Abstract
Previous studies have found that elites differ from the general population not only in terms of their demographic and social composition but also with respect to policy preferences and attitudes. When it comes to civil society organisations, the issue of representation and representativeness of top-level leadership is of particular relevance, since a small number of individuals are expected to act on behalf of groups and causes that are not well acknowledged or even suppressed by other societal actors. Using two survey studies, this study compares the demographic composition and attitudes of the top-level leaders of the most resourceful civil society organisations in Sweden with those of the general population. We find that civil society elites are not demographically representative of the general population and that they also differ in attitudes when it comes to levels of trust, political orientations, satisfaction with democracy, and acceptance of income inequality. What is more interesting, however, is that the differences in attitudes cannot be fully explained by the skewed demographic characteristics of the civil society leaders, indicating that there might be distinctive sets of values and attitudes underpinning a ‘civil society ethos’.
Introduction
Most perspectives on societal elites – defined as ‘those who have vastly disproportionate control over or access to a resource’ (Khan, 2012: 362) – acknowledge that elites control important resources and influence decisions affecting the masses. The observation that we can identify small groups that meet this definition of an ‘elite’ in most societal domains raises an important question: how well does the elite represent the masses? Studies of elites repeatedly find that elites differ from the general population, both in demographic and social composition (Bihagen et al., 2013; Göransson, 2007; Mills, 1959; Zweigenhaft and Domhoff, 1998) and in their policy preferences and attitudes (Dellmuth et al., 2022; Kertzer, 2020; Müller et al., 2012). Reproduction of elites that are not representative of the general population in their social composition and attitudes might maintain or even aggravate unequal access to resources and power positions. Therefore, the presence of an elite-mass gap raises critical issues such as persisting inequality and the problem of democratic accountability (Vogel et al., 2019).
This article focusses on one particular leadership stratum which, until recently, has received scant scholarly attention in research on elites: that is, civil society elites (Meeuwisse and Johansson, 2024). Although the leaders of civil society organisations do not have the same electoral relationships with the general population as political elites do, it has been empirically observed that civil society organisations strive to establish claims of representativeness whereby it is assumed that it is desirable for civil society leaders to reflect the population at large (cf. Meeuwisse and Scaramuzzino, 2023), or for the leadership to at least be representative of the groups whose interests civil society organisations defend (Johansson and Lee, 2014). In fact, while some civil society organisations, such as interest groups, have a clear constituency with potentially clear policy preferences, many others – for instance, solidarity organisations, sports associations, environmental organisations or international aid organisations – tend to have a less well-defined constituency. Hence, at the aggregate level, it is relevant to explore the extent to which the leadership of the civil society sector differs from the general population in both demographic composition and attitudes.
Drawing on elite studies, the gap between the elite and the masses can be problematised in terms of the elite being eager to defend their power position by resisting societal changes that would lead to a more even distribution of resources and more egalitarian outcomes (see e.g. Michels, 2001). In this sense, a small gap between the elite and the masses would be considered an indication of a well-functioning democracy. Civil society studies, however, emphasise that the sector is driven by specific values that are linked to the organisations’ mission (Anheier, 2005). These values can be grounded in political, ethical, religious or other convictions, and are often linked to social movements that have been institutionalised into civil society organisations. From this perspective, a gap between civil society elites and the general population would in itself be seen as an indication of a well-functioning democracy, whereby civil society has a role as a critical voice and as a counterbalancing force to prevailing power structures in society.
Nordic welfare model and corporatism
Following Guldbrandsen and others (Edling et al., 2014; Gulbrandsen, 2019), we argue that the Nordic context brings to the fore the potential conflict between the existence of social elites on the one hand, and issues of representativity, transparency, influence, and so forth – all of which are at the core of democracy – on the other. Nordic countries are particularly interesting in this perspective since they are commonly held to be among the most egalitarian states in the world, with advanced systems of redistribution and broad support for social, political, and economic equality. While elites have been and are driving inequality in the larger picture (Khan, 2012), this has not been the case in the modern history of Nordic countries. The Nordic welfare model came about through a series of important pragmatic and corporatist ‘elite compromises’ (Gulbrandsen, 2019), which led elites in Nordic liberal democracies to focus on ‘equality of outcomes’, rather than ‘equality of opportunity’ (Verba et al., 1987). One could argue that the Nordic corporatist decision-making system provides elites with experiences that foster trust in institutions (Gulbrandsen, 2019).
At the same time, the Nordic model developed largely during the first half of the 20th century, and the last decades have arguably been marked by important structural changes. Changes in globalised markets, technological advances, and migration patterns have put the model under pressure, and elite support for the model should perhaps no longer be taken for granted (Gulbrandsen, 2019). Along the same line of reasoning, the gap between elite attitudes and popular attitudes might be growing. The fact that the egalitarian outcomes of Nordic institutions have weakened over the last couple of decades, not least with respect to growing economic inequality (Hansen and Toft, 2021; The Lancet, 2023), indicates that the material conditions for societal consensus on certain values and attitudes have been eroded.
Moreover, corporatist institutions and the importance of institutionalised arenas for decision-making procedures such as the so-called remiss system have weakened in recent decades, in particular in Sweden. At the same time, lobbying practices and the role of non-elected policy professionals and think-tanks have become increasingly important (Åberg, Einarsson and Reuter, 2021; Lindvall and Sebring, 2006; Svallfors, 2016, 2020). Organised interests however still play an important role in policy-making, and recent interest group studies show that the alignment of public opinion and organised interests matters for the outcome of policy processes (Rasmussen et al., 2021).
Our focus on the Swedish context has a significant bearing on the debates revolving around the democratic problems with elites. Civil society organisations in Sweden have played important roles in what can be described as the ‘Nordic welfare model’, characterised by entrenched distributive institutions with egalitarian goals and a corporatist decision-making tradition, with prominent roles historically played by popular movement organisations (Lundström and Svedberg, 2003; Trägårdh, 2010). At the same time, the Swedish welfare model has undergone a range of changes since the 1990s (Blomqvist and Palme, 2020), including marketisation trends whereby civil society organisations take on more of a service-provision role alongside that of for-profit organisations (Petersen and Hjelmar, 2014). Scholars have discussed the potential shift in the expectations of civil society organisations, from focussing on their advocacy function of representing groups’ interests and community building to focussing on their role as service providers, often on behalf of public authorities (Amnå, 2006; Arvidson et al., 2018; Lundberg, 2012, 2020). The role of civil society organisations has also evolved greatly due to changing patterns of civic participation (Henriksen et al., 2019) as well as increased Europeanisation of civil society (Meeuwisse and Scaramuzzino, 2019; Olsson et al., 2009).
Our study aims to provide new insights and understanding of the changing role of civil society organisations in Sweden, by means of a novel approach focussing on the leadership strata of the most resource-rich civil society organisations. By understanding the social composition and the attitudes of the top-level leaders of the most resourceful civil society organisations, we shed light on what might constitute a ‘civil society ethos’ in contemporary Swedish society and to what extent it conforms to attitudes in the general population.
Differences in political attitudes between elites and the general public
A reoccurring theoretical assumption and empirical result in the literature is that political attitudes differ between elites and the general public (Best, 2012; Blomberg et al., 2017; Teney and Helbling, 2014). For instance, in Western democracies, elites have tended to be more liberal than the population (Holsti, 2004; McAllister, 1991), and for the EU it has been shown that elites show stronger support for European integration than the population (Best, 2012). Research into the explanations of such differences suggests that they are driven not primarily by social class, education, or other demographic factors, but by intra- and inter-elite bonding (Higley and Burton, 2006). However, the fact that there is a certain degree of elite homogeneity regarding political attitudes should not disguise the fact that there is also variation in attitudes among elites. For instance, a recent study of the elites in Brazil highlights the importance of how elites perceive the causes of inequalities when explaining variations in redistributive preferences among elites (López et al., 2022). Similarly, a study of Norwegian elites found that perceptions of the causes behind demographically skewed compositions of societal leadership groups explain the extent to which the elites endorse proactive measures to enhance gender and ethnic diversity (Teigen et al., 2022). Another study of Norwegian elites (Gulbrandsen, 2019), of closer relevance to our present study, shows that there is a lack of elite consensus both across time and between different elite groups when it comes to support for reducing economic inequality; cultural and academic elites stand out in this respect, as critical of increasing inequality.
In this article, we focus on an analysis of the differences in the demographic and social composition and attitudes of the top-level leaders of the most resourceful civil society organisations – civil society elites (Johansson and Uhlin, 2020) – from those of the general population in Sweden. We depart from the widely accepted and normatively prescribed logos of the field; namely, the conception of civil society as a societal sphere representing the plurality of voices, groups, and interests based on voluntarily organised collective actions serving liberal democratic polities (Diamond, 1994; Keane, 2009, 2013). From this conceptualisation follows an expectation about representativity, both in terms of civil society leaders’ demographic composition and in terms of their political attitudes in relation to those of the general population. This Tocquevillian perspective draws on the view of civil society as a counterbalance to the state and can be contrasted with a Hegelian perspective, wherein civil society would rather represent particular interests in society against the state, supposedly the guarantor of the general interest (Trägårdh, 2007). From the Hegelian perspective follows in fact an expectation that civil society leaders would represent the specific groups or values around which civil society mobilises.
There are multiple understandings of what constitutes civil society. We define civil society as ‘a societal sphere separate from the state, the market, and the family which is populated by a variety of collective actors, including organisations, networks, and movements (Lindellee and Scaramuzzino, 2020: 86).’ A considerable fraction of this sphere is organised into formal organisations, complete with administration and management, and it is in such civil society organisations that we identify a group that we call ‘civil society elites’.
A previous comparative study of elite groups in Sweden (Göransson, 2007) showed that Swedish civil society elites are more diverse than other elite groups when it comes to sociocultural composition (Jonsson and Lagerlöf Nilsson, 2007). Like other elite groups, civil society elites are characterised by a strong representation of individuals with higher education and a low representation of women in top positions. This image is partly challenged by more recent studies (Scaramuzzino 2020), which show a more even distribution between men and women in top positions of civil society organisations.
We argue for the relevance of studying the descriptive representation of civil society leaders in relation to the general population mainly from two perspectives. Descriptive representation here refers to the extent to which people who claim representativeness look alike to those represented or the extent to which they share similar experiences (Mansbridge, 2003; Pitkin, 1967).
First, civil society organisations protect and promote societal interests that are not well-acknowledged or are perhaps even suppressed by the ruling elite (Etzioni-Halevy, 2001). This means that civil society leaders claim representativeness based on their authenticity in their intent to speak for others – or for the common good of society – without necessarily sharing the experiences of the oppressed or being part of the group whose voices are marginalised.
Second, many civil society organisations operating in liberal democracies subscribe to democratic ideals, whereby they endorse pluralistic values such as equal opportunities for participation and accountability of those who are in decision-making positions. From this perspective, ensuring that the leadership positions of civil society organisations are open and accessible for all societal groups in equal manners becomes of relevance (cf. Anheier, 2005). This however might also mean that civil society leaders are expected to embody the values and ideals of the organisations they lead rather than the general public opinion as such.
We aim to answer the following research questions:
To what extent do the leaders of resourceful civil society organisations reflect the general population in Sweden in their demographic and social composition and attitudes related to key political and social issues? How can we understand the similarities and differences between civil society leaders and the general population?
The next section describes the methodological approach taken in identifying the civil society elites and for conducting the survey study which has yielded the main data for our empirical analysis. The analysis section focusses on answering the research questions by means of comparison between the civil society leaders and the general population. We conclude the paper with discussions about possible explanations revolving around the concept of civil society ethos.
Methods and data
Interpreting elitism as a consequence of people occupying specific powerful positions is common in elite research. Since the work of Robert Michels (2001) on the ‘oligarchical tendencies of modern democracy’, many strands of elite theory have related elitism to the occupation of top positions in organisations. It was, according to Michels, within the complex, large-scale organisations of the labour movement that individuals monopolised and hoarded organisational means and resources, producing an internal stratification that distanced the leaders from the masses. The so-called ‘positional method’ in elite research is based on a similar assumption, that is that resources and influence are largely tied to positions of leadership in organisations of national relevance (Hoffmann-Lange, 2007), and that elites are those who can exert influence through their strategic positions in powerful organisations (Higley and Burton, 2006). Accordingly, business elites, for instance, have been identified and mapped based on the positions they hold in large business organisations (Bühlmann et al., 2012; Grau Larsen and Ellersgaard, 2018), while political elites have been identified based on the political positions they occupy, for example, in government or parliament (Best and Higley, 2018).
Our study of civil society leaders departs from this positional method in the identification of civil society elites. At the same time, we draw on a novel methodological approach (Scaramuzzino and Lee, 2024) for mapping civil society actors based on their resources and addressing resource stratification in civil society: the Multi-dimensional Measure of Resource Stratification in Civil Society (MMRSC). Our contribution lies in acknowledging the diversity in types of resources that are of importance for the civil society sector, going beyond the economic resources which many studies have focussed on (e.g. Beyers and Kerremans, 2012), and also beyond the state-centred understanding of power (e.g. Binderkrantz and Rasmussen, 2015). By distinguishing two dimensions of economic and political resources (i.e. resources that are internal to civil society and resources that are external to civil society – see next section for more details), we employ a more sophisticated method of mapping civil society organisations. Another contribution of our method concerns our point of departure, in understanding civil society as a sphere in which resources are unequally distributed. Previous scholarly works mapping the civil society sector across countries (e.g. Salamon et al., 2004) have not been attentive to the issue of resource stratification. By addressing resource stratification in terms of unequal distribution of resources within civil society, we pay attention to the most powerful actors that control a disproportionate share of resources and can therefore be considered a corps of elite organisations in civil society.
Identifying civil society elites
To identify resourceful civil society organisations at the national level in Sweden, we have made use of five indicators of resources, set as preconditions for entering our database of organisations. These conditions have been operationalised in the following way (see Scaramuzzino and Lee 2024 for a description of the methodology) (Table 1).
Indicators used in mapping of resourceful civil society organisations (CSOs) in Sweden.
Our choice of indicators is informed by theories widely used in previous research on interest groups and social movements, as well as insights from civil society research. Theories of resource mobilisation (McCarthy and Zald, 1977) are crucial in understanding the type of resources that provide organisations with the capacity to act – in the case of civil society organisations, members, employees, volunteers, and so on. The main tenets of theories of political opportunity structures (Kitschelt, 1986) are also central to our understanding of political resources. We assume that the extent to which civil society organisations have access to institutionalised structures where they can influence political decisions is an important indicator of political resource external to civil society. Having access to policy-making processes granted by public authorities relates to the notion of inside advocacy strategies (Beyers, 2008). This kind of political resource can be relevant either within a specific policy area, or for the civil society sector as a whole – concerning, for instance, policies on state-civil society relations (Scaramuzzino and Lee, 2024). When it comes to operationalising the political resources internal to civil society, we consider the status of umbrella organisations or meta-organisations (Ahrne and Brunsson, 2005) as part of the internal governance of the sector. In this sense, access to decision-making within these organisations is an important resource, as they tend to act on behalf of the sector as a whole or on behalf of a broad spectrum of organisations within certain policy areas (for instance, international aid).
While identifying civil societal organisations using these indicators, we assigned ‘elite scores’ to each organisation based on how many conditions for inclusion a given organisation fulfils. If an organisation entered our pool of the most resourceful CSOs by fulfilling a condition for one indicator, the organisation received a score of 1. If an organisation entered our pool by fulfilling the conditions for three indicators, the organisation received a score of 3, and so on. The elite score thus allows us not only to map out CSOs but also to discern the patterns of resource concentration and stratification.
This method of mapping civil society organisations based on the multi-dimensional understanding of resources is based on several assumptions. First and foremost, we assume that organisations which are highly formalised, established at the national level, and equipped with executive capacity in a rather traditional sense (for instance, by having a large number of employees) are more likely to occupy influential positions in the field of civil society. Second, we assume that the more a civil society organisation is integrated into the policy-making process initiated by the Swedish state (for instance, being represented in consulting bodies and invited to influence legislative processes), the more likely it is that they can exert political influence. Third, we assume that the more a civil society organisation is recognised and an active part of the internal infrastructure of civil society (for instance, through being represented in umbrella organisations for specific policy areas or for civil society as a sector), the more likely it is that it can gain legitimacy and achieve its goals.
These assumptions imply that the landscape of Swedish civil society organisations, mapped using our method, is populated by organisations having an established position in terms of resources, and recognition in civil society or by the state – while the organisations that are less formalised, which use more contentious forms of advocacy, and have a more conflictual relationship with the state are potentially underrepresented in our sample.
Last but not least, our mapping process focussed on the following policy areas to ensure as comprehensive an approach as possible, but also a comparable approach in mapping the civil society sector in relation to existing large-scale mapping exercises (cf. ‘Johns Hopkins Comparative Non-profit Sector Project’; Salamon et al., 2004): old age, culture, disability, environment, gender equality, human rights and democracy, migration and ethnic groups, religion, solidarity, sports and leisure, and third sector policy. The organisational population in our study of civil society elites thus covers a wide range of issues, going beyond the areas generally dealt with in studies of interest representation and interest groups in a narrow sense. For instance, organisations that work within the policy areas relating to the environment, human rights, solidarity and the civil society sector are likely to pursue the distribution of common goods or realising general social progress for all of society, rather than narrowly focussing on interest representation or advocacy activities serving specifically defined social groups. While some organisations, such as pensioners’ organisations, might have a clearer constituency to represent, that is, their members or by extension all pensioners, many organisations, for instance, environmental organisations do not have specific constituencies but rather act in what they perceive to be the general interest. This characteristic of the organisational population we study therefore motivates comparing civil society elites and the general population in discussing issues of representativeness, rather than comparing leaders and specific groups within the population. Also, the level of anonymity of the survey does not allow us to trace back leaders’ answers to the survey to specific organisations or policy areas.
Survey of civil society leaders
To answer our research questions, we analysed two sets of data: (1) a survey of civil society elites in Sweden carried out in the research programme ‘Civil society elites? Comparing elite composition, reproduction, integration and contestation in European civil societies’ in the spring of 2020 and (2) the European Social Survey for Sweden (ESS round 8, 2016).
The population of our survey study includes individuals holding key leadership positions in 394 resourceful civil society organisations that we identified by means of the indicators discussed in the previous section. The leadership positions that we have chosen are based on the governance structure of the organisations. The identified organisations largely shared a similar governance structure, consisting of a board with several members and a chairperson. Some larger organisations also have one or more vice chairpersons. On the administrative side, we often find an employed executive leader – generally a secretary general or a director, and sometimes one or more deputies. When mapping the leaders, we have therefore examined organisations’ websites for the following positions: chairpersons (or the like), vice-chairpersons (or the like), secretary generals (or the like), vice secretary generals (or the like).
Some leaders in our population had more than one position. When checked for the individuals who appeared more than once in our sample, we ended up with 835 unique leaders. We found 768 leaders’ individual e-mail addresses for sending out the survey. For the remaining 67 leaders, we relied on the organisations’ e-mail addresses and contacted the organisations, asking recipients there to forward the e-mail to their leaders. The study was carried out in the first two months of 2020. After having announced the survey, we sent an invitation to participate with a link to our questionnaire. After a number of reminders, we collected a total of 308 responses. Of these, 300 answers were provided by leaders contacted through individual e-mail addresses (39% response rate), 7 were provided by the leaders contacted through organisational e-mail addresses (10% response rate), and 1 answer was provided by a leader by regular mail. The total response rate of the survey study is 37%. Considering that surveying elite populations generally suffers from low accessibility and low response rates (Walgrave and Jeroen, 2018), our study was successful in achieving a good response rate. What is also important to note is that the survey population is not a randomly selected sample of a wider population, but that the study is best described as a census study.
Below we present a non-response analysis to see to what extent our respondents are representative of the total population concerning key variables. As the survey was anonymous, we could only rely on a few variables for this analysis:
Elite score: the number of conditions which a given organisation fulfilled (we divided our survey population into three subsamples: elite score 1 (1), elite score 2 (2) and elite scores 3–5 (3)). This was only possible with the respondents who provided their answers through personal e-mail or regular mail. Position: we compared our survey population with how the respondents answered the question: ‘Which of the following titles best describes your position in the organisation?’ The respondents, many of whom held multiple positions, might however have considered a position other than the one they were selected through. A few chose ‘other position’ as an option. Gender: we compared our survey population with how the respondents answered the question ‘Please indicate your gender.’ We could not determine the gender of some leaders based on the information available.
The next table compares the distributions in the population and for our respondents concerning these three sets of variables (N and percentages).
As Table 2 shows, our respondents represent the survey population quite well concerning these variables. We have a slightly higher representation of respondents from organisations with higher scores compared to organisations with a score of 1. Concerning the positions, we have a smaller share of vice chairs and a higher representation of secretary generals. When it comes to gender balance, we find almost no difference between the survey population and our respondents.
Non-response analysis.
Source: The Civil Society Elite Survey; European Social Survey Round 8 (2016).
Data and variables
In our analysis of the demographic and social composition of and attitudes held by civil society elites in relation to the general population, we use the data from the survey of civil society leaders described above with the data from the European Social Survey (ESS).
First, we compared the leaders and general population with respect to age, gender, country of birth, place of residence, education, and income to see to what extent the leaders in the sample are representative of the Swedish population regarding their social composition. This inquiry is concerned with the issue of homogeneity in the leadership strata of civil society organisations and whether the inequality based on specific social categories (Tilly, 1999) is reproduced in terms of who are allowed to be in positions of power: in this case, the top-level leadership positions of the most resourceful civil society organisations. The variables age, gender, country of birth, and education are measured with identical items in both surveys, whereas the variables place of residence and income level are constructed differently in the two surveys. In the ESS, place of residence is measured via respondents’ self-declaration concerning what type of municipality they live in. In the survey of civil society leaders, we ask which municipality the respondents live in, which is then coded into different municipality types. In the survey of civil society leaders, income level captures individuals’ income before taxes, while in ESS the survey asks for household income after taxes. We expect the former to be lower than the latter in many cases, and as leaders tend to have higher incomes than the average Swedish household, the differences will probably be underestimated in our analysis. We compare the share of respondents with income above the median values in each data set. Median values are used as average incomes can be skewed due to a few extremely high-income earners.
In the second part of the analysis, we compare the level of social trust, political orientations, satisfaction with democracy, and attitudes towards income inequality of civil society leaders with those of the general Swedish population. These variables are measured with identical items in both surveys and are meaningful indicators due to their implications for the role of civil society. For instance, social trust can be discussed in relation to civic engagement (Putnam, 2001), and it is of interest to find out whether civil society leaders differ from the general population in this respect. Political orientation in terms of the left-right spectrum, as well as distributive preferences regarding income inequality, are interesting with respect to the expected role of civil society leaders to speak for the disadvantaged and marginalised on one hand, and the elite positions they occupy in the civil society sector, characterised by resource stratification, on the other hand. Lastly, it is interesting to evaluate how satisfied civil society leaders are with democracy in comparison to the general population, given that of yet another role expected of civil society actors is for them to act as a counterweight to the state (Baxter et al., 2024).
Finally, we use ordinal regression to regress attitudes for each survey on determinants, and qualitatively compare these patterns of correlation between socioeconomic variables and attitudes for civil society leaders to those of the general population. Although done in other studies of elites and professions (Blomberg et al., 2017; Teney and Helbling, 2014), we chose not to merge the two datasets (our survey data and ESS data), due to the different sampling strategies involved. Instead, we chose to conduct the analyses separately for the two datasets. Throughout the analysis, the ESS data is subject to analysis weighting, as discussed in Kaminska (2020). The focus of our analysis is not on directly comparing the strength of correlations across the two models. We rather compare which independent variables function as predictors of the dependent variable in the two populations. In other words, we are interested in whether attitudes in the two groups are explained by the same factors.
Analysis and results
Are the leaders of the most resourceful civil society organisations in Sweden demographically representative of the general population? The comparison of the two surveys points to the fact that civil society is also a societal sphere where we see unequal access among different social groups to powerful positions in organisations. Table 3 presents a comparison between the civil society leaders and the Swedish population at large in regard to gender, country of birth, place of residence, education, income, and age. The number of respondents varies between 304 and 309 for civil society leaders, and between 1846 and 1857 for the Swedish population in the ESS.
Composition of civil society leaders compared to the Swedish population.
Source: The Civil Society Elite Survey; European Social Survey Round 8 (2016).
The most noticeable differences are that civil society leaders are, to a much greater extent, highly educated (77% compared to 21%) and reside in big cities more often than what is the case in the general population (48% compared to 13%). 1 Considering that the civil society organisations included in our study all operate at the national level and likely have their headquarters in the capital, the tendency regarding place of residence among the leaders is not surprising. The share of leaders with backgrounds in higher education is striking, indicating that educational capital is an important condition for becoming a civil society leader in Sweden. Civil society leaders stand apart from the general population in other respects as well: foreign-born individuals are under-represented among civil society leaders (8% compared to 12% in the population), civil society leaders are on average older (55 years old on average compared to a population average of 48 years), and women are slightly over-represented among the civil society leaders in our sample (58% compared to 50%).
Since organisational leadership usually comes with seniority, it is not surprising that civil society leaders tend to be older. The other finding regarding the gender gap is, however, more surprising in the light of how consistently, over time, women have been the minority in the other elite groups in Sweden – for example within the business elite (Bihagen et al., 2013). This over-representation of women in the civil society elite could be partly explained by most employees in the sector being female (Scaramuzzino 2020) as well as the lower societal status of this elite group compared to for instance business or political elites.
Considering the leadership positions that our study population occupies, especially in today's increasingly professionalised landscape of civil society organisations (Selle et al., 2019), the fact that civil society leaders seem to comprise a societal stratum that is demographically skewed in the aforementioned aspects might not be so surprising. Moreover, considering that the civil society leaders we study are not elected representatives such as parliamentary politicians or high-level civil servants who face responsibilities to and are accountable to the general population, one might argue that our focus on the question of descriptive representation of civil society leaders in relation to the general population is less relevant for broader debates on democratic representation. However, civil society organisations’ and their leaders’ legitimacy relies heavily on non-descriptive claims of representativeness (Saward, 2010) in order to be heard in the public arena through their campaigns, lobbying, media exposure, protests, and so on. Their representative claims are however increasingly scrutinised and questioned in the eyes of the public (Kröger, 2016), which is why empirical mapping of civil society leaders’ demographic and social composition is a valuable piece of information in making sense of the contemporary role of civil society organisations in our democratic processes.
Unpacking civil society leaders’ attitudes: Promoting their own interests or acting as critical change agents?
In the remaining sections, we focus on exploring attitudes held by civil society elites. Due to the normative expectations held of civil society organisations in the liberal democratic system, we consider the possibility that civil society leaders’ attitudes would align with the general population, despite rather poorly representing the general population in terms of demographic and social backgrounds.
We compare four attitudinal items from ESS replicated in the survey of civil society leaders: general trust, political position on the left–right-scale, satisfaction with democracy in Sweden, and acceptance of large income differences. 2 General trust, political position, and satisfaction with democracy are measured on a 10-point scale, while acceptance of income differences is measured on a five-point scale. The mean scores for the two groups are reported in Table 4.
Attitudinal differences between civil society leaders and the general population.
Source: The Civil Society Elite Survey; European Social Survey Round 8 (2016).
Compared to the general population, civil society leaders have higher levels of general trust, tend to lean more to the left politically, are less satisfied with democracy, and are less approving of large income differences to reward talent and effort. For all four items, the differences in the mean values between the two groups are statistically significant on the 95% level or higher, using independent sample t-tests. Therefore, we conclude that civil society leaders differ consistently and significantly in their attitudes compared to the general population in Sweden. The differences are particularly salient with regard to levels of trust and distributive preference.
Next, we separately regress the four attitudinal variables on a set of socioeconomic variables for civil society leaders and the general population. The idea is to compare the patterns of correlations among the civil society leaders (CSE), and the Swedish population (ESS) to determine whether socioeconomic factors predict attitudes in a similar way in the two groups. The dependent (attitudes) and independent (socioeconomic factors) variables are the same as those discussed above. Because attitudes are measured with five- or ten-point Likert scales, we use ordered logit regression to estimate the models. The regression analysis does not serve to explain or predict variation in attitudes, but to explore differences and similarities in correlation patterns between the two groups. That being said, the independent variables have previously been found to be reasonable predictors of the four attitudes that we observed.
From Table 5 we see that in the Swedish population, income level consistently and significantly correlates with all four attitudes. People with an income equal to or higher than the median income tend to have higher general trust, lean more to the right politically, be more satisfied with democracy, and be more accepting of meritocratic income differences. Among civil society leaders, the sizes and the signs of the correlation between the dependent variables and income are about the same as in the population, but they are not statistically significant (in two cases, the estimates are significant at the 10% level). Among civil society leaders, there is also a negative correlation between being female and leaning towards the political right, and a positive correlation between being female and rejecting meritocratic income differences; which, as can be seen, are relationships which both also hold true for the Swedish population. Overall, these are the only correlations that are statistically significant for civil society leaders. However, in the Swedish population, three, four, or five of the six independent variables have statistically significant correlations with our dependent variables, and neither the sizes nor the signs of the correlations are consistently mirrored for civil society leaders. We take the dissimilar correlation patterns revealed in Table 5 as an indication that civil society leaders differ from the Swedish population in what could possibly predict their attitudes. While the dependent and independent variables are correlated in an intuitive fashion for the Swedish population, this is not the case for civil society leaders.
Comparison of patterns of correlation between socioeconomic characteristics and attitudes among the general population and civil society elites in Sweden.
Source: The Civil Society Elite Survey; European Social Survey Round 8 (2016). Note: Ordered logit regression is used for the analyses and the table presents unstandardised coefficients and z-values.
p < 0.05.
p < 0.01.
p < 0.001.
Discussion
Departing from the question of the representativeness of civil society elites, this study empirically explores the demographic and social composition of the top-level civil society leadership of the most resourceful organisations in the Swedish context and their attitudes in relation to those of the general population.
The results show that civil society elites differ both in their composition and attitudes compared to the general population. While variables such as age, gender, country of birth, education, and income level explain variation in the attitudes of the general population, the same variables are correlated to the attitudes of civil society leaders to a lesser extent. This means that the differences in attitudes between the civil society leaders in our sample and the general population in Sweden cannot be solely attributed to the particular demographic profiles of the leaders. In other words, our analysis suggests that civil society leaders do not hold different attitudes compared to those of the general population because they represent a skewed section of the population. The results regarding satisfaction with democracy are illustrative of this. In the general population, we find that people with higher education and higher incomes are more likely to be satisfied with democracy compared to people with lower education and with lower incomes. On the contrary, we find civil society leaders to be slightly less satisfied with democracy than the general population, despite the fact that they represent a highly educated section of the general population, with relatively high incomes.
Our second research question, regarding how we can understand similarities and differences between civil society leaders and the general population, raises another issue: namely, whether the occupancy of leadership positions in civil society organisations itself has an impact on attitude formation. In other words, to what extent can discrepancies in attitudes be explained by an organisational position as a civil society leader? To explore this issue, we tested ordinal logistic regression analysis, including both populations (the ESS sample and civil society leaders) in the same model (one for each dependent variable), and added a variable distinguishing the two populations as a factor. Including different survey populations in the same regression model has some methodological drawbacks, although it has been used in previous studies of elites (Teney and Helbling, 2014) and professional groups (Blomberg et al., 2017). The results concerning socioeconomic factors were very similar to the results of the separate models for the two populations presented in this paper. What is interesting to note is that, concerning all four attitudes, the population-related variable turned out to be significant when controlling for socioeconomic factors. The results tended to confirm the bivariate analysis (see Table 4) showing that the general population is less likely – compared to civil society leaders – to trust other people, more likely to be politically oriented to the right, more likely to be satisfied with democracy, and more likely to accept income inequalities.
These differences might be explained by some other intermediary variables at the individual level (for instance, other unknown psychological dispositions, other attitudes, etc.) or at the organisational level (for instance, the ideological orientation of the organisations that the leaders represent). Alternatively, regardless of these particular factors, one could also ascribe the differences we find to the particular role played by civil society leaders. Historically, Swedish civil society has developed as a part of popular movements and in close connection to the Social Democratic welfare system (Lundström and Svedberg, 2003). They have been part of the People's Home project and are therefore ideologically close to the Social Democratic Party, exercising hegemony for at least four decades (Trägårdh, 2010). Accordingly, it might not be surprising that they are, on average, more leftist and negative regarding income differences.
The fact that civil society leaders have higher levels of general social trust compared to the population, is interesting as high trust is often positively correlated with higher levels of satisfaction with democracy. Civil society leaders, instead, although having high trust, tend to be less satisfied with democracy. This specific attitude of dissatisfaction with democracy could be linked to the fact that many of the organisations that the leaders represent work on issues of social rights and have a critical function regarding the state and public policy (Arvidson et al., 2018).
Conclusion
In conclusion, there seems to be a specific civil society ethos whereby civil society leaders’ attitudes are not in line with those of the general population and partly contradict what would be expected by their socio-demographic profile. Hence, we can conclude that civil society leaders’ attitudes neither tend to represent the general population, nor the well-off, highly educated middle class of which they are part. Their attitudes instead tend to mirror the interests of the groups they are supposed to represent or the more general values they are widely expected to do advocacy for as leaders of civil society organisations. This is very much in line with a Swedish understanding of civil society as a critical voice in society and representing marginalised groups and service users’ interests (Arvidson et al., 2018). Although the civil society elite tends to be socio-demographically a privileged group, they seem to hold the values related to their organisational mission and the sector's societal role.
Swedish civil society has for many decades combined a strong advocacy role with cordial and trustful relations with the state (Trägårdh, 2007). In fact, Swedish civil society has played an important role in the development of the Social Democratic welfare state and corporatist model of governance providing space for realising marginalised groups’ rights. From this perspective, civil society elites seem to hold to the value system informing the Swedish model and their representational role, instead of adapting to a changing political climate in times of welfare state retrenchment, neo-liberal reforms and increasing inequalities. This potential mismatch between civil society elites and the political elites could imply changing relations between state and civil society into more conflictual forms threatening the corporativist tradition.
These conclusions can raise questions about the potential intermediary variables and mechanisms for which our study cannot fully account, such as the role that socialisation processes might play in institutions like higher education or in civil society itself. Civil society leaders’ previous (and current) occupational experiences and network positions might also play a role in what political preferences and values they claim to adhere to. It could also be that the pathways to leadership positions in Swedish civil society organisations shape the involved individuals’ political preferences and attitudes in certain ways. What we conceptualise as a civil society ethos is associated with such mechanisms. We envision, for instance, that the particular patterns of political preferences observed in our survey respondents can be results of internal processes in civil society, and not only results of individual dispositions a priori – be they demographic, social, or political.
Footnotes
Funding
The authors disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This work was supported by the Riksbankens Jubileumsfond (grant number M17-0188:1).
