0-0

.129

0.87

90.

000

-0.1

350.

874

0.00

0 A

ge

0.00

71.

007

0.00

00.

007

1.00

70.

000

0.00

61.

006

0.00

0 A

ge d

iffer

ence

0.

000

1.00

0 0.

380

0.00

01.

000

0.88

90.

004

1.00

40.

000

Age

diff

eren

ce^2

0.

000

1.00

0 0.

062

0.00

01.

000

0.57

20.

000

1.00

00.

031

Sam

e et

hnic

ity

0.02

31.

023

0.00

00.

020

1.02

10.

002

-0.0

190.

981

0.11

9 En

coun

ters

0.

240

1.27

2 0.

000

0.24

21.

274

0.00

00.

239

1.27

00.

000

Rec

ipro

city

-0

.118

0.88

9 0.

000

-0.1

200.

887

0.00

0-0

.118

0.88

90.

000

Lef

t-ri

ght o

rien

tatio

n Le

ft-rig

ht o

rient

atio

n -0

.008

0.99

30.

048

x m

ale

0.00

21.

002

0.39

6x

age

0.00

01.

000

0.95

9x

age

diffe

renc

e 0.

000

1.00

00.

220

x ag

e di

ffere

nce

^2

0.00

01.

000

0.74

0x

sam

e et

hnic

ity

0.01

11.

011

0.00

0x

enco

unte

rs

-0.0

010.

999

0.63

8x

reci

proc

ity

0.00

01.

000

0.94

5

Cha

pter

Fou

r

120

Indi

vidu

al (I

) x v

igne

tte

(V)

Man

-0

.010

0.99

00.

213

Man

x m

ale

0.00

71.

007

0.50

7 A

ge x

age

diff

eren

ce

0.00

011.

0001

0.00

0 A

ge x

age

diff

eren

ce ^

2 0.

000

1.00

00.

293

Dut

ch

-0.0

490.

990

0.00

0 D

utch

x sa

me

ethn

icity

0.

047

0.99

00.

000

Mor

occa

n -0

.024

0.99

00.

007

Mor

occa

n x

sam

e et

hnic

ity

-0.0

900.

990

0.12

9 Su

rinam

ese

0.05

40.

990

0.00

0 Su

rinam

ese

x sa

me

ethn

icity

-0

.145

0.99

00.

125

Inte

rcep

t 0.

493

1.63

7 0.

000

0.49

21.

635

0.00

00.

490

0.99

00.

000

-2 lo

g lik

elih

ood

3594

1 27

232

3586

3


121

In other words, older respondents are more sensitive to age similarity, when it comes to helping others, than younger respondents. Since there is no effect of the squared difference, this relationship is linear and not curvilinear. Finally, the preference for ethnic similarity varies across the ethnic groups. While this does not matter for the Surinamese and the Moroccan respondents, the Dutch respondents are more responsive to ethnic similarity (the interaction effect between being Dutch and ethnic similarity is statistically significant).

Intuitive versus deliberate decision making

Our additional analyses showed that conscious, deliberate respondents had an even stronger preference for elder (fictive) persons than intuitive decision makers. They also reacted relatively strongly to reciprocity.

Discussion

Summary of the results

When forced to choose between two fictive persons, respondents show a systematic preference for certain groups of beneficiaries. In the context of the welfare state people show solidarity towards disabled persons and are less inclined to help elderly people on (second pillar) pension benefits. They favour people with under-aged children and many years of working experience. Being willing to do volunteer work enhances the chances of being chosen as well. Given the underlying assumptions of the welfare state (such as neediness and deservingness), we would expect such differences. However, we also find differences that are not in line with the philosophy behind the welfare state. Respondents are slightly more inclined to favour someone from their own neighbourhood. We must conclude that sympathy and/or recognition also play a role, even in the context of the (supposedly) impartial welfare state.

The results regarding the concrete helping situations were quite similar. Here, women, people from one’s own ethnic group, persons that one meets frequently, and those who are willing to reciprocate are more likely to be helped. A major difference, focusing on the vignette characteristics, is that age does play a role here. Older people are more likely to receive help.

We were especially interested in the role of ethnic diversity. Both studies show that people show more solidarity towards people with the same ethnic background. Besides, we checked whether in-group favouritism

Chapter Four

122

varies across ethnic groups. This is not the case for public solidarity, but it is for private solidarity.

When comparing the size of the effects, ethnic diversity matters less for solidarity compared to the other determinants. The conclusion that ethnic background is not the most important predictor of solidarity is especially sound, as the vignette methodology is ideally suited to pick up not only conscious, but also unconscious preferences and attitudes. In this line of reasoning, it is relevant to note that ethnic discrimination could not be explained by differences in the way of decision-making (deliberate versus intuitive).

Interestingly, not only the characteristics of the beneficiaries matter, but also the degree of similarity between the respondent and the (fictive) beneficiary. Besides the results for ethnic in-group favouritism, the following results are found. In the case of public solidarity, similarity in gender, age, and number of under-aged children matter. Age similarity plays a role in private solidarity.

We found support for all five hypotheses. People express more solidarity towards people in need (hypothesis 1), such as the disabled or people with many under-aged children (in the public solidarity study) or the elderly (in the private solidarity study). They were sensitive to the number of working years, a proxy for deservingness (hypothesis 2). The same holds for proxies for reciprocity (hypothesis 3), such as willingness to do volunteer work. All things being equal, people expressed more solidarity towards people with the same ethnic background than towards people from a different ethnic background (hypothesis 4). This effect was even stronger for respondents with a right-wing political orientation (hypothesis 5).

Interpretations and puzzles

We also found some results that were not explicitly predicted. One of these findings was the greater solidarity towards female beneficiaries than towards male beneficiaries. This difference was found in both solidarity studies. A possible explanation is that, even today, women are perceived as needier and less independent than men, for example because, on average, they earn less and are less often financially independent. The inclination to help women more often than men in concrete helping situations might reflect stereotypes about physical strength and technical skills. If these interpretations are correct, the results on gender would fit the neediness-framework of hypothesis 1, but this needs further examination.


123

We found that the number of under-aged children correlated positively with solidarity received, but that solidarity decreased again when a (fictive) parent had more than four children. A possible explanation is that some respondents associate having ‘too many’ children with irresponsibility and undeservingness. Again, this is an interpretation that needs further study.

Although we predicted differences between the public solidarity study and the private solidarity study, we did not specify these differences. We found that solidarity with the elderly is much stronger in the case of private solidarity. Apparently, we do not feel sorry for elderly people with respect to their pension benefits, whereas we do feel that the elderly deserve our help in concrete situations. Possibly, the elderly are seen to be needy in some respects (e.g. physical), but not in other respects (financial).

We also found that knowing someone from the neighbourhood hardly played a role in the public solidarity study, but was very important in the private solidarity study. This makes sense if we think of the welfare state as being impersonal and unbiased. On the other hand, strong personal networks and direct contact seem to be crucial for helping concrete individuals. This underlines that the kind of solidarity varies across contexts (see also Koster, 2013).

Implications

The present study shows the similarities and differences between explanations for public and private solidarity. These analyses suggest that there are a number of motives underlying the solidarity towards others as we find evidence for deservingness, neediness, reciprocity, similarity and self-interest. Hence, rather than trying to pin down one particular explanation, this study suggest that it is much more fruitful to regard solidarity as a combination of these different motives (cf. chapter 1). Statements that solidarity is just a matter of self-interest, or just a matter of pure altruism are overblown and do not do justice to the research outcomes presented here. Future research should take this into account and investigate how these different motives interact with each other.

The results may also provide some input for policy making. This holds in particular for the analysis of public solidarity. Based on the results of that analysis, it may be concluded that cuts in the benefits of particular groups can be easier, in the sense that they are accepted by a larger share of the public than cutting the benefits of other groups. We should, however,

Chapter Four

124

emphasize that these preference are the result of a forced choice that was presented to the respondents. This means that the respondents had to make a choice in the survey and not that they were glad to make that choice. Furthermore, in the present design, we did not take into account the strength of solidarity with each group. Therefore, it cannot be excluded that part of the public will not support cuts in benefits because they do not consider cutting benefits as a preferable policy option.

References

de Beer, P., and Berg, M.C. (2012). Conditions and motives for voluntary sharing; Results of a solidarity game experiment. AIAS Working Paper 124. Amsterdam: University of Amsterdam/Amsterdam Institute for Advanced Labour Studies.

de Beer, P., and Koster, F. (2009). Sticking together or falling apart. Solidarity in an era of individualization and globalization. Amsterdam: Amsterdam University Press.

de Wolf, I., and van der Velden, R. (2001). Selection processes for three types of academic jobs: an experiment among Dutch employers of social sciences graduates. European Sociological Review, 17, 317-330.

Fong, C.M., Bowles, S., and Gintis, H. (2004). Reciprocity, self-interest, and the welfare state. In Jean Mercier-Ythier, Serge Kolm and Louis-André (eds.). Handbook on the Economics of Giving, Reciprocity and Altruism. Amsterdam: Elsevier.

Knoef, M., and De Vos, K. (2009). The representativeness of LISS, an online probability panel. Tilburg: CentERdata.

Koster, F. (2009). Risk management in a globalizing world. An empirical analysis of individual preferences in 26 European countries. International Social Security Review, 62, 79-98.

Koster, F. (2013). Sociality in diverse societies. A regional analysis across European countries. Social Indicators Research, 111, 579-601.

McFadden, D., Bemmaor, A., Caro, F., Dominitz, J., Jun, B.H., Lewbel, A., Matzkin, R., Molinari, F., Schwarz, N., Willis, R., and Winter, J. (2005). Statistical analysis of choice experiments and surveys. Marketing Letters, 16, 183-186.

Putnam, R. (2007). E Pluribus Unum. Diversity and community in the twenty-first century. Scandinavian Political Studies, 30, 137-174.

Rossi, Peter H., and Steven L. Nock (eds.) (1982). Measuring social judgments. The factorial survey approach. Sage.

van Oorschot, W. (2000). Who should get what, and why? On deservingness criteria and the conditionality of solidarity among the


125

public. Policy and Politics, 28, 33-48. —. (2006). Making the difference in social Europe. Deservingness

perceptions among citizens of European welfare states. Journal of European Social Policy, 16, 23-42.

Wason, K.D., Polonsky, M.J., and Hyman, M.R. (2002). Designing vignette studies in marketing. Australasian Marketing Journal, 10, 41-58.

CHAPTER FIVE

MORE OR LESS STRANGERS: DIVERSITY IN THE DUTCH MEDIA DEPICTION

OF ETHNIC MINORITIES AND ITS IMPLICATIONS FOR INTERETHNIC SOLIDARITY

DOROTA LEPIANKA

Introduction: why media analysis?

One of the key themes in this volume is that attitudes and behaviour of citizens towards people with a different ethnic background, including their readiness for co-operation and solidarity, depend strongly on their image of various ethnic groups; an image that might be reflected, but also reinforced by, the mass media. Indeed, the role of the media in contemporary societies is hard to ignore, if only because public perception and evaluation of various social phenomena frequently correspond to their presentation and interpretation in the media (e.g. Bullock, Wyche and Williams 2001; Clawson and Trice 2000; Gilens 1996, 1999; Golding and Middleton 1982; Kennamer 1994; Lemert 1994; Tipton 1994). Researchers emphasize, among others, the framing function of the media, asserting that repeated presentation of specific groups or issues in connection with particular attributes is likely to result in the audience members choosing those attributes in evaluating the groups or issues in question (e.g. Entman 1989; McCombs and Reynolds 2009; McQuail 2010; van Ginneken 2003). In particular, the societal portrayal of minority groups, such as ethnic minorities, with whom direct contract and exchange is very limited, seems to rely on media as the main source of knowledge (e.g. d'Haenens and Ogan 2007; cf. McCombs and Reynolds 2009; Thompson 1995). Not surprisingly, it is repeatedly claimed that through the (stereotypical) portrayals of various groups, coupled with the presentation of the current norms and values and the image of social life in

Chapter Five

128

a particular society (cf. Peeters and d’Haenens 2005), media help to establish and maintain the relations between various social groups, such as between the native majority and ethnic minorities. By the same token, they may also disturb the formation or preservation of such relations (cf. Braun and Koopmans 2010).

In an attempt to shed more light on the nature of solidaristic relations between various ethnic groups in Dutch society, this chapter examines the media-conveyed images of different ethnic minorities living in the Netherlands. The investigation rests on the assumption of an active role of the media in the processes of (re-)establishment of group boundaries and, in consequence, in reinforcing – or vice versa: hindering – the creation and/or maintenance of social relations between (ethnic) groups. Indeed, considering the relative infrequency of inter-ethnic contact between the natives and non-natives in the Netherlands (e.g. Dagevos and Gijsberts 2009), it is plausible that it is predominantly through the media that a socially shared matrix of inter-ethnic relations is constructed and/or reinforced (cf. Lubbers, Scheepers, and Wester 1998); if not directly, by disseminating specific patterns of inter-groups behaviour, then indirectly, by propagating specific group stereotypes.

This chapter is structured as follows: after a theoretical discussion on the media construction of ethnic minorities and the need for a more nuanced and non-generic investigation of ethnic stereotypes, we present the research strategy, followed by a discussion of the results. In the concluding remarks, the implications of the reconstructed media images for inter-ethnic interactions, trust and co-operation in particular, are discussed in the light of the literature on the content of stereotypes as well as the literature on social distance and (social) processes of ‘othering’.

Literature overview and research agenda

Research into the portrayal of ethnic minorities in Dutch media is by no means novel. Presence and depiction of ethnic characters have been investigated in a variety of media, including the daily press (e.g. d’Haenens and Bink 2006; d’Haenens and de Lange 2001; Lubbers et al. 1998; Phalet and ter Wal 2004; Saharso and Lettinga 2008; ter Wal, d’Haenens, and Koeman 2005; Vliegenthart and Boomhaarden 2007) and a variety of TV genres (de Bruin 2005; Koeman, Peeters, and d’Haenens 2007).

More or Less Strangers

129

Media analysts generally agree that, relative to their proportions in the population, ethnic minorities tend to be underrepresented in media reports, especially, albeit not exclusively, in non-ethnic or ethnically neural stories dealing, e.g., with politics and government (ter Wal et al. 2005). Once present in stories about politics, minority actors rarely feature in the high-ranking and prominent roles of politicians and professionals actively involved in political process, but rather as common people (ter Wal et al. 2005; cf. Leurdijk 1997; Phalet and ter Wal 2004) and targets or recipients of policies (ter Wal et al. 2005).

A further consideration is the general underrepresentation of ethnic voices and their limited discursive opportunities (Phalet and ter Wal 2004; cf. d’Haenens and Bink 2006; de Lange and d’Haenens 2002). As noted by Saharso and Lettinga (2008), even in ‘ethnic’ stories, i.e. stories dealing specifically with ethnic or migrant issues, such as debates about the veil, migrants, while present, are not necessarily given a chance to influence the tone, direction or content of the debate (Leurdijk 1997).

Finally, attention has been drawn to the media practices of reporting on ethnic minorities in the context of conspicuous incidents, conflicts, problems and threats (d’Haenens and de Lange 2001; Lubbers et al. 1998; ter Wal et al. 2005; cf. Phalet and ter Wal 2004), with a particular emphasis given to crime, various forms of illegality (e.g. in case of asylum seekers) or their cultural and/or religious ‘otherness’ (der Valk 2002; d’Haenens and Bink 2006; Korteweg and Yurdakul 2009). Ethnic minorities are generally over-represented among criminals and deviants; and stories with an ethnic dimension tend to feature more negative portrayals (of the native and non-ethnic actors) than non-ethnic stories (ter Wal et al. 2005), which strengthens the association of ethnicity with negativity. Furthermore, the negative portrayals are frequently generalised to whole groups (der Valk 2002; d’Haenens and de Lange 2001; ter Wal et al. 2005), e.g. by references to ‘the Moroccan migrants’, ‘the Islam’ or ‘the allochthon’. Interestingly, media focus on problems is more pronounced for groups larger in size (Lubbers et al. 1998). Explicitly positive images are rather infrequent and reserved for recognized celebrities, particularly in sports and/or entertainment. However, indirect positive portrayals are also sometimes evoked, e.g. by presenting minority members as ‘constituent members of society who enrich it’ (der Valk 2002; cf. Leurdijk 1997).

Remarkably, research on media construction of migrants has seldom been done in a set-up that would allow systematic inter-group comparisons.

Chapter Five

130

Few attempts have been made to establish the rank-order of under-representation, lack of prominence and/or negativity in media construction of various groups, or, in other words, to address the issue of diversity in the media depictions of ethnicity. Most studies tend to focus on a generic, homogenized ‘ethnic minorities’ or a specific migrant group, such as asylum seekers or Muslims. A study by ter Wal et al. (2005) regarding the position of ethnic minorities vis-à-vis the (ethnic-Dutch) majority, a study by Lubbers at al. (1998) comparing the media construction of various ethnic groups and a study of de Bruin (2005) on the representations of ethnicity in Dutch TV series constitute in this respect notable exceptions.1 Yet, looking into the differences in the media depictions of various ethnic minorities is of vital importance for a number of reasons.

First, in accordance with the Stereotype Content Model (Cuddy, Fiske, and Glick 2008; Fiske, Cuddy, and Glick 2006), not all seemingly negative/positive stereotypes are equally or similarly negative/positive. In fact, stereotypes are often mixed or ambivalent and their content constitutes a combination of two dimensions – warmth (which entails traits such as fairness, helpfulness and kindness versus hostility and aggression), and competence (which implicates possession of skills, talents and capabilities, or lack of them). Some ethnic minorities, e.g. Jews and Asians in the US, are likely to be seen as highly skilled, talented and determined (i.e. competent) yet at the same time judged asocial, unfriendly, hostile or even aggressive (i.e. cold); other ethnic groups (e.g. Italian and Irish immigrants in the US) are likely to be seen as warm but not particularly competent. In general, unequivocally negative stereotypes, which involve negative evaluations on both dimensions (i.e. being considered both cold and incompetent), are reserved for groups seen as parasitic (e.g. undocumented migrants in the US), whereas unequivocally positive stereotypes are set aside for the in-group and the high-status reference groups (e.g. middle class in the US) (Lee and Fiske 2006).

Second, the specific combinations of ‘warmth’ and ‘competence’ drive the perceivers’ emotional prejudices and behavioural reactions towards the members of stereotyped groups (Cuddy et al. 2008; Fiske et al. 2006). Groups seen as incompetent and cold elicit contempt, rejection and/or disgust, which might lead to distancing, exclusion but also harassment or

1 A comparative perspective is also assumed in the research by Phalet and ter Wal (2004) on the differences in the presence and discursive practices of Turkish, Moroccan and autochthon speakers in press articles related to Muslims and Islam in the Netherlands.


131

even violence. Groups stereotyped as competent yet cold elicit respect but also envy, which evokes reactions of obligatory association and convenient cooperation, yet may result in aggression and scapegoating (in times of instability). Groups perceived as warm but not competent elicit compassion or pity, which feeds altruistic help but may also bring neglect and avoidance. Finally, groups evaluated as warm and competent evoke admiration and pride and elicit reactions of voluntary association and help.

Third, as shown by Ramasubramanian and Oliver (2007), positive (media) stereotyping of one out-group (e.g. Asians in the US) may contribute to the relative diminishing (and as a result, further discrimination) of another out-group (e.g. African-Americans). According to the researchers, this is related to the (discursive practice of) setting one group of migrants apart as a ‘model minority’ and a yardstick against which the performance of all the other groups is measured.

Informed by the insights regarding a relationship between the content of stereotypes and the nature of social interactions, on the one hand, and the active role of the media in the formation and maintenance of stereotypes, on the other, the current study constitutes an attempt to reconstruct the dominant media representations of various minority groups vis-à-vis one other. In the course of the analysis the amount and forms of news media attention offered to various groups are compared, and the differences in their presentation are examined. Specific research questions addressed in the study relate to the visibility and prominence of various groups in the media coverage and the nature of their presentation. The latter is assessed with respect to the homogeneity of presentation and evaluation, where evaluation is explored on two dimensions: warmth and competence. The results of the study are interpreted in the light of literature on social distance and social practices of ‘othering’. The implications of the media depiction of specific ethnic groups and interethnic solidarity are also discussed.

Data and method

The analyses presented in this chapter constitute a part of a broader project on the media representations of various social groups in the Netherlands. In order to assess the dominant media images of the various ethnic groups, a content analysis of daily newspapers, television news editions and an online news engine was conducted in two research periods. The sample consisted of news items from the most widely used sources of political information, including two national newspapers, de Volkskrant and de

Chapter Five

132

Telegraaf; two broadcasting news programmes, NOS Journaal (8 PM edition) and RTL Nieuws (7:30 PM edition), and the most popular Internet news provider nu.nl.

The media sources were selected on the basis of their popularity among different segments of the population and their distinctive profile. Only articles published in the main sections of the dailies were analysed; no articles from media, sport and/or entertainment sections were selected for the analysis. The two research periods selected for the analysis were a period of a pre-election campaign (17 May – 02 June 2010) and a period following the formation of the new government (16 November – 02 December 2010). The relevant articles and television news items were selected on the basis of their verbal content, that is, the presence of specific key-terms. Only articles and news items that mentioned (a) human being(s) (actors), such as ‘asylum seeker’, ‘migrant’, ‘Moroccans’, ‘East-European’ were selected for the analysis. The TV news-items were additionally coded for their visual content, such as presence of clues suggesting a specific context of presentation (e.g. an arrest, a medical intervention, a social occasion, a demonstration).

All news items (215) and actors (332) were analysed by means of content analysis. While content analysis is frequently conceptualised as a quantitative technique that allows a systematic study of large corpuses of text (Neuendorf 2002), a growing number of researchers apply various forms of qualitative, or interpretative, content analysis (cf. Wester, Pleijter, and Renckstroft 2004). In the current study, the two approaches – quantitative and qualitative analysis – are combined.

The coding of the material selected for the analysis was conducted in two steps. In step one, a rigorous quantitative analysis was completed by the author and two trained coders. The coding conducted at this stage could be conceptually divided into two levels. The first (news-item) level involved coding of the general information about the news story and entailed a holistic look at its various elements. At the second (actor) level, the news story was broken into utterances (acts of speech, claims); only utterances and statements that referred to the actors of interest to the study were coded. The inter-coder reliability tests produced Krippendorff alpha ranging from .76 to 1.00 for the measures relevant to the analysis.

In the second step, the statements extracted verbatim from the news messages in step one of the coding (as evidence of a negative, positive or ambivalent depiction of an actor) as well as the comments generated by


133

the coders with respect to every coded actor were re-coded by the author as to the valance (positive versus negative) and the dimension of evaluation (warmth versus competence).

Operationalisation of key variables

Visibility and prominence. Visibility of various groups was measured by the number of news stories featuring a specific type of actor. The prominence of various groups was measured as: (1) a ratio of front page articles (in printed and online news) or headline news (in visual media) to the total number of stories per group; (2) the relative importance of actors within the news stories on a scale from 1 to 3, where 1 = detail and 3 = focus of the story; (3) an index of prominence equal to ‘relative importance’ (on a scale 1-3) multiplied by the length of the story (measured by number of words in the body text of the story or the number of seconds in TV news items) and divided by 100; (4) the proportion of actors within each category that is given the opportunity to express their views; and (5) the proportion of ‘expert’ speakers among the actors with voice (per category).

Homogeneity of presentation. The degree of homogeneity of presentation was measured by the proportions of various types of actors per category. Here, a distinction was made between individual actors, individual exemplars (representatives of a wider category) and group actors.

Evaluation – valance. Negative evaluation was measured through a series of variables, including: (1) featuring in an article with a ‘negative overall theme’, where ‘negative theme’ included public order, conflict and defence and demonstrations in a ‘broad’ conceptualization; and public order and demonstrations in ‘narrow’ conceptualization, and where 0 = no association with a negative theme and 1 = association with a negative theme; (2) presentation as a perpetrator (incl. terrorism suspects), where 0 = not reported as perpetrator and 1 = reported as perpetrator; (3) presentation as a terrorist (incl. terrorism suspects), where 0 = not reported as a terrorist and 1 = reported as a terrorist; (4) presentation as a victim of crime or maltreatment, where 0 = not reported as a victim and 1 = reported as a victim; (5) presentation as a beneficiary of public goods, where 0 = no beneficiary and 1 = beneficiary; and (6) other explicitly negative qualifications, where 0 = no negative qualification and 1 = negative qualifications.

Chapter Five

134

Positive evaluation was measured by a dummy pertaining to explicitly positive qualifications, where 0 = no positive qualification and 1 = positive qualification. Both positive and negative evaluations were re-coded on the basis of statements extracted from the news messages in the process of coding as well as the comments/notes generated by the coders along two dimensions: warmth and competence.

Evaluation – dimension. A positive evaluation was considered indicative of the benevolence (warmth) of the actor if his/her character, behaviour or intentions were described in terms of friendliness, kindness, agreeableness, helpfulness, charity, honesty and trustworthiness; it was considered indicative of the competence of the actor if their character, behaviour or intentions were described as (indicative of being) smart, motivated, devoted, persistent, skilful, industrious, intelligent, etc.

A negative evaluation was considered indicative of the coldness of the actor if (s)he was presented as unfriendly, disagreeable, unkind, dishonest, free-loading, abusive, aggressive, hostile, threatening, etc.; it was considered indicative of the incompetence of the actor if they were portrayed as foolish, unintelligent, careless, forgetful, weak, etc. Any reference to physical disability, infirmness and/or mental or physical vulnerability (e.g. loneliness, addiction) was coded under incompetence.

A number of evaluations were coded as mixed/ambivalent qua dimension of evaluation. These were evaluations which included elements of both warmth and competence or were difficult to classify (e.g. ‘such a fresh girl, a good addition to our team’). Descriptions that included both positive and negative elements were coded twice – as positive evaluations (warm, competent or mix) and as negative evaluations (cold, incompetent or mix).

Findings

Visibility

The relative visibility of the various ethnic groups in the news media coverage can be inferred from the juxtaposed information on the structure of the sample included in the current study and the actual share of the relevant groups in the Dutch population. As demonstrated in Table 1, non-western allochthons, defined according to Statistics Netherlands (CBS) as ‘every person living in the Netherlands of which at least one of the parents was born’ in Africa, Latin America, Asia (with the exception of Japan and Indonesia) or in Turkey (CBS 2000), constitute a clear majority. Described


135

in media messages either by their nationality (40.4%), by being Muslim or coming from a ‘Muslim country’ (16.9%), or by other reference to ‘non-western’ origin, such as continent of origin or skin colour, non-western allochthons amount to 66% of the coded actors. If the group of asylum seekers and illegal immigrants, who usually originate from the same geographical areas as ‘non-western’ immigrants, were added, then the sub-sample of non-western immigrants would constitute 68% of the migrant actors. Of the actors whose nationality is actually reported, Moroccans constitute a clear majority (64 actors), followed by Turks (27) and Iraqi’s (11, 9 of which coded as asylum seekers and 2 of which coded as other/mix of nationalities). Of the Eastern European nationals, the Polish are mentioned slightly more often than other nationalities, yet usually also in combination with Romanians and/or Bulgarians.2

At first sight, the structure of the sample might suggest a serious imbalance in the media attention to various migrant groups. However, considering the sampling procedure, which gave preference to non-Western allochthons and new migrants form Central and Eastern Europe3, the under- and overrepresentation of various groups of migrants, and especially migrants of Western origin, is difficult to evaluate. Some tentative observations might be made only with respect to non-Western allochthons: whereas Moroccans and asylum-seekers appear to enjoy striking conspicuousness in the news media, Turks, Antilleans and Arubans, and especially Surinamese seem to be grossly under-represented. Although usually such lack of visibility is interpreted as disadvantageous to minority populations, given the sampling procedure adopted in the current study (the sample includes only those actors whose ethnic status has been explicitly mentioned), the under-representation of Surinamese and Antillean actors may in fact testify to their ‘commonality’ or ‘domestication’ revealed in the infrequent use of ethnic labelling.

2 Nationals of former Yugoslavia (5) were coded as Eastern-Europeans or, if referred to as such, as asylum seekers, as they were usually reported as members of a broader group of migrants. 3 The search-string used to select the relevant news items included: moslim! OR islam! OR allochto! OR vreemdeling! OR immigra! OR migra! OR asiel! OR vluchteling! OR minderhe! OR Marokka! OR Turk! OR Antillia! OR Aruba! OR Surinamer! OR Surinaams! OR Indones! OR Indisch! OR Moluk! OR Jood! OR Zigeun! OR afkomst OR komaf OR (nieuw! w/1 Nederland!) OR nieuwkomer! OR Pool! OR Polen OR Roem! OR Bulgar! OR (Oost! w/1 Europ!) OR gastarbeid! OR arbeid! OR nationaliteit! OR buitenlander! OR autochto! OR inburger! OR integratie.

Chapter Five

136

Table 1. Sample and relative media visibility of different sub-groups of the migrant population Sample

(N=332) Share in the migrant population

Western 44.7% Eastern European 7.5% (n=25) 33.3%a Western (European) 1.5% (n=5) Indonesians/Indos/Moluccans 1.8% (n=6) 11.4% Non-western 55.3% Jewish 2.1% (n=7) 0.2% Moroccans 19.3% (n=64) 10.4% Turks 8.1% (n=27) 11.4% Antillean and Aruban 0.3% (n=4) 4.1% Surinamese 1.2% (n=4) 10.2% Other or mix of non-western

nationals 9.6% (n=32)

19.0% Other indication of non-western origin

8.7% (n=29)

Ethnic-less Muslims 16.9% (n=56) Asylum seekers and illegal immigrants

12.0% (n=40) 0.5%b

Unspecified/unclear or western- & non-western mix

10.8% (n=36) -

a Of the Western migrants other than Indonesian, Europeans constitute the majority: 31.1% of all foreign population; migrants from Poland, Bulgaria and Romania constitute 10.2% of the European migrants, or 3.2% of the total migrant population. b Asylum requests in 2010 relative to the total of migrant population. Source: CBS (Statline: http://statline.cbs.nl) (Dutch population per month, data for 1 January, accessed November 2013). Nonetheless, the structure of the sample determines the extent of the analyses and group comparisons that can be undertaken in the current study. Groups that are particularly small in number (Surinamese, Antilleans, Indonesians, Moluccans, Jews and Western migrants) are not included in the group-focused analysis other than as a composite reference category of ‘other migrants’. Thus, the group-focused analysis has to be constrained to the more visible groups of migrants: Moroccans, Turks,


137

asylum seekers and Eastern Europeans. To these ethnic groups, the category of ‘Muslims’ is added; it comprises actors whose ethnic origin was not reported and whose non-autochthonous status was inferred from their ascribed religious affiliation. The category of Eastern Europeans includes various actors of Central and Eastern European nationalities and ethnic-less actors described as Eastern European; the category of asylum seekers includes actors described as asylum seeking, regardless of their ethnicity, if reported. For comparative reasons, results with respect to specific sub-groups of migrants are juxtaposed against the composite category of ‘all other migrants’, consisting predominantly of actors of unspecified or mixed ethnicities (e.g. groups forming a mixture of Western and non-Western allochthons, Turks and Moroccans, Muslims and non-Muslims) but also actors belonging to groups that are small in number (e.g. Surinamese, Aruban, Indonesian, etc.). The results are summarized in Tables 2 and 3.

Prominence

In general, the prominence of characters in the news media can be inferred from the prominence of the items in which they feature (e.g. on the front page in the case of newspapers or in the headlines in the case of television news) and their relative importance within the news story (see Table 2). Of all actors coded, only 32 appear in a front page article or a TV news headline, 12 of whom (more than one third) are Moroccan. Least visible, in terms of news-item prominence, are Eastern Europeans (of the 25 actors coded, none features in a TV news headline or a front page article), asylum seekers and Turks. Still, differences between any specific group of migrants and the composite category of ‘all other migrants’ are statistically non-significant.

Strikingly, the hyper-visibility of Moroccans is not reflected in their within-story importance: Moroccan actors appear to attract only slightly more TV news attention and substantially less print media attention than the reference category of ‘other migrants’. Of the analysed migrant groups, it is the Turkish actors who are awarded the most prominent place within media news coverage. The prominence of Turks is reflected not only in the relative space they occupy in the news (15.9 on the index of prominence), but also in the frequency of being given the opportunity to express their own perspective with respect to the events or phenomena reported (55.6% as compared to 32.5% of the reference category, p .05 level). Of all the groups analysed, asylum seekers and Eastern Europeans are awarded least prominence in the news: their presence in headlines and front page news as

Cha

pter

Fiv

e

138

Tab

le 2

. Rel

ativ

e pr

omin

ence

of d

iffer

ent s

ub-g

roup

s of m

igra

nts:

per

cent

age

(abs

olut

e nu

mbe

r) o

f act

ors w

ith a

gi

ven

char

acte

rist

ic.

Oth

er

mig

rant

s (r

ef.

grou

p)

Mor

occa

n

Tur

kish

Mus

lim

Asy

lum

se

ekin

g

Eas

tern

E

urop

ean

Prom

inen

ce

N

120

64

27

56

40

25

H

eadl

ine/

fron

t pag

e ne

ws

(%)

10.8

(13)

18

.8 (1

2)

3.7

(1)

8.9

(5)

2.5

(1)

0

M

ean

with

in-s

tory

pr

omin

ence

2.

2

2.1

2.1

2.

05

2.3

2.

4

In

dex

of p

rom

inen

ce

Pres

s TV

9.5

3.6

8.8

3.

4

15.9

† 2.

4

10.1

1.

9

9.6

2.

6

11.5

0

Act

ors w

ith v

oice

(%)

32.5

(39)

32

.8 (2

1)

55.6

(15)

* 37

.5 (2

1)

17.5

(7)

24.0

(6)

Ex

perts

(%)

5.8

(7)

6.3

(4)

3.7

(1)

5.4

(3)

2.5

(1)

0

Mor

e or

Les

s Stra

nger

s

139

Hom

ogen

eity

a

N

11

4 58

23

48

40

25

Indi

vidu

al a

ctor

s am

ong

non-

inst

itutio

nal a

ctor

s (%

) 39

.5 (4

5)

44.8

(26)

78

.3 (1

8)**

37

.5 (1

8)

27.5

(11)

40

.0 (1

0)

G

roup

act

ors a

mon

g no

n-in

stitu

tiona

l act

ors (

%)

50.0

(57)

34

.5 (2

0)

13.0

(3)*

* 56

.2 (2

7)

72.5

(29)

* 56

.0 (1

4)

In

divi

dual

exe

mpl

ars,

typi

cal r

epre

sent

ativ

es o

f a

grou

p (%

)

10.5

(12)

20

.7 (1

2)

8.7

(2)

6.2

(3)

0 4.

0 (1

)

a Res

ults

with

resp

ect t

o ho

mog

enei

ty a

re b

ased

on

a di

min

ishe

d sa

mpl

e.

Sign

ifica

nce

leve

l: D

iffer

ence

s be

twee

n th

e gr

oups

und

er s

tudy

and

the

refe

renc

e ca

tego

ry o

f ‘O

ther

mig

rant

s’, t

-test

sig

nific

ant a

t p=

.05;

**

sign

ifica

nt a

t p=.

01; *

Par

son’

s Chi

-squ

are

sign

ifica

nt a

t p=.

05.

Chapter Five

140

well as the percentage of actors quoted is far below that of the reference category. Still, statistically speaking, the results obtained for Moroccans, asylum seekers and Eastern Europeans do not differ significantly from those obtained for the reference group of ‘other migrants.’ In statistical terms, only the prominence of the Turks differs significantly (and positively) from the prominence of the reference actors, and only in respect to two measures of prominence: index of prominence in press and proportion of actors with voice.

Homogeneity

Most of the actors analysed are group actors, described as Muslims, allochthon youth, Moroccan boys, Eastern European workers, etc. Of the individual actors, most are individual human beings presented as of interest on their own; a small(er) proportion of actors are individual exemplars, typical representatives of a group to which they belong (e.g. a Moroccan girl presented as a typical inhabitant of a particular neighbourhood). A negligible proportion of actors are presented as members or spokesmen of various (migrant) organizations. There are, however, interesting between-groups differences (Table 2).

In general, actors with unknown, unclear or ‘multiple’ ethnicities (Muslims, asylum seekers, Eastern Europeans) are more often presented as groups than as individuals. Members of the Turkish minority enjoy most individualised presentation: 78.3% as compared to 39.5% of the reference category and 27.5% of the asylum seekers. Consequently, the Turkish (Dutch) are least likely and the asylum seekers most likely to be presented as a group actor (13% and 72.5% respectively). Differences in homogeneity of presentation between other groups of migrants and the reference category are statistically non-significant.

Interesting, although difficult to account for, is the frequency of Moroccan actors presented as typical representatives of their ethnic group (twice as often as the reference category). It may indicate an on-going transition in media representation of Moroccans from a fully homogenized generic category of ‘the Moroccans’ into the individualised-type of presentation, enjoyed already by the Turkish migrants, which might, in turn, be symptomatic of the processes of ‘domestication’. However, alternatively, taking into account the overly negative evaluation of Moroccan actors (see below), the high number of (negatively) evaluated individuals that are ‘presented as representative spokespeople of the entire group’ (ter Wal et


141

al. 2005, p. 938), might bias the overall picture and contribute unduly to the detrimental representation of the whole Moroccan population.

Evaluation

The differences in the media representation of various sub-groups become most apparent when we consider the various aspects of their evaluations. The results presented in Table 3 suggest a higher than average degree of negativity attached to the depictions of Moroccans and migrants identified as ‘Muslims’, a fairly positive evaluation of Turks and an ambiguous representation of asylum seekers and Eastern Europeans.

The negativity manifested in the media reconstruction of Moroccans and, to a lesser extent, Muslims is evident from the frequent associations with threatening themes, the higher than average number of perpetrators (in particular terrorists), the higher than average number of explicitly negative claims and the lower than average number of explicitly positive statements. To be more specific, 48.3% of Moroccan actors, as compared to 30.1% of the reference category of ‘other migrants’ and 20.8% of Turkish actors, feature in a news story whose main theme could be regarded as threatening. Moreover, more than three times as many Moroccan (26.7%) or Muslim actors (28.6%) as ‘other migrants’ (8%), and almost seven times as many as Turkish actors (4.2%), are presented in news items on terrorism (conflict and defence). Moroccan actors are also more than four times as often as Turkish actors and asylum seekers presented as perpetrators or public order offenders (35.0%, 8.3% and 7.5%, respectively). Last but not least, Moroccan actors are more than twice as likely as Turkish actors to be subjected to explicitly negative comment (61.5% vs. 25%) and less than half as likely to be awarded a positively-loaded description (20.0% vs. 58.3%). The results for Muslim actors are roughly similar: 51% of Muslim actors are associated with a negatively loaded characteristic and 18.4% with a positive feature or behaviour.

Cha

pter

Fiv

e

142

Tab

le 3

. Diff

eren

ces

in t

he e

valu

atio

n of

var

ious

mig

rant

sub

-gro

ups:

per

cent

age

(abs

olut

e nu

mbe

r) o

f ac

tors

w

ith a

giv

en c

hara

cter

istic

Oth

er

mig

rant

s (r

ef.

grou

p)

Mor

occa

n

T

urki

sh

M

uslim

Asy

lum

se

ekin

g E

aste

rn

Eur

opea

n

N

113

60

24

49

40

25

Eval

uatio

n

Bro

ada

30.1

(34)

48

.3 (2

9)*

20.8

(5)

34.7

(17)

15

.0 (6

) 28

.0 (7

)

Nar

row

b8

(9)

26.7

(16)

**

4.2

(1)

28.6

(14)

**

5.0

(2)

0

Perp

etra

tors

25

.7 (2

9)

35 (2

1)

8.3

(2)

30.6

(15)

7.

5 (3

)*

20.0

(5)

Te

rror

ists

8.

0 (9

) 25

(15)

**

0 24

.5 (1

2)**

2.

5 (1

) 0

B

enef

icia

ries

22.1

(25)

5.

0 (3

)**

12.5

(3)

12.2

(6)

50.0

(20)

**

12.0

(3)

N

egat

ive

eval

uatio

ns

38.9

(44)

61

.7 (3

7)**

25

.0 (6

) 51

.0 (2

5)

47.5

(19)

56

.0 (1

4)

Po

sitiv

e ev

alua

tions

28

.3 (3

2)

20.0

(12)

58

.3 (1

4)**

18

.4 (9

) 15

.0 (6

) 28

.0 (7

)

Not

e: A

naly

sis (

with

the

exce

ptio

n of

‘the

me’

) con

duct

ed o

n th

e su

b-sa

mpl

es o

f non

-inst

itutio

nal a

ctor

s.

a Bro

ad th

eme

incl

udes

them

es re

gard

ing

publ

ic o

rder

, con

flict

and

def

ence

and

dem

onst

ratio

ns.

b Nar

row

them

es in

clud

ed th

emes

rega

rdin

g pu

blic

ord

er a

nd d

emon

stra

tions

. Si

gnifi

canc

e le

vels:

* si

gnifi

cant

at p

=.05

; **

sign

ifica

nt a

t p=.

01 (P

ears

on’s

Chi

-squ

are)

.


143

The negative and positive valuations of asylum seekers and Eastern Europeans are slightly less straightforward and more difficult to account for. Asylum seekers are least frequently presented in a story whose theme can be considered ‘threatening’ in terms of public order and/or security. They are also least often reported as perpetrators of crime and their share of explicitly negative evaluations is close to average (for similar results regarding asylum seekers, see Lubbers et al. 1998). On the other hand, asylum seekers are at least four times more often than other migrant groups, and twice as often as the reference category, represented as actual or potential beneficiaries of public funds, which, in the context of the negatively slanted discussions on the costs of migration and the social security abuse by migrants, might convey an air of negativity. As far as Eastern Europeans are concerned, their evaluation would not deviate much from the allochthon average but for the explicitly negative evaluations, which are more frequent in the case of Eastern European actors (56%). Indeed, Eastern Europeans are the only group analysed that does not differ significantly from the reference category of ‘other migrants’ on any of the variables tested. In statistical terms, differences between Moroccans and the reference category are most pronounced.

The dimensions of evaluations

A closer look at the nature of negative and positive evaluations (Table 4) reveals further inter-group differences. When presented in a negative light, ethnic minorities are portrayed predominantly as lacking warmth: close to 79% of the negatively evaluated migrants are depicted as hostile, aggressive, dishonest, free-loading, etc., an/or associated with ‘illegality’ in residential and/or employment status (specifically, out of 145 migrants evaluated negatively, 114 are presented as lacking warmth). Unfriendliness, aggression or hostility is most often ascribed to Moroccans and Muslims, and the issue of illegality is most frequently raised in the case of Eastern Europeans and asylum seekers (not reported in the table). Of all the sub-groups analysed, Turks are least likely to be presented as ‘cold’ and most likely to be portrayed as ‘warm’.

Chapter Five

144

Table 4. Frequency of various types of evaluations for ethnic group Negative evaluations Positive evaluations none cold incompetent mix Moroccans (N = 60) none 16 20 4 8

warm 2 2 0 0 competent 4 3 0 0 mix 1 0 0 0

Turks (N = 24) none 6 3 1 0


Muslims (N = 49) none 18 19 2 1


Asylum seekers (N = 40) none 18 14 1 1


Eastern Europeans (N = 25) none 6 11 0 1


All other migrants – reference category (N = 120) none 42 33 3 4


Remarkable is the journalistic or editorial reluctance to evaluate (ethnic) minorities along the dimension of competence: evaluations (negative or positive) that pertain to competence constitute roughly a quarter of all evaluations (54 out of 225). Moreover, positive competence evaluations are most often used to describe migrants in the least specific and most


145

generalised reference category of ‘other migrants’. In contrast, migrant groups that are granted particular conspicuousness in the media, such as Moroccans or ethnic-less Muslims, are less often awarded positive competence evaluations.

However, one has to keep in mind that the results with respect to the dimensions of evaluations are based on a qualitative analysis of a relatively small number of evaluative judgements. Particularly problematic for the coding was the inconsistency between the actor-bound characteristics and the context-bound characteristics, i.e. connotations inherent in the message in which the actor was presented (for similar observations related to the coding of the two stereotype dimensions in the media content, see van Selm, Westerhof, and Wester 2009). While character-bound descriptions constituted the first-choice basis for the current coding, the context of reporting should not be ignored. After all, negative perceptions of and feelings towards the object of description may arise not only from the explicitly negative statements but also from the sheer association with negatively evaluated social phenomena, such as conflict, drama, controversy, violence or deviant behaviour (e.g. d’Haenens and Bink 2006). The results with respect to the content of stereotypes should therefore be interpreted in combination with other indicators of negative and positive imaging.

How well-founded are the media depictions?

The question remains to what extent the evaluations that are upheld in the Dutch news media with regard to ethnic minorities constitute a reflection of social reality, derived from the actual behaviours and status characteristics of the relevant groups. Table 5 presents some figures regarding the profile and performance of two ethnic groups whose evaluation in the media is most contrasting, the Turks and the Moroccans.

The figures clearly show that ‘in the real world’ the two minorities show considerable resemblance on a number of socio-economic characteristics: the educational level, labour market participation and unemployment, income and poverty levels. In fact, the two minorities vary substantially only with respect to figures reflecting their entrepreneurship (percentage self-employed) and involvement in criminal activities (percentage criminal suspects). Turks appear to be more entrepreneurial than migrants of Moroccan descent (7% vs. 4% self-employed, respectively) and less prone to criminal behaviour (3.1% vs. 5.3%, respectively). On the other hand, even under the assumption that these two aspects of social reality –

Chapter Five

146

entrepreneurship and involvement in criminal activities – constitute a sufficient basis for the evaluation of the two groups, the media depictions appear grossly exaggerated in both the positive and the negative descriptions of Turks and Moroccans.

Table 5. Official statistics on selected characteristics of Turks and Moroccans

Turks Moroccans Labour market situation Self-employed (2008) 7.0% 4.0% Employment rate (2010) 52.0% 48.0% Unemployment rate (2010) 11.3% 14.6% In high status jobs (2010) 14.0% 17.0% In low status jobs (2010) 56.0% 53.0% Economic situation Standardized household income

(2009) € 17.000 € 16.600

Living in a household below the poverty line (2010)

18.8% 22.0%

Receiving social assistance benefits (2010)

9.3% 13.7%

Education level Secondary (vocational) education

(2011) 11% 14%

Higher (vocational) education (2011)

28% 29%

Criminal suspects (2009) 3.1% 5.3% Source: Gijsberts, Huijnk, and Dagevos (2012) While there are 1.7 times more criminal suspects among migrants of Moroccan origin than among migrants of Turkish origin, media reports appear to suggest that the difference is 4 times bigger. Moreover, the number of criminal suspects among Moroccans seems much more exaggerated than the number of criminal suspects among migrants of Turkish origin. In the case of Turks, media report their involvement in criminal acts 2.7 more often than it is registered in official statistics; in the case of Moroccans 6.6 times more often.

Similarly biased seems the media account of the competence of the two groups. The evaluations alluding to competence (coded as ‘competence’ and ‘mixed warmth and competence’) are applied to 33% of Turkish and


147

13% of Moroccan actors. Yet, the difference in competence that could be assumed from the official statistics is much less obvious. While Moroccans are indeed almost twice less likely to own their own company (4% of the Moroccans vs.7% of the Turks) and slightly less likely to be employed (the employment rate of Moroccans is 48% as compared to 52% of Turks), they are more likely than the Turks to have a high status job (17% vs. 14%) and slightly less likely to hold a low status job (53% vs. 56%). Also, as reported by Gijsberts and Iedema (2012), Moroccans constitute the migrant group whose intergenerational educational advancement is the greatest, exceeding the advancement of the Turks.4

Summary and discussion: the (un)varying faces of allochthony

Founded on the assumption of a relationship between the content of stereotypes and the nature of social interactions, on the one hand, and the active role of the media in the formation and/or maintenance of stereotypes, on the other, this study investigated the news media images of different ethnic minorities in the Netherlands.

The analysis of 215 news items from five news sources showed an imbalance in the media presence of various groups relative to their share in the Dutch population. While certain groups are granted particular conspicuousness (Moroccans and asylum seekers), others (particularly migrants from the Dutch Antilles, Aruba and Suriname) are grossly underrepresented. While those differences are partly attributable to the sampling bias, they corroborate the results of previous research (e.g. Lubbers et al. 1998).

As far as the content of the presentation is concerned, the analyses showed noteworthy discrepancies in the media coverage of groups investigated: Moroccans, Turks, Muslims, asylum seekers and Eastern Europeans. Most striking are differences in the presentation of Turks and Moroccans – two groups of migrants with a similar religious background, a comparable migration history and a similar degree of cultural conspicuousness in Dutch society. While Turkish characters were relatively out of sight, they enjoyed considerable prominence in the news, measured by their relative within-story importance and the number of characters quoted. Their

4 Admittedly, the Moroccan minority has the highest share that has never gone to school, twice as large as among the Turks (Gijsberts and Iedema, 2012).

Chapter Five

148

presentation was also most favourable: they were least often associated with negative themes or evaluations, and they were granted the greatest share of positive evaluations. Moroccan actors, on the other hand, were awarded exceptional visibility, measured by their overall presence in the news and featuring in a headline or front-page story, yet were not particularly prominent. They were also most often associated with threatening and/or otherwise negative evaluations or themes.

While notable, the differences in the presentation of Turks and Moroccans have not been entirely unanticipated. As already observed by Sniderman and Hagendoorn (2007) on the basis of a national survey conducted in 1998, Turkish immigrants were perceived by the Dutch majority slightly more positively than Moroccans, who ‘are more likely than other minorities [i.e. Turks, Surinamese and refugees] to be characterised as dishonest, violent, intrusive, and not law-abiding’ (p. 48). However, since Sniderman and Hagendoorn observed only ‘minor’ differences in social perceptions of the various migrant groups, the scope of difference in the media construction of Turks and Moroccans is rather surprising. It seems to suggest a growing rupture in the social acceptability (or even desirability) of the two groups, or alternatively, a gap between popular perceptions and media depictions.

Contrasts in the media depiction of other migrant sub-groups were less pronounced. Most interesting is a distinct character of negativity that could be traced in the descriptions of asylum seekers. In comparison to other non-ethnically Dutch actors, asylum seekers were (in 2010) less often associated with criminal or otherwise threatening behaviour, yet more often presented in terms of costs and burdens for the Dutch state and society. They were also slightly more often presented as competent.

Finally, Eastern Europeans appear to resemble most the reference category of ‘other migrants’. Some differences (albeit statistically non-significant) were observable only in case of negative evaluations: Eastern Europeans were more often described in negative terms than ‘other migrants’: they were more often associated with being cold or incompetent and less frequently described as competent. In relation to other groups, Eastern Europeans seem to be situated somewhere in between Turks and asylum seekers, on the one hand, and Muslims and Moroccans, on the other.

If interpreted in the light of the socio-psychological literature, and in particular the literature on social distance, the perceptual hyper-visibility of certain groups of migrants (Moroccans and Muslims), their homogenization


149

(Muslims, asylum seekers and Eastern Europeans) and/or (predominant) negativity of depiction (Moroccans, Muslims and asylum seekers) could be indicative of their (mediated) estrangement and (cultural) ‘otherness’ and suggestive of social distance between them and the (social or cultural) majority (Fine 1994; Krumer-Nevo and Orly 2010; Sniderman and Hagendoorn 2007; cf. Tajfel 1982).5 Indeed, a study by Lubbers et al. (1998) on the Dutch media’s depiction of ethnic minorities showed that the media devote more attention to groups that occupy a lower, or more distant, place in the socially shared ethnic hierarchy and that there is a correspondence between the place the group occupies in the hierarchy and the media tendency to associate it with problems, and crime in particular. It seems, therefore, that hyper-visibility of certain groups in the news media, Moroccans in particular, is inseparable from their negative evaluation and, at the same time, linked to their more distant position in the ethnic hierarchy.

Further light on the media practices of ‘othering’ of ethnic minorities can be shed by analysing the content of group-specific evaluations. Particularly interesting in our results is the predominance of evaluations with respect to the dimension of ‘warmth’ coupled with relative infrequency of evaluations pertaining to the dimension of ‘competence’. This is not a trivial finding.

First, according to social psychologists (e.g., Phalet and Poppe 1997; Wojciszke 1994), in-group members tend to be evaluated primarily with respect to competence, while the evaluation of the members of the out-group appears to be determined by their perceived warmth. The warm-competence imbalance in the presentation of all ethnic groups that is concealed in the media messages, might therefore be seen as a reflection of ‘distancing’ of the non-native population; distancing that seems slightly less apparent in the case of the Turkish minority.

Second, research on the content of stereotypes shows that ‘a person perceived as cold may sometimes behave in moral-sociable ways, but will continue to be perceived as unfriendly and untrustworthy’ (Cuddy et al. 5 Even though the concept of social distance remains routinely beyond the theoretical framework of media analysts, a number of studies on the media construction of various social groups draw attention to the media practices of ‘othering’, that is, practices of setting a social group apart as different. The practices of ‘othering’ identified by media analysts strongly resemble the strategies of ‘distancing’ identified by researchers of social distance. For a discussion, see Lepianka (2015).

Chapter Five

150

2008, p. 92). At the same time, ‘a person perceived as incompetent, who presumably lacks the ability, can never behave competently without challenging the perceived incompetence’ (ibid). Moreover, researchers draw attention to the ‘contagious’ nature of ‘coldness’ and ‘competence’, which tends to pass from one ‘infected’ individual to other individuals in one’s network. It might thus be enough to know just one hostile or non-benevolent member of an ethnic minority to start questioning the goodwill of other members of this minority in one’s network. On the other hand, it suffices to know just one competent minority member to begin to doubt the incompetence of all members of minority in one’s network. The strength and infectious nature of stereotypes related to judgements of hostility and non-benevolence are therefore of paramount importance. It not only anticipates the difficulty minority groups face when trying to break the negative stereotype but implicates as well the stability, if not permanence, of their ‘otherness’ and ‘they-dom’. Our findings indicate that breaking the negative stereotypes might be easier for Turks than for other minority members.

Third, judgements of warmth are more likely to induce specific emotions and behavioural reactions than judgements of competence (Lee and Fiske 2006). Thus, a group that is evaluated as warm is more likely to receive facilitation than a group evaluated as competent. By the same token, a group that is seen as cold is more likely to experience harm than a group evaluated as incompetent. With respect to our findings, this might imply that negative media images of specific ethnic and/or religious minorities are more likely to induce harassment or even violence in case of Moroccans, ethnic-less Muslims, asylum seekers and Eastern Europeans than in case of Turks. In addition, members of the Turkish minority might be more likely to experience facilitation than other minority groups under study.

To summarise, the analyses reported in this chapter indicate noteworthy differences in the Dutch media representation of ethnic minorities that might contribute to the collective drawing and/or reinforcing of group boundaries. While the differences might be sometimes inconspicuous, they are likely to have a lasting influence on the creation and/or maintenance of group relations, including readiness to engage in acts of solidarity. Particularly noteworthy is the relatively privileged position of Turks, who enjoy a much more positive, or less negative, image than other ethnic minorities, including those who – like Moroccans – share a similar religious background and a similar history of settlement in the Netherlands. Such positive stereotyping of just one out-group (in this case


151

Turks) and setting them apart as a ‘model minority’ may contribute to the relative diminishing, and as a result, further discrimination, of another out-group.

References

Braun, R., and Koopmans, R. (2010). The Diffusion of Ethnic Violence in Germany: The Role of Social Similarity. European Sociological Review, 26, 111-123.

Bullock H., Wyche, K., and Williams, W. (2001). Poverty as we know it. Media images of the poor. Journal of Social Issues, 57, 229-246.

Centraal Bureau voor de Statistiek (2000). Hoe doet het CBS dat nou? Standaarddefinitie allochtonen. Index 10, 24-25. Retrieved from: http://www.cbs.nl/NR/rdonlyres/26785779-AAFE-4B39-AD07-59F34DCD44C8/0/index1119.pdf.

Clawson, R. A., and Trice, R. (2000). Poverty as we know it. Media portrayals of the poor. Public Opinion Quarterly, 64, 53-64.

Cuddy, A., Fiske, S., and Glick, P. (2008). Warmth and Competence as Universal Dimensions of Social Perception: The Stereotype Content Model and the BIAS Map. In M. Zanna (Ed.), Advances in Experimental Social Psychology (Vol. 40). Elsevier/ Academic Press.

Dagevos, J., and Gijsberts, M. (2009). Sociaal-culturele positie. In M. Gijsberts and J. Dagevos (Eds.), Jaarrapport integratie 2009 (179-102). Den Haag: Sociaal en Cultureel Planbureau.

de Bruin, J. (2005). Multicultureel drama? Populair Nederlands televisiedrama, jeugd en etniciteit. Amsterdam: Otto Cramwinkel.

de Lange, M., and d’Haenens, L. (2002). Het beeld van asielzoekers in Nederlandse regionale kranten. Tijdschrift voor Communicatiewetenschap, 30, 77-94.

der Valk, I. (2002). The Netherlands. In J. ter Wal (Ed.), Racism and Cultural Diversity in the Mass Media. An overview of research and examples of good practice in the EU Member States, 1995-2000 (287-309). Vienna: EUMC.

d’Haenens, L., and Bink, S. (2006). Islam in de Nederlandse media: Focus op het Algemeen Dagblad. Tijdschrift voor Communicatiewetenschap, 34, 351-367.

d’Haenens, L., and de Lange, M. (2001). Framing asylum seekers in Dutch regional newspapers. Media, Culture and Society, 23, 847-860.

d’Haenens, L., and Ogan, C. (2007). Introduction to the Special Issue: Media and ethnic minorities in Europe. Communications, 32, 137-140.

Chapter Five

152

Entman, R. (1989). How the Media Affect What People Think: An Information Processing Approach. The Journal of Politics, 51, 347-370.

Fine, M. (1994). Working with Hyphens: Reinventing Self and Other in Qualitative Research. In N. Denzin and Y. Lincoln (Eds.), Handbook of Qualitative Research (70-102). London: Sage.

Fiske, S., Cuddy, A., and Glick P. (2006). Universal dimensions of social cognition: warmth and competence. TRENDS in Cognitive Sciences, 11, 77-83.

Gijsberts, M., Huijnk, W., and Dagevos, J. (Eds.). (2012). Jaarrapport integratie 2011. Den Haag: Sociaal en Cultureel Planbureau.

Gijsberts, M., and Iedema, J. (2012). Opleidingsniveau van niet-schoolgaanden en leerprestaties in het basisonderwijs. In M. Gijsberts, W. Huijnk and J. Dagevos (Eds.), Jaarrapport integratie 2011 (76-101). Den Haag: Sociaal en Cultureel Planbureau.

Gilens, M. (1996). Race and poverty in America. Public misperceptions and the American news media. Public Opinion Quarterly, 60, 515-541.

—. (1999). Why Americans hate welfare. Race, media and the politics of anti-poverty policy. Chicago: University of Chicago Press.

Golding, P., and Middleton, S. (1982). Images of welfare. Press and public attitudes to poverty. Oxford: Basil Blackwell.

Kennamer, J. D. (1994). Public opinion, the press and public policy: an introduction. In J. D. Kennamer (Ed.), Public opinion, the press and public policy (1-18). Westport: Praeger.

Koeman, J., Peeters, A., and d’Haenens, L. (2007). Diversity Monitor 2005: Diversity as a quality aspect of television in the Netherlands. Communications, 32, 97–121.

Korteweg, A., and Yurdakul, G. (2009). Islam, gender, and immigrant integration: boundary drawing in discourses on honour killing in the Netherlands and Germany. Ethnic and Racial Studies, 32, 218-238.

Krumer-Nevo, M., and Orly, B. (2010). Critical Poverty Knowledge: Contesting Othering and Social Distancing. Current Sociology, 58, 693-714.

Lee, T., and Fiske, S. (2006). Not an outgroup, not yet an ingroup: Immigrants in the Stereotype Content Model. International Journal of Intercultural Relations, 30, 751–768.

Lemert, J. (1994). Effective public opinion. In J. D. Kennamer (Ed.), Public opinion, the press and public policy (41-62). Westport: Praeger.

Lepianka, D. (2015). Dutch media depictions of older and younger people. Ageing and Society, 35, 1095-1113.


153

Leurdijk, A. (1997). Common sense versus political discourse. Debating racism and multicultural society in Dutch talk shows. European Journal of Communication, 12, 147-168.

Lubbers, M., Scheepers, P., and Wester, F. (1998). Ethnic minorities in Dutch newspapers 1990-5. Patterns of criminalization and problematization. International Communication Gazette, 60, 415-431.

McCombs, M., and Reynolds, A. (2009). How the News Shapes our Civic Agenda. In J. Bryant and M. Oliver (Eds.), Media effects. Advances in theory and research (1-16.). New York: Routledge.

McQuail, D. (2010). McQuail’s mass communication theory. Los Angeles: Sage.

Neuendorf, K. (2002). The content analysis guidebook. Thousand Oaks/ London/New Delhi: Sage.

Peeters, A. L., and d’Haenens, L. (2005). Bridging or bonding? Relationship between integration and media use among ethnic minorities in the Netherlands. Communications 30, 201-231.

Phalet K., and Poppe, E. (1997). Competence and morality dimensions of national and ethnic stereotypes. A study in six Eastern-European countries. European Journal of Social Psychology, 27, 703-723.

Phalet, K., and ter Wal, J. (2004). Moslim in Nederland. Den Haag: Sociaal en Cultureel Planbureau.

Ramasubramanian, S., and Oliver, M. (2007). Activating and suppressing hostile and benevolent racism. Evidence for comparative media stereotyping. Media Psychology, 9, 623–646.

Saharso, S., and Lettinga, D. (2008). Contentious citizenship. Policies and debates on the veil in the Netherlands. Social Politics, 15, 455-480.

Sniderman, P., and Hagendoorn, L. (2007). When ways of life collide. Multiculturalism and its discontents in the Netherlands. Princeton/Oxford: Princeton University Press.

ter Wal, J., d’Haenens, L., and Koeman, J. (2005). (Re)presentation of ethnicity in EU and Dutch domestic news: a quantitative analysis. Media, Culture and Society, 27, 937-950.

Tajfel, H. (1982). Social psychology of intergroup relations. Annual Review of Psychology, 33: 1-39.

Tipton, L. (1994). Reporting on the public mind. In J. D. Kennamer (Ed.), Public opinion, the press and public policy (131-144). Westport: Praeger.

Thompson, J. (1995). The media and modernity. A social theory of the media. Cambridge: Polity Press.

Chapter Five

154

van Ginneken, J. (2003). Collective behaviour and public opinion. Rapid shifts in opinion and communication. London: Lawrence Erlbaum Associates Publishers.

van Selm M., Westerhof, G., and Wester, F. (2009). Test cases, metaphors and role models: elderly people in television commercials. Unpublished Manuscript.

Vliegenthart, R., and Boomgaarden, H.G. (2007). Real-world indicators and the coverage of immigration and the integration of minorities in Dutch newspapers. European Journal of Communication, 22, 293-314.

Wester, F., Pleijter, A., and Renckstroft, K. (2004). Exploring newspapers’ portrayals; a logic for interpretative content analysis. Communications, 29, 495-513.

Wojciszke, B. (1994). Multiple meanings of behaviour. Construing actions in terms of competence or morality. Journal of Personality and Social Psychology, 67, 222-232.

CHAPTER SIX

SOLIDARITY IN A MULTICULTURAL NEIGHBOURHOOD:

RESULTS OF A FIELD EXPERIMENT

PAUL DE BEER AND MAARTEN BERG

Introduction

As discussed in chapter 2, from a theoretical point of view there is little reason to expect a simple, linear negative relationship between ethnic diversity and solidarity. First, it is important to distinguish between the (aggregate) effect of ethnic differences between individual people and the effect of the ethnic composition of a community on solidarity between individual persons, irrespective of their ethnicity. This distinction is crucial for the claim of Putnam (2007) that ethnic diversity may also hurt within-group solidarity, i.e. solidarity between members of the same ethnic group.

Secondly, although very large differences between people may indeed impair solidarity, this does not imply that solidarity will be at its maximum if there are no differences at all. On the contrary, some difference between the benefactor and the beneficiary is generally indispensable for a solidary act to make sense at all. The reason to act solidary is usually that the beneficiary needs some kind of assistance which the benefactor is able and willing to provide. Although this need is usually not directly related to ethnicity, it is possible that ethnicity is correlated with particular needs and skills which foster instead of hinder solidarity. Consequently, it is an empirical question whether there is a relation between ethnic diversity and solidarity and what this relationship looks like.

For the study that is reported in this chapter, we have selected the multicultural working-class neighbourhood Dapperbuurt in Amsterdam as

Chapter Six

156

an interesting case to study how residents with different ethnic and cultural backgrounds interact in everyday life, in particular around the local market. We tested the impact of ethnic diversity on solidarity by conducting a field experiment with citizens who vary in ethnic background in this neighbourhood. Our experiment allows us to test the impact of differences in ethnic background between two participants as well as the impact of the diversity of the group in which they participate on solidarity between two participants. Thus, we are able to test whether group diversity has an independent and different effect on solidarity than the differences between individuals.

We will first introduce the concept of an experimental game and explain the solidarity game that we conducted in the Dapper area. We formulate our main research question and state three hypotheses that will be tested in order to examine the validity of two alternative theories, the conflict theory and the constrict theory. Next, we describe the methodology and the procedure of the experiment in some detail. In the following section we present the results of the experiment. First, we analyse the effect of ethnic differences between two participants and, next, the effect of the ethnic composition of the group in which the participants play the game. The chapter ends with some conclusions regarding the relationship between group diversity and solidarity.

A solidarity game experiment

The solidarity game

The experiment on which this chapter reports, is embedded in the so-called game theory, the discipline of interactive decision theory. A ‘game’ refers to a human interaction that is guided by certain rules that are known by the players. The players know how their decisions, the decisions of others and, sometimes, chance determine the outcome of the game. The interactive element of game theory implies a theoretical focus on cooperation and/or conflict. This can be seen, for example, in the famous prisoner’s dilemma that showed that two people might not cooperate, even though it would be in the interest of both to do so.

Game theory has been used as a way of describing and modelling actual human behaviour in real populations. This approach has been increasingly subject to criticism, as the underlying ‘homo economicus perspective’ has come more and more under attack (e.g. Sen, 1977; Kahneman and Tversky, 1979; Frey, 1997; Sunstein and Thaler, 2008). Human behaviour appears

Solidarity in a Multicultural Neighbourhood

157

to be not only motivated by maximizing individual returns (as is assumed by many of the earlier game theorists), but also by altruistic, other-regarding motives (e.g. Andreoni, 1989; Rabin, 1993; Fehr and Schmidt, 1999; Bolton and Ockenfels, 2000). Recent research therefore focuses both on self-interest and on voluntary sharing and the conditions under which they are prevalent (Smith, 1989; Camerer, 1999). This marks a shift from theoretical game theory to an experimental approach of real human behaviour.

An example of this approach is the solidarity game that was introduced by Selten and Ockenfels (1998). In this game three players partake, who have a 2/3 chance of winning 10 Deutsch Mark (about 5 euro) each and a 1/3 chance of winning nothing. As the outcomes for each player are independent, there are four possible outcomes to the game, viz. no winner, one winner, two winners and three winners. Each player has to state in advance how much (if anything) s/he is willing to share if s/he is the only winner and how much (if anything) s/he is willing to share if s/he is one of two winners. In the case of zero or three winners, the game stops. Defying the assumption of pure self-interest, the great majority of subjects are willing to make substantial (conditional) gifts. The variation in gifts, however, makes it difficult to interpret the results in a straightforward way. A lot of additional research is required to analyse the conditions of voluntary sharing.

The current game is a variant of the original solidarity game of Selten and Ockenfels (1998). The game allows us to study four types of conditions (mentioned below) which may affect voluntary solidarity. Our solidarity game is in several respects an extension of the Selten and Ockenfels game. First, the groups consist of four players rather than three, and there are always two winners and two losers. This enables us to study solidarity both from an intergroup perspective (solidarity between winners and losers) and an intragroup perspective (solidarity among winners). The behaviour of a winner cannot only be influenced by her/his own norms and by her/his perception of the losers, but also by the behaviour of the other winner. Secondly, the dictator role was not always assigned randomly, but in half of the experimental conditions the two players who performed the best on a general knowledge quiz were appointed as winners. This quiz functions as a way to model the role of performance (related to intelligence and earning capacity) in the real world. Thirdly, in half of the conditions, the game was played for four consecutive rounds (with the same players) rather than one. Finally, we informed the participants in the game about some characteristics of their co-players, so that we can test

Chapter Six

158

whether the willingness to share with other players depends on the characteristics of the other player or of the group as a whole. In this way, we examine to what extent diversity of a group affects solidary behaviour.

Main question and hypotheses

Although we include an analysis of the effects of the conditions and the outcome of the game on the sharing between players, our main interest in this chapter is to what extent differences in characteristics between the players and the diversity of a group affects the willingness to share with each other within the group

In a much cited and discussed article, ‘E pluribus unum’, Robert Putnam (2007) claims that “immigration and ethnic diversity tend to reduce social solidarity and social capital.” More specifically, “in ethnically diverse neighbourhoods residents of all races tend to ‘hunker down’. Trust (even of one’s own race) is lower, altruism and community cooperation rarer, friends fewer.” Other scholars have tried to replicate Putnam’s study of American neighbourhoods in other geographical areas, such as Dutch neighbourhoods (Lancee and Dronkers 2008) and European countries (Gesthuizen, Van der Meer and Scheepers 2009; Hooghe, Reeskers, Stolle and Trappers 2009). Some of these follow-up studies support Putnam’s results (Lancee and Dronkers 2008), while others refute them (Gesthuizen et al. 2009) or qualify them (Hooghe et al. 2009). A common element of these studies is that they focus on trust in other people as the main indicator for social solidarity or social capital. It is then often assumed, but seldom tested, that a lower trust in other people will result in less pro-social behaviour, such as helping neighbours or contributing to voluntary work. Since our experiment allows us to observe actual behaviour of participants in which real money is at stake, it offers valuable additional information on the influence of diversity on solidarity. Moreover, since the choice of the neighbourhood in which one lives is, at least to some extent, voluntary, the differences that are observed between the residents of different neighbourhoods may partly be caused by self-selection. People living in ethnically diverse neighbourhoods may differ in some unobserved characteristics from people living in homogeneous neighbourhoods. Thus, it is difficult to establish whether the causal effect runs from the composition of the neighbourhood to individual attitudes and behaviour or in the opposite direction. Since the participants in our experiment were assigned quasi-randomly to groups that differ in heterogeneity, there is no ambiguity about the direction of the causal relationship.


159

Putnam discusses two theories that may explain why there is less social solidarity or social capital in ethnically diverse communities. According to the conflict theory, members from different (ethnic) groups compete for scarce resources and have, consequently, conflicting interests. Thus, members from different groups will trust each other less than members from the same group and, as a result, general trust will be less in a diverse community. However, in this theory it is assumed that in-group trust is not (negatively) affected by diversity or may actually be strengthened by the negative attitudes towards the out-group.

As opposed to this theory, Putnam proposes the constrict theory, which states that ethnic diversity may also erode in-group trust. In an ethnically diverse setting, people may ‘hunker down’ or ‘pull in like a turtle’ (Putnam 2007: 149), which means that they also trust members of their own group less than in an ethnically homogenous setting.

With our experiment in the Dapper area we are able to test both the conflict theory and the constrict theory from a new angle. First, if the conflict theory is correct, we expect less solidarity between participants who belong to different ethnic groups than between participants of the same ethnic group (hypothesis 1). However, this hypothesis is not necessarily inconsistent with the constrict theory. Consequently, if this hypothesis is confirmed, this would offer support for the conflict theory, but it would not prove that the constrict theory is wrong. Conversely, if this hypothesis is rejected, this would be strong evidence against the conflict theory, but not against the constrict theory.

Secondly, we test the effect of group diversity on solidarity. According to the constrict theory, solidarity will be lower in an ethnically diverse community than in a homogenous community, since diversity will result in less solidarity between (ethnic) groups as well as within groups (hypothesis 2). If this theory is confirmed, this would support the constrict theory, but it would not be inconsistent with the conflict theory. However, if the hypothesis is rejected, this would refute the constrict theory, but not necessarily the conflict theory, since group diversity might strengthen intra-group solidarity and, accordingly, need not harm total solidarity.

Thus, hypotheses 1 and 2 can only provide evidence against one of the two theories, but they cannot prove which theory is correct.

Therefore, we will test a third hypothesis which clearly distinguishes between both theories. This hypothesis states that there is a negative

Chapter Six

160

interaction effect between group diversity and solidarity between two individual players. If the conflict theory is correct, we expect a negative effect of group diversity only on solidarity between players from different ethnic groups (hypothesis 3). If the constrict theory is correct, there will also be a negative interaction effect between group diversity and solidarity between players from the same ethnic group. Put differently, according to the conflict theory, group diversity will result in discrimination against subjects with another (ethnic or cultural) background, while the constrict theory claims that group diversity will result in less overall solidarity, but not specifically because of discrimination against subjects with another background.

In addition to the many studies on ethnic diversity, we also test whether other kinds of diversity have similar effects. A priori, it is not self-evident that ethnic diversity has a different effect than, for example, gender diversity or age diversity. Therefore, we will study the possible consequences of these other kinds of diversity simultaneously with ethnic diversity.

Thus, the hypotheses we will test are the following.

Hypothesis 1 (conflict theory): a player shares less with a co-player who differs from her/him in various characteristics than with a co-player who has similar characteristics.

Hypothesis 2 (constrict theory): the larger the diversity of a group, the less players share with each other.

Hypothesis 3 (conflict versus constrict theory): the larger the diversity of a group, the less a player shares with a co-player who differs from her/him in various characteristics.

Data and method

Participants

In our experiment, 180 visitors of the Dapper market in Amsterdam were recruited for immediate participation. All participants received 7 euros for the one shot game and 10 euros for the four shots game (because it took more of their time) as a show-up fee. On top of that, each group of four players earned 40 additional euros. The average player earned 17 euros (including the show up fee) in the one shot game and 20 euros in the four


161

shots game, but actual earnings varied between players and depended on performance, chance and/or the sharing behaviour of the various group members.

Table 1 shows some background characteristics of the participants. Taking the population of Amsterdam as a reference point,1 there is a large overrepresentation of women in our sample. For the other variables, the discrepancy between our sample and the actual population is smaller. Compared to visitors from other neighbourhoods, subjects living in the Dapper-area were dominantly (even more) left-wing oriented.

Table 1. Distribution of subject characteristics Sex Males

(n=52) 28.9%

Females (n=128) 71.1%

Ethnical background

Natives (n=122) 67.8%

Non-natives (n=58) 22.2%

Age 16-19 (n=12) 6.7%

20-29 (n=34) 18.9%

30-39 (n=24) 13.3%

40-49 (n=35) 19.4%

50-59 (n=41) 22.8%

>= 60 (n=34) 18.9%

Political orientation

Left2 (n=99) 78.0%

Right (n=28) 22.0%

Missing (n=53)

Marital status Married (n=42) 23.3%

Cohabiting (n=25) 13.9%

LAT (n=12) 6.7%

Single (n=101) 56.1%

Neighbourhood From Dapper area (n=73)

Not from Dapper area (n=107)

Procedure

A field laboratory was created close to the Dapper market in Amsterdam. There were eight connected computers, making it possible for two groups 1 However, one should recognize that visitors to the Dapper market need not be residents of Amsterdam. 2 People were categorized as ‘left’ if they voted PvdA, SP, GL, D66, Partij voor de Dieren, or CU during the last election, and as ‘right’ if they voted VVD, CDA, PVV or SGP.

Chapter Six

162

of four to play the ‘solidarity game’ simultaneously. Cabins were created to replicate the (anonymous) conditions of an experiment in the laboratory as good as possible, although it was not feasible to completely meet these standards. This was not really problematic when two groups were playing simultaneously (creating confusion about the other members in the groups), but somewhat more problematic when only one group was playing.

After entering the experimental room and sitting behind their own computer, the subjects were asked some questions to generate a profile for each player, consisting of information about the players’ cultural background, age, sex, marital status and the number of years s/he had been living in the Dapper area. This information about all fellow players was visible on the computer screen during the distribution phases of the experiment.

Next, the subjects received further instructions about the experiment. They were told that they would have to participate in a general knowledge quiz (history, science, music, sports, etc.), consisting of ten multiple (four) choice questions. In the second phase, subjects had to indicate how many correct answers (0-10) they expected to have. Also they had to estimate the average number of correct answers of their fellow three players.

In the third phase, subjects had to indicate how much they would be willing to share (of 20 credits) if they would hypothetically (not yet knowing the scores on the quiz) belong to the two winners. In total, 40 credits were shared (20 credits by each winner). The subjects were told that any distribution was allowed. They could keep all 20 credits to themselves, give everything away, discriminate between the other players or treat the other players the same. They were also told that the credits reflected real money, and that they would receive this money immediately after the experiment. In the fourth phase, the results from the quiz were made public. They were also told whether or not they belonged to the two winners. In the fifth phase, only the two winners had to indicate (for the second time) how they wanted to share 20 credits. They were allowed to choose the same distribution as in phase 3, but they were also allowed to make different choices. The two losers did not have to do anything during this phase. In the sixth phase, it was randomly decided by the computer what distributions (those of phase 3 or those of phase 5) would be used. A total distribution was calculated, adding up the relevant distributions of the two winners. Each player was informed about her/his earnings and about the players responsible for these earnings.


163

In some conditions, there were four shots rather than one shot. In the case of four rounds, exactly the same six phases were repeated four times, although instructions were only given in the first round. The two winners might be different for every round, but could also be the same. Each player could win zero, one, two, three or four rounds.

Design and independent variables

Based on the various conditions, the study uses a 2x2-design (Figure 1). The participants were quasi-randomly assigned to one of the four experimental conditions. As in many earlier game theoretical experiments, the subjects of our solidarity game played for real money, giving them a strong incentive to take the experiment seriously. When real money is at stake, sharing behaviour will in general be more informative and less gratuitous, than when subjects are asked about their willingness to act solidary in survey research.

Figure 1: The four conditions of the solidarity game One shot Four shots Performance based Condition 1 Condition 3 Random based Condition 2 Condition 4 In the field study it was difficult to assign the subjects completely randomly to the various conditions of the game. The subjects who were recruited from the market were told in advance about the time investment in the experiment. As the ‘four shots conditions’ took a time investment of about 45 minutes, compared to 25 minutes in the ‘one shot conditions’, and only a minority of all visitors of the Dapper market agreed to participate, it is theoretically possible that the ‘four-shots conditions’ attracted a different pool of subjects than the ‘one shot conditions’.

In the ‘one shot conditions’ an experimental credit had a 1:1 exchange rate with a euro, whereas a credit was worth 25 eurocent in the ‘four shots conditions’. So in all conditions the average earnings per player (excluding the show-up fee of 7 or 10 euros) were 10 euros.

Dependent variable and method of analysis

The dependent variable is the degree of voluntary sharing, as operationalized by the number of credits that players (are willing to) give to other players. In our regression analyses, each actual or intended gift to another player is

Chapter Six

164

taken as a separate observation. Thus, a gift is a dyadic relation between a donor and a (potential) beneficiary. Accordingly, the number of observations for each player varies between three (a loser in a one-shot-game who can only once declare her/his or her intention to donate money to the other three players) and 24 (a person who is a winner in each round of a four-shot-game and, hence, can declare eight consecutive times what s/he intends to give to the other three players). The total number of observations is 1,890 (divided among 180 persons).

Results

Basic analysis

Table 2 displays the average (intended) gift for each combination of conditions, rounds and distribution of the game. An average gift amounts to 2.81 credits, which means that the players (are willing to) donate on average an amount of 8.44 credits from their 20 credits (42.2%) to the other players. The differences between the various conditions of the game are rather small and statistically not significant. Thus, the players are almost insensitive to the kind of game and display a general solidary attitude that is not dependent on the specific conditions. We do not find that players give more in the random based game than in the performance based game, neither do they give more in the first round of the four shot game than in the one shot game. We do, however, find that the average gift declines as the four shot game progresses, though this effect is rather small (and not significant). In general, the winners give less in the second distribution after they have been announced than before they knew they were winners, but again, this effect is small and statistically not significant.

Since the various gifts of one particular player are likely to be correlated, we use a multilevel approach to analyse the determinants of an (intended) gift. The separate gifts are nested within an individual player. Moreover, the gifts of the players of a particular game may also be correlated, for example because the composition of the group affects the giving behaviour of all participants in a particular way. Consequently, the four players of a game are nested within a group. In addition, the gifts within a particular round of the four-shots game may be correlated as well as the gifts within the same distribution (first or second) of a particular round. Consequently, in the multilevel analysis we take account of possible correlations within distributions, within rounds, within individuals and within groups.

Solid

arity

in a

Mul

ticul

tura

l Nei

ghbo

urho

od

16

5

Tab

le 2

. (In

tend

ed) g

ift b

y co

nditi

ons o

f the

gam

e

T

otal

Pe

rfor

man

ce b

ased

Ran

dom

bas

ed

D

istri

butio

n M

ean

N

Std.

D

evia

tion

Mea

n N

St

d.

Dev

iatio

n M

ean

N

Std.

D

evia

tion

T

otal

To

tal

2.81

18

90

2.36

2.87

99

0 2.

34

2.

75

900

2.38

1 2.

87

1260

2.

34

2.

95

660

2.32

2.79

60

0 2.

36

2

2.70

63

0 2.

39

2.

71

330

2.35

2.68

30

0 2.

43

1

shot

gam

e To

tal

3.24

45

0 2.

19

3.

12

270

2.23

3.43

18

0 2.

11

1

3.39

30

0 2.

15

3.

41

180

2.23

3.38

12

0 2.

03

2

2.95

15

0 2.

24

2.

56

90

2.14

3.53

6

0 2.

27

4

shot

s gam

e

ro

und

1 To

tal

2.99

36

0 2.

35

3.

13

180

2.42

2.85

18

0 2.

28

1

2.98

24

0 2.

34

3.

04

120

2.38

2.91

12

0 2.

31

2

3.03

12

0 2.

38

3.

32

60

2.50

2.73

6

0 2.

24

ro

und

2 To

tal

2.83

36

0 2.

28

2.

97

180

2.35

2.69

18

0 2.

21

1

2.93

24

0 2.

29

3.

10

120

2.30

2.76

12

0 2.

27

2

2.64

12

0 2.

27

2.

72

60

2.45

2.57

6

0 2.

10

Cha

pter

Six

166

roun

d 3

Tota

l 2.

45

360

2.34

2.56

18

0 2.

35

2.

34

180

2.34

1 2.

51

240

2.34

2.53

12

0 2.

31

2.

49

120

2.38

2 2.

32

120

2.35

2.60

6

0 2.

44

2.

03

60

2.25

roun

d 4

Tota

l 2.

44

360

2.55

2.44

18

0 2.

30

2.

44

180

2.78

1 2.

42

240

2.50

2.44

12

0 2.

31

2.

39

120

2.68

2 2.

49

120

2.67

2.45

6

0 2.

30

2.

53

60

3.01


167

Actually, two additional levels can be distinguished for each individual player: the intended gifts are nested within a distribution (i.e., the first or second distribution of each round of a game) and these distributions are nested within the round of the game.

Our main interest in this section is whether the differences in characteristics between the players (or, alternatively, their social distance) and the heterogeneity or diversity of the group affect the gifts of the players. The results of our basic analyses are shown in Table 3.

We start with an empty model (A), which includes only a fixed intercept and random intercepts for groups, individuals, rounds and distributions. This model shows that gifts differ significantly between individuals, but not between groups, rounds and distributions. This indicates that group characteristics (such as group heterogeneity) as such do not matter much for individual giving behaviour and that the gifts within one distribution in a particular round of the game are not significantly correlated with each other. Therefore, we exclude these latter levels (group, distribution and round) in the following analyses.

Next, in model B, we retain the individual level in the multilevel analysis and include some characteristics and outcomes of the game as explanatory variables. These characteristics include whether the assignment of the winners of the game is performance based or random based, the round of the game (including a distinction between a one shot game and the first round of a four shots game) and whether the gift is part of the first or the second distribution of each round. The first and last of these characteristics do not significantly affect the average gift. Thus, a winner who is randomly selected does not give more to other players than a winner who has actually performed better than others. It also does not matter whether the player knows if s/he is the winner or not. However, in the four shots game there is a significant effect of the round of the game: the average gift declines as the game progresses. In the first round of the four shots game, players donate each other .9 credits more than in the fourth round. Remarkably, opposed to what we would expect, the players give more to each other in the one shot game than in the first round of the four shots game, although the difference is not significant.3

3 These results differ from the results of the first phase of the experiment that was conducted with students of the University of Amsterdam (see De Beer and Berg, 2012). These differences need further examination.

Chapter Six

168

We also include some outcomes of the game as explanatory variables. The (expected) score on the quiz does significantly reduce the gift of a player. A player who expects (at the first distribution) or knows (at the second distribution) that s/he has a maximum score of ten points donates on average .8 credits less to the other players than if s/he expects or has a score of zero. This suggests that players are less willing to share their profit with others if they really deserve it by having performed well.4 This result is somewhat unexpected, since we found that players are not more generous in the random based games than in the performance based games.

To test whether the players reciprocate gifts that they have received in previous rounds of the game we include the total amount of gifts received by a player in the previous round (of course, this is only relevant in the four shots game). This variable turns out to be highly significant. For each credit received from other players in the previous round, players tend to give .07 credits extra to each of the other players (so, .22 credits in total) in the current round. Thus, there is indeed clear evidence for reciprocity as a motive for gift-giving.

In the third model (C), we add some characteristics of the player, viz. sex, age (and age square to allow for the possibility of a curvilinear relationship), cultural background5 and whether and how long the person has been living in the Dapper area. Of these individual characteristics, age is the only one that matters. Older people tend to give more than younger people, but since age square has a negative sign, this relationship is curvilinear. Up till the age of 50 players give more, but beyond that age their gift declines again. To illustrate the size of the age effect: a person aged 50 will on average give one credit more than a person aged 22. Sex, cultural background and whether the player is a resident of the Dapper area or not do not significantly affect the gift of a player.

4 This effect does not differ between the first and second distribution. 5 The players were asked what their cultural background was, although the answering options actually referred to their ethnic origin, viz. Dutch, Surinamese, Antillean, Indonesian, Turkish, Moroccan, Eastern European, other European, African, Asian, Latin American, North American and Australian, respectively.

Solid

arity

in a

Mul

ticul

tura

l Nei

ghbo

urho

od

16

9

Fixe

d ef

fect

sB

SEB

SEB

SEB

SEB

SEin

terc

ept

2.87

3**

*.2

362.

537

***

.338

-.164

1.17

3.4

841.

206

-1.3

031.

767

cond

ition

(ran

dom

= re

f.)

perfo

rman

ce.0

30.2

85.0

99.2

86.1

24.2

88-.1

74.3

26

roun

d

roun

d 1

of 1

1.24

0**

*.3

001.

396

***

.305

1.37

7**

*.3

061.

348

***

.358

ro

und

1 of

4.9

02**

*.1

29.9

03**

*.1

29.9

02**

*.1

28.8

93**

*.1

28

roun

d 2

of 4

.375

***

.112

.376

***

.112

.376

***

.111

.378

***

.111

ro

und

3 of

4.0

80.1

12.0

80.1

12.0

81.1

11.0

83.1

11

roun

d 4

of 4

(ref

.).0

00.0

00.0

00

dist

ribut

ion

(2 =

ref.)

1

.104

.079

.099

.079

.099

.079

.096

.079

(exp

ecte

d) q

uiz s

core

-.078

***

.027

-.081

***

.027

-.080

***

.027

-.083

***

.027

gifts

rece

ived

.0

73**

*.0

20.0

73**

*.0

20.0

73**

*.0

20.0

70**

*.0

20

sex (

fem

ale

= re

f.cat

.) m

ale

.125

.317

.158

.337

.096

.404

age

.128

**.0

50.1

17**

.051

.121

**.0

57ag

e sq

uare

d / 1

00-.1

28**

.057

-.116

**.0

58-.1

16*

.064

AB

CD

E T

able

3. M

ultil

evel

ana

lys e

s of (

inte

nded

) gift

s in

the

Dap

per

area

Cha

pter

Six

170

cultu

ral b

ackg

roun

d

nativ

e-.2

46.5

37-.2

35.5

45.2

00.6

46

Surin

ames

e.0

75.6

25.0

92.6

29.3

71.6

77

Turk

ish/

Mar

occa

n-.2

46.8

27-.2

69.8

32-.1

68.8

48

Euro

pean

-.960

.780

-1.0

01.7

86-.6

71.8

50

othe

r.0

00.0

00

Dap

per a

rea

resi

denc

e l

ess

than

1 y

ear

.649

.697

.298

.713

.181

.729

fro

m 1

to 5

yea

rs.2

28.4

92-.0

68.5

02-.1

79.5

64 f

rom

6 to

10

year

s.6

12.5

69.3

79.5

76.3

80.6

87 f

rom

11

to 2

0 ye

ars

-1.2

69**

.564

-1.5

26**

*.5

74-1

.671

***

.630

mor

e th

an 2

0 ye

ars

-.225

.459

-.483

.468

-.570

.552

no

resi

dent

(ref

. cat

.).0

00.0

00

sex d

iffer

ence

m

ale

givi

ng to

fem

ale

-.247

.200

-.229

.203

sa

me

-.242

**.1

20-.2

28*

.122

fe

mal

e gi

ving

to m

ale

(ref.c

at.)

.000

diffe

renc

e of

cul

tura

l bac

kgro

und

na

tive

givi

ng to

non

-nat

ive

-.232

.186

-.259

.189

sa

me

.244

*.1

48.2

53*

.149

no

n-na

tive

givi

ng to

nat

ive

(ref.c

at.)

.000

diffe

renc

e of

Dap

per a

rea

resi

denc

e

non

resi

dent

giv

ing

to re

side

nt-.5

45**

*.1

97-.5

58**

*.2

02

sam

e-.3

43**

.161

-.341

**.1

65

resi

dent

giv

ing

to n

on re

side

nt (r

ef.c

at.)

.000

Solid

arity

in a

Mul

ticul

tura

l Nei

ghbo

urho

od

17

1

age

diffe

renc

e (a

bsol

ute)

.020

**.0

10.0

20**

.010

age

diffe

renc

e sq

uare

d / 1

00-.0

55**

*.0

21-.0

56**

*.0

21

sex d

iver

sity

0 m

ales

-.158

.603

1 m

ale

-.191

.561

2 m

ales

.007

.569

3 m

ales

(ref

.cat

.).0

00 4

mal

es-

cultu

ral d

iver

sity

0 n

ativ

es-

1 n

ativ

e.9

60.5

93 2

nat

ives

.562

.482

3 n

ativ

es.9

74**

.416

4 n

ativ

es (r

ef. c

at.)

resi

denc

y di

vers

ity 0

Dap

per a

rea

resi

dent

s-.0

65.7

89 1

Dap

per a

rea

resi

dent

.246

.716

2 D

appe

r are

a re

side

nts

.076

.687

3 D

appe

r are

a re

side

nts

.858

.722

4 D

appe

r are

a re

side

nts

(ref.

cat.)

age

dive

rsity

: Gin

i ind

ex3.

422

2.19

1

Cha

pter

Six

172

*

p <.

1; *

* p<

.05;

***

p<.

01.

Rand

om e

ffect

s (C

ovar

ianc

e va

riab

les)

grou

p.2

76.2

62pe

rson

3.05

6.4

153.

239

.380

3.10

9.3

803.

149

.384

3.09

6.3

90ro

und

.100

.088

dist

ribut

ion

.009

.017

n (g

ifts)

1890

1890

1890

1890

1890

n (p

erso

ns)

180

180

180

180

180

n (g

roup

s)45

4545

4545

-2 R

estri

cted

Log

Lik

eli h

7299

.672

87.1

7275

.372

61.8

7242

.2


173

The impact of differences in characteristics between players

Our first hypothesis states that a player shares less with a co-player who differs from her/him in various characteristics than with a co-player who has similar characteristics. To test this hypothesis, we introduce a number of variables that measure the difference in characteristics between the player and the potential beneficiary (model D). We include three variables that indicate whether the player and the potential beneficiary have the same sex, have the same cultural background and are residents of the same neighbourhood, respectively. These variables take the value 0 if the two persons are identical with respect to the particular characteristic, the value -1 if they differ and the donor is either male, native or a non-resident of the Dapper area, and the value 1 if they differ and the donor is either female, non-native or a resident of the Dapper area. By using these variables we take account of the possibility that the giving relationship between two persons is asymmetric. For example, we allow for the possibility that females tend to give more to males than males tend to give to females. With respect to age, we include both the age difference and the absolute value of the age difference between the donor and the potential beneficiary. By including the age difference we allow for the possibility that older people give more to younger people than younger people give to older people, and vice versa, while the absolute value of the age difference takes account of the possibility that it is not relevant whether the donor is older or younger than the potential beneficiary. Since the age difference turned out to be insignificant (not shown in the table), while the absolute value is significant, we only include the absolute value in the final model. We also add the square of the age difference in order to allow for a non-linear relationship.

In Table 3 the results are shown under model D.

A difference in cultural background between the donor and the potential beneficiary has a highly significant negative impact on solidarity. Natives tend to give almost half a credit (.47) less to non-natives than to natives, while non-natives give .24 credits less to natives than to non-natives. This confirms our first hypothesis as far as a difference in cultural background is concerned: natives and non-natives tend to discriminate each other. This result provides support for the conflict theory, but, as noted above, it is not necessarily in conflict with the constrict theory.

There is also a significant effect of a difference in residency. However, Dapper area residents give on average .34 credits more to non-residents

Chapter Six

174

than to their fellow residents, while non-residents give .20 credits less to Dapper area residents than to other non-residents. Put differently, non-residents receive larger gifts than residents. This result is not easily explained. Perhaps, Dapper area residents want to welcome visitors, while visitors feel a social distance to Dapper area residents.

Sex difference between donor and potential beneficiary has no effect on the average gift. There is no evidence that men or women give more (or less) to players of their own sex than to players of the opposite sex.

With respect to age, both the absolute value of age difference and the squared age difference have a significant effect on gift-giving. Since the first effect is positive and the latter negative, there is a curvilinear relationship. Gifts initially increase with the age difference between the donor and the beneficiary, reaching a maximum at an (absolute) age difference of 18.5 years, and then decline again as the age difference increases further. Thus, solidarity appears to be strongest between persons that differ 18.5 years in age, irrespective of whether the donor is older or younger than the potential beneficiary. By including separate variables for the age difference in case the donor is older and in case the donor is younger than the beneficiary, we tested whether there is an asymmetric pattern of age differences, but this appeared not to be the case.

The impact of group diversity

The second hypothesis states that group diversity or heterogeneity exerts a separate negative effect on solidarity, in addition to the aggregate effect of dyadic differences between the players in the group, which has been tested in the preceding section. Thus, we test whether players in a heterogeneous group are less willing to share money with their co-players than players in a homogeneous group. The results are shown in Table 3 under model E. Although our focus is on ethnic diversity, we also include measures of diversity with respect to sex, age and residency. Cultural diversity is simply measured as the number of natives in the group (ranging from 0 to 4). Similarly, sex diversity is measured as the number of males in the group (ranging from 0 to 4) and residence diversity as the number of Dapper area residents in the group (also ranging from 0 to 4). By using these simple measures and treating them as categorical variables, we allow for a possible non-linear relationship between group diversity and solidarity. For example, it is conceivable that a homogeneous native group has a different effect on giving behaviour than a homogeneous non-native group (which, actually, need not be ethnically homogeneous, since the


175

non-natives can belong to different ethnic groups; we come back to this point later).

Table 4 shows the frequencies of the group composition with respect to sex, cultural background and residency. From this table one can read that there are 35 groups with no male, 36 groups with one male, et cetera. Likewise, there is no group with only non-natives, 16 groups with one native, et cetera.

Table 4. Composition of the groups by sex, cultural background and residency Number of natives / males / Dapper area residents per group

Number of groups by cultural background

sex Dapper area residency

0 persons 0 35 15 1 person 16 36 45 2 persons 23 20 18 3 persons 39 14 14 4 persons 27 0 13 Total 105 105 105 Age diversity is measured by the Gini index of the ages of the four group members (which theoretically ranges from 0, if all have the same age, to 1 if three persons have a zero age and the other a positive age). The Gini index measures the average absolute age difference between all possible pairs of players within a group, divided by the average age of the players in the group.

None of these four measures of group diversity turns out to have a significant effect on the average gift within the group.

Consequently, the experiment does not lend support to the hypothesis that ethnic heterogeneity hurts solidarity. To the contrary, mixed ethnic groups tend to be more generous towards each other than purely native groups, although the differences are not significant. This result contradicts the constrict theory, which assumes an independent effect of group diversity on solidarity, but it is not necessarily inconsistent with the conflict theory.

We also do not find any support for a (negative or positive) influence of age diversity, sex diversity or residence diversity. Group composition does

Chapter Six

176

not seem to matter for solidary behaviour. This confirms our initial finding (model A) that the gifts within a group are not significantly correlated.

It should be noted, that the estimated coefficients of the indicators for group composition measure the pure group diversity effect, since we also include the variables measuring the differences between two subjects. So, average solidarity in heterogeneous groups may be smaller than in homogeneous groups, but this is due to the aggregate effect of the dyadic relations between the individual players and not to the diversity of the group as such. To illustrate, since both natives and non-natives give more to co-players from their ‘own’ ethnic group than to players from the other group, the average gift in ethnically homogeneous groups is larger than in mixed groups. Based on the estimates of these effects in model E, we can calculate that gifts in ethnically homogeneous groups are on average .25 credits larger than in mixed groups. However, this difference is not caused by the diversity of the group itself and, consequently, does not lend support to our second hypothesis, which relates solidarity to the group composition.

Interaction between group diversity and dyadic differences

Hypothesis 3 states that group diversity has a stronger negative effect on the gift of a donor to a player who differs from her/him in some characteristics than to a similar player, To test this hypothesis, we include interaction terms between the dyadic differences and the corresponding indicators for group diversity (model F). For example, we include the interaction term between the sex diversity of the group (i.e. the number of males in the group) and the sex difference between two players. With this interaction term, we test whether gifts between men and women in primarily male groups (i.e. a group with three males) differ from these gifts in primarily female groups (i.e. a group with three females) and in balanced groups (two males and two females).

These interaction terms are not significant for sex differences and age differences, but they are strongly significant for cultural background and residence. Table 5 shows the results for cultural background.

The results are rather puzzling. Since each group counts four players, the cultural diversity is largest if a group consists of two natives and two non-natives and groups are homogenous if they are made up of four natives or of no natives (the latter composition was absent in our experiment). If we compare heterogeneous groups with two natives with fully native groups,


177

average gifts turn out to be larger in the heterogeneous group than in the homogenous group. Moreover, in the heterogeneous groups, the size of the gift is not affected by a difference in cultural background between the players. Players from the same ethnic group give each other just as much as players from different ethnic groups. These results contradict both the conflict and the constrict theory.

Table 5. Estimated effect of cultural diversity and difference in cultural background on giving Cultural difference: donor beneficiary Average

gift Cultural diversity

native native

non-native Æ non-native

native non-native

non-native Æ native

1 native .052 -.107 -.398 -.100 2 -.489 -.420 -.485 -.402 -.447 3 .432 -.213 0a .163 4 -.611 -.611 a reference category However, the picture becomes more complex if we look at the groups with one native or three natives (i.e., one non-native). The average gift in these groups is considerably larger than in the groups with maximum heterogeneity (two natives) or maximum homogeneity (four natives). In groups with three natives and one non-native, the natives give each other on average .65 credits more than they give to the non-native, while the non-native gives the natives also considerably less. In groups with three non-natives (and, thus, one native), the non-natives give each other .45 credits more than the native, although the single native gives the non-natives only slightly less than the non-natives give to each other.

Apparently, the (ethnic) majority of a group discriminates against a single co-player with another ethnic background, but if the ethnic groups balance each other, the players do not discriminate the members of the other group. However, the average size of the gifts is larger in the groups with either one or three natives than in the homogeneous groups or the groups in which the numbers of natives and non-natives are equal. Consequently, it is not so much that a majority of natives penalizes the single minority player (negative discrimination), but rather that the majority players favour each other (positive discrimination). This would mean that ethnic diversity does not harm solidarity, but actually boosts solidarity within ethnic

Chapter Six

178

groups, without hurting solidarity with members of other ethnic groups. However, this in-group favouritism disappears if the different ethnic groups are equal in size.

If we reorder the figures in Table 5, we can depict the patterns of solidarity that emerge from the experiment with respect to group diversity. In table 6 we show for natives and non-natives, the amount of in-group solidarity, out-group solidarity, discrimination and the average gift. In-group solidarity is interpreted as the average gift to a person from the same category (e.g., a gift by a non-native to another non-native), out-group solidarity is the average gift to a person from the other category (e.g., a gift by a non-native to a native) and discrimination is interpreted as the difference between in-group and out-group solidarity. These gifts are related to the total number of members of the player’s category (in-group) in the group. Thus, ‘3 in-group members’ means for a native that the group includes 3 natives, while it means for a non-native that the group includes 3 non-natives.

Table 6 Patterns of giving related to diversity

number of in-group members natives non-

natives In-group solidarity 4 -.61 3 .43 .05 2 -.49 -.42 Out-group solidarity 3 -.21 -.40 2 -.40 -.40 1 -.11 .00a Discrimination 3 .64 .45 2 -.09 -.02 Average gift 4 -.61 3 .15 -.10 2 -.43 -.41 1 -.11 .00b a reference category for in-group and out-group solidarity. b reference category for average gift. Source: calculated from tables 6 and 7


179

The patterns are remarkably similar for natives and non-natives. In-group solidarity is the largest in case of three in-group members. Out-group solidarity is the largest in case a player is the only member of her/his in-group, followed by three in-group members. Discrimination against out-group members is larger in case of three in-group members than in case of two in-group members. Average gifts are the biggest in case of one or three in-group members and the smallest in case of two or four in-group members.

These results contradict both the conflict and the constrict theory. According to the constrict theory, in-group solidarity should decline with increasing group diversity, but table 7 shows that in-group solidarity is larger in a group with one out-group member than in a completely homogeneous group. According to the conflict theory one would expect more out-group discrimination in a group with two in-group and two out-group members than in a group with only one out-group member, while the opposite is the case. And both theories predict that average gifts decline with increasing heterogeneity of the group, which is contrary to the results shown in Table 6.

It is not clear how these patterns should be interpreted. The phenomenon of out-group favouritism by a member of a minority group is not unfamiliar in the social psychological literature, but it is usually related to the low status of (ethnic) minority groups (cf. Tajfel 1982: 9-12). Since we also found out-group solidarity for natives who are a minority within their group, this suggests that it is the size of the in-group and not its status that matters here. Social psychological experiments about the effect of the relative size of the in-group have yielded contradictory results (Sachdev and Bourhis 1991). Leonardelli and Brewer (2001) did not find a significant difference of out-group discrimination between members of minority groups and members of majority groups in case they identify equally strongly with their in-group.

A possible explanation for our results would be that players wish to avoid belonging to the minority within their group. A priori, they belong to their in-group, but in case they show solidarity to out-group members, they may hope or expect to be accepted by the out-group. In case the other three players belong to their in-group, they do not need to do anything to be accepted as a member of the majority group, and consequently, they give very little to their co-players. In case there are no other in-group members, a player tries to be accepted by the three out-group members by sharing much with them. In case of two in-group members and two out-group

Chapter Six

180

members, there is no minority and, consequently, it does not matter which group one belongs to. Therefore, players do not give much to other players and do not have a reason to discriminate. Finally, if there are two other in-group members and one out-group member, it is important to show that one belongs to the majority of in-group members by giving more to them than to the one out-group member.6

The reason that the results for non-residents depart somewhat from this general pattern, is probably that non-residents do not consider themselves to be members of a separate group, since the only thing they have in common is that they do not live in the Dapper neighbourhood. Since living outside the Dapper neighbourhood is not a salient aspect of their identity, they do not have a reason to distinguish between co-players from their in-group and from the out-group, i.e., residents of the Dapper neighbourhood.

Cultural diversity more closely examined

The most salient outcomes of our analysis are the significant negative effect of a difference in cultural background on the size of a gift (confirming hypothesis 1), the absence of an independent effect of group diversity (refuting hypothesis 2) and the non-linear interaction effect between group diversity and the cultural difference between donor and beneficiary (refuting hypothesis 3). However, since we only distinguished between natives and non-natives, it is worthwhile to explore this interaction effect of cultural difference and diversity in more detail.

An interesting question is whether this interaction effect is the same for all ethnic groups. Thus, instead of taking the difference between natives and non-natives as the explanatory variable, we include an interaction term between the cultural background of the donor and the cultural background of the potential beneficiary. This variable distinguishes five ethnic groups, viz. natives, Surinamese, Turks/Moroccans, European immigrants and

6 The alleged wish to belong to the majority group is probably not based on forward looking reciprocity, i.e. that players hope to be rewarded for their generosity in the next round. If that would be the case, we would expect this effect to be absent in the one shot game and in the final round of the four shots game. However, the pattern is more or less similar in the one shot game and in all rounds of the four shots game.


181

other non-natives7. If we add the interaction term between the cultural background of the donor and of the beneficiary in the multilevel regression model, it has a significant effect. Table 7 shows the estimated effects for the various combinations.

Two effects stand out, viz. the significantly smaller gift of natives to Turkish or Moroccan co-players and to other European co-players. However, natives do not discriminate against Surinamese or players with an ‘other’ cultural background, such as Asian or African. Another notable effect is that Surinamese players and players with an ‘other’ cultural background tend to favour themselves and each other.8 Turkish and Moroccan players and non-Dutch European players give less than the other players, but since they never encountered a player with a similar background in the same group, we do not know whether they would have been more generous to a person from their own group.

Table 7. Estimated effect of difference in cultural background on giving donor

beneficiary native Surinamese Turkish/ Moroccan

European other

native .093 .063 -.454 -.893 -.177 Surinamese .023 .556 -.363 -.734 .456 Turkish/Moroccan -.996 .243 -.266 European -.785 .135 -.448 other .116 .804 -.035 -1.292 0a a reference category Since a content analysis of mass media (see chapter 5) shows that the media display a considerably more negative image of people with a Moroccan background than people with a Turkish background, it would be interesting to test whether natives give even less to Moroccans than they do to Turks. However, since only 4 Turks and 4 Moroccans participated in the experiment, their numbers are too small to draw any statistically 7 The latter category includes six players with an Asian background, six with an African background, two with a Latin American background and one with an Indonesian background. 8 Since the category ‘other’ includes only two persons with a Latin American background, the solidarity between Surinamese and these ‘others’ is apparently not related to a similar geographical region of origin.

Chapter Six

182

reliable inferences on solidarity of natives towards Turks vis-à-vis Moroccans.

Since the non-natives in our sample include many different cultural backgrounds, a gift from a non-native to another non-native should not be considered as an intra-ethnic gift. To test intra-ethnic favouritism we construct an additional variable that equals 0 if the ethnic backgrounds of the donor and of the potential beneficiary are the same and 1 if they are different. If the interaction term between this variable and cultural diversity is included in the analysis, it is highly significant. Table 8 shows the estimated effects.

Table 8. Estimated effect of cultural diversity and equality or difference of cultural background on giving Same cultural background? Cultural diversity yes no 1 native .169 -.125 2 natives -.336 -.281 3 natives .644 0a 4 natives -.442 a reference category This analysis affirms the conclusion from the previous analysis: natives favour each other strongly in the presence of one non-native, and non-natives favour a co-player of their own ethnic group in the presence of one native, but intra-ethnic favouritism is absent if the numbers of natives equals the number of non-natives in a group. Moreover, average gifts are smallest in the fully native groups.

Conclusion

In this chapter we examined the effects of individual characteristics and group diversity on the willingness of individuals to share money they have won in a game with other players. We took the well-known and much debated thesis of Robert Putnam (2007) that ethnic diversity harms both inter-group and intra-group trust and solidarity, as our starting point. Our methodological approach differs from other empirical tests of Putnam’s thesis in that we conduct an experiment in which subjects play for real money. Thus, we do not measure attitudes or opinions, but actual behaviour. Moreover, the experimental design allows us to control the


183

conditions under which the subjects act and, thus, makes it possible to draw causal inferences. This experiment also makes it possible to distinguish between the effects of differences between individual players and the effect of the heterogeneity of the group in which they play.

In our experiment, 180 visitors of the Dapper market in Amsterdam participated in a solidarity game, in which they could share an earned sum of money with other players. The players were informed about the sex, the age, the cultural background (ethnicity) and the residency (either in the Dapper neighbourhood or elsewhere) of the other players.

Our main interest is the effect of the characteristics of the players and of the composition of the group in which they play the game on the size of the gift that the subjects are prepared to give to their fellow players. With the exception of age, the personal characteristics per se do not affect giving behaviour, but the differences in characteristics between the players do. In particular, a difference in cultural background between two players has a strong (negative) impact on the gift they bestow each other. Natives discriminate against co-players with a Turkish, Moroccan or European background, but not against Surinamese co-players. Whether the reverse is true is not clear, since there were no groups with more than one Turkish, Moroccan or European player, so we do not know how much they would give to each other. However, Surinamese players and players with an ‘other’ cultural background also demonstrated a bias in favour of players of their own or each other’s group compared to natives, Turks, Moroccan and European players.

This result provides support for the conflict theory, which asserts that members of different (ethnic) groups discriminate against each other.

We also found an effect of the ethnic composition of groups on giving, but there is no straightforward relationship between ethnic diversity and the size of gifts. In general, gifts are largest in groups with either three natives or three non-natives, due to intra-ethnic favouritism within these groups. In case of an equal number of natives and non-natives within a group, however, there is little evidence for intra-ethnic favouritism or discrimination against players from another ethnic group. Remarkably, we found a similar effect with respect to the diversity of residency within the group, i.e. the number of players who live in the Dapper neighbourhood.

We conclude, that in-group solidarity is the largest in case of three in-group members, while out-group solidarity is the largest if a player is the

Chapter Six

184

only member of her/his in-group. Discrimination against out-group members is larger in case of three in-group members than in case of two in-group members (here, the non-residents are an exception) and average gifts are the biggest in case of one or three in-group members and the smallest in case of two or four in-group members.

This is an interesting, but somewhat puzzling result, which deserves further examination. We suggest that an explanation of this result may be that individuals strive to avoid belonging to a minority within their group. If an individual is the only in-group member, s/he wants to be accepted by the majority of out-group members by showing her/his solidarity with them. If there are three in-group members, the individual wants to show her/his adherence with the in-group members by sharing money with them. If all the players belong to the same in-group, however, there is no need to show one’s adherence to the group. Finally, if there are two in-group and two out-group members, there is no minority, and, consequently, it does not matter towards which group one shows solidarity.

These results contradict Putnam’s constrict theory. We find no evidence that group diversity reduces in-group solidarity. On the contrary, in-group solidarity among natives is the lowest in the fully native groups. Moderate (ethnic) diversity increases in-group solidarity, although maximum diversity reduces it again. We find also very little support for the conflict theory. Although there is a clear tendency to discriminate against players from another ethnic group, this does not decrease total solidarity. Total solidarity is largest in moderately diverse groups, despite discrimination against out-group members.

Regarding age, both the age of the players and the age difference between the players matter. Up till the age of 50, the size of gifts rises with age, and gifts are the largest when the players differ about 18.5 years in age, irrespective of who is older and who is younger. Consequently, solidarity seems to be stimulated by a substantial but not too large age difference. This result, too, needs further study and theoretical explanation. The age composition of the group, as measured by the Gini coefficient or by the share of the old or the share of the young in the group, does not affect (the influence of age differences on) the average size of gifts.

To summarize, both differences in characteristics between players and the composition of the group matter for the willingness of subjects to share money with co-players. However, there is no simple linear relationship


185

between social distance and group diversity on the one hand and the average gift on the other. Both with regard to the age difference between two players and with respect to the ethnic and residence diversity of a group a curvilinear relationship exists. Solidarity is largest if there is a moderate age difference (around 18.5 years) between two players and if group diversity is moderate (i.e. one player differs from the other three players in a group). Complete equality of age and complete ethnic or residence homogeneity of the group result in less solidarity, but so does a very large age difference (say, over 30 or 40 years) and maximum heterogeneity (two natives and two non-natives or two residents and two non-residents).

Moreover, generosity and discrimination against a ‘different’ player are positively related. Put differently, discrimination often does not mean that a player gives less to a co-player who is different, but that s/he gives more to a similar co-player. Thus, discrimination does not lead to less solidarity but may even result in more solidarity. As a consequence, even individuals who are discriminated against may be better off than if there would be less discrimination but also less generosity.

If one looks at these results from a moral perspective, it seems that there is a trade-off between solidarity and equality. Those who are in favour of equal treatment of all individuals, irrespective of the (ethnic) group they belong to, may pay the price of less solidarity. Those who wish to maximize the revenue for the least well-off (cf. Johan Rawls’ famous difference principle), may have to accept that some groups are discriminated against.

Testing the effect of group diversity on solidarity by means of a controlled experiment has, of course, a number of limitations. To start, the context of a group of four players is rather artificial. It is not clear whether the effect of the composition of such a small group is similar to the effect of the composition of a larger community, such as a neighbourhood, a city of even a country.

Secondly, we only know a few characteristics of the subjects of the experiment, viz. sex, age, cultural background, residency media use and political preference. It is conceivable that some unobserved characteristics, such as educational level or income, also affect their giving behaviour. If these characteristics are correlated with some of the observed characteristics, some of the relations we found may be spurious. For example, if natives have, on average, a higher education than non-natives, this might explain some of the patterns we have observed.

Chapter Six

186

Thirdly, since most of the passers-by on the Dapper market refused to participate in the experiment, the participants are not a random sample of all the visitors to the market and may therefore not constitute a representative group.

To conclude, the results of this experiment should be interpreted with caution and should preferably be re-examined in a different context and with a different methodological design in future research.

References

Andreoni, J. (1989). Giving with impure altruism: Applications to charity and Ricardian equivalence. Journal of Political Economy, 97(6), 1447-1458.

Bolton, G. E., & Ockenfels, A. (2000). ERC: A theory of equity, reciprocity, and competition. American Economic Review, 90(1), 166-193.

Camerer, C. (1999). Behavioral economics: Reunifying psychology and economics. Proceedings of the National Academy of Sciences, 96(19), 10575-10577.

de Beer, P., and Berg, M. (2012). Conditions and motives for voluntary sharing: Results of a solidarity game experiment. AIAS Working Paper 124. Amsterdam: University of Amsterdam/Amsterdam Institute for Advanced Labour Studies.

Fehr, E., and Schmidt, K. M. (1999). A theory of fairness, competition, and cooperation. The Quarterly Journal of Economics, 114(3), 817-868.

Frey, B. S. (1997). On the relationship between intrinsic and extrinsic work motivation. International Journal of Industrial Organization, 15(4), 427-439.

Gesthuizen, M., van der Meer, T., and Scheepers, P. (2009). Ethnic diversity and social capital in Europe: tests of Putnam’s thesis in European countries. Scandinavian Political Studies, 32, 121–142.

Hooghe, M., Reeskens, T., Stolle, D., and Trappers, A. (2009). Ethnic diversity and generalized trust in Europe. A cross-national multilevel study. Comparative Political Studies, 42, 198-223.

Kahneman, D., and Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263-291.

Lancee, B., and Dronkers, J. (2008). Ethnic diversity in neighborhoods and individual trust of immigrants and natives. A replication of Putnam (2007) in a West-European country. Paper presented at the


187

International Conference on Theoretical Perspectives on Social Cohesion and Social Capital, Royal Flemish Academy of Belgium for Science and the Arts, Brussels, Palace of the Academy. May 15, 2008.

Leonardelli, G.J., and Brewer, M.B. (2001). Minority and majority discrimination. When and why. Journal of Experimental Social Psychology, 37, 468-485.

Maynard Smith, J. (1989). Did Darwin get it right?: Essays on games, sex and evolution. New York: Routledge, Chapman & Hall.

Putnam, R. (2007). E Pluribus Unum. Diversity and community in the twenty-first century. Scandinavian Political Studies, 30, 137-174.

Rabin, M. (1993). Incorporating fairness into game theory and economics. American Economic Review, 83(5), 1281-1302.

Sachdev, I., and Bourhis, R.Y. (1991). Power and status differentials in minority and majority group relations. European Journal of Social Psychology, 21, 1-24.

Selten, R., and Ockenfels, A. (1998). An experimental solidarity game. Journal of Economic Behavior & Organization, 34, 517-539.

Sen, A. K. (1977). Rational fools: A critique of the behavioral foundations of economic theory. Philosophy & Public Affairs, 6(4), 317-344.

Sunstein, C., and Thaler, R. (2008). Nudge. The politics of libertarian paternalism. New Haven.

Tajfel, H. (1982). Social psychology of intergroup relations. Annual Review of Psychology, 33, 1-39.

CHAPTER SEVEN

CONCLUSION

PAUL DE BEER

Introduction

It is tempting to reduce the complexity of social life to some simple rules or regularities. Ethnic diversity is detrimental to co-operation, identification, trust and solidarity is such a simple rule that has been endorsed by many policy makers and social scientists in the past decade. However, the studies that we have reported on in this book demonstrate that the relationship between ethnic diversity and solidarity is much more complex. Thus, we do not merely refute the alleged negative societal effect of ethnic diversity and embrace the opposite, ‘multiculturalist’ proposition that ethnic diversity is good for society. Instead, we conclude that there is not a simple, unambiguous linear relationship between ethnic diversity and social solidarity, although we do acknowledge that ethnic diversity affects solidarity. To start with, this is so because ethnic diversity and social solidarity are both complex, multidimensional concepts, which can be interpreted and measured in various ways.

Ethnic diversity is often interpreted as the share of ethnic minority groups in the population or as the fractionalization of the population between ethnic groups. In the first interpretation, natives or autochthones are juxtaposed against all other, non-native ethnic groups, thus blurring any distinction between these groups. In the second interpretation, it is tacitly assumed that the social distance between all ethnic groups is the same. Both assumptions are too simple to adequately reflect the reality of ethnic diversity.

A second remark with respect to ethnic diversity refers to the level at which it is measured. An ethnically diverse city may be composed of ethnically homogeneous neighbourhoods. Or an ethnically divided country may consist of a number of ethnically uniform regions. As a consequence,

Chapter Seven

190

it matters a lot at which level one studies the effects of ethnic diversity. This may also explain why various studies have found conflicting results regarding the effect of diversity on solidarity or social capital.

Solidarity is even a more complex phenomenon than diversity. In chapter 1 we distinguished between four ideal-typical kinds of solidarity, viz. empathic, bilateral, multilateral and normative solidarity. But these four types of solidarity certainly do not cover all varieties of solidarity. For example, just as with diversity, the scope of solidarity shows a huge variety: it ranges from the family, via the neighbourhood and one’s countrymen to all mankind. Moreover, solidarity can be formal (public) or informal (private), voluntary or mandatory.

Next to the complexity of diversity and solidarity, it is important to realize that the behaviour of the central subject of our study, the individual person, is even more complex. Many studies of human behaviour start from the assumption that individuals are motivated by one overriding goal, be it self-interest or fellow feeling. Contrary to this, our basic assumption was that people are usually motivated by various goals simultaneously. In our view, the interesting question is not whether people are selfish or altruistic, but in which circumstances and under what conditions various motives affect people’s behaviour more or less strongly.

In this concluding chapter we recount the main results of our four empirical studies regarding ethnic diversity, social solidarity and the main motives of individual citizens to act solidary, respectively.

Ethnic diversity

Although the term ethnic diversity is often used as if its meaning is self-evident, the opposite is actually true. The ambiguity of the term ethnic diversity refers to the concept of ethnicity as well as to the concept of diversity.

Ethnicity is a so-called ‘essentially contested concept’. We did not dive into the intricacies of the debate on what ethnicity really means, but we noted two important aspects. First, ethnicity may refer to various characteristics of citizens, such as their citizenship or nationality, their country of birth, their language and their religion. Although these characteristics are often related to each other, they are certainly not identical. Consequently, it may make quite a difference whether a measure of diversity is based, e.g., on citizenship or on language. To give an

Conclusion

191

obvious example: Belgium does not show much diversity with respect to citizenship, but it is split into two (or, actually, three) language communities. Our cross-national study in chapter 3 showed that measures of diversity indeed differ considerably depending on whether they are based on country of origin, language or religion. Consequently, a claim about the harmful effect of diversity on solidarity may be true with respect to one of these aspects but not with respect to the other.

Secondly, an objective measure of ethnicity may not be identical to the subjective self-categorization of citizens. Self-categorization can depend either on whom the subject identifies with or on how s/he is categorized by others. This may or may not coincide with objective characteristics, such as country of birth or citizenship. As a consequence, a measure of diversity that is based on subjective self-categorization will usually differ from a diversity measure based on an objective characteristic.

An interesting question is, of course, which measure of diversity is most relevant when studying solidarity. Unfortunately, we cannot answer this question, because, to our knowledge, there are no studies that analyse this relationship with both a subjective and an objective measure of ethnicity simultaneously. Thus, we cannot draw a conclusion regarding the ‘best’ operationalization of this variable.

As far as the second component of ethnic diversity is concerned, i.e. diversity, the simplest measure is the number of people belonging to an ethnic minority group expressed as a share of the total population. In this way the native or indigenous or autochthone population is juxtaposed against the ethnic or migrant or allochthone population. Effectively, this means that all non-native ethnic groups are assumed to have the same social distance to natives, and that the distance between all non-native groups is assumed to be zero. The more sophisticated fractionalization index, which measures the probability that two persons who are drawn at random from a population do not belong to the same ethnic group, is based on the implicit assumption of an equal social distance between all ethnic groups (including natives). For example, the social distances between native Dutch and Surinamese, between natives and Moroccans and between Surinamese and Moroccans are all assumed to be equal.

However, our analysis of the portrayal of various ethnic groups in the Dutch media in chapter 5 showed that they differ considerably with respect to the characteristics attributed to them, both with respect to their competence and with respect to their ‘warmth’. Moroccans are most often

Chapter Seven

192

associated with threatening and/or otherwise negative evaluations and themes. They are awarded the role of the villain disproportionably more often than Turks – a tendency that does not appear to be grounded in their real-life performance, as reflected by official statistics on crime et cetera. In contrast, Turks are presented rather favourably: compared to the other ethnic groups studied, they are least often associated with negative themes or evaluations, and granted the greatest share of positive evaluations.

The results of our vignette study in chapter 4 and our experiments in chapter 6 are roughly in line with these conclusions from the media analysis. In general, native Dutch have a more favourable or sympathetic attitude towards Surinamese than towards Turks or Moroccans.

As a consequence, any measure of diversity which assumes a dichotomous relationship between citizens (i.e., the social distance between two persons is either zero or one) neglects this ethnic hierarchy and is, thus, not an adequate measure of the average social distance between the members of a population.

Another important element of measuring diversity is the level at which it is measured. There are two reasons why the level matters. First, as mentioned above, the extent of diversity may differ depending on whether we measure diversity at the level of the neighbourhood, the city, the region or the country. Secondly, people may react differently to ethnic diversity in their neighbourhood than to ethnic diversity at the national level. One obvious reason is, of course, that the chance of encountering someone with a different ethnic background is much larger if you live in an ethnically diverse neighbourhood than if you live in an ethnically homogeneous city in an ethnically diverse country. Another reason is that the scope of solidary behaviour differs. Thus, ethnic diversity at the national level may be most relevant in explaining public support for a solidary welfare state, while the ethnic diversity of a neighbourhood may primarily affect informal kinds of solidarity towards neighbours and local residents.

Interestingly, in chapter 3 we found that diversity at the regional level may have a different, although not straightforward, effect on public support for redistribution and trust in formal institutions than diversity at the national level (see below).

Conclusion

193

Solidarity

Since solidarity is a multifaceted concept, it is not surprising that we did not find one overarching mechanism or motive behind solidary behaviour. After all, giving support to income redistribution by the government is quite different from helping a neighbour, and there is a priori no reason to expect both kinds of solidarity to be affected in the same way by, for example, the ethnicity of the benefactor and of the (potential) beneficiary or the ethnic diversity of a country. This was confirmed by our analyses.

The vignette study in chapter 4, in particular, showed that the factors influencing private or particularistic solidarity differ from the factors affecting public or generalized solidarity. Moreover, the small scale experiments on which we reported in chapter 6 reveal some other mechanisms than the cross-country survey analysis in chapter 3.

Nevertheless, we also found some commonalities, especially with respect to the motives underlying solidary acts. In particular, two motives appear to be important in explaining solidary behaviour, viz. reciprocity and neediness.

Reciprocity means that solidarity is in some way returned. Actually, we can distinguish between three kinds of reciprocity. Forward looking reciprocity means that people are willing to help someone, expecting that the other will return their help in the future. Backward looking reciprocity means that people are willing to help others because they have been helped by them in the past. And simultaneous reciprocity means that people support each other, expecting that other people are also offering or returning support at the same time. Although these three kinds of reciprocity are related – for example, forward looking reciprocity in fact presupposes backward looking reciprocity – they are not identical, and in our experiments in chapter 6 we were able to distinguish them from each other empirically. The fact that the willingness to share the profits from a game with other players declines as the end of the game draws near, suggests that forward looking reciprocity is important, since the odds of getting something in return to a gift diminishes as there are fewer rounds to go. We also found that the more a player has received in previous rounds, the more s/he tends to give to others in consecutive rounds, which supports the hypothesis that backward looking reciprocity is a motive for giving. The fact that people also share money with their co-players in a one shot game – even more than in the first round of a four shots game – may indicate that simultaneous reciprocity is also an important motive:

Chapter Seven

194

apparently people expect other players to share their proceeds with others as well.

The vignette study in chapter 4 also confirmed the importance of the three kinds of reciprocity. Respondents were more willing to help someone who promised to repay that help in the future (forward looking reciprocity); in the case of a cut in social expenditure, they preferred to spare people who had contributed for a long time to the welfare state (backward looking reciprocity) and those who were doing voluntary work (simultaneous reciprocity).

That reciprocity is such an important motive for solidary behaviour does not necessarily mean that most solidary acts are based on self-regarding motives. Only forward looking reciprocity is clearly self-regarding; simultaneous and backward looking reciprocity are not purely self-regarding, since the benefactor does not directly benefit from his/her solidary behaviour. Indeed, backward looking reciprocity may be primarily based on a norm of fairness: one should reward someone for his/her solidary behaviour.

Neediness also turned out to be an important motive for solidary behaviour. In itself, this is not surprising, since the fact that someone is in need is often the very reason to show solidarity in the first place. Nevertheless, it is important to note that helping someone who is in need is often not in the self-interest of the benefactor and therefore presupposes other-regarding motives, triggered for example by empathy or a prevailing social norm. Moreover, the vignette study in chapter 4 suggested that there is a ranking of neediness, related either to personal characteristics of the potential beneficiary (in the case of private solidarity) or to the kind of income support that a beneficiary receives (in the case of public solidarity). In the case of private solidarity, women and older people could count on more solidarity than men and younger people. In the case of public solidarity, people on disability benefit were spared most often, followed by those who depend on social assistance (welfare) or an unemployment benefit, while people receiving an old age pension were spared the least.

We also found evidence for some other possible motives for solidary behaviour. In the experiments of chapter 6 the (expected) score of a player – an indicator for her/his performance in the game – had a negative effect on the amount that s/he shared with others, suggesting that deservingness plays a role. To state it differently, the less a person deserves the money s/he wins, the more willing s/he is to share it with others.

Conclusion

195

Effects of ethnic difference and ethnic diversity on solidarity

As explained in chapter 2, ethnic diversity may affect solidarity in two ways. First, a difference between the ethnicity of two persons may (negatively) affect the solidarity between the two persons. Secondly, the ethnic diversity of a group or community may have a separate (negative) effect on solidarity between the members of the group. An important difference between these two mechanisms is that in the second case the solidarity between members of the same ethnic group will also be affected by ethnic diversity.

Although we did find quite some support for the hypothesis that ethnic difference hurts solidarity between two persons, we did not find unequivocal support for the hypothesis that ethnic group diversity affects solidarity within the group negatively.

We studied the effect of ethnic difference in the vignette study of chapter 4 and in the experiments of chapter 6. Both studies showed that individuals tend to show less solidarity towards people who belong to a different ethnic group. However, in line with our findings from the media analysis in chapter 5, we found clear differences between different ethnic groups. First, natives appear to discriminate quite strongly against people with a Moroccan background, but much less against Surinamese. With respect to Turks, the evidence is not clear, since, due to the small numbers, they could not be distinguished from Moroccans in the experiment, while they were not included in the vignettes. In view of the outcome of our media analysis, we are inclined to think that natives would probably have a more positive attitude towards Turks than towards Moroccans.

Secondly, non-natives also seem to discriminate against other ethnic groups, although we must be careful not to draw firm conclusions due to the small numbers of non-natives from various ethnic groups that participated in the experiments and in the vignette analysis. Generally, non-natives appear to discriminate less against natives than natives against non-natives, which may be explained either by the fact that non-natives belong to a minority group or to the fact that the natives have a higher rank in the status hierarchy than (most) non-native groups.

Unfortunately, our study does not shed light on the reasons why different ethnic groups discriminate against each other. Since the vignette study allowed us to control for some other motives (such as neediness and the

Chapter Seven

196

willingness to reciprocate), we can conclude that there is a separate ethnic factor, but it is not clear which mechanism is triggered by ethnic difference. The vignette study demonstrated that the size of the effect of ethnic difference in the case of public solidarity is similar to the effect in the case of private solidarity. This suggests that the negative effect of ethnic difference is not reduced if people have more information about the particular person that asks for help compared to an anonymous person belonging to a specific category, such as the unemployed.

The effect of the ethnic diversity of a group was studied in the international comparative study based on survey analyses in chapter 3 and in the experiments in chapter 6. The survey analyses did not allow us to distinguish between the effects of group diversity as such and the aggregate effect of ethnic differences between pairs of citizens. This study did not find general support for the theses of, among others, Putnam and Alesina that ethnic diversity undermines trust or the welfare state. No relationship was found between heterogeneity and interpersonal trust. Institutional trust turned out to be positively related to ethnic and linguistic heterogeneity at the regional level, but negatively related to ethnic and religious fractionalization at the national level. Thus, with respect to trust in formal institutions, heterogeneity at the regional level may have a counterbalancing effect compared to heterogeneity at the national level. However, with regard to government redistribution, the analyses did corroborate the hypothesis that regional heterogeneity with respect to ethnicity and religion reduces public support. At the national level, we only found a negative effect of religious fractionalization. The overall conclusion is that the link between heterogeneity and human sociality is far from straightforward as is suggested by a large share of the literature. While regional heterogeneity is positively related to institutional trust, it is negatively related to support for government redistribution.

In the experiments of chapter 6, it was possible to study the effects of ethnic difference and of ethnic diversity separately. This showed that the relationship between ethnic group diversity and solidarity is rather complex. Overall solidarity appeared to be the smallest in the homogeneous native groups and the largest in groups with three natives and one non-native or one native and three non-natives. Balanced groups – two natives and two non-natives – were, however, characterized by rather little solidarity. Nevertheless, in the groups with one ‘minority’ member, in which overall solidarity was the largest, the majority discriminated most against the one minority member. In the balanced group with two natives and two non-natives, in which overall solidarity was considerably smaller,

Conclusion

197

there was no evidence of discrimination against the players from a different ethnic group. We suggested that this unexpected and somewhat counterintuitive result may be explained by anxiety of individuals of belonging to a minority. As a consequence, there is less solidarity in homogeneous groups, in which no one has to show that s/he belongs to the majority, and in balanced groups in which there is no minority.

One should, however, take into account that the experiments only allow to study the impact of ethnic diversity of a small, artificial group, whose members interacted only once in this particular game. It is an open question whether the influence of group diversity in such an experiment is similar to the influence of the ethnic diversity of a neighbourhood, let alone a country.

Concluding remarks

Our study has shown that ethnicity does indeed play a significant role in explaining solidarity. We found strong evidence, both from experiments and from a vignette analysis, that individuals show less solidarity towards people from another ethnic group than towards people from their own group. Moreover, we found that the extent of discrimination varies with the ethnic group. Native Dutch tend to discriminate more against Moroccans than against Surinamese. The evidence regarding discrimination by ethnic minorities is less clear-cut, due to the small number of minority members that participated in the experiments and the vignette analyses. But it appears that allochthones also discriminate against other ethnic groups, though less against natives than against other minority groups.

In view of this evidence for interethnic discrimination, one might be inclined to conclude that the ethnic diversity of a group must also exert a negative influence on (average) solidarity in that group. However, this need not be true, since discrimination against a member from another ethnic group may be the result of larger solidarity with members of your own group. If there would be more intra-ethnic favouritism in an ethnically heterogeneous group than in a homogeneous group, this might actually result in more overall solidarity in heterogeneous groups than in homogeneous groups. We did indeed find some evidence for this outcome in our experiments, which showed more discrimination but also more overall solidarity in moderately heterogeneous groups compared to homogeneous and balanced groups. These possible opposing effects may also explain why we did not find unambiguous evidence of a negative

Chapter Seven

198

effect of ethnic heterogeneity at the regional or national level in our cross-country survey analyses.

We conclude that, although ethnicity is certainly an important factor in understanding patterns of solidarity, there is not a simple linear relationship between ethnic diversity and solidarity. Even though ethnic difference in itself may be a source of discrimination, one cannot conclude from this that increasing ethnic diversity will necessarily result in less solidarity. Under particular circumstances, ethnic diversity may even be beneficial for overall solidarity, albeit accompanied by discriminatory tendencies.

Finally, one should keep in mind that the ethnicity of the benefactor and the beneficiary and the ethnic diversity of the group are only two of the many factors that affect the willingness to act solidary. In most circumstances, they are not the most important factors. Although attention for the effects of ethnic diversity on solidarity is certainly warranted, in studying solidarity one should not make the mistake of focussing attention only on this element.

AFFILIATION OF THE AUTHORS Paul de Beer Henri Polak professor at the Faculty of Social and Behavioural Sciences, University of Amsterdam, and co-director of the Amsterdam Institute for Advanced Labour Studies (AIAS), University of Amsterdam Laurens Buijs Lecturer at the Faculty of Social and Behavioural Sciences, University of Amsterdam Maarten Berg Academic Lecturer at the Department of Economics, Leiden University Ferry Koster Associate Professor at the Department of Public Administration and Sociology, Erasmus University Rotterdam and distinguished professor at TIAS School for Business and Society Dorota Lepianka Researcher at the Faculty of Social and Behavioural Sciences, Utrecht University

ethnic diversity and solidarity - ferry koster · 2018-08-16 · british library cataloguing in...

Documents