Solidarity: A European Value?

Abstract

Solidarity is considered essential for the sustainability of societies, both at the level of individual contributions to society and as an aspect of cooperation between countries. As such, solidarity is often called upon by politicians in a declarative way. In contrast to most prior work, we investigate an attitudinal perspective on solidarity, not behavioural or policy-preference perspectives. We look at questions such as the prevalence of solidarity attitudes among European populations, the degree to which the declared norm is shared among these populations, and whether solidarity attitudes have changed with consecutive crises in Europe. We also discuss possible antecedents of solidarity levels. Distinguishing solidarity by close and universal scopes, we find that both are associated with the identification of citizens with communities at different levels. In country and time comparisons, European societies display a good degree of homogeneity and stability. Close solidarity is more pronounced than universal solidarity, and this may have even increased over consecutive crises. There are clear differences in relation to socio-historical region, but only modest associations with the religious composition of countries in terms of denominations. One role of religiosity might be that of an identity marker, where more heterogeneity comes with lower levels of solidarity.

1 Introduction – Is Solidarity in Europe at Risk?

The founders of a united Europe declared solidarity to be one of the core European values described in the Charter of Fundamental Rights. In 2007, the Treaty of Lisbon explicitly added solidarity to the core properties of an idealised European society: the member states share ‘a society in which pluralism, non-discrimination, tolerance, justice, solidarity and equality between women and men prevail’ (art. 2).

With these words, the European Union normatively affirmed shared value orientations that enshrine a common transnational cultural horizon. For a long time, solidarity has been a consensual and apparently foundational value in the political discourse. Many political parties incorporate this value in their narrative. Solidarity is a point of reference not only among those parties inspired by socialism and close to the labour movement. Per the Catholic social teaching, in which solidarity is a key value, it also underpins the ideologies of Christian democratic parties (Stjernø 2005).

This background of declarative political statements may have given solidarity the appearance of a value somehow naturally shared among individuals, just because they belong to the same society. However, according to Grimmel and Giang (2017: 2), solidarity is not a value ‘to be found out there, but must be created’. This perspective points to conceiving of solidarity more as a dynamic process than as a static value that remains stable regardless of changing conditions. Both at societal and personal levels, different conditions may produce different levels of solidarity (van Oorschot et al. 2005; similarly for work values Weiss and Hörisch 2021). In that vein, over recent decades, solidarity has allegedly been challenged by several factors mainly related to the cumulative crises facing the European countries: the banking and financial crises in 2008, with the subsequent economic recession; the refugee crisis; and, more recently, the COVID-19 crisis (Ferrera and Burelli 2019; Hatton 2017; Wallaschek and Eigmüller 2020). While it is quite easy to offer solidarity in times of wellbeing, the challenges to wellbeing and status quo can affect solidarity levels, mainly by drawing on – or even only threatening to draw on – scarce resources so intensely or for so long that readiness to share such resources dwindles (Aschauer and Mayerl 2019; Lahusen and Grasso 2018). The insecurities and fears deriving from these crises would thus open cracks where anti-solidarity sentiments can creep in.

The rise of populist parties and far-right movements, which built their success on these fears, raises concerns for solidarity at different levels (Lahusen and Grasso 2018). While these groups express the sentiments of distrust of parts of the population, disappointed by unsolved issues from unemployment to migration and asylum seeker management, they may also foster the erosion of solidarity at the individual level, by legitimising ethnocentrism, xenophobia, nationalism, and anti-European feelings. With elected populists entering political institutions and offices, they may also affect the shaping of policies and rules towards being more selective and discriminating regarding whom these policies grant the institutionalised solidarity of the state to. The process of de-solidarisation is particularly risky. Political observers argue that the lack of between-groups solidarity is a threat for the cohesion and sustainability of democracies, giving further room for the rise of ethnic and nationalistic exclusiveness (Wallaschek 2019).

With this background, the complexity of European solidarity has attracted a lot of public and scholarly attention. Surprisingly, however, there is very little work that provides an empirical overview of the current levels of solidarity across European countries. This chapter strives to close this gap by investigating what the levels of solidarity are across the countries that participated in the three most recent waves of the European Values Study (EVS).

The chapter will proceed as follows. We first discuss the theoretical foundations of solidarity, very briefly touching upon a variety of aspects discussed in the extant literature. We then lay out the – largely exploratory – methods applied, and why we chose particular factors or variables from the available data. The first results presented will look at how solidarity relates to certain aspects of mass identification, thereby providing an initial validation of our measures of attitudinal solidarity. In the central part of the chapter, we then look at what levels of attitudinal solidarity prevail across the countries participating in the EVS, and how much they have changed, if at all, over time. Moving one step further with the exploratory analyses, we then look at some possible correlates for those levels of solidarity. The chapter closes with a speculative interpretation of these empirical explorations.

2 The Concept of Solidarity

2.1 Socio-structural Foundations for Individual-Level Solidarity

The first challenge for any empirical research on solidarity is focusing on a specific meaning of the term. Despite its relevance, solidarity is a ‘fuzzy concept’ (Rusu 2012: 72) that has been interpreted in a multitude of ways. A first distinction concerns the focus on solidarity as a societal or an individual quality. Durkheim (1893) famously conceptualised solidarity as a societal characteristic that describes different patterns of mutual dependence. For traditional societies, this would be mechanical solidarity, driven by the experience of similarity and a sense of belonging to a close community, which would not least be mediated through religiously based values prevailing in those communities. In modern, differentiated societies, organic solidarity should prevail, driven less by shared values and common collective orientations such as religious beliefs than by the mutual dependency of the societies’ members that results from economic specialisation and societal division of labour. Contemporary scholarship often focuses on solidarity as an individual quality, referring to people’s ‘willingness to help others or to support the group one belongs to, without immediately getting something in return’ (de Beer and Koster 2009: 15), but the relationship to Durkheim’s basic, macro-level distinction is still discernible. Recalling Durkheim’s paradigm of organic solidarity, Steinvorth (2017) argues that solidarity between individuals in differentiated, complex societies would be the product of two opposing forces – while they are individualistic in their motivations, ideally they are also aware of their mutual dependency due to the high division of labour in their societies, and of the vulnerability of the preconditions that allow them to realise their individualistic goals – equality of rights; protection from wilful authorities; and protection from material risks and physical dangers – and all this regardless of most person-level characteristics, which are largely considered to be their own, private affair. By this reasoning, such knowledge of universal neediness in the face of larger challenges should facilitate a foundational willingness to cooperate for the mutual, and in sum collective, benefit (Habermas 1996, 2017).

A challenge with that kind of cooperation out of necessity would be that people might disagree about the right balance between resources invested into collective or cooperative tasks, such as controlling the risks mentioned above, and the short-term realisation of individual goals; further, such cooperation is easily exploitable when people do not contribute their ‘due share’ (Hechter 1984, 2018). These challenges bring the notion of solidarity as a socialised norm or value back in, even for modern, differentiated societies: the more people are intrinsically motivated to offer support and share resources, the easier it will be to sustain the required levels of cooperation. To the degree that traditional and modern societies are successful in socialising their members to be ‘good citizens’, the individual-level emotive expression of the norm of solidarity would then be phenomenologically similar to the ‘mechanical’ solidarity of Durkheim’s traditional societies: in both settings, individuals would feel a sense of obligation to help others in their community if they are in need. What would be different between the idealised society types is how the communities are defined and perceived (per social proximity and similarity in the traditional setting, but likely by more abstract criteria in the modern setting), and how acts of solidarity are actually performed (per personal activities in the traditional setting versus ‘differentiated’, for example, professionalised and formalised methods, possibly through state institutions, in the modern setting).

2.2 Solidarity and Group Identification

That individuals have a sense of being part of some kind of community, of either the traditional, similarity-based type or the modern, universalistic-needs type, would then be essential for any solidarity norms to develop and be activated. But conversely too, it is the willingness to offer solidarity, or a lack of this willingness, that defines where the relevant community or in-group ends and the out-group begins: solidarity and collective identities are in a close reciprocal relationship. Empirical studies of this relationship have initially most often looked at the relation of solidarity as support for redistributive policies within a state and the identification with the imagined community of a nation, and have usually found positive correlations (Miller and Ali 2013), even if some authors have found that the positive effect of national identity on welfare state support is contingent on the inclusiveness of the identity concept (Wright and Reeskens 2013).

This is in line with the theoretical mechanisms sketched above: on the one hand, the existence of binding community norms as implied in the formal citizenship to a state can produce enacted solidarity, but on the other hand, affective feelings of identity are conducive to more solidarity within a state (Brubaker and Cooper 2000; Calhoun 2007). In a more recent body of literature, the same patterns have been confirmed for the larger community of the European Union: citizens’ identification with the EU and their willingness to offer financial (state-level) support across European borders do reinforce each other (Díez-Medrano et al. 2019; Verhaegen 2018).

However, looking at a presumed macro-level correlate of group identification again at the country level, namely ethnic diversity, Janmaat and Braun (2009) surprisingly found that higher ethnic diversity – which they assume to indicate lesser identification – has a positive rather than a negative association with societal solidarity levels. For example, the lower the average ethnic homogeneity of a society, the higher would be the readiness to offer solidarity to the needy in the same society. While the authors can offer no explanation for this unexpected finding, we caution with a view to our own analyses presented later that they employ a different measure of solidarity that might be more vulnerable to policy preferences and ideological viewpoints.

An identity marker of similar relevance as ethnic belonging is religious belonging. Whereas some research has found evidence that the substance of religious beliefs affects the readiness to offer solidarity and concludes that persons with relatively higher and stricter religious commitment would be more ready to include out-groups into their scope of solidarity (VanHeuvelen and Robinson 2017), others have argued that religion has become the most important marker of cultural differences in modern societies (Brubaker 2013). There have indeed been political attempts to recite the Christian heritage of the (European) Occident with a view to mobilising in-group identification against immigration from non-Christian regions that is perceived as excessive (Challand 2009; Foret 2009). It should be noted that the two lines of reasoning are not necessarily contradictory, as the first refers to the intensity and content of religious convictions at the individual level, whereas the latter refers to the – possibly ‘empty’ – boundary-setting functions of declared religious belongings. In sum, the role of religious belonging is likely contingent on the degree to which belonging leads to certain convictions or is just ‘yet another’ identity marker.

2.3 Forms and Recipients of Solidarity

For the sake of completeness and clarification, we need to mention that previous solidarity research has largely focused on a variety of specific forms of solidarity behaviours, in the sense of activities or (agreement to) resource transfers in support of the needy. Much of the comparative literature in the field investigates formal/institutionalised solidarity through welfare systems within states (Fernández Guzman Grassi 2019; Ferrera 2014) and on transnational solidarity between states (Ferrera and Burelli 2019; Grimmel and Giang 2017). Some studies adopting micro-level perspectives (Cinalli and Sanhueza 2018; Kalogeraki 2018; Lahusen and Theiss 2019; Maggini and Fernández 2019; Montgomery et al. 2018) refer to actional components of solidarity as, for example, taking part in actions of solidarity (for example, being an active member of a civil society organisation, marching for the rights of people in need, donating for a group in need, etc.).

One of the core questions specifically in relation to the research related to welfare policies regards the beneficiaries of solidarity and, more specifically, which social groups are considered to deserve help and in which form (Buß et al. 2017; van Oorschot 2006). Particularly in the last decade, in conjunction with the economic and refugee crises, the debate has also paid attention to the transnational dimension of deservingness, where attributions of deservingness apparently are also made towards whole societies (Wallaschek and Eigmüller 2020).

2.4 Short- and Long-Term Trajectories in Solidarity Levels

Although solidarity is not a core concept in Inglehart’s theory of value change, solidarity as a value should follow the trajectory of the ‘post-materialist’ values: as a non-egoistic and less materialistic orientation, it should increase along with a society’s wealth and levels of security (Inglehart 1997). To the degree that solidarity becomes internalised through socialisation, it would then resist short-term fluctuations in economic security and mainly change through cohort replacement. However, Inglehart would also grant the possibility of short-term period effects fluctuating around the ‘general’ trend, so that, for example, economic problems could also allow for some short-term decrease in expressed solidarity orientations. Nonetheless, the stronger effect should be with the socialised and thus per birth cohort stable value of solidarity. It is noteworthy that this pattern could not actually be strongly confirmed in the only direct empirical test we are aware of, which tested whether individual-level post-materialist values would predict solidarity orientations (only weakly), and which further tested whether society-level averages of post-materialism and of solidarity orientations are correlated (not at all) (Janmaat and Braun 2009). Given these results, some caution is in order about the hypothesis that there is a long-term association between solidarity and post-materialist value orientations.

Before the 2008 economic crisis, the general processes of secularisation, individualisation, and globalisation were intertwined with the processes related to EU enlargement, new migration flows, and the decline of welfare systems. In this scenario, the decline of nation states was supposedly leading to a weakening of national solidarity (Radtke 2007). With the economic recession, however, micro worries related to the loss of control over personal outcomes assuring security (Schwartz et al. 2000) began to reinforce solidarity towards the in-group, with stronger attention directed towards people living in the neighbourhood and region, and to fellow countrymen. Similarly, Sortheix et al. (2019) found that benevolence values, siblings of solidarity, but also tradition and security preferences, increased later in the course of the financial crisis. The refugee crisis may have further prioritised in-group solidarity through identity mechanisms by providing a publicly visible ‘out-group’ that was evoked by nationalist/populist politicians, thereby strengthening the salience of more closed identity references (Delanty 2008).

Solidarity at the transnational level apparently encountered different mechanisms. EU enlargement may have increased cosmopolitanism (Outhwaite 2006), but the fact that new member states disproportionately benefited from structural funds challenged the legitimacy of (financial) solidarity at the European level (Grabbe 2006). Furthermore, new migration waves raised fears of new claimants moving into overstretched welfare systems, and have increased ethnic heterogeneity within European societies (Janmaat and Braun 2009). During the economic recession, spending cuts and increases in unemployment rates amplified conflicts about who deserves welfare benefits, challenging, for example, solidarity towards immigrants. The austerity measures imposed at the European level fuelled the first relevant wave of anti-Europeanism, and the failure of the EU in the management of the refugee crisis may have impacted negatively on transnational solidarity (Reinl 2020; Wallaschek and Eigmüller 2020).

3 Methodological Approach of the Chapter

As the discussion above has demonstrated, solidarity research can have many different angles, and most previous studies have chosen the important behavioural perspective of looking at acts that are helpful to others, either directly or in the form of supporting redistributive policies.

An alternative perspective is to look at the attitudinal expressions of solidarity. The possible research questions would in a sense be mostly the same as for the behavioural angle – who, why, how much, and under what conditions is solidarity offered? But they can be approached in a way that needs far less controlling for short-term situational conditions than can behaviours, which are naturally heavily contingent on individual resources, immediate opportunities and cues for action, and situational peculiarities at both individual and societal levels. This chapter therefore focuses on exploring solidarity in its most generic expression as the subjective propensity to offer support to others. Further to this methodological reason, looking at the normative dispositions speaks more directly to the normative angle taken by EU policy-level declarations, which posit ‘solidarity’ as a European value, so we can tentatively check how much those declarations relate to the lived reality of European societies, within and outside the European Union.

The reason for our mostly descriptive approach is that previous literature has hardly ever investigated solidarity as a generic population attitude or ‘value’ across European populations. Rather, the focus has been on solidarity as actual acts of support, and on specific policy areas. Therefore, lacking strong theoretical expectations or well-established prior results on comparative levels of attitudinal solidarity, it seems justified to take a descriptive and inductive view of the matter. This will be informed by referring to the literature presented above, but not in the sense of formally testing an array of hypotheses. Rather, we will more modestly limit ourselves to plausibility considerations, thereby hopefully providing food for future, deeper analyses.

Specifically, we will look at levels of solidarity attitudes in European countries, at the relationship of solidarity levels to subjective identification patterns, at possible regional clusters, and at possible time trends. Drawing from the literature reviewed above, we will tentatively enrich this with information about country-specific economic situations, immigration exposure, and the religious composition of the societies. In terms of analyses conducted, we will first check for the association between solidarity and subjective identification with the EVS 2017 data, we will then describe the trend of two facets of subjective solidarity over three EVS waves (1999, 2008, 2017), and finally we will relate those levels to the external factors just mentioned, in the practical form of very simple macro-level regressions.

4 Data and Measures

4.1 Data Availability

Three consecutive waves of EVS contain the relevant questions on subjective solidarity, which will be described below. The nature of the EVS as a repeated cross-sectional survey enables us to look both at country comparisons and at comparisons of societal averages over a time window of about three decades. To facilitate the latter, we will exclude samples from all countries that were surveyed only once in that time span.

Table 7.1 shows all countries with at least two occurrences over all three waves, which yields 108 samples from 42 different countries, respectively regions. (Northern Ireland is traditionally sampled and treated separately from Great Britain, though both are part of the United Kingdom.) The 1999 and 2017 waves are both less complete in terms of country coverage than the EVS 2008 wave, so that we have a number of instances where a time comparison is only possible in reference to 2008, either looking before or after that wave.

Table 7.1 Overview of countries and EVS waves

4.2 Measuring Solidarity

The subjective concept of solidarity is assessed in the EVS via a battery of nine items (Table 7.2), which was first introduced to the questionnaire with the 1999 wave.¹ The instrument asks the respondents to express to what extent they are concerned about the living conditions of various groups of people.

Table 7.2 Subjective concept of solidarity – battery of items

In previous empirical analyses of the question battery (which are scarce – we are aware of Abela (2004); Lomazzi (2021); van Oorschot et al. (2005); van Oorschot (2006); Rusu (2012)), three substantive dimensions of solidarity have been identified that partly match the theoretical foci of ‘proximity’ and ‘neediness’ built into the items, but in an overlapping way with not entirely clear theoretical interpretations. Using data from the EVS 1999, Abela (2004) identified three dimensions of socio-economic solidarity: as concern for the living conditions of people with social closeness and similarity, for example, people living in the neighbourhood, the region, or the same nation (‘local solidarity’); concern for those in need, like unemployed, older, and sick and disabled people (‘social solidarity’); and concern for people with a larger cultural and social distance (‘global solidarity’), including Europeans, immigrants, and, most generally, humankind. Lomazzi (2021) has recently confirmed that the same dimensional structure can still be found with the EVS 2017 data and has added the crucial information that the battery provides an acceptable degree of measurement equivalence over the large majority of countries in the EVS 2017 sample for the ‘local’ and ‘social’ dimensions, but not for the ‘global’ dimension.

It is important to note that this instrument is measuring a subjective preference of the respondents for potentially offering solidarity to certain kinds of people, where these ‘kinds’ are defined by the varying degree of geographical and social proximity to the respondent for the first five items (Q60) and by aspects of their living conditions for the concluding four items (Q61). In other words, this instrument is a pure attitudinal measure, as it has only affective and cognitive components. Attitude measures can, of course, be partially driven by idiosyncratic factors such as individual personality traits (for example, different degrees of benevolence or neuroticism) and by very short-term situational cues in the interview situation, but these can usually be assumed to be random ‘noise’ when looking at societal averages of those measures. The factors relevant for our purpose, however, would be those that are systematically influenced by societal-level conditions, such as the socialisation experiences that particular groups of people in a given society made at a given period, or the restrictions and opportunities they constantly confront given their society’s welfare system. We should therefore expect that the averaged responses to the EVS solidarity attitudes instrument do indeed reflect the prevalent societal norms about solidarity in a country.

4.3 Sharpening Our Measure of Solidarity

With our interest being in assessing societal preferences for solidarity in its most generic form, we have decided to narrow down the substantive focus of the measures by dropping parts of the question battery from our index building.² Specifically, it seems helpful to exclude the ‘reasons for neediness’ aspect from the indices entirely, so that what remains is the readiness to exert any solidarity at all, only distinguished by the degree of proximity of recipients and respondents, or, in short, the ‘scope’ of solidarity. The ‘scope’ aspect allows us to tap into whether respondents prefer recipients closer to themselves, or whether they are ready to allocate solidarity indiscriminately of proximity, that is, in a universalistic way.³ This distinction adds a relevant angle to our analysis of differences between European countries. Further to dropping all the ‘reasons for neediness’ items from our index construction, we also omit item 3, ‘your fellow countrymen’, from the index building, because this is inherently fraught with an association with national identity. On the ‘wide’ end of the item range, the item on concern about Europeans has similar problems of being overlaid with connotations of identification with the European Union as a political project (which certainly has different levels of support in different countries, partly regardless of the general solidarity dispositions of the citizens of those countries), and of creating a different focus for countries that are geographically and culturally at the core of Europe and those that are on the periphery (for example, Armenia, Turkey). These associations might not only distort the narrow/wide scope distinction by an unwanted element, but they would also induce an overlap of our solidarity measure with the concept of identity, which we understand as an importantly related but still a separate concept and therefore one that needs to be measured in a strictly distinct way. Supporting these decisions, Lomazzi (2021) in her analysis of the EVS 2017 data has found the ‘countrymen’ item and two of the four neediness items to be the most problematic in terms of measurement equivalence across the European countries, so it will likely improve cross-national comparability of our measures if we drop these particular items.

This leaves us with three items from which we derive two indices for each country by year sample: ‘close solidarity’ is computed as the average of responses given to the questions on being concerned about people in the neighbourhood and people in the region, whereas ‘universal solidarity’ is simply computed as the average of the responses to the question on being concerned about all humans in the world. The index values are coded such that higher scores indicate higher levels of solidarity, that is, the original response scale has been reversed.

5 Solidarity and Geographical Identification

The question of identity is highly important for any analysis of solidarity. However, measuring identity – as the subjective identification with large social groups – is inherently difficult, and EVS has introduced measures for this only in its two more recent waves. The questionnaire instrument that we can most directly relate to the solidarity measures with their discrimination by scopes is a battery of questions on ‘geographical’ identification, which asks for the respondents’ subjective belonging to others, within certain abstract geographic borders. However, the response format used was not consistent over the EVS waves. We are therefore limited to using data only from the EVS 2017 wave for the analysis that includes geographical identification. The question battery, shown in Table 7.3, used a rating response format in EVS 2017.

Table 7.3 Geographical identification

We have built a single categorical variable from the five individual items by conducting a Latent Class Analysis (LCA), which yields a compact description of the most frequent combinations of identification patterns preferred by the respondents. We did this using the Latent Gold software, accounting for the multi-level nested structure of the comparative data (Magidson and Vermunt 2004; Vermunt 2003; Vermunt and Magidson 2016). LCA performs a sorting of respondents into ‘classes’, given the prevalence of statistically distinguishable response patterns. Yielding a categorical variable, that procedure can also describe a multi-dimensional construct through assigning cases with overlapping combinations from different traits into their own special categories. In our case, this is useful because the ‘classes’ in the resulting classification appear to be driven by different motivations. One is the preference for identifying with one or several of the abstract social circles at all; the other is the choice of which object of reference, or combination of objects – city, region, country, continent, or world – invites the strongest feelings of closeness. The present LCA yields four classes.

The largest class of respondents of 49% across all countries (country-wise class distributions are shown in the Appendix) is characterised by feeling close or very close to all the reference objects (‘All Close’). The second largest class of 35% feels close or very close to their town or city, to their region, and to their country, but not close or not close at all to Europe or to the world. We thus have a clear preference for conventional and more ‘local’ reference groups in this class versus the wider references (‘Local Close’). The third class comprises only 8% of the respondents, collecting those who feel close/very close to their own country, Europe, and the world, and who feel distant from the more local reference groups (‘Local Distant’). The remaining 8% of respondents in the fourth class expressed closeness to none of the offered reference groups (‘All Distant’). The Appendix provides the Latent Class item response profiles and the country-specific distributions of the estimated identification classes.

As a first piece for the emerging picture of solidarity distributions, we offer a look at relationships of solidarity with our classification of geographical identification. With only data from EVS 2017 available for this, we rely on the visual inspection of the bivariate distributions alone. Therefore, Fig. 7.1 shows the country distributions of solidarity levels over the four different classes we identified (‘All Close’, ‘Local Close’, ‘Local Distant’, ‘All Distant’).

Fig. 7.1

Scopes of solidarity by geographical identity class

The first observation is that there is a very consistent, even if sometimes only slight, slope to the distribution where ‘close solidarity’ levels are the lower, the lower are either the number and/or the ‘closeness’ of reference entities to which the class members felt a sense of belonging. In other words, ‘close solidarity’ is high for respondents who identify with all possible geographical entities, slightly lower if they identify only with the personally closer entities (neighbourhood, town, region, country), even lower if they identify only with the larger/more abstract entities (country, continent, world), and lowest if they identify with none of the entities. The only exceptions to this sloping pattern occur with the two small classes, which have extremely small case counts in some countries, for example, in Norway. This indicates a plausible relationship of close solidarity and local identification, which is not contradicted by the fact that those with high identification with all entities display the highest level of close solidarity.

The distribution of ‘universal solidarity’ by identification classes is similarly plausible. The highest levels of universal solidarity can be observed – in virtually every country except Norway – when either all (‘All Close’) or the more remote (‘Local Distant’) identities are reported as identification objects, whereas the levels of universal solidarity are lower for the two classes that only identify with the closer entities (‘Local Close’) or with none (‘All Distant’).

This reassures us that the distinction between the scopes of solidarity is meaningful, and that the relationship postulated in the literature between solidarity and identification can be observed to some degree, at least at the aggregated level.

6 Country and Time Comparisons of Solidarity Levels

Figure 7.2 shows the country levels of the two scopes of solidarity attitudes per our indices at all available time points. The lines in each country sub-panel thus display the averages of the indices over the two solidarity scopes, for at least two points in time.

Fig. 7.2

Scopes of solidarity over three EVS waves

The panel shows a compelling degree of similarity across most countries in important regards: First, the general mean levels of both scopes of solidarity are in the 2.5 to 3.5 range of the scale, with only Armenia, Austria, Georgia, Montenegro, and Turkey going noticeably above that, and only for ‘close’ solidarity. Nevertheless, between-country differences in levels of solidarity mostly concern both scopes in the same way, that is, if close solidarity is relatively low or high in a given country, so is universal solidarity. Second, the order of the levels is almost universally that ‘close’ solidarity is more pronounced than ‘universal solidarity’, with the notable exceptions of Finland for all time points, Denmark and Great Britain for two time points, and Czechia and Greece for one time point, with ‘close’ solidarity being markedly lower than ‘universal’. Interestingly, these cases appear to have unusually low levels of ‘close’ solidarity rather than unusually high levels of ‘universal’ solidarity. Thirdly, for most countries, both indices are rather stable across time. For a few countries only, changes exceed a magnitude of 0.5 scale points over adjacent time points, with the steepest changes having occurred in Armenia, Austria, Czechia, Denmark, and Republic of Ireland. Regarding the direction of the changes, there are more instances with increases of solidarity than with decreases, and such increases appear to be larger over the second period, that is, between 2008 and 2017. The only countries with decreases for both scopes of solidarity over all time points are Belarus, Northern Ireland, and Croatia. Unfortunately, we do not have the complete time-series for some of the countries with the steeper changes.

Having found that the levels of both solidarity scopes are not usually hugely different, we still need to better isolate the aspect of the spread between solidarity scopes, as this might point at differentiated dynamic aspects for both scopes. As discussed above, some researchers hypothesised that in reaction to the economic and migration crises, solidarity in closer circles would increase at the expense of wider (especially cross-national and out-group) solidarity, thus increasing the overall spread. We have therefore plotted the differences between ‘close’ and ‘universal’ solidarity in a separate figure. If changes in spread are to be understood to be a short-term effect right after the financial crisis, an increase of spread between the two solidarity scopes would have occurred at the occasion of the 2008 to 2010 EVS round. If it were understood as a lasting effect that more permanently shifts the solidarity levels, the spread should not decrease afterwards, or might even further increase for the following observation instance too. However, between the 2008 and 2017 EVS waves, Europe also experienced the ‘refugee crisis’, the effects of which would likely be confounded with any longer-term effects of the financial crisis.

Per visual inspection of Fig. 7.3, we find that for 21 countries (AT, BE, BG, BY, CZ, DK, EE, GR, HR, HU, IT, LT, LV, MT, NL, PT, RO, SE, SI, TR, UA), there is an increase in spread in the earlier period, whereas ten countries (DE, ES, FI, FR, GB/GB-NIR, IE, IS, LU, PL, SK) experienced a decrease in spread over that period.⁴ Looking only at the 24 countries that were observed over all three points in time, we find that for AT, BG, CZ, DK, IT, LT, RU, and SE, the spread continues to rise in the second period. For BY, EE, HR, HU, NL, RO, and SI, the spread decreases somewhat, having risen before, but usually the decrease does not fully compensate for the earlier rise. For DE, ES, FI, and FR, a slight rise follows the earlier decrease. Overall, we therefore see a substantial degree of heterogeneity in the trajectories over time, but there is still a good majority of observations that have experienced an increase in spread for the first period. It might be noteworthy that the highest single increase occurred in Greece, the country that was hit first and arguably hardest by the financial crisis. In sum, the visual interpretation of the results does not contradict the notion that the crisis may have led to a moderate shift of the balance between close and universal solidarity, but this shift would be due more to an increase in close solidarity than to a decrease in universal solidarity. More robust interpretations will, however, require that we take additional factors into account that may overlay, or also correlate with, changes over time.

Fig. 7.3

Difference in levels between close and universal solidarity, over three EVS waves

7 Exploring the Relation of External Factors to Societal Solidarity Levels

If we want to arrive at possible interpretations of the above comparative differences and trends, analyses of potentially correlated variables are necessary. We have done this by a set of rather simple multivariate regression analyses, which allow us to control for ‘confounding’ and joint effects of the relevant external indicators. Only by simultaneously looking at a set of possible correlating variables for our social solidarity types can we disentangle the relative contributions of each of these other variables, net of the other variables.

In view of the introductory discussion and the overall topic of this book, the most relevant indicators besides the time trend relate to the religious composition of the societies, to the economic crisis following the banking crisis that might have put solidarity demands under stress, and to the migration crisis, as indicated by sudden refugee inflows and increasing migrant shares. Further to that, we intend to control for the European socio-cultural region that a country is part of, to check for effects of a joint societal and normative development, and common political traditions and institutional setups, thereby filling the previous description at the country level with some substance. Below, we describe the measures used for each of these potentially relevant areas in more detail.

Religious Composition

We consider three interrelated aspects of religious composition. First, we look at the shares of the most frequent religious denominations/beliefs (Catholic, Protestant, Orthodox, Other (incl. Islam), No denomination) in the population, which we computed directly from the EVS sample data of each wave (European Values Study 2021). This allows us to check for possible effects of particular religious traditions, as far as these are still visible in the distribution of self-reported memberships. Second, we use an index of religious diversity, in short RDI (Pew Research Center 2014), to check whether the presence or absence of religious homogeneity has any effect on prevalent solidarity attitudes in a society. Although the aspect of diversity is technically implied in having the full set of denomination shares, using the RDI as a single measure allows us to separate the diversity aspect from any effects of a particular denominational category being dominant over others. This index relates to data collected by Pew Research from multiple sources with a reference year of 2010, and we apply it to all our cases regardless of time point, under the assumption that religious diversity and its effects do not change on the scale of single decades (Hackett et al. 2012). The fact that it is based on official statistics as well as on survey data should give this index high reliability. Finally, we also tested a measure of ‘active’ religious commitment, using the reported attendance at religious services. For this, we aggregated the reported attendance to a societal mean value of expected weekly attendance from the EVS waves (EVS Trend File: European Values Study 2021), following the methodology laid out in Biolcati et al. (2020). However, we refrained from including this measure into the final models reported below because its effects were largely redundant with the share of Catholics in each country, and its presence in the regression models thus reduced the precision of the individual regression coefficients.

Economic Crisis

As a measure of the economic stress on the overall society, as well as a proxy for individual-level economic insecurity, we draw on the standard measure of the unemployment rate. We use unemployment rates as reported by the World Bank of the year preceding the respective EVS to tap into the publicly perceived state of economic affairs, not the most recent state, which might not have formed the public discourse yet (World Bank 2020b).

Migration Crisis

As measures of possible ‘cultural’ stress on any national in-group identification bases of solidarity, we used two indicators: one is of migrants in the overall population (World Bank 2020a), which accounts for all residents not born in the country of residence, regardless of when and why they migrated to that country. As this is available only in five-year intervals at full and half decades, we used the most proximal values for each EVS wave, that is, data from 2000, 2010, and 2015. The other is the proportion of new asylum applications in the overall population, which, in contrast to the previous indicator, would allow us to capture any short-term effects of the refugee crisis. This data was retrieved from UNHCR (2020), again with the non-annual data series (for 2000, 2007, 2016) being selected per proximity to EVS waves. The raw asylum application counts were converted to shares of the host country population count. This variable has no information for Belgium, Northern Ireland, Republic of Ireland, Serbia, Ukraine, and Turkey, which were therefore excluded from our analyses.

Socio-cultural Region

The five regions (Table 7.4) we distinguish are constructed by a mix of geographic proximity and political/cultural history criteria that might have their own influence on top of the separate country-level indicators described before, for example, in relation to a country’s ‘welfare regime’ (Esping-Andersen 1990). The ‘Northern’ region comprises the Nordic countries, which are characterised by economic wealth, a high degree of political stability, a generous and universalistic welfare state regime, a Protestant religious background, and an early advent of secularisation. The ‘Western’ region comprises the Northwestern and Central European countries, which have a long history of industrialisation, had been most heavily affected by the Second World War, are relatively heterogeneous in terms of Christian denominations, have rather stratified welfare regimes, and were only more recently subjected to intensive secularisation. The ‘Southern’ region comprises the mostly Catholic or Orthodox countries around the Mediterranean (plus Portugal) that were not under Communist rule, but that nevertheless partly saw major political unrest and late democratisation after the Second World War, and which were not early adopters of industrialisation, fostering a rather conservative welfare regime. The ‘ex-communist’ region is heterogeneous in regard to traditional denominations, and with Albania even contains a country with Muslim majority, but the whole region had been subjected to the enforced secularisation of communist rule. Economic wealth is still weaker than in the aforementioned regions, and despite major internal and external (and violent) conflicts, in particular among the Balkan countries, the majority of the countries have achieved minimal stability of their polity. The ‘ex-Soviet’ region, finally, comprises the countries that were formerly part of the USSR and were thus most strongly subjected to the norms and restrictions of Soviet rule. Despite that, they later took different economic and political trajectories, with the three Baltic countries moving economically and politically towards the West, whereas most of the other countries of this group, including Russia, have experienced and/or are still experiencing violent external conflicts. Because of limited economic resources, welfare regimes in these countries are not generous despite the universalist Communist tradition.

Table 7.4 Socio-cultural regions

To add detail beyond the region variable, which implicitly also encodes different institutionalised approaches to providing solidarity, we have also considered a measure of welfare-based redistribution. For this, we used the percentage of the GDP that was redistributed as social benefits to the households, as reported by the OECD (2020). However, this measure was not available for so many countries in our sample that we decided to drop the variable from the final models rather than drop the affected cases, having observed that the associations for the available cases were rather modest.

Finally, we have included the duration of EU membership for the EU members, to indicate integration with the ‘European’ country community. Non-EU members have been coded with a value of ‘0’ on this variable.

A number of the variables described above represent strongly overlapping information, and they are also correlated with other relevant variables such as the Gross Domestic Product per capita that would often be included in such analyses, but which we decided to omit for the sake of parsimony, as it is again related to the region, the duration of EU membership, and the unemployment variables. The correlations between those variables that we do include are mostly below the threshold of r ≤ .5, with the exceptions of RDI and the population share with ‘no denomination/belief’ (.62), and of the share of Catholics and the church attendance average (.67). Our final models included 92 valid observations with full data on all variables, from 34 countries.

Using ‘close solidarity’, ‘universal solidarity’, and the ‘differences by scope’ as dependent variables, we can also gain an impression of the differences between the association profiles among these three aspects of societal solidarity. In particular, the graphical presentations shown above have suggested rather stable level differences between certain countries. Are these related to the socio-cultural history clusters, or to the religious composition of societies? Further, we have observed a modest rise in levels of solidarity for a number of countries, with close solidarity increasing somewhat more than universal solidarity. Can this be traced back to any of the economic or cultural crisis indicators? Table 7.5 shows the core results of the respective regression models.

Table 7.5 Multiple regression results for three aspects of solidarity

Beginning with ‘close solidarity’ and the change over time, we find that indeed the societal levels of close solidarity have increased between the 1999 and 2017 waves, and even in rather constant steps. Per convention, standardised regression coefficients (‘Beta’) above a magnitude of .1 are treated as substantial, and we see a Beta of .219 for the second versus the first wave and a near duplication of that (.397) for the third versus the first wave. For the region variable, where our model specification gives differences of four non-Northern regions versus the Northern region, we find that ‘close solidarity’ is least important in the Northern region, as all the coefficients for the other regions are positive, with the Western countries giving the relatively highest priority to close solidarity, and the ex-communist countries only showing a negligible difference to the Northern countries. EU membership duration has no discernible relationship with ‘close solidarity’, after controlling for the other variables.

Our three crisis indicators – unemployment rates, asylum applications, and migrant share – all have positive associations with the levels of close solidarity. This is despite the time trend (which is moving in parallel to the crises developments and thus likely absorbing some of the effect of the substantive variables!) also having substantial associations. Going by the standardised coefficients, the economic indicator and the sum of the two foreigner-related indicators have a roughly equivalent weight.

Lastly, the religious composition indicators (denomination shares and RDI) have mixed associations with close solidarity. Religious diversity has a strong negative association with close solidarity, which would indicate that the members of more religiously homogeneous societies have a higher subjective motivation for internal solidarity. This effect is present independent of the specific content of the religious beliefs, that is, even very secular societies, if homogeneous in their secularity, should on average have higher solidarity levels than less homogeneous but more religious societies. The beliefs or denomination shares as such have only minor or no associations here with the share of Catholics as reference category, with only the share of Orthodox in the population making a difference beyond the .1 threshold.

Moving on to ‘universal solidarity’, we find that the time trend is almost absent here, with only a moderate increase for the last EVS wave. In contrast, the regional differences are much more pronounced than for close solidarity, with both the ex-communist and the ex-Soviet countries being markedly below the universal solidarity levels of the remaining regions. We stress again that we intentionally removed the ‘concerned with Europeans’ item from the universal solidarity index to reduce a possible conflation of the index with attitudes towards the European Union, which often would not be a positive identification object in many of the Eastern countries, and recently even for some of the EU members in that region. We conclude that populations in the two Eastern regions do indeed have on average lower ‘universal solidarity’ levels, whereas their levels of ‘close solidarity’ are on par with those of the Northern and Southern regions.

For the crisis-related indicators, we observe that universal solidarity apparently moves in the same direction as close solidarity in relation to unemployment rates, whereas the migration/refugee indicators show no association with universal solidarity.

The religious composition has, in contrast to ‘close solidarity’, markedly stronger associations with ‘universal solidarity’. Notably, it is especially higher Protestant shares that are associated with a higher preference of the overall populations for universal solidarity (although it would not necessarily be the Protestants among those populations who foster that attitude most strongly), but societies with higher Orthodox and higher unaffiliated population shares also display higher levels of universal solidarity.

The right-most columns of the regression results in Table 7.5 reflect the contrast between close and universal solidarity in a more concise manner, and therefore the regression coefficients also reproduce the differences between the coefficients for the two components of the scope spread measure. Our simple regression model accounts for the variation in differences between the two solidarity measures even better than for the variations in the absolute levels of the two component measures, with an adjusted R square value of .434 indicating that 43% of the variance in the country-level scope difference data can be related to our indicators. In summary, we observe that in the two more recent EVS waves, the edge of close solidarity over universal solidarity is positive but remains constant; that the Northern and Southern countries have no preference for close solidarity over universal solidarity, while the other regions have a clear preference for close solidarity; that the only ‘crisis indicator’ with a differential relation to solidarity is the asylum applications, but more asylum applications accompany a stronger preference for close solidarity; and that, among the denominations, the more Catholic and more Orthodox societies prefer close over universal solidarity.

8 Discussion and Conclusion

8.1 Limitations of Our Study

Before we proceed to offer some speculative interpretations, it is important to remind readers of the limitations of our study. The first and perhaps central limitation is that we started our study on a surprisingly blank slate, in terms of established knowledge about levels of attitudinal solidarity in European populations, and of the potential causes of between-country differences and changes of such levels over time. Not yet having a body of descriptive core observations and well-developed hypotheses as a starting ground has led us to follow a very explorative approach. Our methodology was geared more at generating hypotheses rather than at testing them. This leads to the second limitation, which is that even for the regression analyses presented, we cannot and do not claim to have provided rigorous tests for any of the substantive statements made in this chapter. Also, for achieving a better understanding of the actual cognitive and emotive processes at work in the respondents, it is desirable to include the individual level in the analyses. We will hint at several relevant individual-level hypotheses in the next sub-section, which would mandate us to conduct multi-level regressions for any future analyses. Such multi-level models would also provide opportunities for a more statistically appropriate specification of the time-related dependencies in our repeated cross-sectional data. Finally, the available set of country cases in our data is on the one hand rather comprehensive in terms of covering Europe, but on the other hand neither provides complete coverage nor is the product of a controlled random sampling process, which makes systematic inference from our results to the whole ‘country population’, that is, to all European countries, an uncertain endeavour.

8.2 A Summary and Some Possible Interpretations

The first core observation was that a plausible association pattern for both of our solidarity indices with different types of geographical identification exists, which on the one hand underlines that the distinction of close versus universal scopes of solidarity is analytically meaningful, and on the other hand demonstrates that solidarity and mass identities are indeed interrelated. Second, we observed that the levels of close solidarity are higher than the levels of universal solidarity in most European countries. It is not unlikely that the positive relationship between solidarity and identification also explains why close solidarity is usually more pronounced than universal solidarity, considering that narrower references for identification are overall much more prevalent than wider references.

Further, we have found systematic variation in the levels of solidarity over the European regions, with the Northern and ex-communist regions having the lowest levels of close solidarity, whereas the levels of universal solidarity are lowest in the ex-communist and ex-Soviet countries. These regional differences remain after controlling for the differential exposition of the countries to some economic, social, and religious factors with their own relationship to solidarity levels, which gives credibility to the notion that solidarity levels are also engrained in the institutions of a society, and/or in long-lasting socialisation patterns that differ between societies. Obviously, we cannot offer a specification of what exactly explains these regional level differences, and it could easily be that very different, region-specific processes have led to such similar solidarity levels. For example, following the reasoning of van Oorschot (2006), the Northern countries might have lower close solidarity levels because of their more generous welfare systems, which ‘substitute’ personal-level solidarity with institutionally provided solidarity, whereas the ex-communist countries, employing different welfare systems, might have arrived at solidarity levels of about the same magnitude because of different in-group identification levels, or because of different values in other realms.

Of special relevance for the focus of this volume is the observation that the religious profiles of our societies – that is, their dominant denominations or confessions – are not the dominant drivers for differences in close solidarity levels, while they may have a certain role in shaping universal solidarity attitudes (VanHeuvelen and Robinson 2017). A puzzling aspect of that role is the difference between the more Catholic societies on the one hand and the societies with higher shares of other Christian denominations plus unaffiliated on the other hand. We observe less universal solidarity in more Catholic countries, without being able to offer an explanation for that.

Another, apparently at least as relevant, aspect of religious composition is its diversity within a country. High diversity has a very distinct negative effect on both forms of attitudinal solidarity, meaning that, in our sample, societies with a more homogeneous religious belief composition also muster higher levels of solidarity. This is interesting also because it runs in the opposite direction of what Janmaat and Braun (2009) found for the association between ethnic diversity and their measure of solidarity in earlier EVS data.⁵ In conjunction with the earlier observations about the relationship between solidarity and subjective identification, this could indicate that the main effects of religious belonging on solidarity are mediated through religion working as an identity marker, with higher religious diversity possibly creating too many perceived internal group divisions for sustaining solidarity directed at in-groups. However, that hypothesis would imply that the negative effect of diversity on solidarity was mainly present for close solidarity and clearly weaker or even absent for universal solidarity, because universal solidarity should be much less sensitive to an identity-based in-group/out-group discrimination. In contrast to the latter expectation, our data still show a quite substantial association of diversity with universal solidarity, although it is indeed weaker than the association with close solidarity. Again, we must note that such hypotheses would be more appropriately tested with individual-level data.

Finally, one core insight of our study is that levels of attitudinal solidarity have not been generally decreasing in the European countries that the EVS surveyed, despite fears that the successive crises had induced such an effect. If changes occurred in the picture of general stability within most countries, those changes were more often increases than decreases in solidarity. The positive association of close solidarity attitudes with our indicators of economic problems and of ‘cultural stress’ per migration inflows then speaks against the suspicion that enduring crises would exhaust the populations’ overall readiness to offer solidarity to others in their own society; rather, the opposite appears to be the case. And even for universal solidarity, which can also be read as solidarity with out-groups, there is no negative crisis effect. Two alternative explanations for the modest increases in expressed solidarity attitudes come to mind. One is that the normative preference for solidarity is increasing over time, that is, solidarity as a ‘moral value’ could have become more important, thus following the trend claimed by some values researchers that values in most contemporary societies are slowly moving towards more benevolent, more universalist, and less materialist orientations along with ever increasing material affluence (Inglehart 1997; Inglehart and Welzel 2005). The other, and quite contrasting, explanation for the increase in solidarity attitudes would be that respondents were in fact expressing a higher concern for others because they perceive such concern to be more justified in times of crisis (see Janmaat and Braun 2009 for a similar line of reasoning in relation to the wealth of countries). Although the positive association between crisis indicators and close solidarity attitudes does not disprove the former interpretation, that association is certainly more in line with the latter interpretation, where the rising ‘concern’ for others would then be a reaction to worsening conditions in the societies. But again, whereas our exploratory analyses have helped us formulate this question, they were not designed to settle it, and more detailed, individual-level analyses that we could not offer in the limited space of this chapter would be needed to better understand the actual process of change in solidarity attitudes.

And even more caveats are in order here. As we stressed above, attitudinal solidarity cannot be equated with enacted, practical solidarity, which therefore may follow very different pathways and may thus still have decreased in reaction to the crises. Also, note that the positive population average association between crisis indicators and subjective solidarity levels would still allow that some smaller segments of the population do experience negative crisis effects with decreasing attitudinal solidarity, and it might be those smaller segments that drive the public discourse and the relative success of populist parties. This might then be triggering reactions in institutionalised solidarity, starting a downward cycle with feedback to individual-level attitudes. Such delayed reactions to the crises are not likely to be visible yet in the European data at large, but the notion might merit closer study of the solidarity trajectories in such countries that have already experienced distinctively populist politicians in leading offices.

8.3 Conclusion

Our study is a first attempt at describing prevailing levels of attitudinal solidarity in Europe. We have found plausible patterns and interesting trajectories. On the one hand, knowing the country-level differences and trends in solidarity attitudes as described here can already be useful in understanding, for example, policy developments in those countries. In that sense, we have already produced a new set of helpful macro data. We have also begun to sketch some speculative hypotheses about what might explain these patterns. With attitudinal solidarity being a construct that is, per its definition as an attitude, necessarily founded in humans’ minds, most of these hypotheses look at individual-level cognitive and affective processes. Therefore, we believe that, on the other hand, the next steps in future research should be to account for these individual-level processes in much more detail than we could achieve in this explorative study, so that the social meaning of the country-level aggregates can be better understood.

Appendix (Figs. 7.4 and 7.5)

Fig. 7.4

LCA response profile for geographical identification (probabilities for responding ‘very close’ or ‘close’, given that a respondent was assigned to class x)

Fig. 7.5

Estimated shares of the classes per country, sorted by size of ‘All-Close’ class