WPS6784 Policy Research Working Paper 6784 “They Are Not Like Us” Understanding Social Exclusion Sadaf Lakhani Audrey Sacks Rasmus Heltberg The World Bank Sustainable Development Network Social Development Unit February 2014 Policy Research Working Paper 6784 Abstract Negative attitudes toward groups in society are toward stigmatized attributes and behaviors; and widespread and underpin systematic processes of intolerance toward specific identity groups. An empirical social exclusion that marginalize people and deny analysis of the determinants of attitudes shows that them opportunities and dignity. This paper looks at country-specific factors are far more important than the processes underlying social exclusion. It uses data socio-economic characteristics. These findings could have covering Eastern Europe and Central Asia to study the important implications for theories about exclusion and responses to a simple hypothetical survey question about for the design of appropriate social inclusion policies. The which specific groups respondents would not like to have authors argue that strategies to address social exclusion as neighbors. Unwelcoming attitudes toward groups need to consider ways to change social norms, attitudes, such as immigrants, ethnic minorities, the poor, HIV+ and behaviors toward disadvantaged groups. The paper individuals, and others are surprisingly common. These explores potential entry points for change within formal attitudes fall into three distinct clusters: intolerance for and informal institutions. the poor and for different lifecycle stages; intolerance This paper is a product of the Social Development Unit, Sustainable Development Network. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at SLakhani@worldbank.org, ASacks@worldbank.org, and RHeltberg@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team “They Are Not Like Us”: Understanding Social Exclusion1 Sadaf Lakhani, Audrey Sacks, and Rasmus Heltberg Key Words: Social exclusion, inclusion, values, marginalization, stigma, discrimination, attitudes, LiTs, poverty, informal institutions. JEL classification: I3 1 This paper was originally prepared as a background paper to the World Bank Social Development Department’s Report ‘Inclusion Matters: the Foundations for Shared Prosperity’, available at http://www.worldbank.org/en/events/2013/09/23/inclusion-matters-foundation-of-shared-prosperity 1. Social exclusion as a process between people Around the world, one encounters negative attitudes toward people belonging to other groups. Such views are often expressed casually, in everyday conversations: “they are lazy” or “they’re not like us” are common expressions people use when referring to members of a particular race, caste, ethnic, or religious group. In isolation, such views may be relatively inconsequential. But once they are aggregated across large segments of a society, negative attitudes toward other groups can have wide-ranging consequences. In fact, aggregated and institutionalized negative attitudes underpin systematic processes of social exclusion. Negative attitudes and discriminatory behavior assign excluded groups an inferior or morally suspect status. Indeed, one group may even dehumanize other groups in order to justify their exclusion. Social exclusion operates in many domains, and involves a range of legal, social, economic, and political mechanisms that hinder development progress (de Haan 1999). Being subject to stereotypes and negative attitudes has profound impacts on the excluded, who may be blocked from accessing a variety of opportunities including employment and education. As a result, the excluded are often more vulnerable to poverty. Sometimes, the excluded internalize the negative attitudes and underperform, as demonstrated by research on lower caste individuals in India (Hoff and Pandey 2008). At other times, they respond with political mobilization, protest movements, or violence. Exclusion can therefore contribute to state fragility and conflict. The purpose of this paper is to examine the values, attitudes and behaviors that drive social exclusion processes in developing and transition countries. We do this using a novel empirical analysis of a large multi-country data set on attitudes to other groups combined with a focused synthesis of the literature. Our starting point is that social exclusion is rooted in everyday social life, specifically in people’s attitudes and behaviors toward other groups they perceive as different to and less valuable than their own group. Using data covering the countries of Eastern Europe and Central Asia, in addition to a few West European comparator countries, we analyze the responses to a simple hypothetical survey question about which specific groups respondents would not like to have as neighbors. We find that unwelcoming attitudes to other groups—immigrants, ethnic minorities, the poor, HIV+ individuals, homosexuals, and many others are surprisingly common and are comprised of three distinct sets of values. This finding allows us to propose a novel categorization of social exclusion attitudes: (1) intolerance for the poor and for different lifecycle stages; (2) intolerance toward stigmatized attributes and behaviors; and (3) intolerance to specific identity groups. In our data set, the second attitude cluster (related to stigmatized attributes and behaviors) is the most pronounced. The empirical analysis of the determinants of the three attitude clusters shows that country-specific factors are far more important than socio-economic characteristics. These findings could have important implications for theories about exclusion and for the design of appropriate social inclusion policies. We use these findings to argue that tackling the norms and values that underpin exclusion should be an essential element of social inclusion strategies. While development policies and programs sometimes acknowledge norms and values, they rarely address them explicitly and systematically. Doing so requires an understanding of the origins, functions, and consequences of norms and values. It also requires a 2 theory of how they change. In this paper, we pursue answers to these questions through empirical analysis of the Life in Transition survey (LiTs), vignettes, and a focused synthesis of the literature. We contend that paying attention to the processes that result in exclusion can lead to more effective inclusion than policies that merely address outcomes such as group differences in health, poverty, housing, education, or labor force participation. Narrowly addressing only outcome differentials ignores the broader underlying values that result in exclusion in the first place. For example, policies aiming for a leveling of the economic playing field and widening opportunities for minority groups can easily be undermined if they do not tackle the discriminatory beliefs and practices of the broader society and, especially, of the staff in agencies that are responsible for implementing the policies. Maria’s story, told below, illustrates many of the points made in the remainder of the paper: exclusion is often the result of prejudiced attitudes and discriminatory behaviors between an individual from one social group, toward another from a different group. However, the values that underlie these attitudes and behaviors are not limited to direct interactions between individuals, but are also embedded within institutions. What might appear to be technical decisions about, for example, provision of health care is actually informed by value-driven notions of who is deserving of treatment, the relationships between indigenous and other people, and culture-specific ideas about what kind of health care is acceptable and desirable. Moreover, even where the formal values of institutions are inclusive, the actions by staff may still be discriminatory and result in exclusion. Maria’s story illustrates the way in which multiple identities often intersect and amplify exclusion: As an indigenous women living in a remote and underdeveloped area, she is particularly vulnerable to exclusion. Furthermore, Maria’s story also illustrates how different exclusionary processes and outcomes reinforce each other. Maria is an illiterate woman who is unable to understand the Spanish language spoken by healthcare providers and who has a husband who precludes her from making her own decisions about family planning. These are all processes and outcomes of exclusion that ultimately contributed to Maria’s death and her family’s subsequent hardships. Poverty, as this story illustrates, is simply one among many other factors in a complex and mutually reinforcing web of social exclusion. The remainder of this paper is structured as follows. Section 2 uses data from the LiTs and the World Value Survey (WVS) to illustrate the high prevalence of negative attitudes toward defined groups. Section 3 presents a statistical analysis of negative attitudes. Section 4 explores the origins of values, attitudes and behavior that we discuss in the statistical analysis and how they can underpin social exclusion. Section 5 explores interactions between norms and formal and informal institutions, while section 6 discusses opportunities for development approaches to tackle the norms and behaviors that underpin social exclusion. Section 7 summarizes and draws out conclusions. CASE: Maria’s Story 2 2 Maria’s story is adapted and consolidated from Alicia Ely Yamin’s work on maternal mortality, detailing the stories of several indigenous women who died of child birth complications in the Altiplano, Peru. 3 Maria, already having six children, was not happy to find herself pregnant again. Since the previous year, food had been scarcer and her husband had been suffering from poor health. But she and her husband, Francisco, were not practicing any birth control as he wanted to have another son. Maria was careful in going to her prenatal check-ups. The nearest health post was 1 km from their house, and Francisco would accompany her, the two of them making slow progress as they walked the poorly-maintained roads. Although the health post was not actually their designated health facility—it was not in their jurisdiction—they liked the resident midwife who spoke Quechua, their native language. The midwife there did not want to turn Maria and Francisco away as she knew how hard it was for them to make the time to go to checkups at all. As Maria had already delivered several babies, her pregnancy was considered high-risk. At the fourth visit, the midwife told them that they would need to go to the hospital which was actually their provider and register and sign a birth plan there, and to go there for the birth instead of coming to the health post closer by. Maria and Francisco were reluctant to go to the hospital which was far away, more than 35 minutes by car on poor roads; they did not own a vehicle, and there was only one in the village that they might be able to borrow. They also felt anxious about the prospect of going to the hospital. The staff there had treated them badly in the past. Maria had also heard from another woman in the village who had given birth at the hospital that they did not permit traditional birth practices, such as a vertical birthing position, or use a sheepskin to cover the floor for the arriving baby, and that they had been rude to her because she was a campesina, a peasant. Nevertheless, they went to the hospital, had a check-up, but were not asked to sign a birth plan or be registered in any way. As they did not understand the procedure of registration, and as neither Maria nor Francisco spoke Spanish, and few of the hospital staff spoke Quechua, they did not inquire any further about the registration and birth planning. Maria went into labor one afternoon several weeks before her due date. She was in pain, and frightened of what was happening. Maria asked Francisco to take her to the hospital but Francisco, who was still upset about the treatment they had received there some years before, decided she should give birth at home. He was also thinking about the money he would have to pay to borrow the vehicle and to pay for gasoline. By early evening, a healthy baby was born, but when Maria did not stop bleeding and became increasingly weak and frightened, Francisco went to get help. There was no telephone in the village so he would have to go himself to get a health practitioner. He borrowed an old motorbike and started out for the hospital. The bad roads had been made worse by the heavy seasonal rain, so it took Francisco almost one hour and 30 minutes to arrive at the hospital. When he arrived, neither the doctor on duty nor the ambulance driver was there. Once they had been called and came back to the hospital, Francisco had to implore the reluctant driver to make the trip in the rain. Finally, after being charged for rubber gloves taken along for the birth—despite the fact Maria was insured under the national health scheme which means she was entitled to two free gloves for each delivery—and a 30 minutes delay, the ambulance set off with Francisco, a doctor and a midwife. Francisco was unable to explain to the doctor or midwife exactly what was the problem and Maria’s history, as he did not speak Spanish well. They had no records for her at the hospital. Approximately 3 hours after Francisco had left the hospital, they arrived at the village but the ambulance was unable to enter into the village due to the rains washing out the mud road. The doctor, driver and Francisco all got 4 out of the vehicle and made their way up the muddy hillside by foot. The midwife, who claimed not to have appropriate footwear, stayed in the ambulance. When they arrived at the house, Maria was not speaking or moving, and there was a lot of blood on the ground. She died a few moments later, before the doctor could examine her. The cause of death was determined as hemorrhage from Uterine Atony, a condition that could have been treated with a simple injection with an inexpensive drug. Following Maria’s death, the family suffered significantly. Francisco took the oldest child out of school so he could help in the fields and with looking after the younger children. The younger children were not well cared for, however, and often stayed home from school, as their brother was lacking in skills and Francisco needed to spend most of his time taking care of the crops and livestock. The new baby, Felix, did not have a birth certificate, as he was not born in a health facility. Without a birth certificate he could not receive his vaccinations or receive assistance through the hospital. Also, without the certificate Francisco could not apply to social programs such as JUNTOS, a cash transfer program to help indigent families. Baby Felix had initially been breastfed by Maria’s sister, but she stopped after a few months as she was also feeding her own child. Felix was given canned milk instead, which was cheaper than formula, but not suitable for infants. Felix died 3 months old. Maternal mortality, low educational outcomes, low life expectancy, infant mortality, low caloric intake, poor health and poverty are just some of the outcomes that stand out in this story. It is a story repeated every day, not just in the Altiplano in Peru, but in many parts of the world. It is a story in which the outcomes are not a result of neutral forces, but are largely derived from the values and discriminatory attitudes held within societies around the world and the way in which they serve to structurally exclude some social groups from a healthy, productive life of their own choosing. 2. Survey data show that discriminatory attitudes are widespread Discriminatory attitudes are widespread, often to a greater extent than an examination of formal institutions, such as law and policy, will reveal (Lawoti 2010, Schwartz 1994, Bertrand and Mullainathan 2004). Often, these prejudices act at a subconscious level, where individuals are unaware of harboring negative attitudes or practicing discriminatory behavior. Through experiments, researchers demonstrate the prevalence of underlying discriminatory attitudes even among those who claim to hold tolerant views. In one such experiment, employers graded fictitious resumes for an advertised position less favorably when the names indicated that the applicant was African-American rather than white (Bertrand and Mullainathan 2004). Employers who listed ’Equal Opportunity Employer’ in their ad discriminated as much as other employers. Cross-section surveys illustrate that discriminatory attitudes are ubiquitous around the world, even in countries where formal institutions embody principles of equality. For the analysis reported in the following section, we rely on the second Life in Transition Survey (LiTS II) which was designed to assess the public’s attitudes (perceptions, beliefs, attitudes and choices) and well-being and the impacts of 5 economic and political change for 29 transition countries and five comparator countries in Western Europe (France, Germany, Italy, Sweden and the United Kingdom). 3 A key question in the survey addresses the normative basis of social exclusion: “Which of the following groups do you prefer to not have as your neighbor?” (see Table 1). We analyze the responses to this question to assess the extent to which negative attitudes toward different groups are held in the different countries covered by LiTS and to identify some of the correlates of those attitudes. We interpret respondents’ answers to reveal not only their housing preferences but also their broader attitudes toward the inclusion and exclusion of specific groups in social life and beyond. Table 1: Response categories and % responding yes (across all sample countries) to the LiTS question: “On this list are various groups of people. Could you please mention any that you would not like to have as neighbors? Please just read out the letter that applies”. Response category % not wanting category as neighbor (all countries) Families with children 4.0 Pedophiles 81.0 Drug addicts 83.3 People of a different race 14.7 People who have AIDS 43.2 Elderly people 4.2 Immigrants / foreign workers 15.9 Homosexuals 54.9 Gypsies 49.3 People of a different religion 10.2 Poor people 5.9 Heavy drinkers 68.8 3 The Life in Transition survey was conducted jointly by the European Bank for Reconstruction and Development and the World Bank in 2006 and 2010. Nationally representative samples of between 1,000 and 1,500 households were interviewed in every country. In total, almost 39,000 households were surveyed in 2010 in 34 countries of Europe and Central Asia. 6 Unmarried couples living together 8.9 Jewish people 11.1 People who speak a different language 7.0 Other (Specify: ) n.a. Source: LiTSII The responses to this question reveal that exclusionary norms are widespread, but also that they differ widely across countries. The basic results are shown in Annex 1; figure 1 offers examples. By a large majority, drug users, pedophiles and heavy drinkers are all unwanted neighbors in almost all the countries covered in LiTS. This is followed by homosexuals and people living with HIV/AIDS, who are undesired neighbors by 55% and 43%, respectively, of those sampled. There is wide variation in respondents’ answers across countries; for example, 77% of Azeris and 4% of Swedes do not want people living with HIV/AIDS as neighbors. For homosexuals, the corresponding figures are 91% and 3.5%, again with Azerbaijan the highest and Sweden the lowest. Similarly, although 49% of respondents do not want gypsies (Roma) as neighbors; this percentage increases to 81% in the Czech Republic and falls to 22% in Bosnia and Herzegovina. Many other groups share the experience of often being unwanted neighbors: immigrants, people with a different religion or of a different race, Jewish, and cohabitating unmarried couples all face intermediate levels of being unwelcome. In Mongolia, Russia, and Turkey, for example, more than 30% do not want to live next to an immigrant. Figure 1: Unwanted neighbors (select groups and countries) 35.0 Which groups would you not like to have as neighbor? Different race 30.0 Immigrants Poor 25.0 Different language 20.0 15.0 10.0 5.0 0.0 Source: LiTS II 7 Some groups are generally more welcome: families with children (not wanted by only 4% of respondents), the elderly (4%), the poor (6%), and people speaking a different language (7%). As with other categories, there is large variation in these attitudes across countries. Twenty-four percent of Turks do not want speakers of other languages as their neighbor. The poor are not particularly welcome neighbors in places like Moldova, Kazakhstan, Armenia, Belarus, Kosovo, and Mongolia (12-19% do not want them). In some countries, even the elderly (17% in Moldova) and families with children (17% in Kosovo) are frowned upon as neighbors. On average, most people indicate 4 to 6 groups they do not want as neighbors, although this number is lower in France and Sweden (2.6 and 2.7, respectively). Dislike of other groups is not confined to Eastern Europe and Central Asia. The World Value Survey (WVS) includes a similarly worded question about unwanted neighbors. In South Africa, 46% of respondents do not want to live next to a homosexual, 14% do not want Indians as neighbors and 9% do not want to live next to a white or a colored person (WVS 5th round, 2007). In Finland, 24% do not want Russians as neighbors. In South Korea, none of the groups asked about are widely accepted: immigrants are unwanted by 39%, people of a different race by 36%, people with AIDS by 94%, and homosexuals 87%. In India, between 40% and 50% of the respondents do not want immigrants, speakers of another language, people with a different religion, race or with AIDS as a neighbor—remarkable given the multiethnic and multi-religious composition of Indian society. Taking a global view, four groups stand out as almost universally untolerated, and often by a large majority: heavy drinkers, drug users, homosexuals, and people with HIV/AIDS. It is not surprising that many people are intolerant of drug users and alcoholics as neighbors (or pedophiles for that matter)— these groups can be associated with risk and noise. The same is not true for homosexuals and people with HIV/AIDS. With the exception of industrialized western countries, these groups are unwanted neighbors by more than a quarter of survey respondents, and in some countries, far more. Discriminatory attitudes are often expressed in behavior These findings do not show whether and how those attitudes are expressed in social and economic life. What they point to is that at some point in our lives, we are likely to interact with people who hold negative attitudes toward us because of the social groups to which we belong. The survey results also illustrate that a large number of people may be subject to amplification of negative attitudes and discrimination flowing from holding multiple identities: e.g. simultaneously being an immigrant, speaker of another language and having another race and religion. Yet negative attitudes measured through the ‘unwelcome neighbors’ survey question are unlikely to remain confined to social contexts. Attitudes to certain groups are likely to be reflections of deeper values that transcend specific situations and contexts (Schwartz, 2009). In the case of frontline service provision, the negative attitudes by providers toward specific client groups often find expression in the manner in which services are provided, even when discriminatory practices transgress the norms and protocols 4 of the providers’ institutions. Teachers in India, for example, have been found to make lower- 4 We use the term ‘off-protocol’ in referring to behaviors that lies outside of the given institutions in which individuals are operating at that time. The term is based on work by Helmke and Levistsky (2004) who refer to 8 caste and dalit children sit on the floor in class, and to talk to them in derogatory terms (Kabeer 2000); doctors and nurses have also been found to withhold care and behave in ways consistent with prejudiced attitudes, as illustrated in Maria’s story. Frontline service providers’ negative attitudes are routinely felt by service users. The LiTS asked respondents about their experiences with health services. Nearly 17% answer yes to a question about ever having been treated disrespectfully by health staff. 5 The Moving out of Poverty study (Narayan, Pritchett and Kapoor, 2009) reported that in the countries included in the study, access to many government programs was conditioned on caste, ethnicity or political party affiliation. Kabeer comments on this aspect of exclusion in the context of the MDGs, citing the reluctance of Indian health workers to visit dalit households as a partial explanation for the incomplete immunization coverage of dalit children. Moreover, people often self-exclude: “both real and anticipated discriminatory behavior and attitudes on the part of health workers deters dalits from using health providers, public and private, particularly for services which involve physical contact, such as giving birth” (Kabeer, 2005 p. 21). A study of dignity in Malawi’s health care system described in detail rude behavior by midwives toward women in labor. One midwife explained that: “patients from the rural areas are treated with less respect but those who are clean and smart are treated with respect because it is assumed that they know their rights”. Another midwife explained that “most patients especially the non-paying patients they don’t know their rights so they do not complain” (Chigwenembe, 2011). Practices that fail to respect patients’ dignity remain widespread even though they violate established professional standards. Thus, negative attitudes toward certain groups appear to be widespread and are likely to find expression in behavior, sometimes leading to exclusion. In order to better understand what might drive intolerant attitudes, we next turn to our statistical analysis. 3. Statistical analysis of intolerant attitudes This section presents an empirical analysis of unwelcoming attitudes and their determinants, using LiTS survey responses. We find three distinct clusters or components of unwelcoming attitudes. This means that the degree to which people are welcoming or rejecting of other groups can be traced back to three distinct sets of values. This allows us to propose a novel distinction of social exclusion attitudes not previously identified in the empirical or theoretical literature: (1) intolerance for the poor and for different lifecycle stages; (2) intolerance toward stigmatized attributes and behaviors; and (3) intolerance to specific identity groups. We infer that that these three distinct value clusters systematically shape attitudes to other groups. Further, our empirical analysis demonstrates that ‘non-institutional behavior’, that is behavior not guided by any norms, and that of Kabeer (2000) who uses the term ‘unruly practices’ in referring to corruption, for example , or discriminatory behavior by teachers. We understand ‘off-protocol’ behavior to be that which is either non-institutional, or that which relies on norms from other institutions. The existence of these behaviors means that even where the norms of a given institution are not inherently discriminatory, or even where institutions attempt to address inequalities, the day to day practices of individuals may nevertheless be discriminatory in nature. 5 The survey question cannot discern what might have caused the disrespectful behavior, however, negative attitudes on the part of the service provider are likely to be an important factor. In comparison, among various complaints about health services, disrespectful treatment by staff ranks broadly similar to bribery and absence of drugs but below problems of long waiting times at health facilities. 9 variation in the presence of exclusionary values is, in part, related to country-specific factors. In contrast, the associations between the socio-economic characteristics of households and individuals’ exclusionary attitudes are weaker than the associations between country-specific factors (history, culture, and other factors that do not vary across households from the same country) and exclusionary attitudes. These results are novel in indicating that three distinct attitude types may drive exclusionary behavior and that country-specific factors are more important in explaining the prevalence of these attitudes than individual-level characteristics. The findings could have important implications for theories about exclusion and for the design of relevant social inclusion policies. Three distinct clusters of intolerant attitudes We begin the analysis by examining clustering among the pairwise correlations of attitudes (see Table 2). A distinct pattern emerges among the correlations. If someone says she does not want to have a gypsy (Roma) as neighbor, chances are she will also not want to have persons of another race or language group, religion, and country of origin (e.g. immigrants) as neighbors. That is, there is a fairly high pairwise correlation of 0.2 to 0.4 between not wanting any of these groups as neighbors. 6 A similar finding emerges for other indicators: for example, there are high correlations between not welcoming drug users, drinkers, and homosexuals. Based on the correlations, three broad clusters of ‘intolerance attitudes’ emerge 7: • Cluster 1: Intolerance for the poor and for different lifecycle stages: not wanting families with children, the poor and the elderly as neighbors. • Cluster 2: Intolerance toward stigmatized attributes and behaviors: not wanting drug users, pedophiles, heavy drinkers, homosexuals, and cohabitating unmarried couples as neighbors. • Cluster 3: Intolerance to different ethnic, religious, and linguistic identity groups: not wanting people of a different race, people speaking a different language, immigrants, people who belong to a different religion, Jewish people, or gypsies (Roma) as neighbors. 8 Next we use principal component analysis to examine whether the twelve indicators of unwelcome neighbor attitudes load on to one distinct measure or separate components. With one exception, the estimates identify the same three distinct components as the simple bivariate correlations discussed above. 9 Thus, each of the three components – intolerance for the poor or different lifecycle states, 6 We cross-checked with other parts of the LiTS survey, including with questions about what kind of people can be trusted. There is a high degree of correspondence in that people who feel that foreigners, people of another religion, or from a different ethnic group cannot be trusted are likely to also not want to have such people as neighbors. 7 The distinctions are largely based on patterns of pairwise correlations, but in some cases a judgment was made so as to maintain similarities across the clusters. 8 Not surprisingly, given the broad nature of these groups, a high share of respondents have a non-zero value of the three indices of intolerance attitudes: around 10% of respondents have a non-zero value of the index of intolerance for the poor and for different lifecycle stages; the same is true for around 95% of respondents in the case of intolerance toward behavioral traits (91% if intolerance of pedophiles is excluded) and for around 85% in the case of intolerance on ethnic, religious, or linguistic identity lines. 9 We dropped attitudes toward unmarried cohabitating couples from our indices, as it did not load on to any of the three factors. 10 intolerance toward stigmatized attributes and intolerance on ethnic, religious, linguistic and identity lines – reflect a unique type of attitude. 10 Although empirically identified, the three clusters make intuitive sense. Cluster 1 relates to lifecycle stages (e.g., the elderly), and most people will at some point in their life belong to one of these groups. Cluster 2 relates to moral judgment (e.g., of homosexuals), and many of the stigmatized attributes and behaviors associated with cluster 2 groups are regarded as ‘socially pathological’ or ‘dysfunctional’ by religious traditions. Cluster 3 relates to the construction of difference based on easily observable traits such as race, language, and country of birth. Figure 2: Bivariate correlations between not wanting any of the listed groups as neighbors To proceed with the analysis we then construct the following indicators of the three clusters: 10 The factor loadings shown in Table 5 are from the three factor estimates are from a varimax rotation. We obtain similar estimates from an oblique ration, which suggests that the three factors are reasonably treated as orthogonal (see Kim and Mueller 1978 for explanations of varimax and oblique rotation). Estimates suggest that that the indicators cannot be used to generate a single latent measure of unwelcome neighbor attitudes. 11 Cluster 1 (Tolerance for the poor and different lifecycle stages): Our first cluster is composed of respondents’ willingness to tolerate the following groups as neighbors: 1) the poor; 2) the elderly and; 3) families with children. Each of the indicators are coded as a “no”/“yes” (or 0,1) variable. We sum respondents’ answers to each of these three questions to create an aggregate index. We follow this same method for cluster 2 and cluster 3. Cluster 2 (Tolerance toward Stigmatized Attributes and Behaviors): Our second cluster consists of respondents’ willingness to tolerate drug users, people living with HIV/AIDS (PLWHA), homosexuals and heavy drinkers as neighbors. Cluster 3 (Tolerance toward Members of Other Identity Groups): The third cluster consists of people’s willingness to tolerate people of difference races, religions, languages, nationalities, as well as Jewish people. 12 Figure 3: Distribution of respondents’ attitudes to the indicators of ‘intolerance attitudes’ Figure 3 shows the distribution of respondents’ attitudes to the indicators that comprise the three clusters. The vast majority of respondents are willing to tolerate living next to the poor and people of different lifecycle stages. By comparison, only a small percentage of respondents are willing to tolerate living next to people with stigmatized attributes or behaviors. These three indicators form the dependent variables for regression analysis of the determinants of intolerance. 13 Independent Variables We construct independent variables to control for respondents’ socio-demographic characteristics, social capital, religious and political values, and economic fragility. Socio-demographic variables: We control for several socio-demographic variables that are likely to affect individuals’ willingness to tolerate members of different groups as neighbors. We control for the level of respondents’ education as we expect educated respondents to be more tolerant of others (Mutz 2002: 112). We also control for respondents’ gender, age income levels, employment status, household size and household ownership, as these factors are also likely to affect individuals’ attitudes toward others. Social Capital: The more an individual interacts with a diverse group of people, the more likely they are to hold tolerant attitudes toward out-group members (Allport, 1954; Cook, 1984; Pettigrew, 1997). We include two measures of respondents’ interaction with others. First, respondents were asked about the frequency with which they socialize with friends. 11 Second, respondents were also probed about whether they belong to any of the following organizations as either an active or passive member: a church or religious organization; a sport and recreational organization or association; an art, music or educational organization; a labor union; an environmental organization; a professional association; and a humanitarian or charitable organization. We coded respondents’ answers to each of these indicators as “0” for non-members; “1” for passive members; and “2” for active members. We then summed these individual indicators to create an index of associational memberships. Religious beliefs: We include a variable, which indicates respondents’ religion in our models, as religion is likely to shape individuals’ tolerance toward people with stigmatized attributes, identities, income levels, and lifecycle stages (Weber, 1930). Individuals who belong to conservative religions are more likely to be intolerant of heavy drinkers and drug users, for example. The extent to which a religion emphasizes the importance of charity and inclusiveness is also likely to affect its followers’ attitudes toward the poor, elderly and other identity groups. Economic fragility: We also expect respondents who are unemployed or who were heavily impacted by the recent economic crisis to be less tolerant of others who they can easily turn to as scapegoats for their economic insecurity (Wimmer 1997). Thus, we include a measure of whether respondents are currently employed and whether they were affected by the recent economic crisis. We also include a direct measure of whether respondents tend to blame other ethnic groups for unemployment in their respective countries. Political attitudes: We include an indicator of respondents’ political attitudes. We expect those who believe in democracy rather than authoritarianism to be more likely to be tolerant of others, since the notions of inclusion and equality underpin democracy. We include an indicator of whether respondents 11 Given that this indicator does not capture the degree of homophily (see McPherson, Smith-Lovin and Cook 2001) of respondents’ social networks, it is a relatively poor indicator of the extent to which individuals interact with members of other groups. Nonetheless, we include this variable in our model as it is the only question that LiTS asks respondents about their socialization with friends. 14 are willing to accept an authoritarian government under certain circumstances or whether they believe authoritarian governments are never acceptable. Regression methods We estimate ordered probit models with fixed effects for countries and robust t-statistics. 12 Rather than presenting regression tables with p-values, we discuss the regression results using predicted probabilities and first differences of attitudes towards neighbors. King (1989) introduced the concept of first differences as distinct from first differences of regressions. First differences and predicted probabilities provide a sense of the size of the effect of variables for quantities of interest. In an ordered probit equation, first differences are estimates of how much the predicted probability would change in response to a particular change in an explanatory variable, while holding the other variables in the equation constant. 13 We present first differences with confidence intervals reflecting the estimation and fundamental uncertainty of the point predictions. To calculate the confidence intervals, we follow the algorithm presented by King (2000) and repeat the algorithm 10,000 times. We graphically present the first differences and the uncertainty around these estimates in coefficient plots (see Figures 4 through 6). In these plots, the first differences are reflected as dark circles and the horizontal lines represent the simulated 95% confidence intervals. For example, Figure 4 shows that moving from Sweden to Moldova corresponds to a percentage point increase of 19.1 in the probability that one is willing to tolerate living next to the poor or people at different lifecycle stages. In the analyses that follow, we calculate first differences for hypothetical scenarios for the average respondent from Sweden compared to Moldova (cluster 1), France compared to Russia (cluster 2) and Croatia compared to Turkey (cluster 3). We calculate first differences that the average respondent will have completely tolerant attitudes (the first category in the ordered probit model) and completely intolerant attitudes (the last category in the ordered probit model). Regression results 12 Technically, the dataset used for this paper has a multilevel structure; individuals are nested within primary sampling units (PSUs), which are nested within countries. The PSUs are the smallest, well-defined geographic units for which reliable population data are available and they tend to be socially homogenous, thereby producing highly clustered data. In most countries, these will be Census Enumeration Areas. Ideally, we should be estimating multilevel ordered logits. Because of the difficulties in implementing this type of model, we estimated ordered logistic models with fixed effects for countries and robust t-statistics. 13 First differences are calculated by holding all variables at their mean and setting one variable first to a low value and next to a high value. We then use an algorithm to calculate the change in the predicted probability with simulated confidence intervals (to give us an estimate of uncertainty around the predicted probability) - the first difference. For example, we calculate the predicted probability that the average respondent from Sweden who is 18 years old won't mind living next to people with specific behaviors. We then calculate the predicted probability that the average respondent from Sweden who is 54 years old won't mind living next to people with specific behaviors. The difference in the two probabilities is the first difference. A hypothetical value of 0.075 can be interpreted as follows: The difference in the predicted probability that an 18 year old versus a 54 year old from Sweden will tolerate living next to people with various different behaviors is 7.5 percentage points. Effect sizes of less than 5 percentage points are fairly small. 15 Overall, the major finding is that country matters: the dominant type of intolerant attitude and how strong it is depends largely on the country in which a respondent is living. Thus, certain intolerant attitudes tend to be widely held by many social strata while showing some tendency to soften with advances in education. However, which attitude and how pronounced it is depends on the country. The implication for inclusion policy is that it needs to be formulated based on an understanding of the national context, and specifically the cultural and historical factors that are shaping the intolerant attitudes that its population tends to hold. Contrary to what we expected to observe, whether an individual was affected by the recent economic crisis does not appear to correlate with their beliefs about whom they are willing to tolerate as neighbors. Similarly, we do not find any evidence that membership in associational activities, our measure of social capital, is associated with tolerance towards other groups. Neither respondents’ gender nor whether they own a home is related to our dependent variables. Cluster 1 (Tolerance for poor/different lifecycle stage): Average Respondent from Sweden compared to Moldova Our first cluster is composed of three indicators: willingness to tolerate the poor, elderly and families with children as neighbors. The results from the ordered probit suggest that respondents’ age, education, income, employment and attitudes toward other ethnic groups are associated with their tolerance for the poor and people at different lifecycle stages. There is no evidence that respondents’ religion or the extent to which they socialize with friends is associated with their tolerance for others. First, country context is highly associated with individuals’ tolerance toward cluster 1. Moving from Moldova to Sweden corresponds to a percentage point increase of 19 in the probability that an individual will be willing to live next to the poor or people of different lifecycle stages. By contrast, moving from Sweden to Moldova increases the predicted probability that an individual is completely intolerant towards the poor and people of different lifecycle stages by a smaller magnitude, 3.5 percentage points. Second, the effects of socio-demographic indicators on attitudes toward the poor and different life cycle stages are larger for Moldovans than for Swedes. Given the lower levels of inequality and higher levels of welfare provision in Sweden than in Moldova, this finding is not surprising. Support for Sweden’s welfare state is likely to go hand in hand with a sense of empathy toward people in different circumstances, including the poor and the elderly. Age is positively associated with tolerant attitudes. The older an individual is, the more likely one will tolerate living next to the poor and people of different lifecycle stages. The difference in the predicted probability of tolerating the poor and people of different lifecycle stages as neighbors for an 18- and 59-year-old from Sweden and Moldova is 1.4 and 5.8 percentage points, respectively. Similarly, the difference in the predicted probability of tolerating the poor and people of different lifecycle stages as neighbors for an employed Swede and Moldovan is 1.0 and 4.0 percentage points, respectively. Education and income are also positively associated with tolerant attitudes in both countries. 16 Third, as we expected, individuals’ attitudes toward other ethnic groups are correlated with their attitudes about whom they are willing to tolerate as neighbors. Individuals who agree with the statement that the presence of other ethnic groups in one’s country increases unemployment are more likely to be intolerant towards the poor and people of different lifecycle stages. However, the effect size of this indicator is quite low in both Moldova and Sweden. Finally, as we also expected to observe, respondents who feel that democracy is preferable to any other political system are also more likely to be tolerant toward the poor and people of different lifecycle stages than those who believe an authoritarian government may be preferable to a democratic one in some circumstances. In Moldova, a change from accepting authoritarian governments to only believing in democracies corresponds to a percentage point increase of 4.4 in the predicted probability of harboring tolerant attitudes toward the poor and people of different lifecycle stages. Cluster 2 (Tolerance toward Stigmatized Behaviors/Attributes): Average Respondent from France compared to Russia Our second cluster consists of respondents’ willingness to tolerate drug users, PLWHA, homosexuals and heavy drinkers as neighbors. The size of the country effects is large. Moving from Russia to France translates into a percentage point increase of 29 in the predicted probability that the average respondent will be willing to tolerate living next to people with stigmatized behaviors and attributes. The relationship between individuals’ age, income levels, and their tolerance toward living next to people with stigmatized behaviors and attitudes is in the opposite direction than what we observed with the first cluster. Younger respondents are more willing to tolerate living next to people with specific behaviors than older respondents. In France and in Russia, the difference in the predicted probability of holding tolerant attitudes is 2.67 and 0.40 percentage points higher, respectively, for an 18-year-old compared to a 59-year-old. Similarly, in France and in Russia, a downward shift from belonging to an upper to a lower income group corresponds to a percentage point increase of 2.15 and 0.33, respectively, in the predicted probability that one holds tolerant attitudes. In both France and Russia, more educated respondents are willing to live next to people with specific behaviors, which is somewhat surprising given the negative correlation between income levels and tolerance. In France, the difference in the predicted probability of holding tolerant attitudes between a respondent who only has primary school education compared to a respondent who has post-secondary education is 5.86 percentage points. In Russia, the predicted probability of holding intolerant views is 6.61 percentage points higher for respondents who only have primary school education compared to respondents who completed post-secondary education. Individuals’ religion also correlates with their tolerance toward others. In both France and in Russia, compared to Orthodox Christians, Protestants or other Christians are more likely to be willing to live next to people with specific behaviors; the first differences are 3.51 (France) and 0.60 (Russia) percentage points. By contrast, compared to Muslims, Orthodox Christians are more likely to tolerate people with stigmatized behaviors and attitudes as neighbors; the first differences are 5.44 (France) and 0.70 (Russia) percentage points. 17 Similar to the findings from the first cluster, respondents’ perceptions toward other ethnic groups are highly correlated with their willingness to tolerate people with specific behaviors as neighbors. In France and in Russia, the difference in the predicted probability of holding tolerant attitudes is 7.33 and 1.25 percentage points higher, respectively, for those individuals who do not agree with the statement that the presence of other ethnic groups increases unemployment. Unlike for the results for the first cluster, there is no relationship between peoples’ attitudes toward democratic and authoritarian governments and their tolerance. Cluster 3 (Tolerance toward Other Identity Groups): Average Respondent from Croatia compared to Turkey The third cluster consists of willingness to tolerate people of different identity groups—race, religion, language, and nationality. Like with the findings from the first two clusters, country context is highly associated with individuals’ tolerance to people of different identity groups. Moving from Turkey to Croatia corresponds to a very large percentage point increase, 43, in the predicted probability that one is willing to tolerate living next to people who are members of a different identity group. Unlike the findings from the first two clusters, there is no association between individuals’ age, as well as income level, and their tolerance of other identity groups. Although employment is associated with attitudes toward neighbors, the effect sizes are relatively small. In Croatia and in Turkey, a change from unemployment to employment corresponds to a percentage point increase of 1.45 and 2.39, respectively, in the predicted probability that one will is willing to tolerate living next to people who belong to different identity groups. Also in contrast to the findings from the second cluster, Muslims are more likely than Orthodox Christians to tolerate living next to people who belong to different identity groups; the first differences are2.35 (France) and 3.73 (Turkey) percentage points. Consistent with the findings from the first two clusters, respondents’ beliefs about other ethnic groups are positively correlated with their willingness to tolerate living next to a member of a different identity group. The difference in the predicted probability of holding tolerant attitudes towards other identity groups is 8.05 and 17.85 percentage points higher for Croatians and Turks, respectively, who believe that the presence of other ethnic groups does not increase unemployment. Also, respondents who believe that democracy is preferable to any other political system are also more likely to be tolerant towards people who belong to different identity groups than those who believe an authoritarian government may be preferable to a democratic one in some circumstances; the first differences are 4.47 (Croatia) and 6.78 (Turkey) percentage points. As we anticipated, our measure of respondents’ participation in social activities is positively associated with their willingness to tolerate members of different identity groups as neighbors. In Croatia and Turkey, a change from never seeing one’s friends socially to seeing one’s friends once or twice a month corresponds to a percentage point increase of 3.27 and 5.27, respectively, in the predicted probability of tolerant attitudes. 18 To conclude, by far the largest and the most consistent finding is that factors associated with the country of the respondent are the most important drivers of attitudes to other groups. The analysis neither indicates what these factors are (but history and culture are good candidates) nor how those attitudes might change. Our findings suggest that in inclusion policy, there may not be a ‘one size fits all’ approach across countries. Instead, countries’ inclusion policies need to consider the question of who is excluded, and why. Our finding that there are three distinct clusters of attitudes may help policy analysts diagnose and set priorities for the types of values that inclusion policy may seek to address. While countering negative attitudes toward specific identity groups is one option for policy makers, the clustering we find may facilitate identification of underlying values that contribute towards negative attitudes to all groups in a cluster. While all groups across the three clusters may suffer exclusion, the different values associated with intolerance to groups in each cluster point to a need for differentiated interventions. Governments, for example, need to evaluate how they can best promote empathy with the poor or the elderly and address the religious and moral traditions that stigmatize specific behaviors or traits. Further, our findings raise the issue of how countries’ norms, values, and institutions may support exclusion – either openly or inadvertently – a question to which we now turn. Figure 4: Regression results (ordered probit) for cluster 1 Note: This figure shows the effect of a change in quantities of interest from their minimum value to their maximum value with 95% confidence intervals on individuals’ tolerance (the first category) or intolerance (the last category) toward living next to the poor and people at different lifecycle stages (cluster 1). The first differences were simulated from an ordered probit model with country-level fixed effects. 19 Figure 5: Regression results (ordered probit) for cluster 2 Note: This figure shows the effect of a change in quantities of interest from their minimum value to their maximum value with 95% confidence intervals on individuals’ tolerance (the first category) or intolerance (the last category) toward living next to people with stigmatized behaviors and attributes (cluster 2). The first differences were simulated from an ordered probit model with country-level fixed effects. Figure 6: Regression results (ordered probit) for cluster 3 Note: This figure shows the effect of a change in quantities of interest from their minimum value to their maximum value with 95% confidence intervals on individuals’ tolerance (the first category) or intolerance (the last category) toward living next to people who belong to out-groups. The first differences were simulated from an ordered probit model with country-level fixed effects. 20 4. What are social norms and what role do they play in social exclusion? This section explores the origins of values, attitudes and behavior and how they can underpin social exclusion. Values underpin people’s attitudes and behaviors and can give rise to exclusion when people from different groups interact. As found in the analysis in the previous section, exclusionary attitudes— and the groups towards which they are directed—are specific to a given society, but grounded in similar social processes. Values and attitudes, behaviors and norms Values lie at the very heart of the process of social exclusion described in this paper; they can be thought of as guiding principles of thought and behavior, serving as standards to evaluate and select actions, policies, people, and events (Schwartz 2009). Values define what is accepted as good, bad, right or wrong, and important. They cause people to take particular positions on social issues (Rokeach 1973). At the same time, as they are communicated to others either explicitly or implicitly via attitudes and behavior, values can affirm to others ‘who you are’ and ‘what you think’ (Raths, Harmon and Simon 1966). As such, their central role in defining oneself and others are critical in the process of social exclusion. Societies develop values through a complex, interactive process involving individuals, culture, ideologies and institutions. Values are found embedded within, and conveyed to us, through institutions such as schools, governments, churches, culture, media, and the family. Our beliefs are molded by values, which in turn give rise to our attitudes about people. Attitudes can be positive or negative, or value-neutral, but once formed, like the values upon which they are based, they are not easily changed (c.f. Roals-Nieto and Segura 2010). It takes new conditions, experiences and information to change attitudes (Wood 2000, Sartorious 2007). Since values are largely drawn upon unconsciously, our attitudes to specific things or people are often based on inconsistencies and incorrect assumptions. Individuals’ behaviors are based on their attitudes. There is a strong, though not always consistent, relationship between values, attitudes and behavior. 14 Behavior that is practiced in common by members of a social group can result in the creation of stable, repeated, patterns of behavior. Acceptance and expectation by the group of these repeated behaviors is what transforms it into a norm. Essentially, norms are what people believe and expect they and others should do. In other words, in any given situation, norms are what are considered ‘normal’ behavior. When individuals repeat their behavior in accordance with norms, the behavior becomes ritualized over time, and some behaviors 14 Often times, individuals’ behaviors don’t match with their personal beliefs- for example when in a situation where strong social norms dictate a behavior that is contrary to a personal belief, an individual may act in accordance with the norm rather than their own belief. The strength of the connections between values, behaviors and norms depends on the strength of values or sets of values in directing behavior. Values held very strongly by an individual are more likely to result in corresponding behaviors. 21 occur unconsciously. Norms also serve to coordinate the behavior of individuals in a group. Coordinated behavior has the effect of providing the group with a sense of shared values, which further reinforces the underlying values, attitudes and behavior. Figure 7 illustrates the relationship between values, behavior and norms. 15 Intergroup dynamics and exclusion Individuals are often classified, by themselves and by others, as belonging to a group or groups. This ‘grouping’ plays an important role in the normative process of social exclusion. When conducting surveys and censuses, researching social phenomena, and executing policies and programs, governments, for example, rely heavily on classifying people into social groups. Yet social groups are not natural phenomena; social groups are constructed by social processes and, as such, are often the product of dominant power relations in society. Defining differences and similarities between people is at the heart of ‘grouping’. This sorting of people into ‘similar’ and ‘different’ exists across all societies and is innate in people (Tajfel and Turner 1985; Crocker and Lutsky 1986). 15 The diagram has been developed based on the VBN (Values-Behavior-Norms) theory proposed by Schwartz (1994) and the description of ‘ritualization’ of gender norms in the classroom found in Moscheta, MacNamee and dos Santos, (2011). 22 Figure 7: The dynamic relationship between values, behaviors and norms is embedded in institutions Political INSTITUTIONS Economic Standardization Coordination Values and attitudes Norms Behavior Ritualization Educational INSTITUTIONS Cultural Classifications help people make sense of and create order out of all the information to which they are exposed; without doing this, people would not be able to make judgments upon which to guide their understanding of the world. When classifying, individuals are often identified based on visual markers or other shared characteristics including skin color, body shape or country of origin. These characteristics may be real or perceived, important or not. However, group classifications often assume a large role in social contexts. That is because attributes of the group are assigned to individual members of the group on the basis of shared characteristics. This phenomenon is popularly known as stereotyping and variously known as ‘typification’ or ‘essentialization’ in academic disciplines. Essentialization has the effect of reducing people down to a few, often easily identifiable and exaggerated markers of difference, and places them within a group of individuals sharing those characteristics. Another aspect of group formation is that group members seek to find negative aspects of other groups, to enhance their own image and to affirm belonging to the group (Tajfel and Turner 1985). Researchers have found that the perception of differing values is an important aspect in the way in which ‘other’ people and groups are constructed (Tajfel and Turner 1986, Young 1990). Applying the idea that ‘other’ groups hold different values, stereotypes are often used to relate certain social groups to societal 23 problems, such as the racial typification of crime in urban USA whereby crime is seen being perpetrated mainly by African-Americans (Free 2003). Likewise, urban problems in many developing countries are often blamed on recent rural migrants and immigrants. The sorting of people and attaching value-based differences to them takes place very early in life. Children have been found to show a considerable negative bias at an early age towards those who do not belong to the same perceived groups as them—along the lines of gender, ethnicity and body-type (Crocker and Lutsky 1986). The environments in which children are socialized – usually the home and school – are the starting points in this process (Powlishta, Serbin, Doyle and White 1994). Teachers often use group labels such as ‘African-American’, ‘male’ and ‘female’, and in other ways reinforce social groupings and stereotypes of the individuals (Bigler and Liben 2007). Further, popular and political representations of groups of people—transmitted through mass media and in other ways—are often absorbed passively, and also influence how members of other groups are perceived (Paluck 2007). The process of defining one’s own group as different from others sometimes takes on extreme forms, characterizing the others as morally suspect, backward or inferior. This is known as ‘othering’. 16 Stigmatization—negative and tenacious labeling—takes place through such devaluing of another social group. Stigmatization and othering are often strengthened as negative emotions become attached to social groups, for example, the fear invoked when thinking of mentally sick people as being dangerous, or certain ethnic groups as being aggressive or violent (Reidpath, Chan and Gifford 2005). Stigmatization embodies various forms of symbolic devaluation – such as found within caste systems in south Asia – which serves to make the other group despised and legitimizes injustices in their treatment. The stigma, once formed and attached to a group, can be self-reinforcing. It is very difficult for individuals from stigmatized groups to break out of the narrow, highly negative way in which stigmatization has defined them. However, while the cognitive processing of differences may be natural, discrimination is not as it is a learned behavior; children learn early in life to react negatively to stigmatized groups (Powlishta and others, 1994). Discrimination also acts in another pernicious way that contributes to exclusion. Values ascribed in the process of assigning difference to social groups can have profound impacts, not only by restricting access to certain resources, but also by being internalized by members of the excluded group. Sociologists and psychologists have long argued that prejudiced attitudes toward individuals and social groups can result in self-fulfilling prophecies whereby individuals from discriminated groups conform to others’ expectations. Studies have found that stereotyping and stigmatization of social groups impede the ability of members to respond to economic opportunities (Hoff and Pandey 2008) and can guide or change professional and life aspirations (Gupta and Bhawe 2007). The academic performance of children belonging to certain 16 ‘Othering’ is the way members of one social group distinguish and distance themselves from, or assert themselves over, another group, by constructing the ‘other’ as being fundamentally different, and crucially, inferior to them (see for example Eyben 2004). Said (1998) documents the way in which ‘othering’ was central to the colonial project because it justified the colonization of people as necessary for bringing religion and civilization to them. 24 stigmatized social groups, for example, is affected by their teachers’ opinions and expectations about their academic abilities (Frier 1993; Rist 2000). In experiments conducted by Hoff and Pandey (2008), lower-caste Indian students performed significantly worse in mental puzzles in situations where they are first reminded of their social status by the announcement of their social identity in the form of their name—a clear give-away of caste—and the presence of higher-caste individuals in the test group. In educational environments, the negativity and lack of self-confidence that arises from this process of internalization affects student learning. Other impacts that have been documented in experiments have been rejection of their social group (Brown and Pinel 1993), a discounting and withdrawing from the task or domain in which the discrimination takes place, (Steele, Spencer and Aronson 2002) and self- blame (Steele and Aronson 1999). Recent research on poverty and the feelings of shame among the poor also highlights the way in which broader societal perceptions of those impacted by poverty have a profound psychological impact on the poor. The impacts of these perceptions include feelings of inadequacy that limit their agency, and experiences of shame associated with poverty which stop some poor people from accessing services. 17 Brain research using neuro-imaging techniques has found that the brain bases of social pain associated with exclusion are similar to those of physical pain. 18 In other words, being socially excluded is a painful experience. Little wonder then that people seek to avoid feeling excluded, for example, by avoiding efforts to join a group or an occupation if they think they might be excluded. 17 See University of Oxford’s global research project on Shame and Poverty, http://www.ox.ac.uk/media/news_stories/2010/poverty_and_shame.html 18 Research subjects were scanned while playing a virtual ball-tossing game in which they were ultimately excluded. Paralleling results from physical pain studies, two parts of the brain called the anterior cingulate cortex and the right ventral prefrontal cortex were active during exclusion and correlated with self-reported distress (Eisenberger et al, 2003). 25 5. Discrimination through social institutions and norms What are institutions and how do they discriminate? Exclusion is grounded in institutions. As the sites where norms are created and maintained, institutions determine the way in which people behave. Institutions are defined as the very ‘rules’ of how we live (North 1990), with ‘rules’ referring to the norms of behavior and to social conventions, in addition to formal rules such as laws and regulations. Laws can be seen as the formalization of norms, implying the use of formal, state sanctions when they are broken. Many norms however are not formalized or codified into law. Norms may be enforced through social sanctions, which can be as rigidly applied as the sanctions for breaking formal laws and regulations. For example, the enforcement of informal norms often takes place through displays of disapproval (e.g. hostile remarks, gossip, and ostracization. Sanctions may also take on more violent (and potentially illegal) forms, such as bullying, intimidation, and even extra-judicial killings (Helmke and Levitsky 2004). The substantive content of institutions is internalized within individuals, and when internalized, norms hold a privileged position in individuals’ cognitive structure. Institutions, whether formal (such as the laws of a state) or informal (such as the caste system, patriarchy, culture, and other social hierarchies and structures) are widespread in all societies. Institutional norms structure many aspects of social life including what is important, who has access to opportunities and how access is obtained. Social exclusion occurs when the way in which institutions assign value and allocate opportunities systematically discriminates against particular groups of people, denying them the opportunities, resources, and recognition that would allow them to participate fully in social, economic and political life (Kabeer 2000). Many anti-discrimination efforts have targeted formal institutions—the constitution, national laws, political and electoral systems, and organizations responsible for service delivery. For example many countries have legislation and policies which enshrine the right of every child to go to school. Yet ‘informal discrimination’ in education is widespread (UNESCO 2010; see case below). This is because the vast majority of social institutions are informal. CASE: Social exclusion and education outcomes: The interaction of ethnicity, language, and poverty 19 Access to education opportunities can help people break out of social exclusion. Yet all too often experiences in school reinforce and perpetuate social exclusion. Children born to parents who are a member of an ethnic or linguistic minority, a particular racial group or a low caste emerge from school with less education and lower achievements than do children from other social groups. Research has found that while poverty matters a great deal in perpetuating educational disadvantage, the non- poverty component is also very important because of the weight of social and cultural discrimination. 19 Source: UNESCO 2010 26 • In Uttar Pradesh and Bihar, India, the caste system still operates in many educational settings. Despite their entitlement to receive a stipend, midday meals and uniforms, very few girls from the musahar or ‘rat catcher’ scheduled caste attend school. Out-of-school girls and their families state that they experience overt stigmatization including ridicule and debasing comments, as well as more subtle discrimination, that reinforces caste hierarchies (e.g. being made to sit on the floor or eat separately from higher-caste children). While government policies have improved, broader social attitudes are slower to change. • The children of pastoralist groups across the world are the least likely to attend school. 85% of young Karamojong pastoralists in Uganda have fewer than two years of formal schooling. Time and distance in accessing schools are factors reinforced by education policy that fails to respond to the needs and realities of their lives. Pastoralists often feel that the curriculum bears little relevance to their lives. They are typically absent from the images and stories in textbooks, or if present, they are often characterized as backward and ignorant, thereby conveying cultural distance between school and the local culture. • In countries where there are minority ethno-linguistic groups that do not speak the official language at home, education contributes to the marginalization of these groups. Cultural and educational inequalities are transmitted through education systems when minority ethnic- linguistic groups experience difficulties in learning and testing in a second or third language. Language policy in education is often a reflection of historical power relations among speakers of different languages. Institutions and the multidimensional nature of exclusion Numerous institutions co-exist simultaneously in any given society, each with some variation in their rules. One of the most basic rules relating to an institution is that of membership and access. Kinship, community and shared values are the common ways in which membership in institutions is defined. These principles of membership define who controls and belongs to the institution. As a variety of institutions operate within a society, individuals may have access to the resources of some institutions and not to others. Membership or belonging may be partial, and exclusion in one domain does not necessarily result in exclusion in another. The relationship between institutions contributes to the intersecting and reinforcing forms of exclusion. For example, exclusion from the political domain can contribute to exclusion in the economic domain (de Haan 1999). This relationship among institutions can set limits to the possibility of one set of institutions counteracting or compensating for the exclusionary effects of another (Wood and Gough 2004). 27 Some highlight the link between cultural exclusion 20 and exclusion in other domains as contributing to patterns of social exclusion. Discrimination against women in the labor market, for example, may result from values derived from religion, the political arena, culture, the structure of the economy, and other institutions in which these values are embedded. In some contexts, these values indicate that while men make better workers and should be the main bread-winners, women should be primarily responsible for child care. Race, ethnicity and sexual orientation are other examples of social groupings where members suffer disadvantage as a result of the external construction of the group, along with a cultural devaluation of traits associated with the group. Social exclusion, conflict, and fragility A growing literature on fragile and conflict-effected environments highlights both the role of exclusion in fragility and conflict, and the importance of ensuring more inclusive political systems and addressing discrimination in post-conflict environments (OECD 2007, UNDP 2011). Researchers emphasize horizontal inequalities (inequalities between social groups) as a key driver of violent conflict (Stewart 2008, 2010) and statistical research finds that levels of discrimination are higher in fragile states than in others. While any type of horizontal inequality— political, economic, social or cultural—can provide incentives for political mobilization and violent conflict, research has stressed the central role of cultural inequalities. Cultural differences – real or perceived – are sometimes exaggerated to bind together excluded groups. This may further increase perceptions of identity differences (Stewart 2010). Much contemporary violent conflict occurs when ethnic or religious differences between groups result in economic and political differences and create resentments that may lead to violent struggles (Stewart and Brown 2007). In some cases, a common identity strengthened through the experience of economic or political inequality is manipulated by elites to mobilize a group to legitimate the use of violence (Aspinall 2007, Croissant 2009). Approaches that emphasize inclusion in political systems do not fully address the underlying causes of conflict. Researchers have found that inclusive political systems alone are not sufficient in tackling the kinds of exclusion that can result in violent conflict. Political inclusion must be accompanied by a social system which widely spreads the benefits of progress, providing socioeconomic growth among the different ethnic and religious groups (Ostby 2008). The specific form of the state structure, however, will not be able to fulfill this role if it is built upon a system that discriminates (Wimmer 2002). Addressing the value-based discrimination that leads to cultural inequalities is the key to preventing violence in fragile and conflict-affected environments. Discrimination is seen as underlying violent conflict. Perceived or real injustices arise from peoples’ identity and the social groups to which they belong. What this tells us is that as well as tackling the outcomes, or manifestations of inequality, it is important to address the underlying values and processes that create discrimination. 20 Cultural exclusion entails the devaluing of entire- or parts of- cultures, associated traditions, beliefs and practices of the cultural group and often the lack of recognition of the existence and importance of such cultures in the formal institutions of the state. 28 6. Addressing exclusion by transforming the normative basis How can development strategies and operations tackle the values and behaviors that result in many forms of social exclusion? Social exclusion is a complex process, involving various social, psychological and political processes and actors. Many development approaches address exclusion at the level of outcomes and emphasize, for example, education and general economic development. Few interventions tackle directly the normative basis of social exclusion. Would it be possible to address the norms and attitudes that underpin exclusion? That is the question to which we now turn. Tackling values, attitudes, and behaviors Values are critical aspects of social life: they give meaning to what we do, and we seek to do what is in line with these values. As earlier discussed, the internalization of values and attitudes takes place in early childhood (see also Singelman and Singleton 1986). Researchers therefore stress the importance of ensuring early exposure of toddlers and children to a balanced set of attitudes about different social groups. Multicultural awareness and tolerance, it is suggested, will prepare young children to act in ways that are more tolerant and harmonious. This aspect of educational content matters greatly and could be emphasized more in education policy. The way in which education is delivered also matters. Ensuring that teachers are not a source of prejudiced attitudes to which children are exposed is of importance in limiting discriminatory attitudes (Vuckovic 2012). Values work in a system where some values are strongly related to and reinforce others (Rokeach 1973, Schwartz 1994). Attitudes and behavior are not guided by single values, but by the possible trade-offs between opposite values. For example, the relative strength of values that emphasize ‘openness’ over those that emphasize security and conformity guide whether an individual is more likely to hold prejudiced attitudes. Sagiv and Schwartz (1995) illustrate that Israeli Jews making social contact with minority Arabs in Israel are those who value ‘change’ more than ‘conformity’. Efforts to bring about changes in behavior toward greater inclusion have to consider the values that might be harmed by transformed behavior. Sagiv and Schwartz’s findings suggest that specific values that contribute to discrimination, such as security, tradition and power, do not need to be tackled head-on; it may be more feasible to seek to increase the importance of an opposing value such as universalism, openness to change or benevolence, and as such arrive at the desired change in behavior. This work could be explored further for application to anti-discrimination policy in developing countries. Although values are durable and hard to change directly by outside intervention, their expression can nevertheless be countered. People’s basic values are largely fixed when they reach adulthood and change relatively little thereafter (Rokeach 1973). The tenacity of values is one of the reasons why efforts to change values take a long time to show results. The process is slow, and most development programs have short-term objectives. However, researchers have noted that people adjust their values to fit their circumstances. They upgrade the importance they attribute to values they can easily attain, compared with those that are blocked (Schwartz and Bardi 1997). This means that laws and sanctions that enforce inclusion, and other ways of blocking channels that transmit values and prejudiced attitudes, can play a role in addressing the normative basis of social exclusion. 29 The formation of stereotypes and negative attitudes towards stereotyped groups plays an important role in social exclusion. Research in psychology and education in particular has found ways of reducing the saliency of existing stereotypes and negative attitudes. Messages that blur group boundaries by illustrating similarities rather than differences between people have been shown to be important in preventing the formation of stereotypes and associated prejudicial attitudes in school settings. Blurring group boundaries also works to disrupt the internalization of existing stereotypes and stigma – and the associated disabling impacts for members of the discriminated against groups. Experiments found that when group boundaries were blurred prior to standardized tests being taken (for example by exercises asking female respondents to write down similarities between men and women), the stereotyped group (women in this example) did far better than those that had not ‘blurred boundaries’. These findings have direct implications for the design of education programs in contexts where groups face social exclusion and strong stigmatization. There are also potential implications for broader development policy. Policies, programs and development research could explore more contingent and less fixed categories of target groups and employ less categorical and less aggregated ways of viewing the nature of groups’ problems (Zetter 1991). The internalization of stigma, thinking and behavior by stigmatized groups that contributes to their own disadvantage is an important aspect of the process of exclusion. Hoff, Kshetramade and Fehr (2006) found that in India, castes that were historically denied rights were likely to continue with behavior that was more favorable to higher-caste group members than to those of their own social groups. This has significant policy relevance: interventions that exclusively target formal institutions addressing historically excluded groups may be more likely to succeed if they are complemented by actions aimed directly at raising those groups’ understanding and perception of their rights and entitlements. Other researchers, looking at stereotyping in education, have highlighted the importance of providing positive role models that affirm the abilities of individuals from stereotyped or discriminated-against groups (Bigler and Liben 2007). Some social scientists are skeptical about the possibilities for changing values, beliefs, attitudes and norms. In this view, discriminatory behavior may be a more promising entry point for addressing discrimination than the underlying values, norms and attitudes (Bem 1970, Sartorious 2007, Wood 2000). While there is no consensus among researchers, the point is that the non-linear, inter-related nature of the exclusion process may yield multiple entry points. Disrupting one part of the process–e.g., behavior—can alter the values held by groups and individuals. Likewise, changing values can lead to changed behavior (Paluck 2007). Yet another potential approach is to transform the institutions in which norms are embedded. We discuss this next. 30 CASE: Addressing LGBT health needs and discrimination in Brazil and India 21 Gender and sexual minorities face some of the most severe and most diverse forms of exclusion among any group worldwide. These groups have far worse health outcomes than heterosexual people. Factors like low rates of health insurance coverage, high rates of stress due to systematic harassment and discrimination, and a lack of cultural relevancy and competency in health care systems, leave gender and sexual minorities at a higher risk for health problems such as mental illnesses, cancers and other diseases. Some governments have taken steps in breaking the stigma and discrimination that contribute to the social exclusion of lesbian, gay, bisexual, and transgender (LGBT) groups. Often these involve integrated strategies that involve changes in policy, better targeting and more attention to needs in the provision of health services, along with social marketing campaigns to tackle stigma among the wider population. In Brazil, the government adopted policies that would assist the adaptation of healthcare to the needs of gender and sexual minorities. This complements a long history of cumulative progress on gender and sexual minority issues such as the promulgation of anti-discrimination laws targeting the actions of businesses and citizens, the legalization of same-sex unions, and widespread public education campaigns to combat prejudice among the general population and in schools. Tamil Nadu state in south India has put in place a series of official measures to address the exclusion of transgender people. The state established a transgender welfare board in 2008, with the social welfare minister as the president. The Board undertook an enumeration of the transgender population in all districts along with the issuance of identity cards with a third gender identity. With these identity cards, the state government has also been able to start issuing ration cards (for buying food and other items from government-run fair-price shops) to transgender people. The Board also addresses a range of critical issues that contribute to the exclusion of transgender people such as education, income generation and social security. Government health services have also been adapted to better meet the needs of transgender individuals. Tackling exclusion in informal institutions Informal institutions are often more salient and may yield a very different set of values and associated norms than formal institutions. Tackling the prejudices in operation in informal institutions is often thought of as a long-term endeavor as they are highly resistant to change (North 1990). This is not always true of formal institutions and organizations, many of which can, literally, be changed overnight via policy and administrative reforms. There are a number of ways in which change may be brought about, possibly quite rapidly. Thus, we highlight the importance of understanding the inter-play 21 Sources: Institute of Medicine (2011), Ottosson (2009), Beyrer, Wirtz, Walker, Johns, Sifarkis and Baral, (2011), UNDP (2010) and Krehley (2009). 31 between formal and informal institutions, and the way in which they can complement, conflict or substitute for one another. Much change in informal institutions takes place through the evolution of societal values, as described above. Changes in the status quo that sustain informal institutions – such as asymmetries in the distribution of power and resources- will also result in change in the institution itself. The two types of change described above tend to be very slow, but incremental, in nature. However, research has also documented rapid changes in culture that have led to the collapse of informal institutions through a shift from one set of norms to another. The discontinuation of foot-binding in China took place fairly rapidly, within a single generation, catalyzed by the creation of an alternative ‘market’ for wives whose feet had not been bound (Shell-Duncan et al 2010). These types of changes have informed an understanding of the nature of the way in which norms governed by informal institutions can change (Mackie 1996, 2000). 22 Emphasis is placed on coordinating between a range of actors to initiate alternative norms, for example gaining a critical mass of individuals to simultaneously break with the norm, along with education and formal sanctions for those who persist with old practices can bring rapid changes in both practices and their acceptability. This approach may have potential for addressing female genital mutilation (Shell-Duncan, Wander, Hernlund and Moreau 2011, Mackie 2009, UNPFA and UNICEF undated). Modern formal institutions can sometimes be used to change informal exclusion. Take for example the case of legal dualism in which customary rules are inequitable and potentially exclusionary. Modern and more equitable and inclusive laws are often ignored. However, Aldashev et al. (2012) show that under some circumstances, modern law can shift the conflicting custom in the intended direction. The trick is for modern laws to offer an ‘exit option’ in the form of an alternative dispute resolution mechanism to which the discriminated group can appeal. This may force the conflicting custom to accommodate change in order to remain relevant and prevent ‘exit’ by the disadvantaged group. Thus, by changing a formal institution, one can change an informal institution and the impact of the change is greatest when the informal and formal institutions are related. Improving the effectiveness of formal institutions can also bring about change in competing informal institutions. Better enforcement of laws and regulations can alter the costs and benefits of following informal rules, such as with US federal enforcement of civil rights legislation and the resulting impact that this had on the informal ‘Jim Crow’ practices that led to continued discrimination against African-Americans. Table 2 offers an overview. 22 Known as Social Convention Model or Social Convention Theory, this approach has been applied to Harmful Social Practices. The relevance of such an approach to wider discriminatory social practices is not discussed explicitly, however, it could ostensibly apply, at the least, to symbolic dehumanizing practices set within a broader set of exclusionary norms. 32 Table 2: Transforming informal institutions Source of change Mechanism of Change Pace of Change Formal institutional change Change in the effectiveness of Variable formal institutions Formal institutional change Change in the design of formal Often relatively rapid institutions Cultural evolution Change in societal values Very slow Changes in distribution of Bargaining Often slow, incremental power, resources Coordinated change approach Shift to new set of norms Rapid Source: Adapted from Helmke and Levitsky 2004, p 733 Religion plays a key role in the generation and transmission of values. Often, religion is a source of prejudiced attitudes about certain groups. Norms and behavior originating from religious beliefs may be seen as difficult to break, as religion is often characterized as being impervious in nature. However, scholars and activists have illustrated that religions evolve, refining and even revolutionizing previous beliefs and norms. While all religions make claims to truth, this truth is mediated by practical, cultural and even legal considerations. As such, development actors should not shy away from addressing religion as a source of discriminatory values. In fact, this point applies not only to religion, but to the range of entrenched culturally-based informal institutions. Even when changes occur within informal institutions due to endogenous or exogenous factors (e.g. changes in formal institutions that result in changes in informal institutions), the effects of past discrimination can persist over time for well-identified groups. Hoff and Pandey (2008) illustrate the persistence of the caste system despite legal reform and how affirmative action policies try to redress caste discrimination. Similarly, the widespread presence of discriminatory attitudes—as found in the survey data presented above—persists even in countries where formal institutions are free from discrimination. How can such informal discrimination be tackled? A first step may be to draw people’s attention to their own prejudiced attitudes and discriminatory behavior of which they may often be unaware. 23 We discuss some ways in which stigma around HIV/AIDS in health care can be addressed. CASE: Improving health-service delivery for people living with HIV 24 23 A simple test has been developed for this. Bertrand, Chugh and Mullainatan (2005) propose the Implicit Association Test, a way of exposing through a series of quick-fire written and visual tests peoples’ attitudes to members of excluded groups. They contend that the test can draw peoples’ attentions to the biased attitudes that they unknowingly hold. Further, they argue, it is a cheap, easy to administrate and direct way of addressing deep- seated and widely-occurring prejudices 24 Sources: Nyblade, Stangl, Weiss and Ashburn (2009) and Feiss, Abebe, Girma and Woldie (2012) 33 Stigma around HIV/AIDS is widespread across the world. The health sector is one of the main settings where people living with HIV (PLWHIV) and those perceived to be infected experience stigma and discrimination. Health workers delivering essential health services to those PLWHIV are among those who hold and perpetuate such stigma. Stigma and discrimination in health care settings and elsewhere contribute to keeping people from accessing services for the prevention, care and treatment of HIV, and from adopting key preventive behaviors. Stigma impacts health and health care: • In Ethiopia, healthcare professionals often refused to treat HIV+ patients. High-levels of misinformation as a consequence of poor training were found among these professionals. • In South Africa, those on ARV treatment often skip or delay doses or misdose themselves by grinding pills into powder in order to avoid detection of their HIV status, even by their families. • In Nigeria and Mexico, infected persons were seen as being responsible for their own infection, and as deserving of infection as punishment for sexual misbehavior. • In Zambia, health workers who become infected with HIV are seen as having ‘failed’, leading them to avoid disclosing their status even to their families. • In Indian hospitals, practices by health workers such as the burning of bedding, charging for the cost of infection-control supplies, using gloves even during non-physical interactions, and informing family members without consent serve to stigmatize HIV patients and deter others from seeking treatment. Research has identified the issue as having multiple dimensions:  Stigma in health care settings and more broadly in society acts as a barrier to testing and to accessing services – both for the general population and for health workers themselves. The consequences of this for health workers and health facilities, when HIV+ health workers delay or avoid care and become seriously ill or die, cause further strain on an overburdened health care system.  There is limited recognition of the link between HIV related stigma and public health outcomes such as patient quality of care, and health workforce capacity. Stigma and discrimination by health workers affects the quality of their care, which is often a critical factor in whether patients adhere to medications and maintain their overall health and wellbeing.  There is insufficient capacity among health care managers regarding how to effectively address stigma and discrimination through programs and policies.  There is a misconception that stigma attached to HIV is too pervasive and strong – often rooted in tenacious cultural and religious beliefs – to effectively change. Successful strategies for tackling stigma A number of strategies have been identified for tackling the stigma of HIV in healthcare settings. 34  Individual level Much of the discrimination is based on a lack of knowledge, even by health care workers, about the transmission of HIV. Fears and misconceptions arise from this lack of knowledge, and fuel the process of stigmatization. Health care workers have also been successfully trained on dealing with stigma in healthcare settings – what it is, how to recognize it, what impact it has on the provision of care and what to do about it.  Environment level Health workers need to have not only the information, but also the supplies and equipment necessary to practice universal precautions and prevent occupational transmission of HIV. This includes gloves for invasive procedures, sharps containers, adequate water and soap or disinfectant for hand washing, and post-exposure prophylaxis in case of work-related, potential exposure to HIV. Researchers find that posting relevant policies and procedures or other critical information in key areas in the health care setting enables health workers to maintain better quality of patient care.  Policy level The lack of specific policies or clear guidance related to the care of patients with HIV reinforces discriminatory behavior. Policies are most successful when developed in a participatory manner and clearly communicated to staff. Stigma reduction activities in hospitals have led to positive changes in health providers' knowledge, attitudes and behaviors, and better quality of care for HIV+ patients. In Vietnam, for example, health centers have eradicated practices such as marking of hospital beds of patients with HIV, or leaving medical charts and instructions in public view.  Monitoring Monitoring of attitudes and behaviors of healthcare workers is an important aspect of tackling stigma and providing better health care for HIV patients. Regulations that mandate monitoring of staff in health care settings, and the setting of benchmarks that health facilities can use for assessing their efforts are examples of successful strategies. This section argues that institutions are a key to a strategy to address social exclusion because they guide and constrain behavior through norms they espouse. However, institutions are durable because they are useful in creating stable expectations of the behavior of others. In other words, institutions work because the rules involved are embedded in shared and repeated thoughts and behaviors of many individuals. Changing these patterns is possible and this section discusses how strategies that address social exclusion may be able to tackle the underlying institutions and related individual behavior that result in discrimination. 35 7. Conclusion and implications for development approaches While social exclusion is both a process and an outcome, this paper has focused on the former rather than the latter. We have characterized the process of social exclusion as one in which individuals are grouped based on certain characteristics, assigned negative attributes based on the grouping, and excluded from full participation in society based on those perceived attributes. As we show with the LiTS and WVS data, while exclusionary beliefs are widespread across space, the causes of and the nature of the beliefs are country-specific. Many current approaches to social exclusion emphasize improving individuals’ and groups’ access to certain public services and enhancing their income-generating activities. Those approaches may usefully be complemented by efforts to address exclusionary processes. As we have argued, these processes are rooted in discriminatory values, attitudes, and behaviors that shape the social meanings and status assigned to various social groups and influence their opportunities and outcomes. Discriminatory norms- -in particular norms that lead to a lack of voice and power--can be addressed directly or indirectly through changing the informal and formal institutions that are producing and perpetuating these norms. We provide examples of development interventions that target discriminatory values, norms and institutions, typically by focusing on formal institutions such as laws, policies, and political systems. Yet, since many informal institutions contribute to social exclusion, they should be central to strategies for tackling social exclusion. For example, in cases where people’s religious beliefs lead them to hold intolerant attitudes toward particular groups or behaviors, development interventions could seek to engage with religious leaders. Informal rules may largely endure changes in formal rules because informal institutions are culturally derived and can take far longer time to change than formal state institutions which can be established or dismantled as quickly as the bureaucracy is capable. Failure to consider the stickiness of informal institutions runs the risk of missing the most salient incentives and constraints to peoples’ behavior. Moreover, efforts to promote inclusion in public services inevitably encounter the problem of discriminatory “off-protocol” practices (such as treating clients from certain groups disrespectfully) by service delivery staff rooted in widely held social norms. Using data covering the countries of Eastern Europe and Central Asia, we studied the responses to a simple hypothetical survey question about which specific groups respondents would not like to have as neighbors. We find that unwelcoming attitudes to other groups are surprisingly common across countries. Many groups, including immigrants, ethnic minorities, the poor, HIV+ individuals, and drug users, are viewed negatively. However, the specific groups facing exclusive attitudes differ markedly across countries, suggesting that each country will need to identify the particular set of values and attitudes in need of revision. Moreover, we find that unwelcoming attitudes can be grouped into three distinct sets of values, allowing us to propose a novel categorization of social exclusion attitudes: (1) intolerance for the poor and for different lifecycle stages; (2) intolerance toward stigmatized attributes and behaviors; and (3) intolerance to specific identity groups. Country-specific factors (history and 36 culture) are far more important than respondents’ socio-economic characteristics for explaining these attitudes. The research surveyed in this paper has important implications for forwarding the understanding of social exclusion, and gives rise to some novel directions for anti-exclusion development interventions. The work of Bigler and Liben (2007), Steele and Aronson (1995), Hoff and Pandey (2008), Mackie (2000 and 2009), Shell-Duncan et al (2011) and others illustrates the potential for transforming stereotyping, the formation of prejudice, the assignment of stigma, and discriminatory off-protocol practices within state organizations. Development practitioners may apply greater care when assigning and using labels for different social groups; consider tackling off-protocol discriminatory behaviors among service delivery staff through sensitivity education and monitoring systems; and reform educational practices to blur social groups’ boundaries in the classroom. The common element is that altering values is fundamental to addressing social exclusion, but requires carefully articulated strategies for targeting and coordinating a wide range of actors and their values, attitudes and behaviors. More specifically, the following implications for anti-exclusion strategies deserve to be highlighted: • Values are at the very basis of social life, structuring society and giving order and meaning to the world. There is a positive feedback link between values, beliefs about other people and behavior towards them. Addressing only exclusionary behaviors may bring about change, and is likely to do so in the short-term and in specific contexts. Simultaneously addressing the values underpinning exclusion and discriminatory behavior may be more effective, longer-term and more far-reaching in impact. • Development approaches often take predefined social groups as their starting points. However, these groupings are often part of the exclusionary processes. Development practitioners need to grapple with how to address the social exclusion of groups of people while being attentive to the ways in which labeling and targeting may contribute to the construction and maintenance of groups and the stigma that often attaches to them. Pre-conceived notions of needs and solutions attached to specific social groups should be questioned. Starting with asking people to identify their challenges with their own words and to facilitate their self-diagnosis of potential solutions has for many years beem a hallmark of good development practice, yet is rarely done well.Formal and informal institutions can offer effective entry points for tackling many exclusionary norms and practices. Strategies to address social exclusion could therefore usefully be informed by an understanding of the different social institutions at work in a given country or area and how these institutions mediate disadvantage, exclusion, and discrimination, including via norms, attitudes and behaviors. • Children form fixed views of groups in early life. Education can therefore play a critical role in addressing the normative basis of exclusion. The focus should be on the relevance of the curricula, the language of instruction, the values transmitted in the content of education, and the way in which teachers deliver the curricula and interact with children, rather than simply ‘more education’. Education policies should emphasize diminishing the creation of rigid group boundaries and the attribution of negative values to social groups. Teacher training could include explicit components to halt potentially discriminatory behavior and to transform prejudiced attitudes. 37 • Negative attitudes and discrimination can be internalized by members of excluded groups who may choose to self-exclude and discriminate against their own group. Strategies to address the normative basis of exclusion should target the attitudes and perceptions of the excluded themselves. • Policy reform and capacity building in state agencies, such as in health or education, should seek to transform the underlying values of the policy and the practices of those who work within the agencies. Efforts to reduce the exclusionary attitudes and off-protocol behaviors of staff include efforts to ensure the compliance of providers with rules and regulations, as well as education that tries to change the staff’s attitudes toward excluded groups. In this regard, targeting providers and state organizations such as teachers, health staff, the police and other law and justice agencies which both have the greatest interface with excluded groups and which are important in embodying and transmitting the values of the state, is a priority. • Formal state institutions and their employees are far from the only actors that play a role in exclusion. Informal institutions and customary leaders are often involved in the maintenance of values and attitudes that contribute to discrimination and exclusion, for example religious figures, cultural leaders, or traditional chiefs. Approaches to shift informal institutions and their leaders toward greater equity and inclusion should be considered. 38 References Aldashev, Gani, Imane Chaara, Jean-Philippe Platteau, Zaki Wahhaj, Using the law to change the custom, Journal of Development Economics, Volume 97, Issue 2, March 2012, Pages 182-200. Alexander, M. 2010. The new Jim Crow: mass incarceration in the age of colorblindness. New York: New Press. Allport,G. W. 1954. The Nature of Prejudice. Reading, MA: Addison-Wesley. Aspinall, E. 2007.‘The construction of Grievance’, Journal of Conflict Resolution, Vol. 51, no. 6, pp950-972. Bem, D. J. 1970. Beliefs, Attitudes, and Human Affairs, Belmont, Calif.: Brooks/Cole. Bertrand, M. and s. Mullainatan, ‘Are Emily and Greg More Employable than Lakisha and Jamal? A Field Experiment on Labor Market Discrimination’, the American Economic Review, Vol. 94, No. 4 (September 2004), pp991- 1013. Bertrand, M., D. Chugh, and S. Mullainathan, New approaches to discrimination: Implicit Discrimination, in The American Economic Review, Vol. 95, No. 2, Papers and Proceedings of the One Hundred Seventeenth Annual Meeting of the American Economic Association, Philadelphia, PA, January, 7-9, 2005. (May, 2005), pp. 94-98. Bigler, R. S., & Liben, L. S. 2007. ‘Developmental intergroup theory: Explaining and reducing children's social stereotyping and prejudice’, Current Directions in Psychological Science, 16, pp162-166 Beyrer, C. A.L. Wirtz, D. Walker, B. Johns, F. Sifarkis, and S.D. Baral. 2011. The Global HIV Epidemic Amongst Men who have Sex with Men, Washington D.C.:World Bank Brown, R. P., & Pinel, E. C. 2003. ‘Stigma on my mind: Individual differences in the experience of stereotype threat’. Journal of Experimental Social Psychology, 39,pp 626–633. Chigwenembe, L. 2011. Dignity In Maternal Health Service Delivery: Cross sectional survey on factors that promote or compromise dignity in maternal health service delivery, University of Oslo Faculty of Medicine, M.Phil dissertation. Crocker, J and Lutsky, N. 1986. Stigma and the dynamics of social cognition in S. C. Ainley, G. Becker, & L. Coleman (Eds.) The dilemma of difference: A multidisciplinary view of stigma, New York, NY: Plenum Press . Croissant, A., U. Wagschel, N. Swank and C. Trinn. 2009. Culture and conflict in Global perspective: The cultural dimensions of conflicts from 1945 to 2007, Gutersloh: Verlag Bertelsmann Stiftung. Daniel, De Haan, A. 1999. ‘Social Exclusion: Towards an Holistic Understanding of Deprivation’, Department for International Development, London. Eisenberger, N. I., M.D. Lieberman and K.Williams. 2003. Does Rejection Hurt? An fMRI Study of Social Exclusion, Science 10 October 2003: Vol. 302 no. 5643 pp. 290-292. Evans, D. G. Human Rights and State Fragility: Conceptual Foundations and Strategic Directions for 39 State-Building Journal of Human Rights Practice Vol 1 Number 2 , 2009 pp. 181–207 Eyben R. 2004. 'Inequality as Process and Experience' in Eyben R., and Lovett J., Political and Social Inequality: A Review, IDS Development Bibliography 20, Institute of Development Studies, Brighton, pp 32-39. Fehr. E, and K. Hoff, ‘Tastes, caste and Culture: the influence of society on preferences’, Policy research working paper 5760, The World Bank Development Research Group, Macroeconomics and Growth team, August 2011, Washington: World Bank. Feyissa, G., L. Abebe, E. Girma, and M. Woldie (2012) Stigma and Discrimmination against eople living with HIV by healthcare providers, Southwest Ethiopia, in BMC Public Health 2012, 12:522 Fraser, N. 1995. ‘From Redistribution to Recognition: Dilemmas of Justice in a ‘Post-Socialist’ Age’, New Left Review (212), 68-93. Free, M. D. 1993. Racial Issues in Criminal Justice: the case of African Americans, Prager: Westport th Friere, P. 1993. (20 Anniversary Edition) Pedagogy of the Oppressed, New York: Continuum. Gupta, V. K, and N. M. Bhawe. 2007. The influence of proactive personality and stereotype threat on women's entrepreneurial intentions. Journal of Leadership and Organizational Studies, 13, 73-85. Helmke, G. and S. levitsky. 2004. Informal Institutions and Comparative Politics: A Research Agenda, Perspectives on Politics, Vol. 2, No. 4 (Dec., 2004), pp. 725-740 Helmke, G. and S. levitsky. 2006. Informal Institutions and Democracy: Lessons from Latin America. Baltimore: the John Hopkins University. Hickey, S. and du Toit, A. 2007. ‘Adverse Incorporation, Social Exclusion and Chronic Poverty’, Working Paper 81, Chronic Poverty Research Centre, University of Manchester Hoff K. and P. Pandey. 2008. 'Economic Consequences of Social Identity: Discrimination, Social Identity, and Durable Inequalities', American Economic Review, Volume 96, Number 2, pp 206-211 Hoff, K., Kshetramade, M. and Fehr, E. 2011. ‘Caste and Punishment: the Legacy of Caste Culture in Norm Enforcement’, The Economic Journal, 121: F449–F475 IDS, ‘The Power of labeling in development practice’, IDS Policy Briefing, Issue 28, April 2006, Brighton, Sussex: Institute of Development Studies. IOM (Institute of Medicine) 2011. The Health of Lesbian, Gay, Bi-Sexual and Transgender People, Building a Foundation for Better Understanding, Washington, D.C., The National Academies Press Kabeer, N. 2005. Social exclusion: concepts, findings and implications for the MDGs King, G. 1998. Unifying political methodology: The likelihood theory of statistical inference. University of Michigan Press. Koster, F. and M. E. Kaminska, ‘ Welfare state values in the European Union, 2002–2008. A multilevel investigation of formal institutions and individual attitudes’, Journal of European Public Policy, Sep 2011 40 Krehley, J. 2009. ‘How to Close the LGBT Health Disparities Gap’, Washington, D.C.: Center For American Progress Lauth, H.J. ‘Informal Institutions and Democracy’, Democratization, Vol 7, Issue 2, 2000 Lawoti, M. 2010. ‘Informal Institutions and Exclusion in Democratic Nepal’, Himalaya, the Journal of the Association for Nepal and Himalayan Studies: Vol. 28: No. 1, Article 2. Lerita, M. and V. Coleman .2006. ‘Stigma an enigma demystified’, in in L. J. Davis, 2006 (ed.) Reader in Disability Studies,pp 131-140 Mackie, G. 1996. ‘Ending footbinding and infibulation: a convention account’, American Sociological Review, 61, 999e1017. Mackie, G. 2000. Female genital cutting: the beginning of the end’, in B. Shell- Duncan, & Y. Hernlund (Eds.), Female "circumcision" in Africa: Culture, controversy, and change, pp. 253-283, Boulder, CO: Lynne Rienner Publishers. Mackie, G. 2009. ‘Social Dynamics of Abandonment of Harmful Practices: A new look at the theory’, Innocenti Working Papers 2009-06, UNICEF. Manski, C. “Economic analysis of social interactions”, Journal of Economic Perspectives, 14, 2000, 115-36 Martin, R. 2009. ‘Classification and the homogeneity of social groups’ Princeton, New Jersey, draft submission for ASA conference. Available at http://faculty.chicagobooth.edu/workshops/orgs- markets/archive/pdf/Ruef_Class&Homogeneity.pdf Moncrieffe J., 2004, 'Labelling, Power and Accountability: How and Why our Categories Matter', in Moncrieffe, J. and Eyben, R., The Power of Labelling, Earthscan. Moscheta, M, S. McNamee, and J. C. Santos, ‘Dialogue and transformation: embracing sexual diversity in the educational context’. Education Review [online], 2011, n.39, pp. 103-122. Mutz, D. C. "Cross-cutting social networks: Testing democratic theory in practice." American Political Science Review 96.1 (2002): 111-126. Narayan, D, L. Pritchett and S. Kapoor, (2009) Moving Out of Poverty Volume 2: Success from the Bottom Up, Palgrave McMillan and World Bank. Nyblade, L., A. Strangl, Weiss, E. and Ashburn, K. (2009) Combattinh HIV stigma in healthcare settings: What works? In Journal of the international AIDS Society, 12:15, available from: http://www.jiasociety.org/content/12/1/15 North. D. C. (1990) Institutions, Institutional Change and Economic Performance, Cambridge: Cambridge University Press. OECD-DAC. Principles for Good International Engagement in Fragile States and Situations, April 2007, OECD. Ostby (2008) Political, environment and civil conflict: Evidence from 55 developing countries in F. Stewart (ed) Horizontal inequalities and conflict: Understanding group violence in multi-ethnic societies, Basingstoke: Palgrave, pp. 136-157. 41 Ottosson, Daniel. May 2009. ‘State-sponsored Homophobia: A world survey of laws prohibiting same sex activity between consenting adults’, International Lesbian and Gay Association (ILGA). Paluck, E. J. (2007) ‘Reducing intergroup prejudice and conflict with the media. A field experiment in Rwanda’. HICN Working paper 34, Brighton: Institute of Development Studies. Panth, S. ‘Changing Norms is the Key to Fighting everyday Corruption’, CommGap Discussion Paper, Communication for Governance and Accountability Program (CommGap), October 2011, World Bank: Washington. Pejovich, S. ‘The effects of interaction of formal and informal institutions on social stability and economic development’, The Journal for Markets and Morality,2, no. 2, 164-181 (Fall 1999) Pettigrew T. F. 1997. ‘Generalized intergroup contact effects on prejudice. Pers. Soc. Psychol. Bull.’ 23, 173–185. Powlishta, K. K., Serbin, L. A., Doyle, A. B., & White, D. R. (1994). Gender, ethnic, and body type biases: The generality of prejudice in childhood. Developmental Psychology, 30, 526 – 536. Raols-Nieto, J.G. and A. Segura, ‘Intergenerational differences in materialism and post-materialism values’, International Journal or Psychology and Psychological Therapy, 2010,10:3, pp499-12 Raths, L.E., M. Harmon, M. and S. B. Simon. (1966) Values and teaching: Working with values in the classroom, E. Merill Books, Columbus, Ohio. Reidpath, D.D. , K.Y. Chan, S.M Gifford, P. Allotey,’ 'He hath the French pox': stigma, social value and social exclusion’ in Sociology of Health and Illness, 2005 May;27(4), pp 468-89. Rist, R.C. (2000) Enduring Dilemmas of Class and Color in America. Harvard Educational Review 70 (3) pp257-301. Rokeach, M. 1973. Understanding Human Values, New York: Free Press. Sagiv, L. and S. H. Schwartz. ‘Value priorities and readiness for out-group social contact. Journal of Personality and Social Psychology, Vol 69(3), Sep 1995, pp437-448. Said, E. 1979. Orientalism, New York: Vintage Sartorious, N. Stigma and Mental Health, in The Lancet, Volume 370, Issue 9590, 8 September 2007, pp 810-811. Schwartz, S. H. ‘Basic Human Values’, Paper Presented at the Cross-national comparison seminar on the quality and comparability of measures for constructs in comparative research: method and aplications, Bolanzo, Italy, June 10-13, 2009. Schwartz, S. H. 2006. Les Valeurs de le Base de Personne, Theorie, Mesure and Application, Revue francaise de sociologie, 42, pp249-248. Schwartz, S. H. and A. Bardi. 1997. ‘Influences of adaptation to communist rule on value priorities in Eastern Europe’, Political Psychology, 18, 385-410. Schwartz, S.H. ‘Are There Universal Aspects in the Content and Structure of Values?’, Journal of Social Issues. 1994, pp 19-45. 42 Shell-Duncan, B, K. Wander, Y. Hernlund and A Moreau, ‘Dynamics of change in the practice of female genital cutting in Senegambia: Testing predictions of social convention theory’, Social Science and Medicine, 73,(2011), pp1275-1283. Shelley, T. 2006. ‘On the government of disability : Focault, power and the Subject of Impairment’, in L. J. Davis, (ed.) Reader in Disability Studies, pp185-196. Sigelman, C. K. and L. C. Singelton. 1986. ‘Stigmatizaton in childhood, a survey of developmental trends and issues’ in S. C. Ainley, G. Becker, & L. Coleman (Eds.). The dilemma of difference: A multidisciplinary view of stigma, New York, NY: Plenum Press . Steele, C. M., Spencer, S. J., & Aronson, J. 2002. Contending with images of one's group: the psychology of stereotype and social identity threat. In M. Zanna (Ed.), Advances in Experimental Social Psychology. San Diego: Academic Press. Steele, C.M. and J. Aronson (1995) Stereotype threat and the intellectual test performance of African-Americans, Journal of Personality and Social Psychology, 69, pp797-811. Stern, P. C. 2000. ‘New Environmental Theories: Toward a Coherent Theory of Environmentally Significant Behavior’, Journal of Social Issues, Volume 56, Issue 3, pp407-424. Stewart, F. HORIZONTAL INEQUALITIES AS A CAUSE OF CONFLICT, A Review of CRISE Findings, World development Report 2011, Background Paper, August 2010, Washington: World Bank Stewart, F. 2008. Horizontal Inequalities and Conflict: Understanding Group Violence in Multi-ethnic Societies, Basingstoke: Palgrave Macmillan Stewart. P and K. Brown. 2007. Greater than the sum of its parts?: Assessing whole of government approaches to fragile states, New York; international Peace Academy. Tajfel, H., & Turner, J. C. 1986. The social identity theory of intergroup behaviour. In S. Worchel & W. G. Austin (Eds.), Psychology of intergroup relations (pp. 7–24). Chicago, IL: Nelson-Hall. UNDP. 2010. Issue Brief, Hijras/Transgender Women in India: HIV, Human Rights and Social Exclusion UNESCO. 2010. Education For All Global Monitoring Report UNFPA and UNICEF, ‘Female Genital Mutilation/cutting: Accelerating Change, Funding proposals, Joint Programme’, (no date) available at www.unfpa.org/public/publications/pid/1294 Vickers, G. 1973. Values, Norms and Policy, in Policy Sciences, Vol 4, no.1 103-111. Vuckovic, A. ‘Making the multicultural learning environment flourish: the importance of the child-teacher relationship in educating young children about diversity’. Australian Journal of Early Childhood. 2008 Weber, Max, P. R. Baehr, and G. C. Wells. 2002. The Protestant ethic and the" spirit" of capitalism and other writings. Penguin Classics. 43 Wimmer, A. 2002. Nationalist exclusion and Ethnic Conflict, Shadows of Modernity; Cambridge University Press: Cambridge. Wood, W. 2000.‘Attitude change: Persuasion and social influence and social influence’, Annual Review of Psychology, 51, 539-570. World Bank. World Development Report 2012, Gender Equality and Development, Washington: World Bank. Yamin, A. E. 2007. Deadly Delays: Maternal Mortality in Peru. A Rights-based Approach to Safe Motherhood. Washington: Physicians for Human Rights. Young, I. M. 1990. Justice and the politics of difference, Princeton Press, NJ. Zetter, R. ‘Labeling Refugees. Forming and transforming a bureaucratic identity’, Journal of Refugee Studies, 4, 1, (1991) pp39-62 44 Annex 1: Statistical results Table 3: Unwanted neighbors Fa mi l i e Pa edo- Drug Di ffe- Peopl e El derl y Immi - Homos Gyps i e Di ffe- Poor Dri nker Unma rr Jewi s h Di ffe- Tota l s wi th phi l es us ers rent wi th gra nts exua l s s rent s i ed rent number chi l dre ra ce AIDS rel i gi o peopl e l a n- (count n n l i vi ng gua ge of 15) togethe country Al ba ni a 6.4 72.2 67.8 12.4 42.5 6.4 4.5 63.1 26.6 4.5 4.5 57.0 6.2 14.4 2.7 3.9 Armeni a 2.1 67.9 88.0 31.3 72.2 4.0 12.4 70.2 34.7 28.0 13.9 67.2 15.4 16.9 8.5 5.3 Azerba i ja n 2.0 94.1 94.9 27.5 77.2 2.5 16.9 91.2 57.3 24.6 2.1 77.2 30.5 37.1 3.5 6.4 Bel a rus 6.1 87.6 91.9 14.3 48.1 6.9 21.3 71.1 69.3 8.4 12.2 83.8 3.2 10.4 7.7 5.4 Bos ni a a nd Herz 8.5 81.6 73.5 10.0 41.1 7.9 7.6 53.6 22.4 7.3 5.1 55.3 7.2 7.8 4.4 3.9 Bul ga ri a 1.5 87.2 92.4 14.6 48.0 1.6 10.9 59.8 59.3 7.1 1.9 75.8 1.2 3.0 3.2 4.7 Croa ti a 0.7 84.4 77.5 6.6 30.9 0.6 7.1 43.2 26.2 2.3 1.1 56.4 1.0 3.3 2.2 3.4 Czech Republ i c 1.7 86.2 89.3 23.4 47.0 3.0 27.0 25.2 80.9 11.0 4.8 76.1 1.8 6.3 9.0 4.9 Es toni a 2.4 85.3 94.4 8.4 31.9 2.0 24.1 45.5 65.9 4.5 2.7 87.1 0.9 5.9 3.4 4.6 Fra nce 0.9 80.8 60.6 4.4 4.3 0.3 6.3 8.1 38.8 1.8 0.3 43.5 0.3 2.1 3.0 2.6 Georgi a 0.1 64.8 93.7 13.1 53.0 0.8 7.8 80.6 33.2 13.1 0.6 71.8 7.3 5.9 4.3 4.5 Germa ny 2.1 74.8 75.1 12.5 13.8 2.9 17.0 15.7 52.6 4.4 2.7 74.8 0.5 5.6 8.1 3.6 Grea t Bri ta i n 2.3 89.3 89.1 5.9 13.6 0.5 17.9 16.7 56.5 2.2 1.7 65.4 1.7 1.7 7.7 3.7 Hunga ry 5.6 71.7 75.2 19.4 45.8 5.2 24.2 51.7 56.4 5.2 4.3 63.9 3.3 6.6 5.4 4.4 Ita l y 0.6 92.4 75.1 13.1 27.7 0.8 13.8 21.7 77.4 5.7 1.9 60.3 0.4 5.5 3.8 4.0 Ka za khs ta n 5.9 85.3 89.2 8.1 48.6 6.2 21.1 72.1 56.3 6.8 15.1 82.2 4.0 2.7 2.3 5.1 Kyrgyzs ta n 3.3 77.1 77.9 20.7 66.9 4.4 15.7 79.6 60.4 23.8 7.8 77.8 20.9 22.5 13.3 5.7 La tvi a 2.9 85.1 96.1 8.3 39.2 1.0 18.0 58.7 54.3 3.8 3.7 85.2 1.0 3.2 0.9 4.6 Li thua ni a 5.5 86.9 88.1 16.8 56.3 4.5 14.3 60.4 60.2 11.9 7.4 44.6 2.3 9.2 2.4 4.7 Ma cedoni a 4.6 92.7 90.3 19.8 66.6 5.5 19.8 75.5 24.6 10.0 5.3 75.1 9.4 15.6 5.7 5.2 Mol dova 4.4 82.3 82.3 31.0 61.7 16.8 23.8 79.5 64.5 29.9 19.0 74.1 21.3 27.5 22.0 6.4 Mongol i a 3.9 87.0 82.7 23.6 50.8 5.6 31.3 57.3 23.3 17.5 13.3 80.0 10.1 19.0 17.2 5.2 Pol a nd 0.8 74.8 70.4 9.6 39.5 1.1 8.3 45.5 40.7 5.6 1.7 50.6 1.2 12.0 5.2 3.7 Roma ni a 3.0 73.8 69.6 7.9 28.0 2.8 9.3 62.8 49.2 4.8 2.9 60.0 4.2 5.6 5.1 3.9 Rus s i a 3.3 86.9 91.7 15.2 47.1 4.1 29.0 72.0 72.4 9.7 7.9 83.9 3.1 5.4 10.9 5.4 Serbi a 3.3 82.8 86.8 7.7 32.9 3.5 8.9 54.3 18.2 5.3 3.8 64.8 3.2 4.0 3.7 3.8 Sl ova ki a 0.9 81.5 85.7 10.1 41.4 2.4 11.7 39.7 73.4 2.9 3.0 71.1 0.3 6.7 3.2 4.3 Sl oveni a 4.1 86.1 81.1 10.8 25.6 3.0 17.6 35.0 58.1 6.4 4.5 47.8 3.1 9.9 6.0 4.0 Sweden 2.5 78.1 85.4 1.9 3.9 0.5 2.2 3.5 28.9 1.7 0.8 58.6 0.2 0.9 0.9 2.7 Ta ji ki s ta n 3.6 86.1 85.6 21.3 62.3 2.0 10.3 81.6 61.8 26.1 5.1 70.6 38.5 21.5 8.7 5.9 Turkey 6.2 83.1 83.4 30.4 56.9 8.2 31.5 72.3 49.5 25.4 8.5 72.6 50.1 35.1 23.8 6.4 Ukra i ne 3.2 86.6 93.5 14.1 47.4 4.5 18.7 68.9 69.8 9.8 9.1 85.7 3.1 6.5 8.9 5.3 Uzbeki s ta n 10.8 42.3 82.6 12.5 60.2 8.8 18.5 57.0 49.6 8.7 10.6 70.6 30.2 22.6 8.6 4.9 Kos ovo 16.8 74.7 75.2 18.3 39.7 10.8 16.4 63.8 27.4 13.3 12.7 68.9 20.1 17.2 14.4 4.9 Montenegro 6.7 82.2 79.5 8.6 40.4 5.1 10.4 63.6 25.5 3.6 5.1 66.8 5.4 8.1 5.0 4.2 45 Table 4: Correlation matrix of intolerant attitudes Cohab Speakers Families Differe With Heavy iting of a with Paedo Drugu nt HIIV/A Immig Homose Different Poor drinke unmar different children philes sers race IDS Elderly rants xuals Gypsies religion people rs ried Jewish language Families with children 1 Paedophiles -0.19 1 Drugusers -0.18 0.34 1 Different race 0.10 0.07 0.11 1 With HIIV/AIDS -0.01 0.16 0.26 0.31 1 Elderly 0.46 -0.13 -0.13 0.18 0.03 1 Immigrants 0.06 0.11 0.15 0.45 0.28 0.11 1 Homosexuals -0.06 0.27 0.31 0.23 0.46 -0.01 0.25 1 Gypsies -0.07 0.20 0.29 0.22 0.25 -0.02 0.28 0.26 1 Different religion 0.14 0.03 0.06 0.44 0.25 0.21 0.33 0.20 0.17 1 Poor people 0.24 -0.01 0.00 0.22 0.15 0.33 0.22 0.10 0.09 0.26 1 Heavy drinkers -0.12 0.20 0.42 0.12 0.23 -0.07 0.16 0.27 0.26 0.10 0.02 1 Cohabiting unmarried 0.16 0.00 0.05 0.27 0.23 0.20 0.20 0.21 0.06 0.35 0.15 0.07 1 Jewish 0.11 0.04 0.07 0.36 0.28 0.18 0.29 0.23 0.15 0.42 0.18 0.09 0.439 1 Speakers of a different language 0.11 0.02 0.06 0.38 0.21 0.18 0.35 0.16 0.16 0.43 0.25 0.09 0.283 0.383 1 Legend: cluster 1 Intolerance for the poor and for different lifecycle stages Cl us ter 2 Intolerance toward specific behavioral traits Cl us ter 3 Intolerance on ethnic, religious, or linguistic identity lines Table 5: Factor loadings for intolerance attitudes Intolerance for Intolerance Intolerance on poor/different toward ethnic, religious, lifecycle stigmatized linguistic, stage attributes identity lines Factor Factor Loadings Loadings Factor Loadings Families with children 0.60 Elderly 0.70 Poor People 0.48 Drug Users 0.56 PLWHA 25 0.54 Homosexuals 0.60 Heavy Drinkers 0.49 Different race 0.60 Different Religion 0.62 Jewish 0.54 Different Language 0.56 Foreign 0.49 25 PLWHA is an abbreviation for People Living with HIV/AIDS. 46