The ISSP 2017 Survey on Social Networks and Social Resources: An Overview of Country-Level Results

Abstract This contribution complements the special issue on the 2017 ISSP survey on social networks and social resources by presenting a comprehensive overview of its results at the country level. Our analysis comprises a summary of the respondents’ views on social trust and cohesion, the size and diversity their social network, their subjective health and life satisfaction, as well as the frequency of their contacts and their feeling of social isolation. When interpreting our results, we refer to the views of Putnam, Uslaner, and Larsen regarding the nature of social trust, its sources, and its consequences. In this regard, our findings show, for example, that country levels of social trust are line with prior findings—with Nordic societies such as Denmark, Iceland, Finland, and Sweden confirming their high-trust status. We, however, also observe an inverse relationship between objective inequality and the perception thereof in central-east European post-communist countries and China.


Introduction
Social cohesion is one of the core issues of the contemporary social sciences. It can be seen as a recurring dilemma between diversity and solidarity, between pluralism and the search for a common identity, as well as a dynamic between individualism and collectivism. The processes of change such as the "threatened social middle" (Koppetsch 2013) and responses to "external shocks" such as migration and the global financial crisis (Streeck 2017) appear to be the key challenges for social cohesion-especially as far as the support for the political and social welfare system in the western countries and the younger Eastern European democracies is concerned (Larsen 2013: 3-18).
The study of social trust and its origins, types, and effects allows for deeper insights in social cohesion and related processes of diversity and solidarity as well as an understanding of inequality, polarization, segregation, and corruption. Research considers trust a social glue for communities, a "chicken soup of social life" that brings cohesion and other positive social effects (Uslaner 2002: 1). It is complex and sometimes mysterious and can be considered an individual assessment of the moral standards of a society (Delhey and Newton 2004: 4). Trust is also understood as an informal social institution and a source of social solidarity (North 1998in Rothstein 2013: 1011 as well as an indicator of social cohesion and a fundamental economic condition of capitalist markets. Trust is also a fundamental ingredient of democratic pluralism and an ontological basis of our social life and action (Hawley 2012).
The International Social Survey Program (www.ISSP.org) data on social networks and resources is an ideal source for studying social cohesion and related questions. It includes a total of 60 items on the nature and size of the social network of the respondents, the perceived social trust, and other factors that are important for social cohesion. In this vein, our contribution offers a first overview of country-level results and brief interpretations using the lens of social capital, trust, and cohesion research as expressed in the works of Robert Putnam, Erik Uslaner, and Christian Larsen. This introduction continues with a brief overview of the theoretical background and our research questions. We then briefly describe the 2017 ISSP data and subsequently present results on the diversity and size of the social network across countries, indicators of social trust, and selected indicators on health care and subjective health. The final section concludes that most of our results are in line with previous findings but also points to a need for longitudinal research.

Main concepts of social cohesion and social trust
Considering the vast literature on differences, trends, and changes in social capital, networks, cohesion, and trust, we concentrate on the works and prevailing hypotheses of Robert Putnam, Eric Uslaner, and Christian Larsen. Our focus on these three authors is also influenced by Andrew Abbott's (2004: 8-39) observation that perspectives of these researchers invite further critical and empirical investigation given that their ideas are also partially contradictory.
Putnam's social capital research and the relevance of the "civicness" of community Robert Putnam defines the central notion of social capital as "social networks and the norms of reciprocity and trustworthiness that arise from them" (Putnam 2000: 19). Socioeconomic inequality, trust, and corruption play a supporting role in his narrative of changes in social capital. He locates and explores social trust mostly in the context of civic community culture and governmental performance (Putnam 1993), social capital and the indicators of its decline (Putnam 2000), and-probably even more publicly accessible-the erosion of American community life and the significance of opportunity gaps for the coming generations (Putnam 2015).
The different forms of civic participation, community networks, and general trust in fellow citizens represent the relevant research dimensions, when analyzing the differences in social capital between countries or social groups. Loosely following Putnam's explanatory emphasis and his impressive mix of various and highly diverse measures of reciprocity, trust, civility, honesty, and social capital 1 (Putnam 2000: 142-147), we can address the following themes in the context of the available ISSP data: The structure and the extent of social networks; the presence and differences in social trust; the distinction between social trust and the perception of social fairness; different forms of political, civic, and religious participation; and the link among civic participation, volunteering, and social trust.

Uslaner and his concept of moralistic trust
Eric Uslaner (2002Uslaner ( , 2012 offers an alternative view on social capital and civic networkswith trust, and not civic engagement, as the core social phenomenon and the moralistic foundation of social cohesion. While Putnam emphasizes the difference between bonding as the generator of an in-group social capital and bridging as an out-group social capital mechanism in stabilizing cohesion in a larger society and highly diverse communities, Uslaner stresses the conceptual distinction between strategic, experience-based, reciprocal, and particularized forms of trust. In addition, he also emphasizes the more general, valuebased, and moralistic elements. Uslaner's emphasis on trust as faith in strangers-following the socio-psychological research tradition (Rosenberg 1956, Allport 1979)-seems to oppose the more rationalist and economic views of trust, which consider it a proxy for cooperation, strategic evaluation processes, and underlying reciprocal social transactions.
Uslaner's concept of trust comprises strong ethical beliefs and egalitarian ideals, intrinsic to democratic cultures and civic traditions, and a sense of control over one's life and the optimism that things can get better: The conventional wisdom is that we trust other people because we know a lot about them. Instead, I argue that we can and do trust strangers. Indeed, the 'standard' trust question (most people can be trusted) really is about trusting people we don't know. There are different types of trust. Putting faith in strangers is moralistic trust. Having confidence in people you know is strategic trust. The latter depends on our experiences, the former does not. (Uslaner 2002: 4) Given that social or moralistic trust is considered based on an egalitarian culture, the differences in trust among countries should, to a certain extent, coincide with the level of socioeconomic inequality. The level of trust thus should be higher in more equal societies. Vice versa, large or rising income inequality could have an eroding effect on trust and social cohesion. Reviewing Uslaner's arguments and empirical evidence, we can address the following themes when analyzing the ISSP data: The relation between trust and equality with special attention to social trust and income inequality; the differences between trust and fairness or-to be more specific-among general fairness, social fairness, and general social or moralistic trust; the differences between social trust and political or institutional trust; and the relation between social trust and the loss of control, general health, and life satisfaction.
Larsen's social constructivist theory of economic inequality and trust Christian Larsen's social constructivist theory builds on Benedict Anderson's influential sociological concept of imagined communities. Larsen focuses on the perception of actors, their life in a (non-)egalitarian, middle-class society, and the gradual construction of the social "other." Larsen sees one of the motivations of his research in the question of why many Britons and Americans came to consider the (perceived) sizeable 'bottom' of society as untrustworthy, undeserving, and even dangerous, while many Swedes and Danes came to consider the (perceived) small group of citizens at the bottom of their society as trustworthy, deserving, and peaceful. (Larsen 2013: 3) Further, Larsen adds that a focus on changed public perceptions of society and its imagined members is needed to explain changes in trust (2013: 101).
Another theoretical revision worth mentioning is Larsen's review of the concept of a shared moral community, often representing the theoretical foundation of social cohesion. The constitutive part of shared moral community in modern societies is not only about sharing the same ideals or attitudes toward religion, family, political system, or sexual orientation. It is more about the presumed general ethical integrity of your fellow citizens, or in Larsen's words: "The most important aspect of social cohesion is that citizens believe they share the norm of not cheating each other" (Larsen 2013: 11). Following Larsen's arguments regarding the processes involving trust and inequality, we can consider these two topics: The relationship between social trust and the perception of inequality and the overall cohesiveness of societies through the lens of social trust using Larsen's main indicator of social cohesion, which is included in the ISSP data.

Data and methods
Our analysis is based on public opinion data from the 2017 ISSP module on social networks and resources from a total of 30 countries. The data consist of random samples of the adult population in each country and was collected either face-to-face or in selfcompletion mode. The sample sizes are usually between 1,000 and 1,500 respondents in each country. The total sample size is 44,492 respondents.
The ISSP 2017 data include 60 questions on social network-relevant aspects. We organized the description of the results in three sections: diversity and size of the social network, indicators of social trust, and selected indicators on health. All results are based on the weighted samples. More information on question wording and categories as well as the construction of indices are reported in the tables and figures in the results section.
In order to test the relationship between relevant dimensions, we also correlated the country-level mean values and percentages derived from the ISSP data with three macro-level indicators: (1) the Gini index on income inequality derived from the World Bank, (2) the Satisfaction with Life index by Adriane White (2007), which combines UN, WHO, CIA and other data, and (3) IMF data on GDP per capita (PPP). 2

Diversity of the network, number of contacts, and feeling of isolation
The ISSP survey started with a question on the diversity of the respondents' social network and whether they have members of a certain profession among their family members, close friends, someone else they know, or not at all. Table 1 shows the percentage   of respondents who have each profession among their family members or close friends. Across all countries, a nurse (41.2%) and a schoolteacher (39.6%) are mentioned the most often and a human resource manager (19.2%) and a lawyer (22.2%) the least often.
Considering the scope of the network, Suriname (42%), the United States (41.4%), and Iceland (41.3%) are the countries in which respondents are connected to the widest array of occupations, while Thailand (13.2%), Japan (15%), and Taiwan (18.9%) represent the other end. The social network thus is larger and more diverse in the former countries. Alongside the size of the network, ISSP also asked respondents about the frequency of their social interactions and their feeling of isolation (see Figure 1). We summarized all related items in three indices, after controlling for sufficient internal consistency using Cronbach's alpha. The index "frequency of organized activities" is measured by the question on how often the respondents took part in activities of (1) groups or associations for leisure, sports, or culture, (2) political parties, groups, or associations, and (3) charitable or religious organizations that do voluntary work. The index "frequency of interactions" is based on how often the respondents go out to eat or drink with at least three friends and on how often they have contact with their parents, siblings, children, other family members, and close friends. The index "feeling of isolation" consists of the questions on how often respondents have felt (1) a lack companionship, (2) isolated from others, and (3) left out within the last four weeks.
As for the organized group activities, respondents in India and Thailand are the most active and respondents from Japan, Russia, China, and Lithuania are the least active. Contacts to family and friends are reported most frequently in Israel, Slovenia, South Figure 1. Group participation and social contact versus perceived isolation (mean values). Mean values of indices. Group participation ranges from 0 "never" to 10 "once a week"; contact to family and friends ranges from 0 "never" to 10 "daily"; and feeling of isolation ranges from 0 "never" to 10 "very often." N varies between 43,172 and 43,759. Weighted data. Source: ISSP 2017.
Africa, and Spain and the least often in Sweden, Finland, and Japan. Overall, it is noteworthy that visits to family and friends occur more often than the organized activities with an average of 6.5 compared to 2.1 (with 0 ¼ never and 10 ¼ daily). Finally, Figure 1 also shows the feeling of isolation based on the three different indicators of exclusion. The reported isolation is the lowest in Slovenia, Switzerland, and Germany and the highest in Croatia, Thailand, and Israel. We also considered the correlations among these three variables at the country level. The correlation is À.28 between subjective isolation and group activities; À.15 between isolation and family interactions; and .11 between family and group interactions. The feeling of isolation thus correlates more strongly with group activities than with family interactions at the country level.
Considering Putnam's assumption that civic engagement-which involves participation in charity organizations and volunteering-tends to be linked to strong informal social networks and social trust, we also tested the correlation between variables for social trust (see next section on these variables) and participation in charitable or religious organizations that do voluntary work. Both variables represent relevant indicators of the strength of social capital and thus should correlate positively. The results at the country level demonstrate the opposite: a weak, negative, and statistically not significant correlation between these two variables (À.12). This poor relationship between these two variables is also visible when considering specific countries. The countries with the largest volunteering networks in the ISSP sample are Thailand, India, the United States, New Zealand, Suriname, and the United Kingdom. Yet the social trust levels are highly variable in these countries and range from 20% in Suriname to 71% among Indian respondents. This suggests that high-trust or highly cohesive societies do not necessarily possess dense civic volunteering networks.
Social trust, social fairness, and institutional trust Social trust Figure 2 demonstrates the levels of general social trust and fairness for individual countries based on two related, but conceptually distinct questions: perceived trust or trustworthiness and perceived fairness of other people. The item on trust in other citizens is often described as the standard general trust or general social trust question: "Generally speaking, would you say that people can be trusted or that you can't be too careful in dealing with people?" 3 The second question asks for the respondents' evaluation of the fairness of other people: "How often do you think that people would try to take advantage of you, if they got the chance, and how often would they try to be fair?" Figure 2 shows the percentage of respondents who rather think that other people can be trusted and that they try to be fair most often.
The country-level comparison shows that the overall level of perceived fairness is higher than the level of general social trust, except for India. It, however, also indicates that the gap between the perceived fairness and general social trust is rather small and around or even below than 10% in two Scandinavian high-trust democracies (Denmark and Sweden) and in two post-socialist democracies (Czech Republic and Lithuania) as well as in Spain and Thailand. The results also show that the cross-national variation of fairness is lower (in the range of approximately 50%-90%) than that of social trust (in the range of approximately 20%-80%).
At least two observations need to be noted regarding our theoretical background. First, the differences in trust and fairness correspond visually quite well with Uslaner's idea of two distinct concepts, which are often mistakenly swapped or placed side by side: "Trust is trust. It isn't helpfulness, fairness or even honesty" (2002: 72). In this vein, the level of general trust reflected in the statement that most people can be trusted represents the concept of moralistic trust, whereas the percentage of respondents believing that people are generally fair relates to the concept of fairness. The distinct nature of trust and fairness is probably best be reflected in the extreme cases of India, Taiwan, and the Philippines, where we observe large differences between general trusters and respondents believing in fairness.
Second, when applying Larsen's social constructivist view 4 and assuming that the levels of social trust are an indication of social cohesion, we can report that high levels of trust occur only in a minority of our 30 countries. There are merely six countries in the ISSP sample, in which at least two-thirds of the respondents demonstrate general trust in other people-Denmark, Iceland, Sweden, India, Finland, and Switzerland. Only a few societies thus should be considered highly cohesive societies. Further, faith in strangers remains a very rare and thus precious currency in Suriname, the Philippines, Taiwan, South Africa, Russia, Slovenia, and Japan, where less than 35% of the respondents report such trust. Figure 3 offers an overview of country differences in the views on social fairness. The findings are based on an index consisting of three items: (1) perceived large income inequalities, (2) a preference for small differences in living standards, and (3) a desire that the government reduces income inequalities. These three items have a relatively high internal consistency (Cronbach's alpha ¼ .73) and can be considered an indicator of perceived social fairness and thus cohesion within their society. Figure 3 displays the differences in country means, with taller bars indicating that respondents perceive their society as socially and politically fair.

Social fairness
The results indicate that respondents in post-socialist countries (with the exception of Spain) such as Russia, Lithuania, Hungary, the Slovak Republic, Slovenia, and Croatia express a strong preference for a fair society in terms of low inequality and living standards. The other extreme-with the exception of the Philippines-are seasoned democracies and affluent societies such as the United States, Denmark, Australia, New Zealand, and Sweden in which respondents do not call as strongly for social fairness measures. . Mean values of index on perceived social fairness based on statements such as differences between people should be small in a society. Scale ranges from 1 "strongly disagree" to 5 "strongly agree." N ¼ 43,228. Weighted data. Source: ISSP 2017.
Of course, we also need to consider that these societies are rather wealthy and that even the poorer groups are able to afford a rather decent life. Overall, we can thus note an influence of the political history and the level of affluence on the preferences for social fairness measures.
The third figure in the context of social trust displays the respondents' trust in national courts and major private companies (see Figure 4). The question wording is: "Using the following scale ranging from 0 to 10, where 0 means "No trust at all" and 10 means "Complete trust," please indicate how much trust you personally have in … ?" In most countries, respondents express more trust in national courts than in major private companies. The highest percentage of trust in national courts (approximately 60% to more than 75%) can be found in Denmark, Switzerland, Finland, Austria, and-curiously enough-China. Respondents in Croatia, Mexico, the Slovak Republic, Taiwan, Slovenia, and Russia demonstrate lower levels of trust in both types of institutions.
Our findings on the different indicators of social trust and related concepts offer a fertile ground for further discussion and research. One unanswered aspect is differences and connections between the different elements of trust. To shed some light on this question, we ran a correlation analysis with all indicators mentioned in this section. This correlation shows that the linkage between general social trust and trust in national courts is quite strong (.60); the connection between social trust and trust in major private companies, on the other hand, is weak and statistically insignificant (.21). Social trust in the sense of general, deeply embedded moral perceptions of other people is thus stronger related to institutional trust in courts than in economic enterprises. We thus can hypothesize that the latter institutions are evaluated differently, perhaps based on short-term, everyday experiences, and cognitive evaluation of actors.

Country-level influences on social trust and cohesion
Socioeconomic inequality is considered the most relevant underlying factor of social trust and cohesion by Uslaner and Larsen. In order to briefly test their premises and ideas, we correlated social trust from the ISSP data with different measures of inequality. Our results at least partly confirm Uslaner's analyses. The correlation between the Gini index and social trust is À.48 and À.32 between social trust and the aggregated perceived income inequality taken from the ISSP survey. The correlation between social trust and the wealth of a country in the form of GDP is .34. The strongest and most convincing link of social trust with inequality could in our case be detected with the same item Larsen has used. In this item (Q14a from the ISSP 2009 survey), the respondents were asked to choose one out of five diagrams that best describes the society they live in. Larsen used the share of the respondents choosing a Type D diagramwhich according to the ISSP questionnaire represents "a society with most people in the middle"-as a measure for the perception of living in a middle-class society. The correlation coefficient of this item with social trust from the ISSP 2017 survey is .79 5 and thus stronger than the previously mentioned correlations.
Health care, obligations to help, and subjective health Our concluding results section presents the respondents' reported subjective health and life satisfaction, preferences for different providers of health care, and related aspects. Figure 5 shows the responses to the question "Who should provide health care?" with the answer possibilities: "Government," "Nonprofit organizations," "Family, relatives, or friends," "Private companies/for-profit organizations," and "religious organizations." Among these options, the government was selected the most often. Across all countries, 81% of the respondents prefer the government as health care provider, followed by 9% expressing a preference for family help. Taiwan is the only country in which respondents chose the family over the government. Alongside this question on health care, the ISSP also asked "Who should provide care for the older people?" Here, more respondents opt for the family (24%). Yet, the government is still the leading preference, with 64% of the respondents choosing this option.
The respondents' preferences change when asked who they would turn to for "help in the household or garden," "in the household when ill," "in depression," "in giving advice about family problems," "in enjoying a pleasant social occasion with," as well as in situations such as "borrowing a large sum of money," "finding a job," "administrative problems or official paperwork," "finding a place to live," or "looking after oneself if seriously ill." Family and close friends are now the most common choice for most of these problems. Public services are mentioned only in connection with job issues and similar welfare problems, but only in Austria, Finland, France, Germany, and Japan.
Overall, we see a turn from a preference for institutional care and help to a call for more personal help in everyday problems. While the government is considered the most important provider of health care, respondents rather turn to family and close friends when they need actual help. The social network and friendship thus are an important pillar for this type of help. Yet, respondents are not just the receiver of help, they also might get asked to provide help for their family and friends. The ISSP thus also includes questions on the perceived norms and demands with regard to support. Figure 6 provides an overview of the perceived demands for help. The fewest demands are reported in Austria, Switzerland, and Germany and the highest demands in the Slovak Republic, Mexico, and Finland. Considering the correlations between preference for different health care providers, perceived demands, and preference for the family becoming active shows that the more often demands are mentioned in a society, the fewer respondents would like to have the family responsible for health care and vice versa.
The final figure displays the perceived health and life satisfaction at the country level (see Figure 7). These results are based on a set of questions covering the areas subjective health, how often respondents felt depressed or overwhelmed by demands, their overall life satisfaction, and whether they felt that they are able to achieve their goals. All these items form a single dimension and were thus summed up in a single index for our overview at the country level. Here, Austria, Switzerland, and the United States are the countries with the highest level of satisfaction, and Japan, Lithuania, and China the countries with the lowest satisfaction.
The subjective life and health satisfaction, the sense of control over one's life, and the general optimism that one can get ahead in life are seen in close relationship to the social embeddedness and social trust by Uslaner (2002: 79). The correlation at the country level between social trust and subjective health/life satisfaction is .53. The strength of this correlation is similar to the strength of the correlation with an alternative measure of life satisfaction, the Satisfaction with Life index by Adrian White (2007). The Pearson correlation coefficient of social trust with White's measure is .55.
Following the argument that general social trust is largely based on an optimistic view of one's world, partly connected to the general sense of control over life, but rather independent from short-term experience such as life difficulties (Uslaner 2002: 77-104), we also examined the strength of the relationship between trust and perceived life difficulties. In the ISSP 2017 questionnaire, the respondents were asked "During the past four weeks … how often have you felt difficulties were piling up so high that you could not overcome them?" The expectation-when based on the assumption that generalized social trust incorporates a strong moral component how the respondents see the world and is not necessarily directly shaped by their individual, everyday experience such as current life difficulties 6 -is that the relation is not very strong and, if existing at all, rather negative. The correlation between general social trust and perceived short-time and unsurmountable life is indeed weak, negative (À.24), and statistically not significant. Our expectation is thus confirmed.

Discussion and conclusions
This contribution complements the special issue on the 2017 ISSP survey on social networks and social resources by presenting a comprehensive overview of its results at the county level and by testing some ideas of the social trust literature.
Our overview started with a look at the diversity of the social networks of our respondents. It showed that nurses and schoolteachers are most commonly part of the social network of our respondents and human resource managers and lawyers the least often. Further, we presented information on preferences in health care and related issues. The analysis revealed a rather homogenous picture across all countries as far as the provision of health care is concerned: An overwhelming majority prefers the government. The results regarding subjective health and life satisfaction showed that a majority of the respondents is quite satisfied. Finally, as for the frequency of interactions, our results showed that respondents see their family and friends very often, with little variation across countries. The reported level of loneliness thus is rather low in all countries.
Our analyses placed more emphasis on the topics of social trust and cohesion. As for social trust, our findings on the levels of social trust are line with prior findings in the sense that the Scandinavian societies of Denmark, Iceland, Finland, and Sweden are characterized by high levels of social trust. The percentage of respondents who express social trust is around 70% and even 80% in Denmark. Social fairness is considered very high in Germany, Denmark, Finland, and Iceland. Overall, the majority of the democratic and capitalist countries quite predictably find themselves somewhere in the above-average trust and the above-average fairness area.
Alongside these descriptive findings, we also attempted to test various assumptions and ideas of Robert Putnam, Eric Uslaner, and Christian Larsen. Here, our results corroborate Uslaner's finding on the relationship between inequality (here measured by Gini) and the level of social trust and Larsen's finding that trust is related to perceived inequality. Putnam's view on the connection to civic engagement, however, is not confirmed, given that volunteering is not correlated to social trust at the country level. Our findings are thus in line with some previous results but also point to a few inconsistences. We, however, need to be aware of the limitations that arise from our data and indicators. First, there are some differences in the measures. Larsen's item on the society shapes was taken from the ISSP 2009 survey data set and thus reflects the perceived structure of the society, while the ISSP 2017 items on the perceived size of inequality also includes an assessment and evaluation of the magnitude of income differences in a society.
Further, we observe an inverse relationship between objective inequality and the perception thereof in the seven central-east European post-communist countries and China. These countries are characterized by a low objective level of inequality. Yet, a large percentage of the respondents agree that the income differences in their society are too high. Given that the expressed level of social trust is also low in these countries, trust seems to be more strongly related to perceived inequality than to actual inequality.
On the other hand, the ISSP data suggest that the level of social trust levels has increased considerably in these post-communist countries over the last years. The average level of social trust for the eight countries with the legacy of communism-China included-is 44.6% in 2017. The comparison between ISSP 1998 and 2017 for those five post-communist countries that took part in both waves shows an increase of 14.2 percentage points-which equals a relative increase of 40%-among the group of trusters. Future research thus should explicitly consider changes over time, the role of perceived and objective levels of inequality, and the relationship between trust and fairness given their very different levels in some countries.