WPS5945 Policy Research Working Paper 5945 Life Satisfaction, Social Capital and the Bonding-Bridging Nexus Maurizio Pugno Paolo Verme The World Bank Middle East and North Africa Region Economic Policy, Poverty and Gender Unit January 2012 Policy Research Working Paper 5945 Abstract The paper investigates the relation between social capital variable the locus of socializing and use it to explore the and life satisfaction focusing on the distinction between relation between social capital and life satisfaction across bonding and bridging. Using the latest version of the world citizens and across groups of similar countries. combined World and European Values Surveys, the The results indicate that people with extreme bonding or authors first address the question of measurement of bridging attitudes are less happy than people with more social capital by means of a multi-step factor analysis. balanced attitudes. Unlike the literature on social capital Through this procedure, they nd that proxies typically and economic growth that finds bridging attitudes more used for social capital tend to polarize around two desirable than bonding attitudes, they nd that bonding dimensions interpreted as bonding and bridging. These attitudes are at least as important as bridging attitudes two dimensions are in fact associated with a single latent for life satisfaction. This suggests that the social capital variable with opposite signs suggesting that they describe dimensions important for economic growth may not two sides of the same latent variable rather than two necessarily coincide with the social capital dimensions independent latent variables. The authors call this latent important for life satisfaction. This paper is a product of the Economic Policy, Poverty and Gender Unit, Middle East and North Africa Region. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at pverme@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Life Satisfaction, Social Capital and the Bonding-Bridging Nexus Maurizio Pugno∗ and Paolo Verme† Abstract The paper investigates the relation between social capital and life satisfaction focusing on the distinction between bonding and bridging. Using the latest version of the combined World and European Values Surveys, the authors �rst address the question of measurement of social capital by means of a multi-step factor analysis. Through this procedure, they �nd that proxies typically used for social capital tend to polarize around two dimensions interpreted as bonding and bridging. These two dimensions are in fact associated with a single latent variable with opposite signs suggesting that they describe two sides of the same latent variable rather than two independent latent variables. The authors call this latent variable the locus of social- izing and use it to explore the relation between social capital and life satisfaction across world citizens and across groups of similar countries. The results indicate that people with extreme bonding or bridging attitudes are less happy than people with more balanced attitudes. Unlike the literature on social capital and economic growth that �nds bridging attitudes more desirable than bonding attitudes, they �nd that bonding attitudes are at least as important as bridging attitudes for life satisfaction. This suggests that the social capital dimensions important for economic growth may not necessarily coincide with the social capital dimensions important for life satisfaction. JEL: A13, D6, I3, Z1. Keywords: Life Satisfaction, Social Capital, Bonding, Bridging. Sector Board: Poverty (POV) ∗ Department of Economic Sciences, University of Cassino, Italy. † The World Bank and Department of Economics “S. Cognetti de Martiis�, Torino, Italy. 1 Life Satisfaction, Social Capital and the Bonding-Bridging Nexus 1 Introduction Social capital is a relatively new concept in economics, but the literature on the subject is already rich and controversial. It has been recognized that this concept captures a phenomenon important for economics, in that it relates to trust and cooperation, and hence to efficiency and economic growth. At the same time, economists have brought several criticisms against the concept of social capital, such as its vague de�nition and the difficulty of its measurement (Durlauf and Fafchamps 2005; Dasgupta 2005; Paldam and Svendsen 2000). The de�nition of social capital varies across the social sciences. In sociology and in political science, social capital is often de�ned as a concept pertaining to organizations, or even nations, because it has been mainly referred to social relationships and supporting structures (Putnam 2003; Fukuyama 1997). Economics, and in part Bourdieu (1986), have usually de�ned the concept of social capital, as it was originally conceived by Loury (1977), and then developed by Glaeser et al. (2003), as pertaining to individuals. Social capital has also been criticized for the difficulty of its measurement. The concept emerges from the literature as multidimensional and as including immaterial components. Economists seemingly prefer to focus on some speci�c variables usually considered to be indicators of social capital, like trust (e.g. Glaeser 2000; Fehr 2009; Alesina and La Ferrara 2002), or membership of social organizations (Glaeser et al. 2000). Instead, when the concept of social capital has been explicitly treated, it is generally considered as a latent variable captured by various proxies (Beugelsdijk and Smulders 2009). Recently, social capital has attracted growing interest, also among economists, in relation to individual self-reported life satisfaction.1 This relationship has been usually found to be positive, and in some cases very signi�cant, depending on the indicators used for social capital (Bjornskov, 2006, Helliwell, 2006, 2008, Ram, 2010, Leung et al., 2011). However, the relationship between social capital and economic growth, which has been studied in the economic literature since the seminal article by Knack and Keefer (1997), 1 For a discussion on the concept of self-reported life satisfaction see Layard (2005), Kahneman and Krueger (2006), and Veenhoven (2007). 1 is not as consistent in predicting a positive role of social capital. For example, Olson’s (1982) economic analysis of interest groups and Ban�eld’s (1958) sociological analysis of “amoral familism� �nd a negative relationship between social capital and economic growth. Putnam (2000) has in fact recognized that social capital may display positive or negative externalities and proposed the distinction between bonding and bridging social capital in order to study these divergent effects. Bonding social capital emerges when trust and cooperation are restricted within groups, so that negative externalities may ensue for other people. Bridging social capital emerges when new linkages between groups arise, so that cooperation increases. Recent exploratory �ndings in the economic literature seem to con�rm that bonding and bridging social capital can be considered as two distinct variables, and that the former has a negative effect on economic growth and the latter has a positive effect (Knudsen et al. 2010; Beugelsdijk and Smulders 2009). Moreover, some studies show that family ties, usually included in bonding social capital, causally reduce trust in people, which is the basic indicator for bridging social capital (Alesina and Giuliano 2009; Ermish and Gambetta 2008) while other studies show that both family ties and trust in people are positively correlated with people’s life satisfaction (Alesina and Giuliano 2007, Helliwell and Wang 2010). Two main problems thus clearly emerge. First, how to distinguish between the bonding and bridging dimensions starting from a variety of proxies for social capital, and how the bonding dimension is related to the bridging dimension. Second, how these two dimensions of social capital are related to individual life satisfaction: in particular, whether they relate positively in any case, thus departing from the mixed results on the relationships of bonding and bridging social capital with economic growth. In addressing these two problems, and by using a large world dataset, this paper obtains two main surprising results. First, proxies for bonding and bridging social capital appear to describe two sides of the same latent variable, rather than two distinct latent variables. This would imply that bonding and bridging cannot be conceived as two distinct dimensions of social capital but should instead be seen as two opposite sides of a single dimension of social capital, which is a broader concept possibly including other dimensions. This result prompts us to call the bonding/bridging dimension the locus of socializing, in order to emphasize that it is where people mostly enjoy social relationships, whether within their kinship and existing ties, or across new social ties. 2 The paper also reports a second surprising result: that often balanced attitudes to- wards bonding and bridging predict life satisfaction better than extreme bonding or bridging attitudes. For example, individuals who attach more importance to the family than to friends, or more importance to friends than to the family, appear to be less sat- is�ed with their lives than individuals who take a more balanced view. The same applies for the importance of work with respect to leisure. Individuals who hold balanced views on the importance of work and leisure seem happier than individuals who have unbal- anced preferences. Overall and with one exception, bonding variables tend to deliver more happiness than bridging variables, a �nding consistent with previous literature on social capital and life satisfaction. These results are obtained by carefully examining the available literature and the available proxies for bridging and bonding social capital in the latest version of the combined World and European Values Surveys (WVS-EVS) and by applying a multi- step factor analysis for the identi�cation of the bonding and bridging latent variables. This approach has induced us, in particular, to pay more attention than in the past to self-reported values and beliefs such as the importance attached to family, work and children’s qualities. This is in line with the theoretical and empirical arguments on the role of culture in economics (Guiso et al. 2006, Tabellini, 2010). The paper is organized as follows: section 2 conducts an in-depth review of the literature on bonding and bridging social capital in order to evidence the motivation of the paper; section 3 describes the data and the model used; section 4 analyses how bonding relates with bridging indicators; section 5 sets out the results on the relationship between the bonding/bridging dimension and individuals’ life satisfaction; and section 6 concludes. 2 Motivation and literature review The importance of social capital for economists is evident from the variety of economic phenomena that have received better explanation with the use of the concept of social capital, such as economic growth (Knack and Keefer 1997), and, among others, individual self-reported life satisfaction and happiness. The literature on what is now called ‘happiness economics’ has been growing rapidly, often by basing the analysis on psychology, and by yielding implications for public policy. A variety of evidence on the validation and reliability of subjective data of this kind has 3 also been provided, so that happiness economics can be regarded as both a �eld and a method of research (Stutzer and Frey 2002, 2010; Konow and Earley, 2008). One of the most robust �ndings in happiness economics is that marriage and social life play a key role in individuals’ life satisfaction, and that it is not less important than household income (Layard 2005; Nickerson et al. 2003; Wilson and Oswald 2005). More speci�c studies on the role of social capital in individuals’ life satisfaction show a number of results: • First, social capital, as captured by the three main indicators of trust, networks, and norms (Putnam, 1993:167), contributes greatly and signi�cantly to explaining the cross-country variance of individual life satisfaction. Besides trust in people, typical proxies used for social capital in Putnam’s sense are ‘membership of social organizations’, ‘having someone to rely on’, ‘donating money or time to an organi- zation’, contacts with family, friends and neighbors, ‘justifying cheating on taxes’, ‘importance of religion’, ‘religious practice’, and other proxies for governmental quality (Helliwell 2008, 2006). Helliwell et al. (2009) even conclude that in all [world] regions [...] social support is tightly linked to life satisfaction, with a global coefficient that exceeds that on log income.2 • Second, time devoted to social relationships causally impacts on life satisfaction in a positive and relevant way (Meier and Stutzer 2008; Becchetti et al. 2008; Bruni and Stanca, 2008). Proxies for the time devoted to relations are ‘frequency in performing volunteer work’, ‘attending social gatherings’, ‘cultural and religious events’. • Third, the deterioration of social capital in the US over the most recent decades, as claimed by Putnam (2000) and con�rmed by others (Costa and Kahn 2003; Robin- son and Jackson 2001), is associated with the decline of individuals’ happiness in the same country, as reported by Layard et al. (2009), Easterlin and Angelescu (2009), and Stevenson and Wolfers (2008). This linkage seems to be signi�cant and relevant, even after controlling for individuals’ income and comparison income (Bartolini et al. 2008; see also Bjørnskov 2008). 2 Similar conclusions are found by Winkelman (2009), who has used panel data for Germany, and by Leung et al. (2011), who have used non-panel data for Canada, while Ram (2010) has found more fragile results by using country data instead of individual data. 4 Social capital is a broad term and economists have typically criticized the concept for its vague de�nition and the difficulty of its measurement (Durlauf and Fafchamps, 2005; Dasgupta 2005; Paldam and Svendsen, 2000). Economics has thus preferred to focus on speci�c observable variables rather than latent variables (Glaeser, 2000, Fehr, 2009, Alesina and La Ferrara, 2002) and on individuals rather than societies at large (Loury, 1977; Glaeser et al., 2003). The sociological literature recognizes the problem of the multidimensionality and ambiguity of social capital but proposes the distinction between bonding and bridging social capital. Putnam (2000:22-24), in particular, devotes some interesting theoretical considerations to this distinction: “Of all the dimensions along which forms of social capital vary, perhaps the most important is the distinction between bridging (or inclu- sive) and bonding (or exclusive). Some forms of social capital are, by choice or necessity, inward looking and tend to reinforce exclusive identities and homogeneous groups. [...] Other networks are outward looking and encompass people across diverse social cleav- ages. [...] Bonding social capital, by creating strong in-group loyalty, may also create strong out-group antagonism, [...] and for that reason we might expect negative external effects to be more common with this form of social capital. Nevertheless, under many circumstances both bridging and bonding social capital can have powerfully positive social effects.� This distinction has been taken up and discussed by other authors, so that two as- pects clearly emerge (Gittel and Vidal 1998; Woolcock and Narayan 2000; Wallis et al. 2004). Whereas bonding social capital arises when individuals strengthen existing ties within a community, thus reinforcing trust for speci�c people and speci�c actions, bridging social capital arises when individuals explore new ties and links with other communities, thus facing a greater risk. Whereas bonding social capital may generate negative externalities among different communities due to the pursuit of sectarian inter- ests, bridging social capital generates positive externalities among different communities, because of the creation and exploitation of new opportunities. The reason for the interest in the distinction between bonding and bridging social capital regards the externalities effects, which make it possible to predict whether the effects of social capital are positive or negative. Distinguishing between bonding and bridging would also be instrumental in separating the de�nition of social capital from its effects. The negative effects of closed interest groups on the economy are well-known in 5 both economics (Olson 1982) and sociology (Ban�eld 1958), so that one would expect to �nd a great deal of empirical research on the distinction between bonding and bridging social capital. Instead, there are few studies on this distinction, and they are especially rare in economics. Putnam (2000: 23) frankly admits that “I have found no reliable, comprehensive, nationwide measures of social capital [for the US] that neatly distinguish ‘bridgingness’ and ‘bondingness’.� The few economic studies to have considered the bonding/bridging distinction in the analysis of the positive effects of social capital focus on economic growth. A much-cited study in this regard is Knack and Keefer (1997), which is based on the World Value Sur- vey (WVS). Its main �nding is that trust of the survey’s respondents impacts positively and signi�cantly on a country’s economic growth. The subsequent literature has largely con�rmed this result (e.g. Zak and Knack 2001; Beugelsdijk et al. 2004; Dearmon and Grier 2009). Knack and Keefer then attempt to apply the bonding/bridging distinction of social capital, though they do not use this terminology. In fact, they distinguish people’s participation in social organizations depending on the kind of organization, i.e. whether it is of the Putnam variety, like cultural organizations which enhance coop- eration and solidarity, or of the Olson variety, like political parties, which are rather motivated by rent-seeking behavior. However, although the estimated coefficients in the overall GDP growth equations differ between the two kinds of organization, they are not signi�cant. Two other very recent studies focus on regional growth in the US and the EU respec- tively. Knudsen et al. (2010) apply factor analysis to distinguish among the different kinds of participation in social organizations. However, they only �nd that participation is not of the same kind, so that they are obliged to �x an a priori criterion, and to plead for “experts� for its application. They can thus distinguish organizations into two groups depending on whether these organizations are exclusive or inclusive with respect to new entrants. The authors again apply factor analysis to several indicators of social capital, and �nd two substantial factors that can be attributed to bridging and bonding social capital respectively. The �rst factor has positive and substantial loadings on the indicators of ‘diversity of friendships’, ‘inclusive organizations’, and ‘political activity’, while the second factor has positive and substantial loadings on the indicators of ‘ex- clusive organizations’ and ‘faith-based engagement’. The other loadings in both factors are instead negligible. After the application of other tests, it is �nally concluded that 6 bonding and bridging are two distinct dimensions. When Knudsen et al. (2010) estimate growth equations based on the results of the factor analysis, an interesting result emerges: the bridging composite indicator is pos- itively correlated with regional income growth, but the bonding composite indicator is negatively correlated. A similar result with composite indicators is obtained by Beugels- dijk and Smulders (2009). However, their distinction between bonding and bridging proxies is a priori, like that of Knack and Keefer. They select bridging social organiza- tions on the basis of presumed non-rent-seeking activity, while the bonding composite indicator is built on the survey questions concerning the ‘importance of family, friends and acquaintances’. The result that the bridging and bonding types of social capital display a correlation with economic growth with opposite signs has been further con�rmed for some speci�c indicators. For example, Tabellini (2010) selects four social capital indicators: gener- alised trust, ‘feeling free and controlling over own life’, and the importance attached to child’s respect for other people, and obedience. A certain bridging/bonding distinction emerges from the principal component analysis, which shows that the �rst principal component of these four indicators displays a positive correlation with the �rst three, and a negative correlation with the obedience indicator. The estimate of growth rates in the EU regions again shows that the �rst three indicators display a positive sign, and the obedience indicator displays a negative sign. Guiso et al. (2010) speci�cally address the civic component of social capital, which may be included in bridging social capital. The selected survey questions regard the justi�cation for ‘claiming government bene�ts when not entitled’, for ‘cheating on taxes’, and for ‘accepting a bribe in the course of own duties’. The authors �nd that these three indicators are positively correlated with their �rst principal component, but that the principal component is not signi�cantly correlated with economic growth, although its correlation with some proxies for government efficiency is positive and signi�cant. A particular aspect of the distinction between bonding and bridging social capital that has been speci�cally investigated is the relationship between family ties and gen- eralised trust. This relationship would be ambiguous on a priori grounds, because, on the one hand, children learn trust within the family as a necessary proving ground for extending trust to other people, as developmental psychologists argue (e.g. Ainsworth et al. 1978; Kafetsios and Nezlek 2002) while, on the other hand, family ties may limit 7 opportunities and motivation for outward exposure, thus limiting trust in strangers (Er- mish and Gambetta 2008). Family ties would thus display a positive relationship with generalized trust in the former case, and negative in the latter case. The paper by Alesina and Giuliano (2009) shows that the relationship between family ties and generalised trust is negative when the strength of family ties is measured on reported scores for the ‘importance of the family in an individual’s life’, the ‘duties and responsibilities of parents and children’ and the ‘love and respect for ones parents’. This result is based on the WVS and has been con�rmed by experimental data by Ermish and Gambetta (2008), who add that the negative causality runs from family ties to generalised trust. Another aspect of the distinction between bonding and bridging social capital which has been investigated is the relationship between religiosity and generalised trust. Guiso et al. (2003), on the basis of the WVS, �nd positive and signi�cant correlations be- tween generalised trust and some measures of religiosity, i.e. ‘raised religiously at home’, ‘attending religious services’, and ‘believing in God’. This result is con�rmed by Helli- well (2008). However, religious denomination matters signi�cantly (Glaeser et al. 2000; La Porta et al. 1997). For example, Guiso et al. (2010) show that the proportion of Catholics in the population is negatively related with trust in strangers. Regarding eco- nomic growth, Barro and McCleary (2003) show that this seems to respond positively to the extent of religious beliefs, but negatively to church attendance. Generalised trust is clearly included in bridging social capital, but the survey question normally used to measure generalised trust in many studies has raised some problems in this regard. The question, which also appears in the WVS, is “Generally speaking, would you say that most people can be trusted or that you can’t be too careful in dealing with people?� The answer could be either “Most people can be trusted� or “Can’t be too careful�. Gleaser et al. (2000) argue, on the basis of a mix of survey and experimental data, that the survey question about trust predicts trustworthiness and not trust. If this result were con�rmed by the literature, this question would give ambiguous information about bridging social capital. However, Fehr (2008), who also uses experimental evidence, �nds that the trust question captures a component of trust as a social preference, and not just as a belief about others’ trustworthiness and as a risk preference. As regards the impact of social capital on individuals’ life satisfaction, there are 8 no studies in the economic literature that are speci�cally focused on the distinction between bonding and bridging. Bartolini et al. (2008) consider, besides generalized trust, people’s participation in social organizations, whether of the Olson or Putnam variety, thus following Knack and Keefer (1997). The impact on life satisfaction of the former type of organizations emerges as negative and of little signi�cance, while the impact of the latter emerges as positive and highly signi�cant. Other studies focus on the impact of speci�c indicators of social capital on life sat- isfaction, but their �ndings are not necessarily in line with the results discussed above. Family ties are found to be positively, signi�cantly, and sizeably correlated with life satisfaction (Alesina and Giuliano 2007), which is consistent with the recurrent result of the positive association between marriage and life satisfaction. However, this same result is not easily reconciled with the negative correlation found between family ties and generalized trust, which is positively associated with life satisfaction. A similar but less clear issue emerges from religiosity. This variable appears to be positively correlated with life satisfaction in several studies (Dolan et al. 2008; Clark and Lelkes 2009), but it also depends on the indicators of religiosity used (Helliwell 2008), and (according to psychologists) on the indicators of self-reported well-being used (Lewis 2005). In particular, if the life satisfaction scale is of the ‘Cantril ladder’ type, as in the Gallup Poll dataset, then the importance of religion does not show a signi�cant correlation with life satisfaction (Helliwell 2009). The conclusions from this brief review of the literature (summarized in Table 1) are thus the following. First, social capital cannot be taken as a homogeneous concept. In fact, some exploratory and partial analyses show that social capital can be captured by speci�c measurable indicators, which �t nicely with the de�nitions of bonding and bridging as two dimensions of social capital, and which seem to relate negatively to each other in some important instances. Therefore, further analyses on the relevance of this distinction should especially investigate the relationship between these two dimensions of social capital. Second, the distinction between bonding and bridging indicators of social capital is important because of their distinct role in growth and life satisfaction. The literature on social capital and growth has shown that speci�c individual indicators or composite indicators often relate positively with economic growth if they are of the bridging type, but negatively if they are of the bonding type. Similar studies on the role of bonding and bridging indicators for life satisfaction are still scarce. 9 In the sections that follow, we will use the latest version of the WVS-EVS to ad- dress the two questions stated earlier: the de�nition and relation between bonding and bridging indicators; and the relation between these indicators and life satisfaction in the context of world regions characterized by different histories, cultures and levels of development. [TABLE 1] 3 Data, model and variables The data we use include all rounds of the WVS-EVS carried out between 1980 and 2008. The data set was created by combining the 1981-2004 World and European integrated data �le (version 2006.04.23) and the last round of the World Value Survey following the instructions provided by the World Values Survey organization (www.worldvaluessurvey.org). The resulting dataset includes �ve rounds of the World Values Surveys and three rounds of the European Values Surveys for a total of 98 countries, 1639 regions, 24 recorded years and 355,298 observations. This is the largest dataset of its kind to date. Our �nal objective is to assess the impact of bonding and bridging attitudes on life satisfaction worldwide. For this purpose we use what can be described as a standard model when working on happiness cross-country and longitudinally with the WVS-EVS. The model is as follows: Hi = α + βBi + γXi + ηCc + δYt + µi (1) where H stands for happiness, B is a vector of variables representing bonding and bridging attitudes, X is a vector of control variables representing individual and house- hold characteristics, C is a vector of countries dummies, Y is a vector of years dummies, α, β, γ, η, δ, are the parameters to be estimated and µ is the error term. The subscript i stands for individuals and the subscripts c and t stand for countries and time (years) respectively. Therefore, in the most general model, we pool all available observations for all countries and years and estimate the happiness equations on individuals with countries and years �xed effects. The same equation is applied to the full set of coun- tries available and to 16 separate groups of countries representing different geographical areas, recent histories, cultures and levels of economic development. All equations are estimated with robust standard errors (Huber-White estimator) and regional clusters 10 (within countries regions) relaxing in this way the assumption that observations within regions are completely independent. As a measure of happiness (H) we use life satisfaction. The question asked is “All things considered, how satis�ed are you with your life as a whole these days?� Answers are on a ten step scale where ‘1’ is equal to ‘Dissatis�ed’ and ‘10’ is equal to ‘Satis�ed’. This is a standard question in happiness studies, and it is also a question widely tested across the social sciences in terms of reliability of answers. As our dependent variable is categorical and ordered, all equations are estimated with an ordered logit model. Our key variables are proxies for bonding and bridging attitudes and behaviors (B), which we consider as latent variables. This is a recurrent approach in happiness and values studies and also an obligatory choice with the dataset we use. As already dis- cussed in the previous section, there is no recognized single measure of bonding and bridging attitudes and behaviors in the literature, and our dataset does not measure these variables directly. The next section will discuss in detail the procedure we followed to identify the proxies for our bonding and bridging latent variables. We also use a set of controls in all equations (X). These are age (simple and squared) and dummies for the following variables: female; low education and upper education (‘middle education’ is the base category); married (‘married’ and ‘living together as married’); families with no children and families with three or more children (families with 1-2 children is the base category); subjective good health (respondents who replied that health was ‘good’ and ‘very good’) and subjective bad health (respondents who replied that health was ‘poor’ and ‘very poor’. Subjective ‘fair’ health is the base cate- gory); breadwinner; part-time workers, self-employed workers, unemployed and inactive individuals (the base category is ‘employed full-time’); low income people (�rst to third step in a ten step income scale) and high income people (eighth to tenth step in the ten step scale. The base category is middle income people - fourth to seventh step in the same scale). Therefore, when we assessed the role of bonding and bridging in a life satisfaction equation, we controlled for work status and income status in addition to the standard individual characteristics. 4 The nexus between bonding and bridging We started from the assumption that it is possible to characterize bonding and bridging attitudes so as to form two groups of measures. The theoretical guiding distinction 11 was that values and beliefs indicative of the strengthening of existing ties within a well- de�ned community are associated with bonding attitudes, whereas values and beliefs indicative of the exploration of new ties and links with strangers are associated with bridging attitudes. In this section, we report an exploratory analysis conducted to test whether the theoretical distinction between the concepts of bonding and bridging holds with empirical data. More precisely, this exploratory analysis was expected to show: (i) whether the set of measures selected was able to predict two different latent variables, so that two separate groups of measures loaded positively on two different distinguishable factors; or alternatively (ii) whether the two groups of measures emerged as negatively correlated and loaded on the same factor, suggesting that these two groups of measures were in fact two sides of the same latent variable. In other words, we attempted formally to test Putnam’s (2000:23) contention that “bonding and bridging are not ‘either-or’ categories into which social networks can be neatly divided, but ‘more or less’ dimensions along which we can compare different forms of social capital.� To test this hypothesis, we followed a procedure articulated into the four steps now described. Step 1. We were initially guided by theory, the existing literature, and the availabil- ity of variables with a sufficiently large number of observations (not all variables were available in all waves and for all countries in the WVS-EVS) to select a reduced set of proxies for bonding and bridging attitudes from the over 900 variables available in the database. In this way, we identi�ed a �rst set of variables that are typically used in sim- ilar analyses to proxy bridging or bonding. These variables were grouped grouped into four areas: 1) the importance attributed by respondents to a set of items (family, friends, religion, politics, leisure and work); 2) the importance attributed to selected child qual- ities (independence, hard work, responsibility, imagination, tolerance and respect for other people, thrift for saving money and things, determination and perseverance, reli- gious faith, unsel�shness and obedience); 3) trust (trust in people and trust in a set of institutions including churches, armed forces, press, labour unions, police, parliament, civil service, television, government, political parties and major companies); and 4) whether respondents considered as justi�able a set of behaviors (receive social bene�ts, avoid public transport fees, cheat on taxes, accept bribes, homosexuality, prostitution, abortion, divorce, euthanasia and suicide). Compared to the literature reviewed, the 12 only important area that was missing concerned membership in organizations. Accord- ing to several authors, this is an important variable that reveals bonding or bridging behaviors. However, the use of this variable in the WVS would have reduced the usable sample by half and would have severely restricted our possibility to investigate bonding and bridging attitudes across groups of countries, which was a central concern of our analysis. Step 2. In a second stage, we used factor analysis within each of the four groups of variables selected in order to try to reduce the number of variables and assess whether bonding and bridging variables best relate to two separate diemensions or to one dimen- sion only. This allowed us to discard some of the variables with little loadings on the latent variables and also to exclude those variables that simply replicated each other. For example, it emerged rather clearly that attitudes towards homosexuality, prostitution, abortion, divorce, euthanasia and suicide captured the same types of individuals. As a consequence, we retained only one factor that provided a large loading and also a signi�- cant number of positive answers (‘justi�ed homosexuality’). It also emerged quite clearly that, within each group of variables, the variables selected polarized around two latent factors, which we hypothesized to be bonding and bridging attitudes. These variables were importance of family, importance of religion, importance of work, child obedience and trust in institutions for the latent variable we considered as ‘bonding’, and child independence, child imagination, justi�ed homosexuality and trust people for the latent variable we considered as ‘bridging’. These results are largely in line with previous studies (see section two).3 One might consider trust in institution as a bridging rather than bonding variable but the factor analysis we conducted puts this variable together with the other variables that are typi- cally associated with bonding. It could be argued that people who have greater trust in institutions such as the church or the armed forces are typically those people who also put greater emphasis on the institution of marriage and family. However, the point of Step 2 of the analysis was to be driven by the data rather than by ex-ante assumptions and the data suggests to treat trust in institution together with other variables typically associated with bonding. Step 3. Next, we used again factor analysis to address the question of whether bonding and bridging attitudes could be considered as two separate variables or two 3 Results for this stage of the analysis have been omitted but they are available on request. 13 sides of the same variable. Are people less bonding also invariably more bridging (and vice-versa) or can people be more (or less) bonding and bridging at the same time? To address this question we carried out a factor analysis with the variables selected in Step 2. The results are shown in Table 2. It can be seen (top panel) that only the �rst factor has an Eigenvalue sufficiently high to be retained.4 In other words, the analysis suggested that our proxies for bonding and bridging attitudes were jointly related with only one latent variable in some meaningful way. It can also be seen (bottom panel) that the �ve variables that we considered as proxies for bonding attitudes all load positively against the �rst factor while the four variables that we considered as proxies of bridging attitude all load negatively against the same factor.5 This �nding suggests that the two sets of variables that we identi�ed could in fact represent two sides of the same latent variable. This is a new result in contrast with those in the existing literature. For example, Knudsen et al. (2010) �nd, on the basis of US data, that two factors should be retained, and that the proxies for bonding and bridging attitudes correspondingly load on the two factors, thus concluding that bonding and bridging are two separate latent variables (see section 2). [TABLE 2] Step 4. As a further test for the preliminary �nding of step 3, we carried out a third factor analysis exercise (Table 3). This time we assumed that bonding and bridging attitudes were two sides of the same variable and we constructed proxies for a single latent variable by building variables with low scores corresponding to bonding attitudes and high scores corresponding to bridging attitudes. We constructed �ve of these variables using some of the same variables we used in step 3 as well as some of the variables discarded in step 2 but useful for this last exercise. The �rst variable (importance of family vs. importance of friends) was constructed by attributing a value of ‘1’ to those people who equally valued family and friends, a value of ‘2’ to those people who valued family more than friends and a value of ‘3’ to those people who valued friends more than family. We used the same construction procedure for the variables ‘importance of work 4 Different authors tend to use different cut points to decide how many latent factors should be retained. Here we use a cut point of 0.7, which is a rather standard choice, and also look at the distance between factor loadings. 5 Note that in table 2 and 3 we report results for the �rst four factors, even if we retain only the �rst factor. This is to show how none of the other factors provides a clear interpretation relatively to the variables selected in Step 2. 14 vs. importance of leisure’ and ‘importance of religion vs. importance of politics’. In this way, we could use in the factor analysis the value of ‘1’ as a base category and we were able to distinguish well between bonding and bridging extremes. One other variable was constructed by attributing a value of ‘1’ to those people who valued child independence but did not value child obedience. A last variable was constructed giving a value of ‘1’ to those individuals who have great trust in people but not in institutions. In the same factor analysis, we also added the variable ‘justi�ed homosexuality’. We expected people more tolerant of homosexuality to be more bridging and less bonding. In short, all variables were constructed with low values representing ‘bonding’ and high values representing ‘bridging’. More details on the construction of the variables are given in Table 1 in the Annex. The results in Table 3 (top panel) show again that only the �rst factor could be safely retained in the analysis and that the variables constructed are all related to only one latent variable. The table also shows (bottom panel) that all six variables constructed are positively associated with the latent factor. Indeed, low values (bonding) of the constructed variables show negative loadings, and high values (bridging) show positive loadings with respect to the middle values, which represent equal weights attributed to bonding and bridging attitudes. We can also see that the highest loadings are associated with the variables ‘religion vs. politics’, ‘work vs. leisure’ and ‘justi�ed homosexuality’. It follows that the concepts of bonding and bridging can be considered as two sides of the same concept, and that treating these two concepts as one may be useful in empirical applications. These results would suggest coining a new expression for the bonding/bridging di- mension of social capital. We propose the term locus of socializing 6 , which would stress two important aspects: 1) that social capital is a broader concept that includes other 6 Locus of socializing recalls a term used in social psychology, namely the locus of control. The locus of control indicates the locus (place) where people think their decisional sphere is situated. People who believe that everything that happens to them is due to faith or destiny are called ‘externals’, and those who think that everything that happens to them is due to their own behavior are called ‘internals’. Social psychology posits that the locus of control can be measured on one monotonic and increasing scale where the ‘internals are on the right hand side of the scale and the ‘externals are on the left-hand side. Similarly, we posit that the locus of socializing is the locus where people invest most of their socializing time. Bonding types tend to devote most of their time to relations with family, church, work and close kin, while bridging types tend to devote relatively more time to relations outside their close kin. See Rotter (1966) for the concept of the locus of control and Verme (2009) for an analysis of the locus of control in the context of happiness and freedom. 15 dimensions besides bonding and bridging7 , and 2) that people face a trade-off in enjoy- ing social relationships, either within their kinship and existing ties, or across new social ties. [TABLE 3] 5 Bridging, bonding and life satisfaction In this section, we turn to the life satisfaction equations which we use to try determining how bonding and bridging attitudes relate to happiness. We �rst use the pooled world sample with the disaggregated proxies for bonding and bridging attitudes, assuming these variables as two separate concepts (Table 4). We then use the constructed proxies of the locus of socializing. We assume bonding and bridging attitudes as two sides of the same variable and test these proxies on the world sample and across groups of similar countries (Tables 5, summarized in Table 6).8 Table 4 shows that all bonding indicators are positively and signi�cantly correlated with life satisfaction, and with a sizeable coefficient as compared to the standard controls. The most important indicators are ‘importance of family’, ‘importance of religion’, and ‘trust in institutions’. But also ‘importance of work’ and ‘child obedience’ are positively and signi�cantly correlated with life satisfaction. The positive correlations of the bonding indicators used with life satisfaction are not new results with respect to those in the literature (see section 2). However, our previous analysis and the estimations in Table 4 show in more detail how different and interlinked components of bonding can affect life satisfaction. The results for the bridging indicators are more mixed. We �nd ‘trust in people’ to be positive and signi�cant, ‘homosexuality is justi�able’ negative and signi�cant, while ‘child independence’ and ‘child imagination’ are non-signi�cant. Whereas the �nding on trust is not new, the �nding on tolerance of homosexuality is rather surprising. The proportion of homosexuals in the population has been used in some studies as a proxy for bridging attitudes in the composite ‘creativity index’, which, in its turn, has been 7 For example, Woolcock (2001) considers ‘linking social capital’, besides bonding and bridging social capital. 8 The choice of world regions rather than single countries is dictated by the number of observations available given the type of equation. Table 2 in annex provides the full list of countries and number of observations for each world region. 16 found to be a good predictor of economic growth (Florida 2002). In our estimations, people who report that ‘homosexuality is justi�able’ do not appear to be relatively more satis�ed, which suggests that there may be a trade-off between the objectives of life satisfaction and growth. The negative relation between ‘homosexuality is justi�able’ and life satisfaction is also quite consistent across the world regions. The regional equations (Table 5, summarized in Table 6) show that this variable carries a negative and signi�cant coefficient in 10 of the 16 equations, while in �ve equations the variable is non-signi�cant. Only in one region of the world (CIS countries) does ‘homosexuality is justi�able’ carry a positive and (weakly) signi�cant sign. CIS countries include the former Soviet Union countries with the exception of the Baltic countries. We know that attitudes towards homosexuality in these countries are very conservative, and the explanation of this sign may be that people who are more open towards homosexuality are in fact a restricted and self-selected group of happier people. [TABLE 4] Table 5 reports the results for the world and for groups of similar countries using the composite indicators of what we called the locus of socializing. Our purpose here is to understand whether people who are more bonding or more bridging are comparatively happier. The composite indicator ‘importance of family vs. importance of friends’ shows a negative and signi�cant coefficient for both high and low values, as compared to the middle level, where the importance of family is equal to the importance of friends. This means that individuals who attach a great deal of importance or little importance to the family with respect to friends also report a lower satisfaction than those who maintain a more balanced view.9 The coefficient of ‘importance of family vs. importance of friends’ exhibits the same pattern in the great majority of the 16 regions considered in Table 5, while it is non-signi�cant in the remaining regions (see Table 6). Given the cultural and historical diversities of these groups of countries, our results seem rather strong. People with a more balanced view between bonding and bridging attitudes tend to be happier than people polarized around the extremes of the locus of socializing.10 9 Note that people who give an equal weight to family and friends - the base category - could show different degrees of satisfaction. However, the point here is to see whether those who give different weights to family and friends exhibit opposite coefficients in relation to life satisfaction. If, on average, the happiest (or least happy) category was made of people who give equal weight to family and friends, then the other two categories would show the same sign, which is not the case. 10 This �nding is consistent with the research in developmental psychology showing that both hyper- 17 Some studies have found that the ‘importance of family’ indicator may be detrimental to economic growth, possibly through the negative effects on trust in people (see section 2). Joining this result with our result would suggest that too much emphasis on the family may be detrimental to both individuals in their life satisfaction and their trust in people. The composite indicator ‘importance of work vs. importance of leisure’ exhibits a similar pattern to that of ‘importance of family vs. importance of friends’. The coefficients of this variable are negative for both people who attribute a great deal of importance to work relatively to leisure and, vice-versa, for people who attribute a great deal of importance to leisure and little to work. The regional equations summarized in Table 6 con�rm the pattern of the two negative coefficients for all regions, although the coefficients are signi�cant in only 8 of the regions for the bonding ‘types’ and in 6 of the 16 regions for the bridging ‘types’. This result is consistent with the argument that too much emphasis on work, to the point of workaholism, may undermine individual health (Hamermesh and Slemrod 2008). It is also consistent with the �nding that uninteresting jobs seem to lead to less satisfaction on the job (Helliwell and Huang 2005; Clark 2005). People with balanced views between work and leisure seem happier overall. The composite indicator ‘importance of religion vs. importance of politics’ exhibits a different pattern. People who attach a great deal of importance to religion as compared to people with a more balanced view between religion and politics are happier (Table 5). This is con�rmed by the fact that the same result persists in 10 of the 16 regions of the world, whereas in the remaining six regions (mostly former socialist or African countries) the coefficient is non-signi�cant. Instead, people who express a preference for politics as compared to people with a balanced view between religion and politics do not seem to display any particular pattern. In two of the regions, the sign of the coefficient is positive and signi�cant; in two regions it is negative and signi�cant; and in the rest of the regions the coefficient is non-signi�cant. The �nding about religion is very consistent with the literature on happiness and religion (see section 2) while the �nding on people with a preference for politics cannot be clearly interpreted. Perhaps the low number of people in this category explains the difficulty of �nding a consistent and signi�cant sign. The coefficient for the variable ‘child independence vs. child obedience’ is negative protective and dismissing parental child-rearing and education is detrimental to the healthy development of the child and to children’s abilities in social relationships (Ainsworth et al. 1978; Kafetsios and Nezlek 2002). 18 and signi�cant for the pooled world sample. This is con�rmed by the regional equa- tions, where the coefficient of this variable is negative and signi�cant in over half of the regions and non-signi�cant in the remaining half. Attaching more importance to child independence than to child obedience predicts less life satisfaction. Interestingly, Tabellini (2010) has found that attaching more importance to child obedience predicts less economic growth. These two �ndings together are again indicative of a trade-off between the objective of increasing happiness and the objective of boosting economic growth. A further new �nding of Tables 5 and 6 regards the positive and signi�cant coefficient of the variable ‘trust in people vs. trust in institutions’. This result is consistent across the regions of the world, with the coefficient being signi�cant in 7 of the 16 regions and non-signi�cant in the remaining 9 regions. Regions where the coefficient is positive and signi�cant are mainly the former socialist countries of Europe, where we know that trust in institutions used to be at very low levels, especially during the early years of the transition process (which is captured by our data). In substance, individuals with relatively more trust in people than in institutions are happier than those with a more balanced view. But this is the case only in those countries where trust in institutions was very low. Indeed, when the two variables trust in people and trust in institutions are treated separately (Table 4), they both carry a signi�cant and positive sign and also show similar sizes in coefficients and standard errors. [TABLES 5 and 6] 6 Conclusions The paper has reconsidered the concepts of bonding and bridging introduced by Putnam (2000) in order to gain a better understanding of their nature and their association with life satisfaction. An in-depth review of the literature has shown how the concepts of bonding and bridging are considered to be latent variables: they are not directly measured, and they have been empirically studied using a large number of proxies such as trust, attitudes towards social relations, various beliefs and behaviors. In all cases, the proxies for bond- ing and bridging have been distinguished a priori rather than on the basis of evidence so as to describe two separate aspects of the multidimensional concept of social capital. 19 In this paper, we have endeavored to restrict the number of variables that could well depict bonding and bridging attitudes by means of a factor analysis conducted in four steps. As a result of this analysis, we have found that bonding and bridging attitudes are better described as two sides of the same latent variable, rather than as two different latent variables. We have called this new latent variable the locus of socializing to distinguish it from the broader concept of social capital. In fact, we have argued that the locus of socializing can be seen as a speci�c dimension of social capital and that different individuals or population groups can be more bonding or more bridging when they socialize. Based on these �ndings, we have then studied the association between bonding and bridging and the locus of socializing with individual life satisfaction. We �nd that happier people are those who tend to have more balanced attitudes towards family and friends and towards work and leisure, compared with those who focus on only one type of socializing (bonding or bridging). We also �nd that more religious people are invariably happier than people more dedicated to politics; that people with a greater appreciation of child obedience are happier than people with a greater appreciation of child independence; and that people who do not justify homosexuality are happier. These are all indications that bonding attitudes may deliver more happiness than bridging attitudes. Instead, more trust in people, as opposed to more trust in institutions, yields more happiness, at least in some world regions. This we have interpreted as a sign that more bridging attitudes may deliver more happiness. If we compare our results with those on social capital and economic growth reviewed in section 2 of this paper, we �nd that the recipe for happiness does not necessarily coincide with the recipe for economic growth. In particular, while bridging attitudes seem more bene�cial for economic growth than bonding attitudes, a more balanced approach towards bonding and bridging may be conducive to more life satisfaction and, in some cases, bonding may deliver more happiness than bridging. Our �ndings are very preliminary, but they are likely to be crucial for better understanding of such longstanding issues as the “Easterlin paradox�, i.e. the argument that economic growth is not necessarily associated with growth in happiness. References Ainsworth, M.D.S., Blehar, M.C., Waters, E., Wall, S. (1978) Patterns of Attach- ments: a Psychological Study of the Strange Situation. Hillsdale (NJ): Erlbaum 20 Alesina, A., Giuliano, P. (2007) The power of the family, IZA DP, No. 2750 Alesina, A., Giuliano, P. (2009) Family ties and political participation, Harvard Institute of Economic Research. Discussion Paper No. 2171 Alesina, A., La Ferrara, E. (2002) Who trusts others?, Journal of Public Economics 85(2), 207-234 Ban�eld, C.B. (1958) Moral basis of a backward society. New York : Free Press Barro, R.J., McCleary, R.M. (2003) Religion and economic growth, American Soci- ological Review 68(5), 760-781 Bartolini, S., Bilancini, E., Pugno, M. (2008). Did the decline in social capital depress Americans’ happiness?. Quaderni del Dipartimento di Economia Politica, n.540, University of Siena. Becchetti, L., Pelloni A., Rossetti, F. (2008) Relational goods, sociability, and hap- piness, Kyklos 61(3), 343363 Beugelsdijk, S., Smulders, S. (2009) Bonding and bridging social capital and economic growth. Discussion Paper No.27, University of Tilburg Beugelsdijk, S., de Groot, H.L.F., van Schaik, A.B.T.M. (2004) Trust and economic growth: a robustness analysis, Oxford Economic Papers 56(1), 118-134. Bjornskov, C. (2008) Social capital and happiness in the United States. Applied Research Quality Life 3, 43-62 Bjornskov, C. (2006) The multiple facets of social capital. European Journal of Political Economy, 22, 22-40 Bourdieu, P. (1986) The forms of capital. In: John G. Richardson (ed.): Handbook of Theory and Research for the Sociology of Education. New York: Greenwood Press, 241-258. Bruni, L., Stanca, L. (2008) Watching alone, Journal of Economic Behavior and Organization 65(3-4), 506-528 Clark, A. (2005) What makes a good job? Evidence from OECD countries. In Bazen, S., Lucifora, C., Salverda W. (Eds.), Job quality and Employer Behaviour, New York: Palgrave McMillan Clark, A., Lelkes, O. (2009) Let us pray: religious interactions in life satisfaction, Working paper no. 2009/1, Paris School of Economics Costa, D.L., Kahn, M.E. (2003) Understanding the decline in social capital, 1952- 1998. Kyklos 56:17.46. 21 Dasgupta, P. (2005) Economics of social capital, The Economic Record 81(255), 2-21 Dearmon, J., Grier, K. (2009) Trust and development, Journal of Economic Behavior and Organization 71(2), 210-220 Dolan, P., Peasgood, T. and White M. (2008) Do we really know what makes us happy? A review of the economic literature on the factors associated with subjective well-being. Journal of Economic Psychology, 29, 94-122 Durlauf, S.N., Fafchamps, N. (2005) Social capital. In:Aghion, A., Durlauf, S.N. (Eds) Handbook of Economic Growth. Amsterdam: North-Holland, 1639-99 Easterlin, R.A., Angelescu, L. (2009). Happiness and growth the world over: time series evidence on the happiness-income paradox. IZA Discussion Paper No. 4060, March. Ermish, J., Gambetta, D. (2008) Do strong family ties inhibit trust?, Working Paper, no.37, Institute for Social and Economic Research, Nov. Fehr, E. (2008) On the economics and biology of trust. IZA Discussion Paper no.3895 Fukuyama, F. (1997) Social capital and the modern capitalist economy, Stern Busi- ness Magazine 4(1) Gittell, R., A. Vidal (1998) Community Organizing: Building Social Capital as a Development Strategy. Thousand Oaks, CA: Sage Books Glaeser, E.L., Laibson, D., Sacerdote, B. (2002) An economic approach to social capital. Economic Journal 112, F437-58 Glaeser, E.L., Laibson, D., Sacerdote, B., Soutter, C. (2000), Measuring Trust, Quar- terly Journal of Economics 115(3), 811-46 Guiso, L., Sapienza, P., Zingales, L. (2003) Peoples opium? Religion and economic attitudes, Journal of Monetary Economics 50, 225282 Guiso, L., Sapienza, P., Zingales, L. (2006) Does culture affect economic outcomes?, Journal of Economic Perspectives 20(2), 23-48 Guiso, L., Sapienza, P., Zingales, L. (2010) Civic capital as the missing link, Prepared for the Social Economics Handbook, edited by J. Benhabib, A.Bisin, M.O. Jackson Hamermesh, D.S., Slemrod, J.B. (2008) The economics of workaholism, The B.E. Journal of Economic Analysis and Policy 8(1) (Contributions), Article 3 Helliwell, J.F. (2008) Life satisfaction and quality of development, NBER Working Paper No.14507. Helliwell, J.F. (2006) Well-being, social capital and public policy: what’s new? Eco- 22 nomic Journal 116, C34-45 Helliwell, J.F., Putnam R. (1995) Economic growth and social capital in Italy, East- ern Economic Journal 21, 295-307. Helliwell, J.F., F., Barrington-Leigh, C.P., Harris, A., Huang, H. (2009). Interna- tional evidence on the social context of well-being, NBER Working Paper No.14720. Helliwell J.F., Wang, S. (2010) Trust and well-being, NBER working paper, No. 15911 Kafetsios, K., Nezlek, J.B. (2002) Attachment styles in everyday social interaction, European Journal of Social Psychology 32(5), 719-35 Kahneman, D., Krueger, A.B. (2006) Developments in the measurement of subjective well-being, Journal of Economic Perspectives 20(1), 3-24 Knack, S., Keefer, P. (1997) Does social capital have an economic pay-off?. Quarterly Journal of Economics 112(4), 1251-88 Knudsen, B., Florida, R., Rousseau, D. (2010) Bridging and bonding: a multi- dimensional approach to regional social capital, University of Toronto (mimeo) available at: http://www.creativeclass.com Konow, J., Earley, J. (2008) The hedonistic paradox, Journal of Public Economics 92(1-2), 1-33 Larsen, L., Harlan, S.L., Bolin, B., Hackett, E.J., Hope, D., Kirby, A., Nelson, A., Rex, T.R., Wolf, S. (2004) Bonding and bridging. Understanding the relationship between social capital and civic action, Journal of Planning Education and Research, 24, 64-77 Layard, R. (2005). Happiness: Lessons from a New Science. New York: Penguin Press Layard, R., Mayraz, G., Nickell, S. (2009). Does relative income matter? Are the critics right?. CEP Discussion Paper No. 918, March (in Diener, E., Helliwell, J.F., Kah- neman, K. (Eds) International Differences in Well-Being, New York: Oxford University Press, forthcoming) Leung, A., Kier, C., Fung, T., Fung, L. and R. Sproule (2011), Searching for happi- ness: the importance of social capital, Journal of Happiness Studies, 12(3), 443-462 Loury, G. C. (1977) A dynamic theory of racial income differences. In: Wallace, P., LaMont, A. (Eds) Women, Minorities and Employment Discrimination,, Lexington: Lexington Books. 23 Meier, S. and Stutzer A. (2008) Is volunteering rewarding in itself?, Economica 75(297), 39-59 Nickerson, C., Schwartz, N., Diener, E., Kahneman, D. (2003) Zeroing in on the dark side of the American dream. Psychological Science 14(6), 531–6. Olson, M, (1982) The rise and decline of nations. New Haven, CT : Yale University Press Paldam, M., Svendsen, G.T. (2000) An essay on social capital. European Journal of Political Economy 16:339-66 Putnam, R.D. (1993) Making Democracy Work. Princeton NJ: Princeton University Press Putnam, R.D. (1995) Bowling Alone, Journal of Democracy, 6(1),65-78 Putnam, R.D. (2000) Bowling Alone. New York: Simon and Schuster Ram R (2010) Social capital and happiness, Journal of Happiness Studies, 11(4), 409-418 Robinson. R.V., Jackson, E.F. (2001) Is trust in others declining in America? Social Science Research 30:117-45 Rotter, J. B. (1966). Generalised expectancies of internal versus external control of reinforcements. Psychological monographs 80 (609) Stevenson, B., Wolfers, J. (2008). Economic growth and happiness. Brookings Pa- pers on Economic Activity, May Tabellini, G. (2010) Culture and institutions: economic development in the regions of Europe, Journal of the European Economic Association, 8(4), 677-716 Veenhoven, R. (2007) Subjective measures of well-being, in: McGillivray, M. (Ed.) Human Well-being: Concept and Measurement, Palgrave /McMillan, Houndmills, New Hampshire, USA, 214-239 Verme, P. (2009) Happiness, Freedom and Control. Journal of Economic Behavior and Organization. Vol. 71, Issue 2, August 2009, pages 146-161. Wallis, J., Killerby, P. Dollery, B. (2004) Social economics and social capital, Inter- national Journal of Social Economics 31(3), 239-258 Woolcock, M. (2001) The place of social capital in understanding social and economic outcomes. In J.F. Helliwell (ed.), The Contribution of Human and Social Capital to Sustained Economic Growth and Well-being: International Symposium Report, Human Resources Development Canada and OECD. 24 Woolcock, M., Narayan, D. (2000) Social capital: Implications for development the- ory, research, and policy, World Bank Research Observer 15(2), 223-49 Wilson, C.M., Oswald, A.J. (2005) How does marriage affect physical and psycho- logical health?, IZA DP No. 1619 Ram R (2010) Social capital and happiness, Journal of Happiness Studies, 11(4), 409-418 Winkelmann R (2009) Unemployment, social capital, and subjective well-being, Jour- nal of Happiness Studies, 10(4), 421-430 Zak, P.J., Knack, S. (2001) Trust and Growth. Economic Journal 111:295-331. 25 Table 1: Summary Table of the Literature on Bonding and Bridging Social Capital Study Indicators of social capital Outcomes (Indicators are identified as bonding or bridging only when they are labelled as such in the original articles, when they are clearly defined in theory, or when they are thus found in the present article.) - composite measure of: availability of sports and cultural associations; newspaper Positive associations with regional per Helliwell and Putnam (1995) readership; turnout in referenda; incidence of preference voting capita output growth in Italy - composite measure of institutional performance - citizen satisfaction with government - trust in people Positive association with cross- - sum of the scores from questions on justifiability of: national per capita output growth cheating on taxes; avoiding a fare; claiming government benefits; keeping found money; Knack & Keefer (1997) failing to report damage to a parked vehicle - Olsonian group membership Non significant association with cross- - Putnamian group membership national per capita output growth - religious beliefs Positive association with cross- national per capita output growth Barro and McCleary (2003) - church attendance Negative association with cross- national per capita output growth - factor analysis identifies exclusive groups, faith-based engagement Negative association with cross- Knudsen et al. (2010) regional per capita income growth in the US - factor analysis identifies diversity of friendships, political activity, inclusive groups Positive association with cross- regional per capita income growth in the US - factor analysis identifies a one-dimension measure of the importance of family, friends Negative association with cross- Beugelsdijk & Smulders and acquaintances regional per capita income growth in (2009) Europe Positive association with cross- - membership of selected non-rent-seeking groups regional per capita income growth in Europe Guiso et al. - the first principal component positively correlates with claiming government benefits when Non-significant association with cross- (2010) not entitled, cheating on taxes, accepting a bribe in the course of own duties national per capita output growth - importance placed on child’s obedience Negative association with cross- (also identified in the first principal component) regional per capita output growth in Tabellini (2008) Europe - trust in people Positive associations with cross- - importance placed on child’s respect for other people regional per capita output growth in - people’s freedom and control over their lives Europe (also identified in the first principal component) Negative association with trust in Alesina & Giuliano (2009, - the first principal component of family importance, duty to love parents, parents’ people and with political participation 2007) altruism Positive association with individuals’ life satisfaction - importance of religion - having someone to rely on Helliwell et al. (2009), - attending church Helliwell (2008, 2006) Positive association with cross- national individuals’ life satisfaction - proxies for governmental quality - donated money to an organisation - donated time to an organisation - helped a stranger - trust in people - trust in police - group membership - trust in institutions - contacts with relatives Positive association with individuals’ Bartolini et al. (2008) - trust in people happiness over time in the US - contacts with neighbours and friends - Putnamian group membership Negative association with individuals’ - Olsonian group membership happiness over time in the US Meier and Stutzer (2008) Positive effect on individuals’ - frequency in volunteer work happiness over time in Germany - attending social gatherings Positive association with individuals’ Becchetti et al. (2008) - attending cultural events life satisfaction over time in Germany - performing volunteer work - participating in sports - attending religious events - volunteering in union organisations Non-significant association with cross- - volunteering in political organisations national individuals’ life satisfaction Bruni and Stanca, (2008) - volunteering in professional organisations - volunteering in environment organisations - volunteering in church organisations Positive association with cross- - volunteering in sport organisations national individuals’ life satisfaction - volunteering in art organisations - volunteering in charity organisations Guiso et al. (2003) - raised religiously at home Positive association with cross- - attending religious services national individuals’ trust - believing in God 26 Table 2: Factor Analysis - Bonding and Bridging (Step 3) Factor Eigenvalue Difference Proportion Cumulative Factor1 0.87608 0.61096 1.3965 1.3965 Factor2 0.26512 0.14275 0.4226 1.8192 Factor3 0.12237 0.11105 0.1951 2.0142 Factor4 0.01132 0.05488 0.018 2.0323 Factor5 -0.04355 0.08164 -0.0694 1.9628 Factor6 -0.1252 0.01579 -0.1996 1.7633 Factor7 -0.14099 0.01337 -0.2247 1.5385 Factor8 -0.15435 0.02913 -0.2461 1.2925 Factor9 -0.18348 . -0.2925 1 Variable Factor1 Factor2 Factor3 Factor4 Uniqueness importance of family 0.2327 0.2274 -0.0894 0.0413 0.8844 importance of religion 0.483 0.136 0.0282 -0.013 0.7473 importance of work 0.3344 0.2072 -0.1157 0.0114 0.8317 child obedience 0.3375 -0.1833 0.0527 0.0459 0.8476 trust in institutions 0.2145 0.1159 0.2269 -0.0076 0.889 importance of child independence -0.2882 0.2334 -0.0378 -0.0441 0.8591 importance of child imagination -0.235 0.1675 0.0311 0.0189 0.9154 justified homosexuality -0.3776 0.0456 -0.0389 0.0653 0.8496 trust in people -0.1897 0.1427 0.2049 0.0242 0.9011 Observations: 251718; Correlation method: Principal factors rotation; Number of parameters: 30 27 Table 3: Factor Analysis - Bonding and Bridging (Step 4) Factor Eigenvalue Difference Proportion Cumulative Factor1 1.10566 0.54352 0.965 0.965 Factor2 0.56214 0.43651 0.4906 1.4556 Factor3 0.12563 0.0482 0.1096 1.5652 Factor4 0.07743 0.12878 0.0676 1.6328 Factor5 -0.05135 0.03205 -0.0448 1.588 Factor6 -0.0834 0.00515 -0.0728 1.5152 Factor7 -0.08855 0.14675 -0.0773 1.4379 Factor8 -0.2353 0.03116 -0.2054 1.2326 Factor9 -0.26646 . -0.2326 1 Variable Factor1 Factor2 Factor3 Factor4 Uniqueness Importance of family > importance of friends (a) -0.245 0.2356 0.2034 0.0231 0.8426 Importance of family < importance of friends (a) 0.1951 -0.13 -0.2386 -0.0161 0.8879 Importance of work > importance of leisure (b) -0.3329 0.4101 -0.1049 0.0433 0.7082 Importance of work < importance of leisure (b) 0.3181 -0.3733 0.1216 -0.0285 0.7438 Importance of religion > importance of politics (c) -0.5551 -0.2788 -0.0124 0.0759 0.6083 Importance of religion < importance of politics (c) 0.5239 0.3167 0.0197 -0.0746 0.6192 Child independence Vs. Child obedience 0.2769 0.0645 0.0005 0.1329 0.9015 Trust in people Vs. trust in institutions 0.2087 0.0065 -0.0153 0.1526 0.9329 Justified homosexuality 0.3046 -0.0051 0.0283 0.1471 0.8848 Observations: 256629; Correlation method: Principal factors rotation; Number of parameters: 30 (a) Base category: Importance of family = importance of friends (b) Base category: Importance of work = importance of leisure (c) Base category: Importance of religion = importance of politics 28 Table 4: Life Satisfaction Equation - Pooled World Sample lifesat Bonding Importance of family 0.256 (13.87)** Importance of religion 0.229 (16.75)** Importance of work 0.04 (3.68)** Important child qualities: obedience 0.071 (5.73)** Trust in institutions 0.215 (16.09)** Bridging Important child qualities: independence -0.005 -0.46 Important child qualities: imagination -0.02 -1.7 Trust in people 0.206 (14.48)** Justified homosexuality -0.106 (7.57)** Controls female 0.057 (5.65)** age -0.041 (18.42)** age2 0 (20.55)** lower education -0.088 (4.01)** upper education 0.098 (6.50)** married 0.307 (24.18)** no children 0.112 (8.03)** children three or more 0.032 (2.33)* good health 0.591 (36.57)** bad health -0.77 (31.29)** breadwinner -0.021 -1.78 employment status - part-time -0.045 (2.13)* employment status - self-employed 0.03 -1.56 employment status - unemployed -0.413 (16.13)** economically inactive 0.029 (2.22)* low income group -0.368 (20.32)** high income group 0.293 (9.68)** Observations 240811 Robust z-statistics in parentheses. * significant at 5% level; ** significant at 1% level Countries and years dummies omitted. 29 Table 5: Life Satisfaction Equations - Pooled World Sample and World Regions (Cont.) 1 2 3 4 5 6 7 8 Central- North Central South North Central South Eastern World America America America Europe Europe Europe Balcans Europe Importance of family > importance of friends (a) -0.138 -0.293 -0.157 -0.163 -0.232 -0.183 -0.142 -0.157 -0.137 (13.35)** (8.22)** (2.48)* (5.43)** (9.63)** (6.35)** (4.97)** (4.09)** (4.85)** Importance of family < importance of friends (a) -0.283 -0.496 -0.266 -0.206 -0.343 -0.421 -0.346 -0.615 -0.224 (13.26)** (3.25)** (2.04)* (3.39)** (7.02)** (5.46)** (4.62)** (3.28)** (3.62)** Importance of work > importance of leisure (b) -0.1 -0.084 -0.102 -0.133 -0.093 -0.134 -0.107 -0.05 -0.038 (8.97)** -1.25 (2.04)* (5.25)** (3.02)** (4.02)** (2.76)** -0.69 -1.2 Importance of work < importance of leisure (b) -0.07 -0.046 -0.251 -0.109 -0.109 -0.194 -0.116 -0.019 -0.047 (5.50)** -1.19 (2.88)** (3.12)** (3.30)** (4.40)** (2.60)** -0.27 -1.44 Importance of religion > importance of politics (c) 0.114 0.09 0.119 0.144 0.197 0.134 0.167 0.161 0.089 (10.51)** (2.21)* (2.46)* (4.80)** (4.78)** (3.31)** (5.75)** (5.24)** (2.77)** Importance of religion < importance of politics (c) -0.027 -0.074 -0.096 -0.112 -0.001 0.008 -0.005 0.187 -0.035 (1.97)* -1.89 -1.05 (3.17)** -0.04 -0.17 -0.12 (3.12)** -1.08 Child independence Vs. Child obedience -0.06 -0.109 -0.113 -0.179 -0.057 -0.051 -0.09 0.06 0.021 (5.44)** (2.93)** (2.67)** (4.94)** (2.32)* -1.31 (2.75)** -0.98 -0.76 Trust in people Vs. trust in institutions 0.07 0.027 -0.002 -0.038 -0.01 0.198 0.138 0.012 0.229 (4.99)** -0.47 -0.03 -0.77 -0.44 (4.52)** (3.06)** -0.19 (6.14)** Justified homosexuality -0.127 -0.171 -0.221 -0.25 -0.285 -0.105 -0.208 -0.132 -0.035 (8.93)** (4.17)** (5.62)** (7.44)** (8.79)** -1.8 (4.92)** (2.93)** -1.34 female 0.069 0.126 0.104 -0.037 0.164 0.03 0 0.139 0.098 (6.70)** (2.71)** (2.04)* -1.63 (5.40)** -0.83 -0.01 (2.79)** (3.93)** age -0.041 -0.03 -0.051 -0.016 -0.053 -0.036 -0.045 -0.063 -0.059 (17.95)** (4.15)** (4.58)** (2.56)* (8.80)** (5.31)** (7.36)** (6.05)** (12.52)** age2 0 0 0.001 0 0.001 0 0 0.001 0.001 (20.46)** (5.74)** (5.30)** (3.51)** (9.32)** (6.23)** (6.71)** (6.16)** (11.46)** lower education -0.09 0.114 0.112 0.105 -0.073 -0.227 -0.184 -0.22 -0.24 (4.26)** -1.29 -1.87 (2.28)* (2.02)* (4.42)** (3.70)** (3.74)** (5.11)** upper education 0.103 0.051 -0.071 0.065 0.025 0.075 0.131 0.128 0.32 (7.06)** -0.78 (2.12)* -1.59 -0.8 -1.53 (2.76)** (3.45)** (9.49)** married 0.333 0.6 0.297 0.274 0.555 0.422 0.55 0.356 0.365 (25.54)** (12.53)** (11.28)** (9.11)** (19.37)** (8.96)** (14.35)** (6.79)** (9.57)** no children 0.098 0.159 0.076 0.162 -0.001 0.087 0.096 0.113 0.161 (7.12)** (2.83)** -1.43 (4.07)** -0.03 (2.36)* (2.65)** (2.20)* (3.59)** children three or more 0.048 0.184 0.011 0.032 0.058 0.012 -0.024 -0.007 0.031 (3.57)** (5.16)** -0.19 -1.02 -1.94 -0.23 -0.79 -0.11 -1.07 good health 0.604 0.921 0.656 0.476 0.897 0.833 0.638 0.711 0.551 (36.87)** (18.87)** (24.06)** (12.45)** (22.42)** (14.12)** (13.98)** (12.58)** (18.27)** bad health -0.772 -0.84 -0.497 -0.693 -0.999 -0.991 -0.95 -0.714 -0.668 (31.64)** (6.32)** (5.10)** (9.51)** (11.73)** (17.91)** (9.38)** (9.24)** (14.83)** breadwinner -0.019 0.024 0.022 -0.02 -0.035 -0.108 0.046 0.011 -0.007 -1.64 -0.57 -0.52 -0.61 -1.02 (2.13)* -1.29 -0.22 -0.22 employment status - part-time employee (d) -0.044 0.039 -0.142 0.047 -0.065 -0.04 -0.168 -0.092 0.088 (2.08)* -0.71 (2.07)* -1.28 (2.09)* -0.6 (2.29)* -0.66 -1.21 employment status - self-employed (d) 0.026 0.063 -0.018 -0.021 0.021 -0.05 0.143 0.153 0.074 -1.37 -0.85 -0.26 -0.38 -0.37 -0.73 (2.45)* (2.03)* -1.02 employment status - unemployed (d) -0.412 -0.34 -0.146 -0.319 -0.766 -1.13 -0.639 -0.259 -0.638 (16.50)** (2.50)* -1.49 (6.29)** (12.68)** (11.61)** (9.67)** (4.13)** (11.65)** economically inactive 0.033 0.195 0.013 0.018 -0.022 0.097 -0.018 0.064 0.062 (2.61)** (3.72)** -0.21 -0.36 -0.44 (2.09)* -0.44 -1.47 -1.82 low income group -0.365 -0.254 -0.188 -0.212 -0.234 -0.24 -0.178 -0.574 -0.402 (20.29)** (3.52)** (3.35)** (5.48)** (6.25)** (5.10)** (3.70)** (7.09)** (9.16)** high income group 0.296 0.259 0.215 0.243 0.129 0.08 0.089 0.543 0.391 (9.96)** (3.60)** (2.10)* (5.22)** (4.33)** -1.67 -1.79 (5.81)** (8.16)** Observations 240811 10443 11039 23527 21678 17101 21177 10561 24749 Robust z-statistics in parentheses. * significant at 5% level; ** significant at 1% level. Countries and years dummies omitted. (a) Base category: Importance of family = importance of friends (b) Base category: Importance of work = importance of leisure (c) Base category: Importance of religion = importance of politics (d) Base category: employment status - full-time employee 30 (Table 5 - Cont.) 9 10 11 12 13 14 15 16 Common wealth of North- Independe Africa and Sub- nt States Middle- Saharian South China and Baltics (FSU 12) East Africa East Asia Asia Vietnam Oceania Importance of family > importance of friends (a) 0.072 -0.106 -0.023 -0.101 -0.113 -0.135 -0.009 -0.338 -1.49 (3.48)** -0.4 (3.10)** (2.33)* (2.27)* -0.17 (6.42)** Importance of family < importance of friends (a) -0.076 -0.246 -0.264 -0.168 -0.574 -0.234 -0.349 -0.577 -0.95 (4.03)** (2.07)* -1.53 (5.23)** (2.11)* (2.38)* (3.60)** Importance of work > importance of leisure (b) -0.112 0.009 -0.098 -0.131 -0.06 -0.119 0.016 -0.126 (3.40)** -0.19 (2.82)** (2.90)** -0.92 -1.56 -0.29 -1.23 Importance of work < importance of leisure (b) -0.129 -0.004 -0.072 -0.056 0.051 -0.054 0.061 -0.255 -1.39 -0.09 -1.21 -0.7 -1.17 -0.43 -0.55 (4.26)** Importance of religion > importance of politics (c) 0.08 0.061 0.152 0.037 0.259 -0.035 0.015 0.064 -1.49 -1.69 (3.11)** -0.98 (4.67)** -0.62 -0.18 -0.77 Importance of religion < importance of politics (c) 0.018 -0.026 -0.022 -0.249 0.03 -0.27 0.229 -0.074 -0.41 -0.72 -0.26 (3.47)** -0.5 -1.66 (3.51)** -0.86 Child independence Vs. Child obedience -0.103 0.011 -0.101 -0.144 -0.021 -0.138 0.04 -0.038 (2.58)** -0.23 (2.89)** (2.71)** -0.36 (2.01)* -0.46 -0.73 Trust in people Vs. trust in institutions 0.234 0.102 -0.105 0 0.115 0.068 -0.024 0.123 (3.94)** (2.11)* (1.98)* 0 -1.13 -0.8 -0.27 (2.34)* Justified homosexuality -0.031 0.092 -0.161 -0.017 -0.301 -0.05 -0.181 -0.285 -0.5 (2.30)* (1.96)* -0.31 (4.11)** -0.64 (2.38)* (4.30)** female 0.078 -0.055 0.235 0.079 0.208 0.153 0.12 0.149 -1.52 -1.14 (5.83)** (2.68)** (3.22)** (3.87)** (2.50)* (2.17)* age -0.102 -0.048 -0.034 -0.031 -0.043 -0.014 -0.022 -0.066 (5.94)** (5.06)** (3.71)** (4.88)** (3.57)** -1.31 (2.54)* (3.02)** age2 0.001 0 0 0 0.001 0 0 0.001 (6.47)** (5.18)** (3.97)** (7.05)** (4.09)** (2.13)* (3.10)** (4.24)** lower education -0.255 -0.082 0.004 -0.154 -0.103 -0.209 -0.099 -0.107 (3.43)** -1.39 -0.09 (3.34)** -1.01 (3.16)** -1.33 -1.62 upper education 0.311 0.182 0.104 0.064 0.175 0.132 0.241 -0.063 (3.95)** (3.83)** -1.81 -1.31 (3.61)** (2.79)** (3.71)** -1.11 married 0.204 0.336 0.252 0.19 0.458 0.074 0.402 0.669 (4.57)** (5.30)** (3.50)** (4.29)** (6.88)** -1.22 (4.41)** (8.30)** no children 0.001 0.218 0.025 0.15 0.029 0 0.07 -0.035 -0.02 (2.26)* -0.47 (3.01)** -0.37 -0.01 -0.88 -0.45 children three or more 0 0.157 -0.013 -0.05 -0.018 0.003 0.152 0.146 -0.01 (3.02)** -0.22 -1.28 -0.32 -0.04 (2.24)* (2.05)* good health 0.644 0.524 0.47 0.699 0.763 0.576 0.668 1.186 (11.24)** (12.58)** (5.03)** (11.70)** (12.58)** (7.49)** (10.81)** (10.59)** bad health -0.776 -0.666 -0.564 -0.817 -0.792 -0.601 -0.524 -1.139 (10.23)** (14.27)** (6.24)** (11.67)** (7.31)** (5.27)** (4.96)** (10.62)** breadwinner -0.043 -0.063 -0.063 -0.045 0.059 0.041 -0.115 0.003 -0.69 -1.66 -1.21 -1.13 -1.08 -0.51 (2.22)* -0.02 employment status - part-time employee (d) 0.149 0.012 -0.101 -0.192 -0.124 -0.194 -0.013 -0.118 (2.40)* -0.18 -1.45 (2.42)* -1.59 -1.91 -0.11 -1.16 employment status - self-employed (d) 0.216 0.255 -0.062 -0.087 -0.073 0.069 0.044 -0.06 (2.83)** (3.30)** -1.01 -1.51 -1.1 -1.16 -0.47 -0.34 employment status - unemployed (d) -0.639 -0.272 -0.393 -0.367 -0.433 -0.092 -0.629 -0.406 (5.43)** (4.25)** (4.70)** (6.38)** (4.40)** -1.31 (2.95)** -1.92 economically inactive -0.009 0.101 -0.081 -0.057 0.017 -0.028 0.003 0.029 -0.16 (2.09)* (2.03)* -1.13 -0.26 -0.46 -0.03 -0.23 low income group -0.404 -0.464 -0.288 -0.579 -0.426 -0.425 -0.611 -0.222 (5.91)** (5.75)** (5.11)** (9.76)** (4.55)** (5.32)** (6.86)** (2.61)** high income group 0.395 0.454 0.214 0.609 0.629 0.647 0.624 0.244 (6.48)** (7.66)** -1.83 (8.11)** (6.78)** (7.41)** (5.93)** (3.37)** Observations 7417 21203 13679 24543 9124 13615 6908 4047 Robust z-statistics in parentheses. * significant at 5% level; ** significant at 1% level. Countries and years dummies omitted. (a) Base category: Importance of family = importance of friends (b) Base category: Importance of work = importance of leisure (c) Base category: Importance of religion = importance of politics (d) Base category: employment status - full-time employee 31 Table 6: Summary of Life Satisfaction Equations - Pooled World Sample and World Regions World Regions + sig. - sig. non sig. tot. Importance of family > importance of friends (a) -0.138 0 13 3 16 (13.35)** Importance of family < importance of friends (a) -0.283 0 14 2 16 (13.26)** Importance of work > importance of leisure (b) -0.1 0 8 8 16 (8.97)** Importance of work < importance of leisure (b) -0.07 0 6 10 16 (5.50)** Importance of religion > importance of politics (c) 0.114 10 0 6 16 (10.51)** Importance of religion < importance of politics (c) -0.027 2 2 12 16 (1.97)* Child independence Vs. Child obedience -0.06 0 9 7 16 (5.44)** Trust in people Vs. trust in institutions 0.07 7 0 9 16 (4.99)** Justified homosexuality -0.127 1 10 5 16 (8.93)** 32 ANNEX Table 1: Construction of key variables No. Variable WVS Code Construction 1 A001=1 1 Importance of family A001 0 A001>1 and A001<5 1 A002=1 2 Importance of friends A002 0 A002>1 and A002<5 1 A003=1 3 Importance of leisure A003 0 A003>1 and A003<5 1 A004=1 4 Importance of politics A004 0 A004>1 and A004<5 1 A005=1 5 Importance of work A005 0 A005>1 and A005<5 1 A006=1 6 Importance of religion A006 0 A006>1 and A006<5 1 Important 7 Important child qualities: independence A029 0 Not mentioned 1 Important 8 Important child qualities: obedience A042 0 Not mentioned 1 Important 9 Important child qualities: imagination A034 0 Not mentioned 1 if A001=A002 10 Importance of family vs. importance of friends A001, A002 2 if A001A002 1 if A005=A003 11 Importance of work vs. importance of leisure A003, A005 2 if A005A003 1 if A006=A004 12 Importance of religion vs. importance of work A004, A006 2 if A006A004 1 if A029=1 and A042=0 13 Child independence vs. child obedience A029, A042 0 Otherwise 1 Most people can be trusted 14 Trust in people A165 0 Can't be too careful (originally=2) 1 0