INTRODUCTION WHAT IS SOCIAL CAPITAL?

CONTENTS Introduction... 2 1. Trust... 5 1.1 Interpersonal trust... 5 1.2 Interpersonal trust by affective proximity... 7 1.3 Aspects of trust... 8 1.4 Diversity and trust... 9 1.5 Trust in professional groups... 11 1.6 Trust in institutions... 13 1.7 Trust and perception of corruption... 15 2. Personal networks... 17 2.1 Size of household... 17 2.2 Family network... 18 2.3 Friendship network... 19 2.4 Characteristics of friendship networks... 20 2.5 Total network size... 25 2.6 Contact with networks... 26 3. The effects of networks... 30 3.1 Asking for help in cases of need... 30 3.2 Networks, the family and situations of need... 31 3.3 Exchanging favours... 33 3.4 Seeking work... 35 4. Distribution of social capital... 36 4.1 Groups without friends... 37 4.2 The population aged 65 or over... 37 5. Involvement with associations... 40 Conclusions... 44 Technical features... 45 References... 46

INTRODUCTION WHAT IS SOCIAL CAPITAL? Every individual is in constant contact with others. The quality and type of these interactions are important in how effective each individual is in achieving their objectives. Social capital is a resource at the disposal of individuals and groups which serves to oil the wheels of social interaction. An individual with high social capital is better equipped to move in different social spheres, including work, daily life and economic life. Conversely, an individual with low social capital will tend to rely more heavily on formal channels and public or private services. Social capital has become a much used concept occupying an increasingly wide area of interest for academics, economists, politicians and regulators. Although the concept itself has attracted growing attention, there is still much debate and discussion about the exact components that make it up and the relationship between them, its causes and effects and, above all, the best way to measure its various dimensions. Ambiguities apart, the literature concurs that social capital is a multidimensional concept comprising the resources of an individual or community that derive from trust, the rules of reciprocity, and connectivity or social involvement (social networks and associationism). The Organisation for Economic Cooperation and Development (OECD) defines it as networks, together with shared norms, values and understandings which facilitate cooperation within or among groups. The distribution of social capital among large numbers of individuals has important social implications. In a society where people trust each other, the costs of negotiating and reaching agreements are automatically reduced. Likewise, individuals greater connectivity through their personal networks facilitates information flow and the ability to satisfy individual and collective needs. In contrast, a society of atomised and mutually suspicious individuals will be characterised by relations based on rigid contracts, with little information sharing and high access barriers to social resources and opportunities. The social capital concept is associated with a wide range of benefits at both the individual and collective level. Social capital has been identified by various authors as a key driver of economic development (Fukuyama 1995: Knack & Keefer 1997), democracy (Putnam 1993) and school learning performance (Coleman 1988), among other issues. In his book Bowling Alone (1995), Robert Putnam, one of the names most closely associated with the analysis and measurement of social capital, pointed out the decline in informal socialising and political and association membership in the United States since the middle of the 1960s. The author relates this decrease in civic involvement to factors like changing work and family structures and the rise of new technologies, which brought about a social transformation between the generation growing up before the second world war and the post-war generation. Sociology too has contributed keynote studies of social capital, by authors like Pierre Bourdieu (1985) and James Coleman (1988). Coleman sees social capital as a resource facilitating action. Like other forms of capital, social capital is productive, making possible the achievement of certain ends that would not be attainable in its absence. One of his central contributions has been to analyse the importance of social capital in human capital formation and in children's school learning performance. There is also a large body of literature that acknowledges interpersonal ties as a determining factor for the mobility of individuals. Granovetter (1973) developed the 2

theory of strength of weak ties, which stresses the importance of indirect influences outside the immediate circle of family or friends. Finally, many authors focus on the adverse effects of social capital, such as promoting the exclusion of those who are not in the group, making too many claims on group members, or restricting individual freedom (Portes 1998). International organisations and government agencies have also directed their efforts to the analysis and measurement of social capital, in the belief that a better understanding of this phenomenon can usefully inform public policy making. We can cite here the programmes of the World Bank, the OECD and the national statistics offices of Australia, the United Kingdom and Canada. INNOVATIONS OF THE BBVA FOUNDATION SURVEY Social capital is not easy to measure given the multiple definitions of the concept itself and its multidimensional nature. With no consensus existing as to its effective, reliable measurement, the surveying of social capital in 13 countries posed a major methodological challenge. Little progress has been made to date on creating social capital indicators applicable to different countries. Among the developed countries, most efforts have been led by the universities or national statistics institutes, while among the developing countries the work of the World Bank is among the first that comes to mind. The BBVA Foundation survey applies a single conceptual design to countries of differing cultural traditions and levels of socioeconomic development, and represents a significant advance in the construction of a more comprehensive and refined social capital metric valid for international comparison. The study s main innovations include: a) Number and variability of countries studied: although other international studies have looked at facets of social capital 1, this monographic study on social capital innovates by including countries at different development stages and with different cultural traditions. It covers 13 countries in three continents, with a total of almost 20,000 cases, breaking down as approximately 1,500 per country. Participating countries are the United Kingdom, Spain, Germany, Denmark, France, Italy, Japan, Russia, Israel, Turkey, the United States, Mexico and Chile. b) Issues covered: the study covers an especially broad range of social capital dimensions and indicators. Traditional measurements are joined by a closer study of selected aspects of social capital, using indicators which allow a more precise quantification of social networks and others exploring different dimensions of trust. c) Methodological innovations: as with other BBVA Foundation studies, the questionnaire makes extensive use of 0 to 10 scales in preference to 1 to 5 (Likert). This is a significant innovation offering wider possibilities with regard to the type and complexity of analyses. A 0 to 10 scale means a larger number of response categories and, in theory at least, allows opinions to be gauged more accurately. The fact that the "conceptual distance" between one score and another is the same in every case also has a series of advantages for comparative studies (Mohler 1998: Smith 2004), minimising the measurement errors that can arise in the process of translation and from applying a standard tool to different cultures. 1 See studies like the World Values Survey or European Social Survey. 3

ABOUT THE REPORT This report centres on two key dimensions of social capital, trust and networks. Trust is an intangible resource, a psychological property that facilitates social interaction while bringing down transaction costs. Networks are the material expression of trust; a web of social relationships based to a greater or lesser extent on mutual confidence. The study looks at interpersonal trust in general and in relation to groups of people within individuals close environment, trust in professional groups and institutions, and how these different levels interrelate. It focuses on: how trust extends outwards, or not, from closer to more distant spheres; degree of trust in groups different from ourselves and how the perception of sameness influences trust; different facets of trust, like competence and honesty, and the importance ascribed to each. On the subject of networks, the study seeks both to quantify them and characterise them by their properties: personal networks: family, friends, neighbours, work colleagues; diversity, length and geographical dispersion of friendships; frequency of contact with personal networks; use of networks to ask for help or look for work; involvement with associations. 4

1. TRUST The trust between a society's different components individuals, groups and institutions is a key ingredient of that society's social capital. Social interaction always involves an element of risk: the fact that the other may act in a manner contrary to our wishes or expectations. Trust is based on the assumption that people or institutions will act as we expect them to, and is an essential element in social relations. Its effects are both individual and collective. By reducing transaction costs, it encourages cooperation and, ultimately, makes social interaction more fluid and predictable. Fukuyama (1995) argues that the presence of mutual trust brings down transaction costs, which in turn has a visible boost effect on economic growth. From the standpoint of social capital, trust is a key factor in all kinds of transactions, including loans, employment contracts, investments, etc. A high level of trust means we need to devote fewer resources to protecting ourselves from outside harm. Written contracts are less of a necessity and lawsuits less frequent (Knack and Keefer, 1997). This chapter of the study examines interpersonal trust in general and with regard to groups with differing degrees of affective proximity, identifying the specific attributes that shape feelings of trust. It also looks at trust in institutions and professional groups. The general level of trust is fairly high in all the societies analysed. However it also seems to be directly related to the respondent's proximity to and knowledge of the object. Hence we find very high levels of trust in people's immediate circles family and friends, while trust in people in general and different professional groups and institutions is variable but generally lower. Moral or emotional aspects are more important in generating interpersonal trust than technical or competence factors. Among some sectors, proximity and also sameness are major factors in generating interpersonal trust. In general, Denmark and the United States stand in the upper part of the trust map, while Russia, followed by Chile and Japan, are those least ready to invest their trust. 1.1 Interpersonal trust Most of the population trust the people around them. The level of general trust exceeded the midway point (5 on a scale of 0 to 10) in all countries except Chile and Turkey, and was highest of all in Denmark. Looking at the total distribution of responses and not just the average value, high trust scores (7 to 10) are most prevalent in Denmark followed by the United Kingdom, the United States, Germany and Russia, while in remaining European countries, Japan and Israel intermediate positions tend to dominate. Table 1.1: Generally speaking, would you say that most people can or cannot be trusted? Mean on a 10-point scale where 0 means most people cannot be trusted and 10 means most people can be trusted (Base: all cases) DK 1 UK USA DE RUS ES IT JP FR IL MEX CH TR Mean 7.2 6.5 6.1 5.9 5.9 5.9 5.8 5.8 5.6 5.4 5.4 4.6 4.4 Most people cannot be trusted (0 to 3) 3.5 6.4 13.0 16.8 16.2 10.1 13.9 10.5 14.9 20.4 21.7 33.3 34.6 Intermediate scores (4 to 6) 22.8 35.3 37.2 36.6 38.0 49.6 45.9 50.1 49.3 43.5 38.7 41.4 44.7 Most people can be trusted (7 to 10) 72.1 53.5 49.4 44.7 44.6 38.8 38.4 35.3 34.5 36.0 38.2 22.9 20.3 DK/NA 1.6 4.8 0.5 1.9 1.2 1.6 1.8 4.1 1.3 0.1 1.5 2.4 0.4 1 In this and all tables: Spain(ES), France (FR), Germany (DE), United Kingdom (UK), Denmark(DK), Italy (IT), Turkey (TR), Russia (RUS), Israel (IL), Japan (JP), United States (USA), Mexico (MEX), Chile (CH). 5

As well as measuring the general trust felt in most people, a number of indicators were used to gauge how individuals view human conduct and how far this perception influences trust. A consensus exists that human beings are basically selfish and look out for themselves: an opinion which meets with the majority support of citizens in every country except Denmark, the United Kingdom and Japan. This vision of rational beings seeking to maximise their opportunities a characteristic tenet of the economic sciences does not, however, imply the absence of interpersonal trust or lead automatically to the idea that human beings are essentially ill-meaning. In several countries, such as the United States or Spain, general interpersonal trust is medium to high even if the majority see humans as essentially selfish beings. This trend, however, does not hold in all countries. In Denmark trust is very high, and rests on the belief that most of the time people try to help others (44%). Table 1.2: Which of the following two phrases do you agree with most? (Base: all cases) IL TR IT CH RUS FR ES MEX DE USA DK UK JP Most of the time people look out for 71.4 70.3 66.2 65.4 64.2 63.0 61.0 60.1 56.9 53.1 41.0 39.2 37.1 themselves Most of the time people try to help 13.0 12.2 6.6 25.2 17.4 15.4 15.5 24.4 23.1 25.7 44.4 16.7 14.2 others Sometimes people look out for themselves and sometimes 14.6 16.7 26.3 9.0 17.2 19.6 21.7 15.2 15.8 20.3 13.6 42.4 45.0 they try to help others (DO NOT READ) DK/NA 1.0 0.8 0.9 0.4 1.2 2.1 1.8 0.4 4.2 0.9 1.0 1.7 3.7 Opinions come out more divided in a supplementary indicator which asks if most people try to take advantage of others or most try to be fair. This indicates that not all people who believe human behaviour is basically selfish also think it is hostile to others. The idea that people try to take advantage of others finds majority support among Chileans, Mexicans and Turkish and convinces a relative majority among the Israelis, Spanish and French. In contrast, in Denmark, especially, but also the United States and Russia, a majority think that people try to be fair. The Japanese and British, again, along with the Italians adopt an intermediate position. Table 1.3: Which of the following two phrases do you agree with most? (Base: all cases) CH MEX TR IL ES FR IT DE USA RUS JP UK DK Most people try to take advantage of 70.8 58.8 50.7 42.3 40.7 39.2 36.6 32.3 28.2 20.1 19.1 17.3 8.4 others Most people try to be fair with others 15.4 22.2 27.2 36.2 22.6 24.2 16.9 38.2 49.3 47.9 23.3 31.9 77.1 Some people try to take advantage of others and some try to 12.8 18.8 21.2 19.2 34.7 31.2 45.4 23.8 21.9 27.7 51.3 47.4 13.0 be fair (DO NOT READ) DK/NA 1.0 0.2 0.9 2.3 2.0 5.5 1.1 5.7 0.7 4.2 6.3 3.4 1.5 6

In contrast to the findings on individuals inherent nature, those responding that most people try to take advantage of others exhibit significantly less interpersonal trust. Thus trust in people is far more affected by the moral quality ascribed to humanity in general (the desire to be fair to others, or its absence) than by economic rationality (the desire to extract some personal gain from social interchange). Table 1.4: Generally speaking, would you say that most people can or cannot be trusted? According to view of human conduct Mean on a 10-point scale where 0 means most people cannot be trusted and 10 means most people can be trusted (Base: all cases) Most people try to take advantage of Most people try to be fair with others others DK 5.7 7.5 UK 5.4 7.1 DE 4.6 7.0 IT 4.9 6.9 USA 4.6 6.9 ES 5.2 6.8 RUS 4.2 6.8 FR 4.6 6.5 IS 4.3 6.5 JP 5.1 6.2 MEX 5.1 6.0 CH 4.2 5.4 TR 4.0 5.0 1.2 Interpersonal trust by affective proximity As well as their feelings of trust for most people, respondents were asked about the trust they felt towards groups of people more or less close to them, like family members, friends, work colleagues and people from their own city or country. The results show degree of closeness to be a clear determining factor for trust. The specific trust felt for those groups with whom the respondent has a close bond (friends, family) does not radiate out with the same intensity to more distant collectives. Trust diminishes as we move out from closer to more remote circles, starting clearly with family members and friends, followed by the people respondents work or study with, if they are working or studying. According to the survey data, trust in family members is equally high in societies with very different cultures, traditions and religions, which underscores the importance of general trust and trust in specific social groups as a differentiating feature between one society and another. One striking result is the fairly low level of trust felt towards neighbours in certain countries (Russia, Italy, Chile, France, Israel), which tend to assign them a similar score to more distant circles like people of the same city or the same country. It appears that, in these cases, the ties formed with neighbours do not lift them above the category of people in general as generators of trust. Nor is there much discrimination between the trust expressed for people of the same town or city and people of the same country, with average values similar in all cases. In some countries, trust in the primary circle extends fairly easily to other groups. Trust in friends is at times as high as trust in family members. This is the case in Denmark, United States, France and United Kingdom. In other countries, however, trust in the family does not radiate out to remaining circles: in Japan, Chile, Russia or Turkey, for example, the family merits significantly more trust than any other group, including friends. 7

We can aggregate trust in each group to obtain a global measurement of interpersonal trust, distinguishing between the trust felt in closer and more distant networks, as shown in the lower part of the next figure. The summated scale of close networks was constructed from trust in friends and family members and has a range from 0 to 20, while the scale of more distant networks was constructed from trust in neighbours, the people of your town or city and the people of your country, with a range from 0 to 30. These scales facilitate a comparative reading across countries. They show that although trust in close networks is high in all countries, Denmark stands first in the upper part of the map, while Chile occupies the opposite extreme. Denmark also stands out from the group by the measure of trust in more distant networks, where Spain too records a high level. The countries showing least trust in more distant networks are Russia, Israel and Chile. Table 1.5: Overall, how much do you currently feel you can trust the following groups and professions? Mean on a 10-point scale where 0 means you feel you cannot trust them at all, and 10 means you feel you can trust them a great deal (Base: all cases (no. of cases)) DK USA UK DE ES JP RUS IL FR IT TR MEX CH Family members 9.2 8.9 8.8 8.9 8.7 9.1 9.3 8.9 8.2 8.9 9.2 8.9 8.2 Your friends 1496 1502 1483 1581 1506 1483 1600 1500 1561 1491 1503 1491 1494 9.0 8.4 8.3 8.3 8.1 7.7 7.5 7.5 7.8 7.0 6.5 7.2 7.0 1497 1492 1484 1560 1461 1448 1586 1473 1539 1492 1465 1310 1128 People you work or study with, if you are working or studying 7.8 1076 6.7 1179 6.7 994 6.5 1052 6.9 867 6.7 894 6.5 1188 6.5 1051 6.1 957 6.1 934 7.2 620 6.6 822 5.6 815 Your boss, supervisor or teacher, if you are working or studying 7.7 995 6.7 1120 6.3 938 6.2 943 6.0 786 6.4 836 5.4 1124 6.1 936 5.6 858 5.8 774 6.6 465 6.5 814 5.8 765 Your neighbours 7.2 6.3 6.4 6.2 6.0 5.7 4.9 5.1 5.3 5.2 6.0 6.2 5.0 1385 1469 1432 1527 1481 1425 1606 1471 1540 1460 1491 1493 1476 The people of your town 6.8 5.9 5.5 5.8 6.4 5.5 4.7 5.0 5.3 5.4 5.2 6.4 5.1 or city 1461 1478 1404 1504 1479 1356 1566 1454 1506 1457 1489 1487 1498 The people of your 6.5 5.7 5.4 5.5 6.2 5.1 4.8 4.5 5.2 5.3 5.8 5.7 4.7 country 1482 1484 1412 1483 1480 1357 1542 1486 1522 1465 1494 1484 1487 Summated scale of trust in closer networks (family 18.2 17.3 17.0 16.9 16.5 16.5 16.3 16.2 15.7 15.7 15.5 15.1 13.4 and friends) Summated scale of trust in more distant networks (neighbours, people from city, country) 2 19.7 17.7 16.6 16.6 18.2 15.4 13.9 14.1 15.3 15.6 17.3 17.9 14.3 2 People you work or study with and your boss or teacher have been excluded from these scales as these indicators only apply to the smaller sampler of individuals who are in work or study. 1.3 Aspects of trust Although opinion surveys tend to measure trust as a one-dimensional concept, trust in people draws on a variety of factors. This BBVA Foundation study seeks to explore different aspects of trust and to ascertain which attributes contribute most to its formation. Table 1.6 shows the importance people attach to different qualities when deciding whether someone can be trusted. It is clear that although respondents tend to see all 8

these attributes as important, the highest scores go to those referring to a moral quality; honesty/reliability (people who tell the truth and do what they say). Another key aspect of a more technical nature is competence ( competent ). Although respondents regard it as very important, they rank it below the dimension of honesty. Thus honesty matters more than competence in all countries surveyed when it comes to generating trust. The same holds true for other facets less moral or more pragmatic in nature, such as similarity to respondents or a good appearance. Turkey and Russia are the countries that attach most important to "being like me" while a good appearance scores highest in Mexico. Table 1.6: How important are each of the following qualities in deciding whether to trust a person? Mean on a 10-point scale where 0 means not at all important and 10 means very important (Base: all cases) TR DK USA DE IT CH IL RUS E S UK MEX JP FR That s/he be honest 9.6 9.5 9.4 9.2 9.2 9.3 9.1 8.8 9.0 9.0 8.9 8.8 8.0 That s/he be reliable That s/he be truthful 1510 1497 1502 1601 1503 1499 1503 1614 1508 1473 1498 1488 1544 9.6 9.1 8.6 9.1 9.0 9.2 8.9 8.8 8.6 8.2 8.7 8.8 8.4 1508 1493 1501 1596 1501 1496 1502 1617 1508 1473 1493 1488 1568 9.3 9.2 9.4 9.4 9.2 9.5 9.2 8.7 8.9 8.8 9.1 8.9 8.7 1507 1494 1502 1601 1503 1498 1502 1616 1508 1477 1497 1470 1571 That s/he be competent 7.5 6.9 7.7 7.2 7.5 7.0 6.1 7.0 6.7 5.6 7.7 5.6 6.1 or qualified 1502 1429 1486 1570 1492 1487 1496 1589 1500 1434 1490 1441 1553 That s/he be like me 8.7 5.0 5.9 5.6 6.8 6.6 6.5 7.4 6.4 6.0 6.8 6.7 6.2 1507 1447 1490 1584 1483 1490 1501 1588 1486 1444 1485 1456 1543 That s/he look after his or 6.0 6.0 6.0 5.7 6.2 6.2 4.6 5.6 5.4 4.9 7.1 6.1 5.2 her appearance 1507 1484 1493 1580 1487 1494 1493 1585 1504 1453 1490 1461 1539 That s/he be generous 7.8 5.0 7.1 5.7 8.0 8.5 8.3 7.8 7.8 5.8 8.2 8.0 7.5 1502 1464 1493 1589 1497 1498 1501 1606 1505 1454 1494 1457 1555 That s/he be concerned 8.5 8.0 8.3 8.3 8.4 8.7 8.4 8.4 8.1 7.8 8.4 8.1 7.6 for others 1505 1474 1501 1598 1499 1493 1502 1608 1498 1457 1483 1471 1560 1.4 Diversity and trust Some characteristics like age, sex, ethnic background or religious beliefs can increase or reduce the trust felt in others. Having religious beliefs is a support factor for trust among most Turks, half of Mexicans and Americans and a large percentage of Chileans. In remaining countries, most respondents state spontaneously that a person's religious beliefs are not a factor in deciding whether or not to trust them. Likewise, in most countries over half the population spontaneously state that race or ethnic background is not a factor in deciding whether to trust someone. The country dissenting most from this pattern is again Turkey, where half of respondents say it is easier to trust someone of your own race or ethnic group, and Mexico, where 43% think the same. In other countries, notably Japan and Russia (38%) followed by Denmark, United States and Italy (31%), a sizeable segment cite the same race or ethnic background as a stimulus to trust. These data suggest that sameness plays an important role in certain countries in deciding who to trust. A far larger number of respondents believe that age and gender have an influence on trust. In Chile, Mexico, the United States, Israel, Russia and Turkey age is a strong support 9

factor for the generation of trust, with the majority stating that older people are more to be trusted than young people. Likewise, a relative majority in the two Latin American countries and the United States see women as more trustworthy than men. The European countries surveyed dissent from this pattern in that most people tend to see these factors as irrelevant in deciding whether or not to trust someone. Spain is one of the countries where respondents least single out a particular quality as influencing trust. Table 1.7: Who would you say you can trust more? (Base: all cases) CH MEX USA IL TR RUS DK DE UK FR JP ES IT Older people or young people? Older people 66.2 66.3 59.6 53.6 47.2 46.8 42.5 40.4 33.2 33.0 32.9 30.2 25.5 Young people 7.4 7.2 5.9 12.2 17.0 8.5 3.4 5.4 3.9 4.6 4.3 13.1 12.6 It is irrelevant (DO NOT READ) 25.4 24.7 31.6 31.4 33.4 41.1 51.3 43.9 56.8 57.5 56.8 53.7 60.0 DK/NA 1.0 1.8 3.0 2.8 2.4 3.6 2.8 10.4 6.1 4.8 6.0 3.0 2.0 Women or men? Women 42.1 43.5 37.8 27.4 34.5 30.0 20.7 20.4 23.7 16.2 20.2 21.4 20.5 Men 21.9 14.1 20.1 31.4 26.5 24.3 13.3 15.5 9.2 11.4 18.7 16.2 19.6 It is irrelevant (DO NOT READ) 35.0 40.7 38.6 37.2 36.4 43.0 63.4 51.1 61.1 66.3 54.5 59.6 57.5 DK/NA 1.0 1.7 3.5 4.0 2.6 2.7 2.6 13.0 6.0 6.1 6.7 2.8 2.5 Someone who holds religious beliefs or someone who does not? Person with religious beliefs 41.8 50.1 49.0 28.8 61.7 38.6 12.3 20.2 13.8 12.3 11.3 16.2 22.9 Person without religious beliefs 8.2 6.4 5.7 15.4 3.2 4.3 13.8 6.4 6.7 5.4 14.0 6.2 4.0 It is irrelevant (DO NOT READ) 48.6 40.9 41.5 51.9 32.6 51.5 68.7 52.9 71.7 72.2 59.1 72.3 69.1 DK/NA 1.5 2.8 3.8 4.0 2.4 5.5 5.2 20.5 7.8 10.1 15.5 5.3 4.0 Someone of a different race or ethnic background or someone of your own race or ethnic background? Person of a different race or ethnic 6.6 8.5 4.7 5.9 3.4 1.7 1.6 2.5 2.0 1.7 0.9 4.9 3.9 background Person of the same race or ethnic 26.9 42.6 30.5 25.0 49.7 38.4 30.8 23.6 17.2 17.6 38.9 24.3 30.5 background It is irrelevant (DO NOT READ) 63.1 44.8 60.4 66.6 42.6 56.6 60.6 51.2 72.1 69.6 46.8 65.7 62.0 DK/NA 3.4 4.1 4.4 2.6 4.2 3.2 7.0 22.6 8.6 11.1 13.5 5.1 3.6 There is also a link between the profile of the individual and the weight assigned to each characteristic in deciding who to trust. More specifically, sameness or similarity tend to favour trust. The belief that older people are more to be trusted tends to increase with respondents' age. Likewise, both men and women tend to express greater trust in their own than the other sex, and people holding religious beliefs are likelier to trust those professing some religion. Table 1.8 sets out the age differences in perceptions of trust in older as opposed to younger people. Age similarity as a stimulus to trust is stronger among older than among younger people. Those aged 65 or over tend to see older people as more trustworthy to 10

a greater extent than younger people. Among the younger age group, conversely, more respondents feel they can trust older people than young people. Table 1.8: Who would you say you can trust more, older people or young people? (Base by age: all cases) MEX USA CH TR IL DK DE RUS UK FR JP IT ES Older 18-24 58.2 54.8 53.4 48.4 46.9 45.8 35.9 35.3 31.0 29.7 25.9 23.8 20.7 people 65 and + 77.6 62.5 73.0 34.9 57.1 35.6 48.3 65.7 38.2 41.8 45.1 31.7 43.1 Young 18-24 11.0 8.6 15.5 20.2 20.1 7.9 11.3 20.9 5.6 11.4 10.0 18.6 28.5 people 65 and + 4.3 3.8 3.7 18.6 12.6 3.6 3.0 2.0 1.1 2.0 2.8 10.4 6.2 It is 18-24 29.4 33.7 30.3 30.3 29.7 43.6 41.6 39.2 57.4 53.1 59.2 54.5 48.9 irrelevant 65 and + 14.8 30.3 22.0 43.1 29.8 56.8 40.6 29.9 53.4 51.9 46.5 56.7 46.9 Table 1.9: Who would you say you can trust more, women or men? (Base by sex: all cases) MEX TR USA CH RUS IL JP ES UK DE IT DK FR Women Women 59.9 51.7 50.6 47.7 43.9 35.0 33.3 31.9 30.6 28.1 27.5 21.3 20.7 Men 25.8 15.4 24.5 36.3 13.4 19.6 6.3 10.0 16.4 12.1 11.6 20.0 11.3 Men Women 6.7 13.0 12.5 21.2 14.9 24.2 6.9 8.6 5.2 9.0 14.2 13.8 6.9 Men 22.1 41.6 28.0 22.5 35.4 38.8 31.2 24.4 13.4 22.4 26.4 12.8 16.4 It is Women 31.8 32.4 33.1 30.0 38.9 36.5 52.7 56.9 58.0 49.8 55.7 62.8 66.9 irrelevant Men 50.3 40.9 44.3 40.2 48.0 37.9 56.3 62.5 64.5 52.4 59.7 64.1 65.7 So while in some countries a large percentage believe that a person's demographic characteristics are not a factor in deciding to trust them or otherwise, there are aspects which act to stimulate trust for certain segments of the population and among them sameness or similarity plays an important role. 1.5 Trust in professional groups This section analyses the trust felt in different groups of professionals who play a key role in our society. It is interesting to know from a social capital standpoint if there is a significant gap between trust in particular professionals and trust in the institutions where they do their work. In line with the findings of other studies, doctors, scientists and teachers, segments clearly associated with the worlds of knowledge and education, attract higher levels of trust in every country. Police officers and judges also tend to find favour, while politicians then civil servants frequently appear towards the "cannot trust them" end of the scale. Some countries, however, exhibit trust patterns differing from this general map. Russia, Chile and Mexico all stand out for the little trust felt towards certain law and order professionals; concretely the police (Russia, Mexico) and judges (Chile, Russia). Respondents in Russia also express little trust for economic actors, both retailers and business people, and for the public sector as represented here by civil servants and other government employees. United Kingdom respondents stand out for their low trust in journalists. Among the countries occupying the upper reaches of the trust map, some interesting differences emerge in the trust rankings of professional groups. In Denmark, where all groups stand above the midway point on the scale, civil servants earn a high trust rating in contrast to the position they occupy in other countries. 11

In the United States, respondents express a great deal of trust in religious and military personnel. This last group also attracts an extremely high trust rating in Turkey, while religious personnel score very high in Mexico. In order to capture greater variability and obtain a more robust measure of trust in professional groups, we have constructed a summated scale for the 12 groups together. This also gives a more specific picture of the trust felt for each group. As groups are rated on a scale from 0 to 10, this summated scale runs from 0 to 120. This allows us to look from another angle at the differences between trust readings in different countries. The mean value of the trust-in-professionals scale varies from one country to another. Once again, the highest values are found in Denmark and the United States, and the lowest in Russia, then Japan and Chile. Table 1.10: Overall, how much do you currently feel you can trust the following groups and professions? Mean on a 10-point scale where 0 means you feel you cannot trust them at all, and 10 means you feel you can trust them a great deal (Base: all cases) DK USA TR FR MEX IT ES UK DE IL CH JP RUS Medical Doctors 8.0 7.7 7.2 7.5 7.6 6.9 7.1 7.5 7.1 7.3 7.0 6.3 5.6 Judges Police officers Religious personnel Teachers Scientists Civil servants and government employees 1489 1493 1498 1573 1499 1504 1502 1474 1587 1505 1501 1463 1617 8.0 6.7 6.6 5.9 5.1 5.7 5.4 6.4 6.6 6.7 4.4 6.4 4.1 1417 1464 1399 1512 1438 1440 1450 1397 1514 1453 1477 1329 1496 7.9 6.4 4.9 5.7 3.8 6.4 5.6 6.0 6.1 4.9 5.3 5.0 3.1 1482 1496 1489 1545 1498 1492 1493 1466 1574 1484 1500 1457 1602 7.8 7.4 6.5 5.9 7.1 6.2 4.4 6.4 6.4 5.1 5.6 4.6 4.8 1357 1469 1491 1439 1486 1471 1476 1361 1453 1441 1494 1284 1455 7.4 7.9 8.1 6.7 7.6 6.5 6.9 7.0 6.5 6.6 7.2 5.9 6.6 1455 1490 1498 1542 1497 1473 1490 1435 1520 1477 1498 1406 1585 7.3 7.2 7.8 6.9 7.0 7.4 7.3 6.0 6.7 7.5 6.6 6.0 7.2 1402 1421 1435 1538 1355 1472 1448 1305 1481 1427 1437 1251 1522 6.8 5.4 4.6 4.7 3.6 4.6 4.8 4.7 4.7 3.7 3.9 4.1 2.9 1484 1474 1482 1533 1481 1487 1473 1409 1556 1490 1491 1438 1607 Military personnel 6.4 7.2 7.6 5.8 6.5 6.4 5.0 5.8 5.7 6.7 4.7 5.7 5.9 1356 1463 1481 1501 1459 1452 1449 1317 1415 1449 1481 1295 1533 Shopkeepers or 6.1 6.2 4.6 6.1 5.8 4.8 5.7 5.8 5.2 4.6 5.2 5.5 3.3 retailers 1462 1468 1459 1563 1472 1481 1470 1425 1550 1492 1486 1356 1574 Businessmen 6.0 6.0 5.4 4.8 5.4 4.9 4.9 5.0 4.4 4.5 4.5 5.1 3.2 1452 1457 1408 1499 1423 1407 1455 1395 1524 1457 1483 1294 1504 Journalists 4.8 5.4 4.4 5.3 5.8 5.1 5.5 3.8 4.6 4.4 5.1 4.9 4.2 1473 1471 1474 1544 1450 1468 1482 1398 1547 1490 1491 1324 1589 Politicians 4.8 3.9 2.5 3.9 3.4 3.3 3.7 3.6 2.8 2.8 2.6 3.2 2.5 1481 1477 1473 1537 1487 1481 1476 1439 1568 1492 1488 1442 1605 Summated scale of trust in professional groups (0 to 120) Cronbach's alpha 0.87 0.89 0.85 0.90 0.87 0.88 0.88 0.90 0.90 0.84 0.84 0.92 0.86 Mean 78.2 76.0 68.0 66.9 66.9 66.6 65.0 64.5 63.3 63.0 61.2 57.4 51.0 Median 80 78 69 69 69 68 66 67 65 64 61 60 51 12

1.6 Trust in institutions Trust in institutions is vital for the proper functioning of society and the democratic process, for the rule of law and for civic health. Without it, the belief goes, the services that institutions provide would be underused, and far more resources would have to be diverted to the enforcement of laws and government policies. The term institutional trust refers to the level of trust that people feel in institutions such as government authorities, the security forces, international organisations, etc. This BBVA Foundation study asked respondents how far they trusted a range of 20 political, economic and social institutions. This provides us with a broad if not exhaustive picture of how people view the chief institutions with an impact on social organisation. The list includes institutions especially relevant to social capital in that their legitimacy and effectiveness depends to some extent on the trust they inspire as service providers or guardians of the norms of coexistence or political organisation. A distinction can be drawn between those operating mainly on a national basis, like the army, government agencies or trade unions, and those that, without necessarily being multinationals, have an international presence. These include NGOs like the Red Cross or Greenpeace, as well as religious authorities of diverse creeds. Table 1.11: Overall, how much do you currently feel you can trust the following institutions? Mean on a 10-point scale where 0 means you feel you cannot trust them at all, and 10 means you feel you can trust them a great deal (Base: all cases) DK USA TR MEX IT ES FR CH UK DE IL RUS JP The Constitutional 8.1 6.7 7.1 5.2 5.6 5.7 5.5 4.6 5.5 6.5 6.9 5.0 Court 1431 1449 1428 1414 1321 1362 1232 1402 1131 1515 1471 1449 The courts of justice The police The army The government 7.8 6.4 6.6 5.1 5.5 5.5 5.7 4.6 5.9 6.2 6.4 4.3 6.1 1443 1445 1439 1400 1438 1432 1495 1432 1342 1548 1476 1534 1261 7.9 6.6 5.5 4.3 6.5 5.8 6.0 5.4 6.1 6.4 5.6 3.3 5.5 1485 1491 1485 1490 1486 1489 1527 1490 1457 1576 1486 1601 1447 6.8 7.1 7.9 6.4 6.5 5.2 5.7 5.0 6.1 5.8 7.3 5.2 5.8 1374 1448 1478 1463 1436 1436 1473 1469 1327 1466 1445 1562 1296 5.9 5.7 7.3 5.3 4.3 5.2 4.6 5.6 4.8 3.6 4.7 4.0 4.5 1485 1488 1486 1480 1482 1477 1533 1499 1412 1572 1495 1610 1432 The regional 6.5 5.9 5.9 5.5 4.7 5.2 5.4 5.2 4.9 4.1 4.6 4.0 5.0 government 1475 1487 1456 1490 1463 1469 1457 1449 1414 1571 1497 1597 1417 Parliament The mass media The central bank Trade unions 6.4 5.7 5.7 4.8 4.8 5.4 5.2 4.4 4.9 4.4 4.2 3.0 4.2 1468 1445 1454 1364 1414 1410 1365 1413 1340 1538 1446 1596 1341 5.5 5.2 4.9 6.4 5.2 5.5 5.2 5.8 4.2 4.7 4.7 4.6 5.3 1490 1476 1470 1480 1481 1489 1550 1489 1412 1583 1490 1611 1389 7.9 6.4 6.7 5.9 5.2 5.0 5.5 5.9 5.8 5.2 5.4 4.9 5.7 1407 1355 1370 1387 1379 1402 1470 1316 1334 1489 1398 1540 1274 6.4 5.6 4.9 4.7 4.9 5.1 5.3 4.8 5.5 4.5 5.1 4.0 5.1 1432 1262 1282 1391 1404 1412 1450 1379 1249 1485 1367 1470 1084 Red Cross/Red 7.2 8.1 7.0 7.8 7.5 7.3 7.0 7.6 7.3 7.1 7.9 6.6 6.4 Crescent 1457 1475 1476 1477 1466 1483 1493 1443 1385 1571 1477 1476 1213 Greenpeace 5.9 6.3 5.6 5.3 6.4 6.6 6.1 6.6 6.1 6.2 5.6 5.8 4.8 1352 1209 977 987 1309 1403 1328 1171 1280 1503 1096 1345 703 13

Table 1.11 cont.: Overall, how much do you currently feel you can trust the following institutions? Mean on a 10-point scale where 0 means you feel you cannot trust them at all, and 10 means you feel you can trust them a great deal (Base: all cases) DK USA TR MEX IT ES FR CH UK DE IL RUS JP Non Governmental 5.1 5.8 6.3 4.6 5.6 6.6 5.6 4.8 5.0 4.6 4.7 3.4 5.2 Organisations 1137 1283 1300 1224 1213 1418 1315 1116 1110 1247 1318 1250 882 United Nations Large corporations 6.3 5.5 4.0 5.1 5.5 5.5 4.8 4.7 5.3 5.2 3.6 4.7 4.1 1375 1345 1288 1241 1364 1376 1378 1148 1231 1449 1311 1255 1035 5.6 4.9 5.2 5.0 5.2 4.5 5.1 4.9 4.6 4.0 4.4 3.3 4.9 1385 1435 1291 1236 1403 1384 1471 1226 1253 1523 1317 1274 1213 The European 5.2 4.9 4.2 3.9 5.3 5.3 5.0 4.5 4.7 4.5 3.2 4.4 3.8 Commission 1351 967 1280 1033 1312 1329 1341 1037 1157 1412 1099 1128 513 Protestant Church Catholic Church 7.2 6.5 4.4 3.6 2.7 4.6 4.8 5.3 5.3 2.0 3.6 1351 1254 1259 813 1062 951 1378 1100 1361 747 555 5.3 5.9 2.0 7.6 6.4 4.6 5.4 6.0 5.1 4.9 2.1 3.1 3.9 993 1318 979 1476 1451 1421 1227 1482 1115 1366 841 975 579 Jewish religious 5.4 6.0 2.2 3.8 3.9 2.7 4.6 4.1 5.0 3.8 4.4 2.8 3.0 authorities 965 1080 1043 1125 792 992 937 944 880 1129 1296 967 493 Muslim religious 3.9 4.0 6.9 3.7 3.0 2.3 3.9 3.5 3.9 2.4 1.8 2.6 2.8 authorities 1070 1101 1411 1066 856 1063 1003 879 946 1177 1159 1026 511 Orthodox Church 5.7 1481 Buddhist religious 4.7 authorities 776 Overall, institutional trust stands in the medium-high range, with trust greater in national than international institutions. The institutional trust map can be drawn as follows: The most trusted institution is the Red Cross/Red Crescent, which scored between 7 and 8 points on a scale of 0 to 10 in all countries surveyed. In second place come institutions related to justice and the upkeep of law and order: the courts of justice and the Constitutional Court, the army, the police, and the central bank (scores from 5 to 7). The medium-low band of the map of institutional trust is occupied by national and regional government, parliament, the multinationals, the media and the trade unions. Finally, the place of least trust goes to religious authorities. In line with the findings for trust in religious believers, religious authorities only generate trust where they are in a majority. For instance, in a traditionally mono-religious country like Italy, only the Catholic Church inspires trust while remaining religious authorities score an average of 3 to 4 points. This pattern is repeated in multi-religious countries but in an attenuated form: hence the Protestant Church generates most trust in the United States and the United Kingdom, with other religious authorities figuring at around 5 on the trust scale, with the exception of Muslim religious authorities. Some institutions attract a high rate of non responses. In addition to the large percentage of non responses given for religious institutions, in some countries a highly significant percentage declined to rate the European Commission, United Nations, the Constitutional Court and non governmental organisations like Greenpeace. The country 14

with the highest percentage of non responses was Japan, followed by Russia and France. The countries with the highest levels of institutional trust are Denmark and the United States, where few institutions stand lower than the mid-point on the scale (only the Muslim religious authorities in both cases, and the European Commission and the multinationals in the United States). Most of the European countries surveyed occupy the middle ground of the institutional trust map. Among the other European countries in the survey, Italy, the United Kingdom and, especially, Germany stand out for their low trust in legislative and executive institutions (survey field work was conducted between February and April 2005). Spain joins Denmark as the only European country where the national government obtains a score above 5. In contrast, trust in Spain is at critical levels with regard to religious authorities, including the Catholic Church and large corporations. Turkey stands out for the high trust accorded to national institutions, from the army and central government to parliament and the police. However, international institutions score fairly low with the exception of the Red Crescent, and especially low in the cases of the European Commission, United Nations and non-muslim religious authorities. Japan and Israel, meantime, exhibit medium or high trust in their various national institutions, and significantly less in international ones. Chile, Russia and Mexico again share some common features: trust in certain law and order institutions tends to be low (courts of justice in Chile and Russia, police in Mexico and Russia). Furthermore, in Chile and Mexico trust in the government tends to be a little higher, while in more developed countries the relation is reversed. Finally, institutional trust is at its lowest levels in Russia, where 15 out of the 20 institutions mentioned earned an average score below 5 points. 1.7 Trust and perception of corruption Corruption, by altering the rules of play, undermines people's expectations and hence their trust in the conduct of persons and institutions. Asked about the frequency of possible corruption cases in their country when a municipal employee, a judge or a police officer accepts a bribe in return for a contract, a favourable sentence or not imposing a fine, respondents revealed a large degree of inter-country variability. The situation suspected to be most frequent in all countries was that of a local government employee accepting a bribe in exchange for a contract or permit. This situation was seen as very frequent by most Mexicans and Russians, and a relative majority of Italians and Israelis. Conversely, the same situation was regarded as exceptional by a majority of Danes. A police officer accepting a bribe to excuse someone a fine or other penalty was also seen as a frequent occurrence by a large majority of Mexican and Russians and a significant percentage of Turks. Finally, a judge being bribed to give a favourable sentence was considered more of an occasional occurrence in most countries. Thus Mexico and Russia evidence the most acute perception of corruption while at the other end stands Denmark, where the suspicion of corruption is clearly far less prevalent. The map of perception of corruption coincides to some extent with the map of trust: societies with more global trust Denmark followed by the United States tend to perceive less corruption. In the cases of Mexico, Russia and Chile, low trust in specific institutions the police and courts of justice - squares with the perception that they harbour a great deal of corruption. 15

Table 1.12: How often do you think the following situations currently arise in our country? (Base: all cases) MEX RUS IL IT CH JP FR TR USA ES DE UK DK A local government employee accepts a bribe in exchange for a permit or contract Always/ Often 64.2 60.9 48.2 45.5 43.0 40.8 35.9 34.4 30.9 28.7 26.9 25.1 5.5 Sometimes 24.3 24 31.1 35.8 37.4 42.1 31.2 32.7 42.2 39.1 37.9 41.5 28.4 Rarely/ Never 9.7 5.2 14.1 13.5 15.1 9.5 22.9 19.4 24 22.7 28.4 24.6 64.3 DK/NA 1.8 10 6.7 5.3 4.6 7.7 10.1 13.5 2.8 9.5 6.9 8.7 1.9 A police officer accepts a bribe in exchange for not imposing a fine or penalty Always/ Often 79.0 74.2 22.9 15.1 34.4 23.4 20.4 44.2 21.6 19.7 9.0 14.1 2.3 Sometimes 16.8 16.6 34.6 39.9 36.6 49.3 35.3 30.3 42.5 38.9 31.4 36.2 8.5 Rarely/ Never 3.7 4.0 33.6 38.1 25.4 17.6 33.2 15.2 33.0 31.2 51.4 41.9 88.0 DK/NA 0.7 5.3 8.9 7.0 3.7 9.7 11.1 10.5 2.9 10.1 8.2 7.8 1.3 A judge accepts a bribe in exchange for a favourable sentence Always/ Often 54.7 46.6 8.7 19.1 31.4 6.6 15.3 11.3 14.4 16.2 4.3 7.8 0.6 Sometimes 25.4 29.3 21.1 38.6 35.7 24.6 29.5 29.9 35.6 37 19.5 23.0 2.4 Rarely/ Never 16.0 10.5 60.6 34.2 28.0 52.5 42.8 39.5 46.7 36.3 63.7 57.2 95.0 DK/NA 4 13.6 9.7 8 4.9 16.3 12.4 19.3 3.3 10.5 12.4 12.1 2 16

2. PERSONAL NETWORKS Individuals' membership of formal and personal networks is a core dimension of social capital, since these networks are a materialisation of trust. In this regard, their study provides useful insights into social interactions. Networks may have individual or collective effects as resources which strengthen the sense of belonging and relations of reciprocity and bring down transaction costs, favouring mutual support and information sharing between individuals and groups. They are indicators of the individual's ability to make his or her way in society and interact with others. From a social capital standpoint, we can distinguish between networks formed within a group and between groups (the strong and weak ties, of Granovetter 1973, or the bonding and bridging of Woolcock 2000). The ways these two kinds of networks mix and match can produce both beneficial and adverse effects. The ties formed within groups favour a sense of belonging, cohesion and the achievement of shared goals. However some authors point out the harm that can be caused by concentrating all one s relations within the same tight group. These are, basically, sectarianism, exclusion and the failure to generalise trust, i.e. to invest trust in people in general and not just the small group in which one is immersed (Fukuyama 1998; Yamagishi 1998). At the same time, the ties formed between groups (weak ties), can surmount divisions based on class, religion and ethnicity, promoting inclusion and tolerance. Some authors also stress the benefits these ties entail for the spread of information and social mobility (Granovetter 1973). From this standpoint, the characteristics of a person's network (size, diversity) can influence the amount and quality of the resources at that person's command. This section looks at a series of networks defined by differing degrees of proximity family members living with you, family members not living with you, friends, neighbours, work colleagues, and examines some of their characteristics including size, degree of closeness, diversity, durability and frequency of contact. In general, people have a similar number of family members and friends, with a positive relationship between the two networks. In most societies, family and friendship networks are wide and contact is on a regular basis; far more so than contact with other collectives like neighbours or work colleagues. Meantime the diversity of networks exhibits considerable variability between countries. 2.1 Size of household The number of members in the household is the key to the composition of each individual s closest network. Household size varies from one country to the next. Small households predominate in Europe and the United States, while the most numerous are found in Mexico, Turkey and Chile. In 4 out of the 6 European countries in the study, over 50% of individuals live in households composed of one or two persons. Denmark, Germany and France stand out for their large numbers of single-person households, a fact which may have major implications for network composition, social life, attitudes to asking for help, etc. Living alone is something clearly linked to the life cycle itself: in most European countries and the United States, more than 30% of those aged over 65 live on their own. In France, Germany and Denmark, the phenomenon extends to over 30% of the population aged under 30. In Spain, conversely, only 5% of this age group live alone. 17

Table 2.1: Household size Percentage of individuals (Base: all cases) DK DE FR UK US RUS ES IL IT JP CH TR MEX 1 person 30 29 28.2 20.2 17.7 15.4 14.6 12.6 10.8 8.7 8.1 6.4 4.8 2 persons 41.2 37 32.6 30.9 33 27.8 30.3 24.2 24.4 20.1 16.6 16.5 14.2 3 persons 12.3 16.1 16 20 20.8 26.3 25.6 16 26.4 22.1 20.9 20.9 18.5 4 persons 12.3 13.3 15.1 20.2 15.9 20.3 21.2 18.8 27.9 29.3 21.7 25.8 24.4 5 or more 4.2 4.6 8.1 8.7 12.6 10.2 8.3 28.4 10.5 19.8 32.7 30.4 38.1 persons 2.2 The family network In addition to the family members they live with, respondents were asked abut the number of relatives with whom they are in regular contact. Family networks in this sense tend to comprise over 8 members in all countries except Germany, Russia and Japan 1. The total size of the family network does not necessarily correspond to the size of the household. In a number of societies people living in small households may nonetheless have an extensive family network. Spain and Denmark combine small households with a big family network outside the home. In Japan, conversely, the size of both is more evenly matched. Figure 2.2: Composition of the family network Average number of family respondent lives with and average number of family with whom respondent maintains regular contact without sharing a home (Base: all cases) No. of family members you live with No. of family members with whom you maintain regular contact Turkey 2,8 8,1 10.9 Spain 1,8 9,1 10.9 Mexico 3,2 7,3 10.5 Is r ae l 2,5 7,7 10.2 Denmark 1,2 8,9 10.1 Italy 2,1 7,9 10.0 Chile United States France United Kingdom 2,8 1,8 1,5 1,7 7,2 7,5 6,5 7,0 9.8 9.0 9.0 8.2 Germany Russia Japan 1,3 1,9 2,4 5,8 4,7 3,5 7.1 6.6 5.9 0 2 4 6 8 10 12 14 16 The number of family members a person can rely on can be defined in different ways. In the past, the size of family networks was calculated on the basis of degree of contact. The present survey tries to demarcate this concept a little more clearly by asking about family members the respondent has a close relationship with, defined as 1 All personal network indicators are presented without extreme values, defined as those above the 75 percentile by more than three times the value range of 50% of observed cases. For example, if 50% of respondents score 2 and 4, extreme values are all those higher than 10. 18