108 Denlers, Gabriel, and Torcal are due to different signs of the factor loading coefficients. Interpreted in a substantial sense, the three concepts of citizenship are positively correlated. 6 Far this purpose we have computed three indices, one for each of the dimensions resulting from the previously reported factor analyses. Each index therefore is based on the sum score of two items. All items were rescaled so as to have a minimum of 0 and a maximum of 1. These three indices will also be used for subsequent analyses. 7 This is also confirmed by the relatively modest standard deviations for the mean support rates. For law-abidingness, with a mean score of 0.70, the standard deviation was 0.21; for criticism and deliberation, the mean was 0.77 and the standard deviation 0.18. And finally for solidarity, the mean was 0.70 and the standard deviation 0.20. If we look at comparable figures per country, we get similar results. 8 See Coleman 1988; Staub 1989; Putnam 1993, 2000; Verba el al. 1995; Feldman and Steenbergen 199B; Warren 1999a, 2001; Claibourn and Martin 2000; Dekker and Uslaner 2001; Gabriel el al 2002; Hooghe and Stalle 2003b; and Stolle 2003. 9 In this chapter we will not analyse the potential effect of people's involvement in different types of organisations. Although previous work (for example, Putnam 1993; Foley and Edwards 1996; Cohen 1999; Putnam 2000: 31-115; Stolle and Rochon 2001; Stolle 2003) has shown that not all types of organisations will have the same impact on the transmission of civic virtues, a detailed analysis of this topic would be beyond the scope of this chapter. 10 Information on operationalisation is provided in Chapter 2 for social trust and in Chapter 3, note 13, for the measures of organisational involvement. 11 This may be seen as problematic, because there is likely to be reciprocal causality between the dependent and the independent variables. For our purposes, however, it is not necessary to get an exact estimate of the relative strengths oF the reciprocal causal effects; therefore, we can ignore this complicating factor. 12 Therefore, die results of the extended models will be examined only: 1 to see if controls affect the impact of our two key explanatory variables; 2 to establish the reladve explanatory power of these two factors vis-á-vis the control variables; 3 to gauge the gains in additional explanatory power by introducing these additional factors, 13 The use of this composite index is justified by both the relative strength of the correlations between the three factors and by the basic similarities in the analytical results reported in Table 4.4. The composite index is computed as the average score of the three separate indices. 14 A low score indicates a position on the far left; a high score indicates a position on the far right. 5 Political and social tolerance Jose Manuel Leite Viegas Introduction In the final quarter of the twentieth century, a considerable number of countries made the transition from totalitarian or authoritarian regimes to democratic government, particularly in southern Europe, eastern Europe, and Latin America. While the introduction of democracy does not automatically imply respect for the rules and the law that the constitutions stipulate, comparative empirical studies show that support for democracy has risen among the citizens of these new democracies (Klingemann 1999; Norris 1999b) as has the acceptance of basic democratic values (Dalton 1994; Thomassen 1995). At first glance, we might think that tolerance would be included in these trends. After all, political tolerance springs naturally from the egalitarian and pluralistic principles of liberal democracy: respect for individual liberties, equality before die law, acceptance of diverse opinions and lifestyles, rejection of arbitrary conduct in complying with die law, and the endorsement of minority rights. The problem, however, is in fact somewhat more complex. Democratic pluralism is based on the recognition that modern societies are heterogeneous and diat social differentiation may entail positive results in terms of social and political progress. But the majority principle, another rule of democracy, may also jeopardise respect for the opinion and behaviour of less popular or minority groups in society, a risk that Alexis de Tocqueville ([1835, 1840] 2000) clearly emphasised. It was recognised early on tiiat political intolerance is not restricted to non-democratic societies. It represents a serious problem, which has been analysed empirically in contemporary democracies, particularly in the postwar period. Until die work of Sullivan et al. (1982), empirical analyses concurred with die thesis that political tolerance had tended to increase in democratic countries in the postwar period, mainly due to higher levels of education among their citizens (Stouffer 1955; Davis 1975; Nunn el a!.. 1978). Later studies confirm this tendency. They also confirm the importance of education and date of birth, both of which are of course associated with 110 Leiie Viegas social and economic modernisation (Davis 1975; McClosky and Brill 1983; Sniderman el al. 1989; Wilson 1994; Thalhammer et al. 2001). This theoretical orientation, which we may term the 'standard theory', has been contested by the so-called 'revisionist' theory. Sullivan et al (1982) question earlier results, in particular the seminal work by Stouffer (1955), based on what they consider to be a theoretical as well as methodological bias. For the progenitors of the 'revisionist theory', 'tolerance implies a willingness to "put up with" those things one rejects or opposes' and, on a political level, 'it implies a willingness to permit the expression of ideas or interests one opposes' (Sullivan el at 1982: 2). Consequently, the empirical analyses of discriminatory attitudes towards a single target group (in the American studies, a left-wing group), without considering how close the respondents were to this group, were not, strictly speaking, measuring tolerance but rather tire lessening, in time, of the rejection of people with such ideological tendencies. I believe, however, that these theories are flawed by a rationalistic positivism that is inconsistent with the real situation. Tolerance does not just develop after one gets to know another person and his or her ideas. It also involves a predisposition towards understanding someone else's motives and accepting them, albeit conditionally. The attempt to understand other people's ideas in their context and on the basis of their complex causality - 'the sober second thought', as some analysts term it-can mitigate the overall rejection of an individual or a group, with this process becoming an integral part of tolerance. It should be added tiiat other empirical studies question the linear nature of the relationship between the degree of intolerance and the proximity to members of a given social group (Sniderman et al. 1989; Gibson 1992b). In analysing attitudes of political tolerance, we are not simply dealing with a diversity of tastes in relation to distinctive social groups. If this were the case, we would be concealing the existence of exclusion and discrimination processes involving certain unpopular and minority target-groups. Political tolerance has to be measured according to the acceptance by the individual of the rights and liberties of people belonging to unpopular or minority groups, on equal terms with all other citizens. Political tolerance begins, therefore, in the sphere of citizenship. My analyses in this chapter, however, will not focus only on political but also on social tolerance. The latter is defined as the extent to which acceptance in the various areas of social life is accorded to potential targets of discrimination such as political and religious extremists, people belonging to various types of stigmatised groups, and ethnic minorities. The rise of new phenomena or ones with new configurations or dimensions, such as AIDS, drug addiction, crime, terrorism, and ideological fundamentalism, and the increase in ethnic minority groups in developed societies have awakened feelings of insecurity and reinforced the stigmatisation of Political and social tolerance 111 different social groups. These processes are known to nourish attitudes of intolerance. The question is how these new situations manifest themselves at the beginning of the twenty-first century with respect to attitudes of (in) tolerance in democratic societies, where social, cultural, and lifestyle differences are increasingly accepted. In this chapter, attitudes towards political exclusion and social discrimination registered at a given moment are taken to be the product of two factors: 1 the beliefs and attitudes acquired during primary socialisation (Sears 1993: 121) or, in a more comprehensive formulation, the habitus (Bourdieu 1980: 88), that is, the symbolic and ideological structures generated during the experiences acquired in different life trajectories; 2 the impact of conjunctural problems (terrorism, drugs, crime, religious and political extremism), which largely depends on the level of information on these problems. I also presume that the structure of beliefs and values acquired in socialisation and life trajectories will determine the different ways in which individuals perceive situations of insecurity and, in particular, their degree of tolerance. Where the wilhngness to accept social difference is limited, any aggravation in a target group's situation tends to provoke general attitudes of social exclusion and discrimination. As Sniderman et al. (1989: 4) put it 'For the people only loosely attached to the value of tolerance, it can suffice that a group is out of the ordinary or merely unfamiliar to excite an intolerant response'. Where there is a greater willingness to accept difference, citizens respond to situations of insecurity in more selective ways: they might react to a given target group without serious repercussions on other target groups, and distinguish between social discrimination and political exclusion, for civic reasons as well as for reasons related to the perception of danger. These ideas will guide the empirical analysis. I shall therefore seek to distinguish attitudes towards social target-groups that only differ in their socio-cultural characteristics or lifestyles, even if unpopular, from attitudes towards social target-groups that, on the other hand, are seen as a social threat. In the next section of this chapter, the levels of political and social intolerance in our sample of countries will be analysed in a comparative perspective. I will also present a theoretical framework of macro-social characteristics that will allow us to understand the cross-national differences in tolerance. In a subsequent section, I mil attempt to explain measures of tolerance on the individual level by relating them to other aspects of social and political experiences and thinking. 112 Leite Viegas A comparative analysts of political exclusion and social discrimination Some scholars have advanced the economic development in modern societies as one of the preconditions for the establishment of democracy (Lipset 1960). From a more general point of view, there are other conditions for the institutionalisation of democratic values and governments: the social division of labour, functional interdependence, the creation of markets, free cooperation between people, the development of individual responsibility, the spread of scientific and technological knowledge, and a greater exchange on the individual as well as societal level (Lipset 1994). More open, pluralistic, and interdependent societies promote feelings of trust in others, which is generally accepted as a basis for tolerance. This developmental point of view would not be complete without mentioning that, by destroying the old forms of integration in communities, capitalist development generates new forms of social and political inequality and exclusion. Relationships of integration and citizenship must therefore be interpreted, together with the often conflicting capitalist development and democratisation of the state (Marshall [1963] 1973). It is in this matrix of social relationships that the principles of liberalism, citizenship, and democracy have been created and broadened. But it is the latter aspects, these political principles, that more directly determine the levels of tolerance in a society. Democratic relationships institutionalise die interplay of peaceful discussion and deliberation between groups with differing ideas and interests. With the necessary respect for the political opinions of minorities, democratic relationships thus help to spread a culture of tolerance and participation (Muller and Seligson 1994). Let us begin with an analysis of die factors of democracy and citizenship. The main idea that I intend to develop is clearly and concisely expressed by Peffley and Rohrschneider (2003: 5) who argue that 'citizens in a more stable democratic nation have more opportunities to practice or observe toleration through elections, pluralistic conflicts of interests, and this should increase citizens' appreciation of tolerance'. Within this framework, Peffley and Rohrschneider consider the 'number of years that the country has lived under a stable democratic regime' and whether the system is 'federal' or 'non-federal' as factors explaining political tolerance. The existence of a democratic regime and its durability over time will be an important dimension to consider. But it must be complemented by other democratic indicators. By necessity, a legal, democratic framework includes some of the requirements already mentioned - electoral competition, the freedom to oppose, and respect for the law - which, irrespective of the different democratic institutional formulae, immediately produce, to a certain extent, the effects in question. However, the effect on tolerance will also depend on the type of democratic institutions and the way they operate. Consequently, Peffley and Rohrschneider add Political and social tolerance 113 federalism as another indicator of 'checks and balances' in die way democracy operates. I will introduce yet another aspect of democratic functioning: the degree of political inclusion. It could easily be hypothesised that die greater the number of people involved in these processes, the more pronounced the effects of tolerance will be. In odier words, tolerance will depend on how far the rights of citizenship go and how far tiiey are effectively exercised. Therefore, my first hypothesis is that more enduring and inclusive democracies will show higher levels of tolerance. This hypothesis involves two dimensions: the duration of democracy and the degree of political inclusion. The indicators used are: 1 the number of years uninterrupted democratic rule; 2 die average voter turnout in national parliamentary elections between 1945 and 2000 (only considering democratic elections); 3 die year in which women were given the right to vote. I acknowledge from the outset that the last two indicators are not per-fecdy ideal. To begin with, political involvement cannot be reduced to electoral participation. Furthermore, formal inclusion could extend, for example, to die political rights granted to non-nationals or the ease with which they can acquire formal citizenship in their country of residence. In practice, however, the choice of indicators is largely dictated by availability. The second hypothesis is related to socio-economic factors. As we know, the development of the welfare state succeeded in creating higher levels of education, social security, remuneration, and employment, minimising die effects of social exclusion that result from die process of development. With higher standards of living and higher levels of education, citizens tend towards values more centred on the individual and his or her well-being. In this context, we also find an increasing acceptance of different lifestyles (Inglehart 197Y). A similar idea was presented already in Stouffer's seminal study of tolerance. According to Stouffer (1955: 236): great social, economic, and technological forces are operating slowly and imperceptibly on the side of spreading tolerance. The rising level of education and the accompanying decline in authoritarian child-rearing practice increase independence of thought and respect for others whose ideas are different. Analyses of the processes of modernisation may incorporate, at a more specific level, different options of social development. Everything else equal, a high level of social integration is more likely in countries where it is supported by continuous public policies than in countries where it is not. The 'Nordic model' of the welfare state, which is inclined to address citizenship in all its aspects, in particular die political aspect, fosters 114 Leite Viegas participation more than the 'Continental' model, which tends to address the materialistic values of family security (Esping-Andersen 1991). My second hypothesis, which focuses on the end product of these policies, is that higher levels of social well-being and socio-economic integration promote tolerance. Two aspects should be considered: the level of social development and well-being on the one hand, and the extent of integration as indexed by the levels of social marginalisation on the other. As a measure of the former, I employ the Human Development Index (HDI), which is in turn based on three indicators: life expectancy, education, and GDP. As a measure of the latter, I use the average rate of unemployment in the decade preceding our survey. Certain technical reservations are in order with respect to the second of the two measures. The progenitors of the HDI refused to include it into their index on die ground that conditions of comparability are met only for countries within the OECD. In addition to this problem, data for the most recendy established democracies cover only a few years. Despite these limitations, the indicator is included in the model, since it matches the theoretical concept Overall tolerance levels by country I intend to analyse attitudes of political exclusion and social discrimina-tion towards different social groups: extremist groups, stigmatised groups, and ethnic minorities. To monitor these attitudes, respondents were asked whether any of the social groups presented to diem: 1 should be allowed to hold public meetings; and 2 would be acceptable as neighbours. The groups included 'Christian fundamentalists', 'Islamic fundamentalists', 'left-wing extremists', 'right-wing extremists', 'racists', 'people with a criminal record', 'drug addicts', 'people with AIDS', 'homosexuals', 'immigrants', and 'people of a different race'. The tendency to exclude politically and discriminate socially with respect to die entire set of groups is presented in Table 5.1, which indicates the mean number of groups mentioned on a scale ranging from 0 (no group) to 1 (all groups). Evidendy, there are significant differences between countries with regard to political exclusion as well as social discrimination. In line with the results obtained by others (Gibson 1992a; Thalhammer et al. 2001; Peffley and Rohrschneider 2003), the macro-social characteristics of each country appear to exert a strong influence on overall levels of political and social tolerance. Comparing the two types of intolerance, the figures tend to be higher for social discrimination than for political exclusion, especially in Denmark, Sweden, the Netherlands, Political and social tolerance 115 Table 5.1 Political exclusion and social discrimination" Country Political exclusion Social discrimination Denmark 0.10 (0.16) 0.1S (0.18) Portugal 0.15 (0.21) 0.17 (0.18) Sweden 0.19 (0.20) 0.32 (0.23) The Netherlands 0.21 (0.19) 0.29 (0.19) Norway 0.22 (0.21) 0.34 (0.24) West Germany 0.25 (0,23) 0.32 (0.26) Spain 0.27 (0.22) 0.26 (0.28) East Germany 0.29 (0.21) 0.31 (0.22) Switzerland 0.32 (0.21) 0.37 (0.21) Russia 0.38 (0.24) 0.40 (0.25) Romania 0.43 (0.33) 0,44 (0.35) Moldova 0.46 (0.35) 0.47 (0.36) Note a Means with standard deviations in parentheses. Countries are ordered and grouped by the degree of political exclusion. Slovenia lias been omitted since only some of the required information was collected in that country. Norway, West Germany, and Switzerland. In countries where political exclusion is more prevalent, this difference is not evident. We will return later to these variations. In any case, it is clear that die countries can be divided into three groups based on the level of political tolerance. The first group includes Denmark, Portugal, Sweden, the Netherlands, and Norway, the second West Germany, Spain, East Germany, and Switzerland, and the third Russia, Romania, and Moldova. How do the two hypothesis fare based on this grouping? Table 5.2 shows the level of political tolerance along widi the values of the political and socio-economic factors singled out by the hypotheses. Let us first consider the impact of the political characteristics. Switzerland excepted, all countries with a continuous democratic life of at least 80 years belong to the most tolerant group. The contrary holds true as well. The most recent democracies (less than ten years old) all belong to the group of least tolerant countries. Countries that are intermediate in terms of the duration of democracy (more than ten years but less than 80) are intermediate in terms of political tolerance as well, except for Portugal, which in spite of its relatively short recent democratic history falls into the most tolerant group. The remaining political characteristics may shed some light on the deviant behaviour of the Portuguese and Swiss cases. Switzerland displays the lowest average turnout, and was the last country to give women the vote. These characteristics undermine the value of the first indicator (years of democracy) and may help explain why Switzerland belongs to the second rather than the first group in terms of political tolerance. While Portugal has a relatively high turnout, it is clear diat an explanation Political and social tolerance 117 .i gtS 13 - e -a. a feg s s c ■a 68 ■2 §§ w ft Ol s fa § 6^ b 5 1 « -H m in N 1-^ it) \ó in tfl O H if] pi (n co * CO Cn co CD Ol O O O O O lC) o —* —i CC —i on co H itj H t-t cm ca i> o en ci ^^^ tO ID 13 I> O CO^lEJt*' □QCfíCT^ c0x>cocdco co r-- j> »n m i> o to o a o (m co cm o o o co K (N 00 CD co m CM ^ CO i—f + + -i- -ä- O m Q H IM iCJI-CiCM CO CO O H H H (M W CM CM CM Cf) CO ^ ^ o o o d odd o o o o o c ť Ö u O 5 n ti t3 i s - - Ü K .£3 ß . P Ph tn H Z |la1 i> w W «3 pí S G-a -i, go a' •O CL Ji ■5 i C 3 5 2 Ca •B.H o s. a E u U u *í U t3 u S OJ IÜ "5 u bo CL S C o c = s 'a — w aj o u •5^-5 -° «3 C •O "O rt u tj E 0.3 c 3 3 U o o ü íq S °w TO TJ »- t> fa -ri □ 0 CL, B g " B-S g »§s| Pí 3 T3 *^ l?fl I si S iT 5 ■° 13 S Ěf \ 15 u a t u 03 O. N« i ž c a ni ^ u Ills uú E £ « .5 sil* u 7] > E tj i.f?É 3 to -5 5 ci S 'S U-fl " 2 u u E 00 ti CL tl □ is w oj 3 J3 g s s e s § (jEI E5 c g (O O CL - p o ISS« A.%<£ E up ? u u ^ »1 Um Mu ■a «a S ■» ta o, B i e5 o a u OJ .£ .ti bo u o 5 & II'S c g E 5 8 |,g 5 s E .y 3 o • 3 i-1 o i- sj n S ^ bí= >• S a .s o .s 1 .§ ■a n 1° S u E ts ! cow 3 li, 13 73 , Sei " US » E s '" - S o o - rt ? 04 la II f§ ■= li g.E-3 g of the Portuguese exception must talie into account the impact of conjuncture! factors, to which we will return later on. Let us first, however, consider the impact of the last two political characteristics more systematically. The Netherlands, Denmark, West Germany, and Norway display the highest levels of voter turnout, all equal to or greater than 80 per cent With the exception of West Germany, all belong to the most tolerant group. In Russia and Switzerland, turnout is notably lower, with the former belonging to the least tolerant group and the latter to the intermediate group. The countries that first granted women the right to vote, such as Norway (1913), Denmark (1915), Russia (1918), the Netherlands (1919), and Sweden (1921) belong, with tire exception of Russia, to the most tolerant countries. The case of Russia illustrates the comparability problems associated with this indicator. In fact, we are comparing recognition of the right to vote in liberal democracies with its recognition in countries where democracy was nonexistent. But die hypothesis is not compromised as it is based on three indicators. With regard to the other two, Russia remains in line with the conjecture. Let us now proceed to the second hypothesis and die impact of socioeconomic characterisdcs. In descending order: die countries widi HDI values above 0.9 are: Norway, Sweden, the Nedierlands, Switzerland, Denmark, and Spain. With die exception of Switzerland and Spain, all belong to die most tolerant group of countries. Moldova, Russia, and Romania, all with HDI values of around 0.7, belong to the least tolerant group. Portugal registers an intermediate HDI value (0.880), which is not fully in line with its top position in terms of tolerance. To shed some light on the exceptions, we may consider the second indicator: the average unemployment rate. The figure for Spain (19.1 per cent) is signiiicandy higher than for any odier country. In Portugal, by contrast, the unemployment rate is relatively low (third from die bottom). In bodl cases, a combined reading of the two measures (HDI and unemployment) helps to explain die position with respect to tolerance. The extremely low unemployment rate in Switzerland, however, makes the Swiss case even less intelligible than if it was on the basis of the HDI alone. An analysis of the correlations between the rate of political exclusion and die political and socio-economic characteristics considered shows that only the HDI and the number of years as a democracy yield statistically significant coefficients (-0.82 with j!í<0.01 and -0.67 with p<0M, respectively). If, in addition, we regress political exclusion on these two variables, only the HDI shows a statistically significant effect (^9= —0.10; p< 0.003; iť = 0.68). In stricdy statistical terms, we might thus conclude that a direct effect of die other political and socio-economic characteristics considered cannot be confirmed. Nor, however, are we in a position to rule out their importance. The fact that we are dealing with a limited number of cases 118 Leite Viegas obviously limits the reliability of the findings. A strict reading of the statistical results indicates that a direct effect can be confirmed for the HDI only. But we must not forget that die number of years as a democracy, when considered individually, correlates closely with the level of tolerance, and that the other factors singled out by our two hypotheses may be important as auxiliary explanations. It is evident that these hypotheses require more extensive testing before more definitive conclusions can be drawn. Conjunctural factors and tolerance Can we say, in view of the theoretical considerations and empirical findings presented above, that civil and political rights are no longer a problem in established democracies, as there is a consensus regarding the social and political acceptance of others, with all the diversity of social status, religion, and ideological and political orientation? Not exacdy. Empirical research shows that attitudes of political intolerance persist in established democracies. There are even cases in which an increase has been recorded, as is confirmed in particular by the European Values Surveys (EVS) of 1990 and 1999 and die Euro Barometers of 1997 and 2000 (EB 47.1 and 50). The report from EB 50 (Thalhammer et al. 2001) states right at die beginning diat attitudes toward minority groups have developed with contradictory signs since 1997. If, on the one hand, certain results revealed more positive attitudes towards minorities, in particular a rise in support for policies of integration for diese social groups, there was, on the odier hand, an increase in the fears of deterioration in public services, particularly in the area of social protection, with the causes being partly attributed to minority groups. In a response to these apparently inconsistent patterns, Sniderman et al. (1989) established a distinction between 'principled' and 'situational' tolerance. For diem, it is essentially a matter of distinguishing consistent attitudes with regard to principles of tolerance from situation-dependent expressions of (in)tolerance not based on an underlying set of principles. 'Situational tolerance' reveals itself in assessments that depend on the specific relationship between the majority and the different groups that are potentially subject to discrimination. While it may not be necessary to actually diink in terms of two different concepts of tolerance, these considerations lead to a hypothesis about the effects of conjunctural factors. A rise in insecurity and political conflict that the majority of citizens associate with a certain social group tends to increase political and social discrimination against that group. However, the impact will depend on the overall (and hence more 'principled') level of tolerance. In countries with high levels of intolerance, subjective perceptions of conflict and insecurity tend to increase discrimination against the target group without distinction. In countries with low levels of intolerance, the effects are more selective. They are most visible for rights that, Political and social, tolerance 119 supposedly, put citizens' security at risk, but are less evident with respect to other rights of tile target group. To test this hypothesis we need to disaggregate political exclusion and social discrimination results based on the target groups. The results are presented in Tables 5.3 and 5.4, where the individual groups are divided into four categories: ideological extremists, religious fundamentalists, stigmatised groups, and ethnic minorities. With regard to my hypothesis, note to begin with diat die percentages for social discrimination against 'drug addicts' are significantly higher than the corresponding percentages for political exclusion, except in diose countries where the general level of intolerance is highest: Russia, Romania, and Moldova. In some cases, the differences are quite large: 52 percentage points for Norway, 44 for Sweden, 40 for the Netherlands, 28 for Denmark, 24 for West Germany, 25 for Spain and Switzerland, and 17 for Portugal. The same pattern is evident for 'people with a criminal record' and 'people with AIDS', although in these cases the differences are smaller. For 'homosexuals', 'immigrants', and 'people of a different race', the differences between political exclusion and social discrimination tend to be very small. In these cases people do not seem to think that their personal security is threatened. This pattern seems to support my hypodiesis. Before drawing any general conclusion, however, it is important to consider tile results for the five extremist groups. Generally speaking, the order of die countries with regard to the political exclusion of 'Christian fundamentalists' and 'Islamic fundamentalists' is similar to that for political exclusion in general. However, the figures are generally higher than those for die target groups previously considered, particularly for 'Islamic fundamentalists'. If we compare the political exclusion rates for 'Islamic fundamentalists' with those for social discrimination, we find the values to be quite similar for Russia, Romania, and Moldova, In other countries, Portugal and Spain excepted, the rate of social discrimination is higher than that of political exclusion. Countries with relatively high levels of general tolerance seem reluctant to accept physical proximity to these religious extremists, although they are relatively ready to accept their participation in political life. Once again, the hypothesis is confirmed. Generally tolerant societies differ in their attitudes toward political exclusion and social discrimination. With regard to die political exclusion of ideological extremists, die picture changes significandy. With regard to 'right-wing extremists' and 'racists', die exclusion rates are relatively high for almost all countries (Denmark and Portugal excepted), and die borderlines between the three groups of countries blurred. It is worth reflecting on some particular situations. In Portugal, the political exclusion of religious and political extremists is low. Although die exclusion rate for 'racists' is higher than for other extremist groups, it is low in comparison with odier countries. This explains the country's Table 5.3 Political exclusion of different social groups" Social group Ideological extremists Left-wing extremists Right-wing extremists Racists Religious extremists Christian fundamentalists Islamic fundamentalists Stigmatized groups People with a criminal record Drug addicts People with AIDS Homosexuals Ethnic minorities Immigrants People of a different race CH DK EG WG ES MD NL NO FT RO RU SE Si1' 49 10 41 37 42 45 36 25 9 45 16 21 33 69 24 72 65 48 37 51 45 11 38 67 35 54 73 32 67 57 55 54 61 60 32 49 60 57 65 32 7 24 17 29 57 12 21 12 49 12 17 31 47 20 33 31 52 55 29 34 20 55 50 .26 47 31 4 32 24 17 58 11 27 17 43 50 27 _ 14 S 22 18 22 65 13 10 20 64 62 17 - 6 2 6 6 6 49 3 3 10 10 26 2 - - 5 15 8 10 59 5 9 16 16 49 6 26 13 4 9 6 8 37 4 6 5 30 12 3 10 8 2 5 4 6 34 2 3 8 22 9 2 7 Notes a Entries are the percentages of respondents who want to exclude individuals from the social group in question from participation in public life, b For Slovenia, only partial information is available. Table 5.4 Social discrimination against different social groups" Social group CH DK EC WG ES MD NL NO PT RO RU SE SI* Ideological extremists Left-wing extremists 52 17 38 42 35 42 39 28 7 41 14 27 Right-wing extremists 69 37 73 71 42 35 59 53 8 36 60 47 _ Racists 73 40 61 61 50 53 69 63 34 47 50 64 63 Religious extremists Christian fundamentalists 41 13 25 21 24 58 16 30 10 47 13 26 _ Islamic fundamentalists 55 47 36 41 48 56 38 46 18 52 47 39 - Stigmatized groups People with a criminal record 45 7 40 37 26 61 22 55 22 46 54 60 22 Drug addicts 39 33 33 42 47 70 53 62 37 66 80 61 52 People with AIDS 7 3 10 12 14 58 7 10 19 38 49 7 15 Homosexuals 14 5 13 12 9 61 5 11 17 58 54 8 24 Ethnic minorities Immigrants 10 2 9 7 10 39 5 12 2 31 12 6 9 People of a different race 4 10 6 7 7 36 4 6 10 26 10 5 6 Notes a Entries are the percentages of respondents who reject members of the social group in quesdon as neighbours, b For Slovenia, only partial information is available. 122 LeiteViegas position on the general scale of political tolerance. In addition, and in contrast to the pattern found in other countries, the rate of social discrimination against these particular groups is lower than the rate of political exclusion, except for 'racists1. These results seem to be consistent with my ideas about the effect of conjunctural factors on tolerance. In Portugal, there is little political conflict with either religious or political extremists, as illustrated by the absence of xenophobic political parties. Switzerland also displays certain interesting peculiarities. The values for political exclusion and social discrimination are high not only for religious and ideological extremists, but also for individuals with a criminal record, which reveals the primacy of the values of order and security. For the other target groups, however, exclusion and discrimination rates are very close to countries with average levels of tolerance. It may also be worthwhile to consider the development over time of social discrimination towards different target groups in order to clarify certain points and confirm others. For this purpose I have selected some countries from the EVS in 1990 and 1999 (Table 5.5). With regard to ideological extremism, diere is a general trend towards increased social discrimination rates in all countries, except Portugal, where there has been a marked decrease, and Germany, where there has been a slight fall of 3 percentage points in the case of the extreme left When it comes to 'drug addicts', the discrimination rates are high and relatively stable, again with the notable exception of Portugal. For all other target groups, the trend is towards a decrease in social discrimination, especially in Portugal, where there has been a fall of 20 percentage points for 'people with AIDS' and 26 percentage points for 'homosexuals'. These findings add to our understanding of the particularities of the Portuguese case. A micro-level model of political tolerance At various points in the previous section, I have hinted at individual-level characteristics favouring or inhibiting die development of tolerance. It is now time to provide a more systematic presentation of these factors for the purpose of constructing and testing a micro-level model of tolerance. Let us begin with the dependent variable. According to well-established methodological principles, the more homogeneous the object to be explained, the more successful the explanation will be. The theoretical assumptions originally put forward and, to a certain extent, confirmed by the empirical results for different target groups favour a division of these groups into two main categories, the first containing extremist groups and the second stigmatised groups and ethnic minorities (see Table 5.3). My focus is on the political exclusion of these groups.1 Hence, two dependent variables were constructed, one for each category of groups. Both variables indicate die mean number of groups excluded on a scale ranging from 0 (no group) to 1 (all groups). Political and social tolerance 123 Table 5.5 Social discrimination against different social groups according to the world values surveys 1990 and 1999" Social group and year Denmark Germany The Netherlands Portugal Spain Left-wing extremists. 1990 6 51 47 30 25 1939 9 48 50 13 25 Riglil-wing extremists 1990 7 62 52 28 28 1999 20 74 69 15 29 Muslims1' 1990 15 20 14 19 12 1999 16 13 12 8 11 People with a criminal record 1990 28 27 28 57 37 1999 31 25 32 43 32 Drag addicts 1990 54 60 72 61 56 1999 60 58 73 46 53 People with AIDS 1990 9 28 15 45 34 1999 6 12 8 25 21 Homosexuals 1990 12 34 11 52 29 1999 8 14 6 26 16 Immigrants/'foreign workerŕ 1990 12 16 9 10 9 1999 11 11 5 3 9 Notes a Figures are percentages. b The CID survey provides data on Islamic fundamentalists rather than Muslims, c The CID survey provides data on immigrants. Foreign workers are not mentioned. Regarding the independent variables, we should distinguish the long-term structural factors underlying the formation of basic predispositions about social and political life from conjunctural effects deriving from perceptions of social conflict2 Widi regard to the structural factors, the literature allows us to distinguish three main dimensions. The first focuses on socio-economic and socio-cultural levels of modernisation in industrial societies. The second covers social attitudes and identities, as expressed through people's perceptions and feelings about others and themselves (social identities). The third is more explicitly political and encompasses values, identities, and behaviour directly related to political life. Among the factors within the first dimension, education and age (or generation) are most prevalent in the literature (Stouffer 1955; Nunn el al. 1978; Sullivan el al. 1982; Bobo and Licardi 1989; Sniderman et al. 124 Leite Viegas ■■■■■■■> 1989). The reasons advanced for the importance of education are twofold. j On die one hand, education puts people in touch with different ideas and -j different lifestyles. On die other hand, more educated people are better j equipped to relate to abstract democratic norms in specific situations. Age > (or generation), gender, place of residence (in our case operationalised on a scale from rural to urban), and church attendance are variables .,-..> which belong to the same theoretical construct. ■ | Within this set of variables, my hypothesis is that gender will not matter j much (Thalhammer et al. 2001) but that we can expect higher tolerance '• among the well-educated, the young, the religious, and those living in an ; urban environment These expectations refer primarily to the political : j exclusion of stigmatised groups and ethnic minorities. For the other dependent variable, my expectation is that these variables will wholly or partly lose their explanatory power. With regard to the second dimension (social attitudes and identities), it was recognised at an early stage diat perception of danger associated with a certain group contributed to intolerant attitudes towards this group. These effects, however, depended on the characteristics of the person. Stouffer (1955) referred to the optimistic characteristics of persons favouring tolerance, and McClosky and Brill (1983) considered flexibility J and self-esteem to operate in a similar way. Within the fields of sociology and political science, die amount of trust placed in others (see Chapter 2) ! is somewhat analogous to die psychological constructs just mentioned. In j their study on tolerance, Sullivan et al. (1982: 162) mention the relatively i high correlation (r= 0.20) between social trust and tolerance. My hypoth- j esis is that social trust has greater explanatory power with regard to the | exclusion of stigmatised groups and ethnic minorities than with regard to the exclusion of extremist groups. Within diis dimension, I also consider the degree of attachment to one's religion and culture, for which a negative relationship with political tolerance is expected. As shown by several studies (e.g. Gibson and Gouws 2000), strong identifications with particular groups tend to boost intoler- j ance with respect to other groups. However, it is conceivable that religious j identification could increase die political tolerance of stigmatised groups j and ethnic minorities. As long as extremist and fundamentalist threats to j religion are not at issue, religious principles of social solidarity may be v:j operating. j Witii regard to the third dimension finally, the macro-analysis carried | out above underlined die importance of the political sphere in shaping j attitudes of tolerance. In many micro analyses, acceptance of democratic : | values appears as the main explanation (Lawrence 1976; Sullivan et al. j 1982). There is, however, a certain amount of controversy about the j results obtained. For Prodiro and Grigg (1960) and McClosky (1964), for j instance, die empirical results do not indicate a relationship between j acceptance of democratic standards and tolerance. j Political and social tolerance 125 While no stricdy equivalent indicators appear in my model, I do consider confidence in institutions (as operationalised in Chapter 2), that is confidence in political institutions and actors as well as institutions associated with the rule of law). My expectation is that trust in institutions favour tolerant attitudes with respect to both dependent variables. Authoritarianism and dogmatism have often appeared in the literature as explanations of political intolerance (Stouffer 1955; Sullivan et al. 1982). While my model does not include anything strictly equivalent to diese psychological constructs, I do consider the norms of good citizenship presented in Chapter 4. My hypothesis is that citizens stressing law-abidingness are more intolerant, particularly widi regard to stigmatised groups and ethnic minorities. For those who emphasise solidarity, I expect the contrary, that is greater tolerance, at least for stigmatised groups and ethnic minorities. For those who underline the importance of criticism and deliberation, finally, a positive effect on tolerance might be expected for both dependent variables. Ideological position has also been considered in the literature. The results indicate that liberals (in die United States) and people on the left (in Europe) tend to be more tolerant (Sniderman et at. 1989; Thalhammer et al. 2001). However, some European studies have pointed to an increasing lack of ideological identification as well as a blurred understanding of the traditional notion of left and right (Schweisguth 1999). Nevertheless, my hypothesis is diat those who place themselves to the left are more tolerant toward stigmatised groups and ethnic minorities. However, I expect litde explanatory power with regard to extremist groups, since in this case, the threats to democratic principles will tend to neutralise primary inclinations of tolerance. With respect to political involvement, early results pointing to a positive and significant relationship (Stouffer 1955) have later been contested. With other variables controlled, in particular education, Nunn et al. (1978) found that political involvement had litde impact. Nevertheless, the matter certainly merits further testing. The measure I use is the extent to which people discuss politics, and my hypotiiesis is that those who do so more frequently are less prone to political exclusion, particularly widi respect to stigmatised groups and ethnic minorities. With regard to conjunctural factors, migratory flows in Europe and the spread of drugs and AIDS have produced new minority and stigmatised groups. This situations may breed perceptions of insecurity, which in turn might favour the rise of intolerant movements. The same sense of insecurity may pardy explain the increased political exclusion of extremist groups in traditionally tolerant countries. To explain attitudes towards groups tiiat pose a supposed threat to people's security, we have only one variable, that is, the amount of exposure to information about politics and social affairs obtained through newspapers, radio, and television. My hypotiiesis is diat a higher level of political tolerance is to be expected 126 Leitc Viegas among those who receive more information, at least with regard to stigmatised groups and ethnic minorities. For the other dependent variable, my expectations are die reverse. The results of my analyses are presented in Tables 5.6 (extremist groups) and 5.7 (stigmatised groups and ethnic minorities). Note to begin with that, for most countries, the variance explained is lower for the first dependent variable than for the second. This is in line with my general hypothesis that many of the predictors are likely to have less impact for extremist groups than for stigmatised groups and ethnic minorities. Even in die latter case, however, a predominant part of the variance remains unexplained. Another general pattern is that the results show a considerable amount of variation across countries. This is true with regard to die proportion of variance explained as well as with regard to the impact of die individual predictors. As we shall see, this does not mean that the effects are generally inconsistent or out of line with my hypotheses. But the extent to which a particular effect is present varies considerably across countries. In some cases, however, there are indeed signs of inconsistency in the sense that the same predictor shows a positive and statistically significant effect in some countries but a negative one in others. Among these is die very first predictor: the size of the locality in which the respondent lives. For the first dependent variable, the effect of living in a more urban environment is negative (as hypodiesised) in five countries but positive in three others, all of which located in eastern Europe (East Germany, Romania, and Russia). For the second dependent variable, the three significant are all negative, but the expectation that place of residence would matter more for this aspect of tolerance than for die first is hardly borne out In line widi our expectations, gender does not have much impact In five out of 24 cases, there is a weak but statistically significant difference, usually indicating slightiy greater tolerance among women. As far as age is concerned, the expected pattern of greater intolerance among older citizens manifests itself with a fair amount of regularity. The effect is statistically significant, and sometimes fairly strong, in 14 cases, all but one of which (Portugal for the second dependent variable) showing the expected sign. The results for education, on the other hand, are surprisingly mixed. For stigmatised groups and ethnic minorities, statistically significant effects are found in only four instances, all of which carrying the expected sign and some reaching respectable strength. For extremist groups, the effect is significant in five instances, but in all but one of these (Norway), education is positively rather than negatively associated with die rate of political exclusion. The effect of church attendance, finally, resembles that of gender. It is statistically significant in only three of 24 cases, two of which showing the expected sign. oooooooooooooo III II I lt o 1M n h O O T-* CM —t o o o o d Oriqqqqqq I o o o o d d d d I I o t T-t cm t-H t a i-h im i d d o I i m o i p p t-j i o o o I I i pi (O m i O O CM ! d d d I I o o o o ooooooooooo II ] I I ooooooooooo f I II o o o o o o I III. o o o o ooooooooooo I II I oaoooooooooooo I II I i en cn i—i cm I p o o o ^ ^ d d d I I I o a a o o i W lfl CO C I O 1-1 CTi O i O O d O O t-. t-h d d d I I o o o o I I o o o o ta O C3 O I I * * * i toujMno I O f-H f-H f-1 f-^ I o o c o c * * a-O o o ci H H odd I i O C4 CO em i—i O i—i a ci a d I I * o o o d d d iPIONCNHOHinbCO^COOiN |pppppppr-1i—IOOOOO i I 111 I ooooooaoaaoaoo I III O Tj< 8 £ ■a js u 5 c y p u C E B J3 -° £ = a a M 5 8 = 3 a h Q5 b 'a a a c u v - a 'a -a -c u u u .£ S3 O u a aa-Sr: c o o □ £ "0 • SB! " " £ " IS -a so 2 js o 1S- § ° |"3 •a .S " ii 3 « S Su . -S > - " '32 = >,-o § a la. 3 E s E « S*a U r C ^ -* -1 CO U P .2 c « c £ "S !l 8.5 o g e 6P n 0 "3 - - = i S a S -lis i =1 =• i -q =: 3 J3 n F „ E m a J. c.y ii. ri ^ c 6- .H^ S u fee 0 2j5 o 2 rt -s Political and social tolerance 129 pi -i ^ in o w m in o o > w a h ^ 00<—i 1—i-OOO-hOOO^OOO ddddciddddddo'doo II I I III I II rl rl im Ifl W CI cq if) O O Ifl PI PJ O O CM O ooooppooo dodo oooooööcjo III II * * * . ~& do o m a I o o o o p ! d ö ö ö ö I I o o o o o I I I ©OOOOOOOi II [ I oooo © O O O ' III I I i o o o o o I I * i ) O O O O p ( i d 6 d d d I I I I o o I o o o I I , © o o I I O O O O I I I oooo o —« . tO Cl O O . © ©* I I O T-i ) Ö Ö I O O ü0 t O ri O I © Ö Ö ( i o © © a o o o o © © o o o 0 o o 1 I * ft * * < O O < I I r- to i a •-—I i ö d I o O C4 -—* T-J odd I I I * * CI ID CQ ^ O O O i-h dodo t I I * # O to * * * * * « * * * I h Ol O c W i o o I 0 o o 1 I o o I o o o I oooo I I p p d d I I o o I I o o I ft ft ft ft ft ft 1 T—I 00 d d I O IN IM ■ © © O I ddd< I I oooo I I I in Ki o a o o ö d ö I I Q U Q I o O O i O O O ' I U EU C u C u C 5 H — 5 = 28 I o o o O ' I I 11*3! e -■s .3 f 11 e o a i 13 3 45 E 1 - § 1 b &J 5 u s u m k n c \a -o "n «J Tl - - 3 S Ö u u .SP « x) S3 O u 7 to ~G u — , 3 *rt tn t3 t3 ö ö ö^-d .H, ,j p< S 1* id S "i 2 u .y S"1 D. 2 u » u 9 Z a J= Let us now proceed to the second of our three general tiieoretical dimensions, diat is, social atdtudes and identities. For religious attachment, statistically significant effects are visible in only five cases, four of which refer to die second of die two dependent variables. Three of these effects are positive, indicating a greater rate of exclusion among respondents with strong religious attachment. Notably, die other two cases, where die effect is negative, both pertain to East Germany. For cultural attachment, the pattern is somewhat stronger as well as more consistent. Nine of the effects are statistically significant, all of which in die expected direction. Six of them refer to the second dependent variable. However, the most important factor within this general dimension is undoubtedly social trust. In this case, 15 effects are statistically significant, all of which having the expected negative sign. Again, most of them (two-thirds) refer to the second rather than die first dependent variable. The first representative of the third, more explicidy political, dimension is confidence in institutions. The pattern in this case is not quite as strong and consistent as for social trust, but still readily discernable. Eleven of the effects are statistically significant and nine of them have the expected negative sign. The two deviations are East Germany on the first dependent variable and West Germany on die second. Note that in this case a slighdy larger proportion of the significant effects (six of 11) refer to the first rather than the second dependent variable. Widi respect to the three norms of good citizenship, those regarding law-abidingness and criticism and deliberation both have a moderate but consistent impact. Seven of the effects are statistically significant in the former case, six in die latter. All carry the expected sign, positive for law-abidingness and negative for criticism and deliberation. All but two of diem (both for law-abidingness) refer to the second radier than the first dependent variable. For solidarity, the pattern is weaker as well as less consistent. Only four of die effects are statistically significant (two for each dependent variable) of which tiiree carry the expected negative sign. The deviating case is East Germany with regard to the second of the two dependent variables. Left-right placement displays a picture fairly similar to that of law-abidingness. Six of the effects are statistically significant and all carry the expected positive sign. Two-thirds of diem refer to the second radier than the first dependent variable. The impact of political involvement as measured by the frequency of political discussion is not nil but quite inconsistent. While six of the effects are statistically significant, only half of them carry the expected negative sign. The positive effects (for die Netherlands, Russia and Spain) all pertain to the first rather tiian the second dependent variable. Regardless of whether we see it in die light conjunctural effects, as I did in my dleoretical discussion above, or take it to be yet anodler indicator of political involvement, the pattern for political media exposure is very 130 Lúle Viegas similar to that for political discussion. While eight of the effects are statistically significant, five of diem carry a positive radier than negative sign. In tins case, positive effects are found for the first (Moldova, Norway, and Portugal) as well as the second (Moldova and Portugal) dependent variable. Conclusions I have demonstrated that there are significant differences between the countries studied widi regard to the level of political and social tolerance. Two hypotheses regarding the impact of political and socio-economic factors were advanced to explain the level of political tolerance across countries. Although the measures at our disposal represent a simplification, they indicate that the level of tolerance is closely associated with social development and well-being on the one hand, and the duration of democracy on the otiier. We can divide the tiieoretical positions on tolerance into two camps. Based on dieories of modernisation, the first argues that the process of development produces greater tolerance. The second, relying on the most recent information on conflicts in relation to certain social groups and the increase in xenophobia, claims dial intolerance in democratic societies is rising, contradictory to the supposed structural trends. My response was not to take sides in a dichotomous fashion but to try to specify the conditions under which tiiese positions are defensible. The theories of modernisation are correct if we consider groups diat represent other cultures, social characteristics, and lifestyles, but are not perceived as a social danger. My group-by-group analysis of political exclusion and social discrimination as well as my attempts to explain political exclusion on die micro-level helps to buttress this conclusion. Although our data represent but a single point in time, they could neverdieless bear witness to the effect of new forms of insecurity on the rate of social discrimination. Among other things, I have shown that the most tolerant countries discriminate socially against certain target groups to a greater extent than they exclude them politically, in particular with respect to 'drug addicts', 'people with a criminal record', and 'Islamic fundamentalists'. In these countries, attitudes of rejection are more selective and can be assumed to coincide witii perceived insecurity. In the most intolerant countries, rejection is equally strong in the political as in the social domain, revealing what we could call low structural tolerance. It also proved worthwhile to consider the absolute rejection levels for different target groups. With respect to political exclusion, the groups most strongly rejected are 'right-wing extremists', 'racists', and 'Islamic fundamentalists'. In these cases, the differences between countries are very small, with die exception of Denmark and Portugal, where the rejection levels are noticeably lower. At the odier end of the scale, die groups : Political and social tolerance 131 least commonly excluded are 'people with AIDS', 'immigrants', and 'people of a different race', aldiough in Moldova, Romania, and Russia the rate of political exclusion of these groups remains high. These results are consistent widi my hypothesis diat the effect of perceptions of insecurity with regard to certain social groups depends on the overall level of tol-i erance. In countries that are generally more tolerant, the impact is selective. It does not extend to all target-groups and implies a distinction between social discrimination (which is higher) and political exclusion. Witii regard to the micro-level model of political exclusion, I decided : to divide die different target groups into two categories: extremist groups on die one hand and stigmatised groups and ethnic minorities on die other. The regression analyses produced complex, varied but nevertheless systematic results. To begin with, and as conjectured, the proportion of variance explained tends to be lower for the first dependent variable dian the second. Further, the results show considerable variation across countries with respect to the variance explained as well the impact of specific predictors. None of diem has a clear effect across die board. Quite a few, however, show a significant impact in the expected direction in a substantial subset of countries. In oudining the model, I distinguished between diree dimensions as far as die long-term, structural factors are concerned. The first focuses on levels of modernisation, the second on social attitudes and identities, and .-! die tiiird on more explicidy political characteristics, i Among the predictors belonging to the first dimension, age (or genera- ! don) provides die strongest pattern, with die elderly being significantiy more intolerant in more than half of the cases. Somewhat surprisingly, the impact of education is not quite as visible. Furthermore, die direction of die effect varies widi the dependent variable: negative widi respect to die exclusion of stigmatised groups and ethnic minorities but usually positive with respect to the exclusion of extremist groups. The remaining predic-| tors within the first dimension - size of locality, gender, and church atten- dance - tend to show but weak and/or inconsistent effects, i As far as die second dimension is concerned, the strongest predictor is ■I social trust, which has a significant effect in the expected (negative) direc- j tion in well above half of the cases. Cultural attachment also shows die i hypothesised (positive) effect in a relatively large number of instances i whereas the impact of religious attachment is radier weak and inconsistent With respect to the diird dimension, finally, confidence in institutions, ■i degree of support for the norms of criticism/deliberation and law- .:■.! abidingness, respectively, and left-right placement all display quite consis- ! tent effects in die expected direction (negative for die first two, positive i for the outers) in a moderately large number of instances. For die third J norm of good citizenship, solidarity, as well as for political involvement in ; the form of political discussion and media exposure, die pattern is weaker J and/or less consistent. 132 LeiteViegas Notes 1 Regression analyses were also carried out for social discrimination against these two categories of groups. Inasmuch as they do not add much to the results, they are not presented. 2 This is a classical distinction in sociology and is cited in many studies on tolerance (see for example, Marcus el al. 1995: 15). There are also authors who overestimate the importance of situational factors either in general (Downs 1957; Converse 1964) or in some specific analyses (McClosky 1964; McClosky and Brill 1983). Part II Voluntary organisations and social networks