Who is represented? Exploring the demographic structure of interest group membership. Marcel Hanegraaff, Joost Berkhout, and Jens van der Ploeg

Transcription

1 Who is represented? Exploring the demographic structure of interest group membership. Marcel Hanegraaff, Joost Berkhout, and Jens van der Ploeg Paper prepared for ECPR General Conference September , Oslo Abstract Research shows that people with low socio-economic status or age, women, and people with a migration background are only marginally politically involved. This systematic finding in studies of biased political participation is however only marginally connected to the study of interest groups. This is surprising because the issue-focus and the strength and structure of political organization is likely to affect the unequal representation of group interests before government. Moreover, observed inequalities may be amplified in organizations because politically capable subgroups in the population form the breeding ground for relatively specialized, professionalized and politically resourceful associations. In this paper we explore, first, whether different demographic segments are represented by interest groups in proportion to their respective presence in society. Second, we examine whether groups with different membership bases vary in their ability to effectively organize political action. Our data stems from a large scale interest group survey project in the Netherlands (n=787). We find that distinct groups, especially lower educated, in society seem underrepresented in the group system. However, these inequalities are not systematically strengthened by more vigorous representative organizational capacities of the interest organizations of the relatively active segments of the population. Our results speak to the broader debate on whether or not some demographic groups (based on ethnicity, education, gender) and their issues, are represented proportionately in politics.

2 Introduction The political participation of citizens in interest groups is an important mean to the representation of groups of citizens in the public sphere and before government. But the translation of citizen participation into political organizations is neither automatic nor equal. Political participation is unequally distributed among citizens, and interest groups selectively represent certain political causes. This leads to biased representation by interest groups. Schattschneider (1960, 32) notes that such bias in interest representation is shown by the fact that even non-business organizations reflect an upper-class tendency and large areas of the population appear to be wholly outside the system. He points out that citizen-surveys show that people with low socio-economic status, women and those living in rural areas are hardly ever members of interest associations. Recent surveys show similar results, such as high discrepancies in political participation across citizens with varying levels of education (add references). This systematic finding in studies of biased political participation is however only marginally connected to the study of interest groups. Rather, interest group scholars tend to analyze how bias affects mobilization towards private or public actors, whether policymakers provide biased access towards particular types of interest groups, or whether the influence of groups is evenly distributed or skewed towards certain interests. Bias in membership, nonetheless, has only been studied in a few instances (e.g. Skocpol 2004; Strolovitch, 2006; Schlozman et al 2012). This is surprising because individuals do not come together at random but simultaneously share various characteristics. This implies that groups potentially combine weak and strong citizens (Strolovitch, 2006), combine politically interested with socially engaged individuals (Salisbury, 1969; De Bruyker et al 2017) and create connections between distinct social or economic problems and policy fields. In the extent to which they manage to do 1

3 that, groups could contribute to the creation of social cohesion and political equality, or, social fragmentation and political bias. Studying representation bias seems even more pressing because the few studies we do have on this issue all indicate that deprived citizen groups are far less represented than citizens with a favorable social and economic position (Skocpol 2004; Strolovitch, 2006; Schlozman et al 2012). In this paper we aim to add to this debate in two ways. First, we expand studying the inequality of citizen representation to a Western European context, i.e. the Netherlands. As Skocpol argues, comparing across countries and time is crucial to understand the roots (and potential solutions) to unequal participation and representation of citizens in politics (Skocpol 2004: 12). The Netherlands serves as an excellent case to expand these studies at it is a least likely case to find similar results. First, all in contrast to the US, the Dutch the welfare state is highly developed, the funding system for NGOs by the government is among the most generous in the world, and the interest group community is considered as one of the densest systems on the globe. This means collective action problems are less severe than in the US, hence the chance of more equal representation should be higher. Second, next to the empirical addition, we also have a theoretical contribution to the literature by asking: does variation in membership structure affect the political opportunities of interest groups? As Salisbury notes, differences in political capabilities of potential members directly shape the nature of the offer that interest group leaders can propose to potential members (Salisbury, 1968). That is, high capability individuals should be less likely to be in need of social incentives and more likely to be offered expressive, political incentives in return for membership. This potentially leads the groups with more capable members to be relatively weakly politically organized, while groups with less capable members are better positioned to 2

4 have an impact on political decision-making. We assess the link between participatory membership with the resources of a group, the level of professionalization and the level of specialization. Last, we extend the idea that policy shapes politics and that post-material policies lend themselves better to the mobilization of certain groups from society than other policies. Our data stems from a large scale interest group survey project in the Netherlands (n=787) whereby we asked leaders of membership organizations about the demographic nature of their membership. Given the focus on individuals as members or supporters, we exclude from our analysis all groups that do not have citizens as members, including business interest associations. In what follows we first provide an overview of the literature on unequal representation. We then argue how this might affect organizational and political capacity of interest organizations. This leads to a set of four hypotheses. After describing the research design, we test our hypothesis. We end with some final thoughts and an avenue for new research. Theory: Political participation and political organization While the participatory capabilities of citizens in the interest group literature have only been sporadically studied, it is a mainstream issue in political science. The well-established literature on political participation shows that individuals differ among each other in the degree and nature of political participation (e.g. Smets and van Ham, 2013). First, most classically, this refers to the differences among groups of distinct social-economic status (SES) with low-income groups being less likely to vote, protest and participate politically in other ways. Second, more recently, scholars have also pointed out that civic skills affect this relationship (Brady et al, 1995). Through work experience, voluntary work or other collective social activities, people build 3

5 civic capacity, whichmakes them more likely to participate politically. People with more social capital participate more. These experiences take time to materialize, which leads young people to participate to a lower degree than old people. Third, recent studies indicate that gender and migration background affect the political capabilities of individuals and consequently produce differences in political participation (Marien et al, 2010). Most importantly, all these differences refer to turning up to vote in elections but also include non-conventional means of political participation such as protest, online action, politically-motivated consumer boycotts and membership of interest groups. The distinct pattern of political behavior of individuals in groups is partially translated into political organization. First, as regards political parties, in social-structural accounts of party behavior, distinct groups in society will form political parties or make sure that existing parties recognize their interests. The party system and electoral outcomes will consequently largely reflect the social-economic composition of society (Lipset and Rokkan, 1967; Lipset, 1960). For instance, recent economic changes such as the (global) integration of markets may have created groups of economic losers of globalization who favor welfare-chauvinist policies, voiced by radical right-wing politicians (Kriesi et al, 2012; 2006). However, note that this social structural model is challenged as insensitive to actual political preferences among the public opinion, the organizational difficulties of mobilisation and the structuring qualities of party-political conflict more broadly (e.g. van der Brug and van Spanje, 2009). Second, scholars of social movements such as Edwards and McCarthy (2004) note that middle-class groups remain privileged in their access to many kinds of resources, and, therefore, not surprisingly social movements that resonate with the concerns of relatively privileged social groups predominate. At the same time, they are quick to note that social bases are only part of the story. Movement leadership and 4

6 strategies, and political context, also determine which issues are voiced through movements. Their resource mobilization theory is at the root aimed at better understanding how groups are able to overcome prevailing patterns of resource inequality in their efforts to pursue social change goals. Third, scholars of interest groups were so deeply affected by Olson s work that they mostly considered non-participation as a norm (Olson, 1965; Oliver, 1993). This implies that they largely assumed that there is no, or only limited, direct relationship between the individuals in society whose socio-economic status is potentially affected by policy change on a given policy issue and the number of interest groups on the issue (Lowery et al, 2015). More recently we see a revived interest in the membership structure of interest groups, mostly initiated by Theda Scocpol s seminal work on the inequality of American interest representation (2004). Her expansive work shows that while we have witnessed an increasing growth of civic organizations advocacy since the 1960s and 70s, this has not led to more equality in representation. In her own words: the gains in voice and public leverage have mainly accrued to the top tiers of U.S. society while Americans who are not wealthy or higher educated now have fewer associations representing their values and interests, and enjoy dwindling opportunities for active participation (2004: 12). Building on these findings, Dara Strolovitch highlights that interest groups often prioritize the interests of their most advantaged members, such as male rather than female racial minorities, or affluent rather than poor women. In line, Sclozman and colleagues (2012) demonstrate that American democracy is marred by deeply ingrained and persistent class-based political inequality. The well-educated and affluent are active in many ways to make their voices heard, while the less advantaged are not. Interest groups provide no remedy to this problem. Rather they enhance the problem through biased representation of privileged groups at the expense of more marginalized groups in society. 5

7 We build on these insights in two ways. First, empirically we expand the argument to a West-European context. This is important because, as Scocpol argues (2004: 12), to understand why representation is biased we need to compare mechanisms underlying bias across time and countries. This is especially relevant because United States and Western-European vary a great deal on various important dimensions, such as the development of the welfare state, the way interests mediation is constructed, and the type of incentives governments provide to interest organizations. Second, theoretically we expand Scholzoman s theories on the organization of bias (see also Schattschneider s 1960). We test whether and how unequal representation of marginalized groups affects the organizational and political capacity of groups. To put it differently, we test whether the underrepresentation of groups in politics is enhanced, diminished, or not affected when interest groups enter the equation. The membership of interest groups: from representation to political orientation We build our arguments in two parts (see also figure 1). In the first part we explore whether interest group membership can be seen as a proportional representation of several key demographic groups in society. In the second part we explore the relationship between the membership of organizations and organizational political capacities and orientation. Theoretically, we depart from the idea that individuals possess political resources and that interest groups survive by bringing these resources together for some collective purpose. This may be understood as a mutually beneficial exchange relationship between members and associations (Salisbury, 1969; Berkhout, 2013). Furthermore, we assume that organizations, at least to some extent, consider membership demands and capabilities when deciding on organizational goals and in developing political strategies. 6

8 [INSERT FIGURE 1 AROUND HERE] We start with the first hypothesis, which deals with the coherence between membership of interest organizations and the demographic structure of society. The resource model of political participation assumes that individual political capabilities, usually indicated by time, money and civic skills, largely originate from socio-economic status and are additionally shaped by additional experiences in life such as in work and voluntary groups (Verba et al 1995). We consider the five types of demographic groups that have received the most attention in the literature on political participation. Each structures groups in society into politically more capable and less capable citizens: education, age, gender, migration background, and political orientation. To start, people with higher income have money at their disposal to donate to political causes. But, and this is heavily emphasized in this approach, people develop civic skills in various environments and this provides them with resources that may be employed in public action. Education is very important in this regard, as it leads to higher income and gives people the non-material resources, such as political awareness, critical citizenship and voicing skills, that increase the political capabilities of the higher educated. As already noted in early public opinion studies (e.g. Rosenau, 1961; Almond, 1960; Key, 1961), the attentive public, from which group members are recruited, tends to be relatively stable in composition and comes disproportionately from the more educated and higher income groups (Cobb and Elder, 1983, 107). Older people should also have more resources as their life experience has given them more opportunities to gather them. There should not be a difference in the level of political resources 7

9 between men and women after controlling for socio-economic status (Burns et al 2001, 47-48). It seems, however, that different life experiences such as distinct work-family roles seem to produce different types of political resources for men and women. Marien et al (2010, 193) note that men seem more likely to participate in a more conventional manner than women. Membership of interest groups can be conventional or non-conventional, depending on the activities of the group. Other things equal, men and women should be more or less equally capable of interest group membership. Furthermore, migrants commonly do not receive full citizen rights and consequently have fewer opportunities to develop civic skills in their receptive country. This leads them and their descendants with relatively few political capabilities. According to some scholars, individuals who identify themselves as left-wing are more politically capable than right-wing individuals. Kriesi et al. (1995, also see Kriesi, 1993) note that the new social movements in the seventies and eighties mobilized especially left-wing oriented groups in society. Their left-wing associated appeals for a better environment, human rights and peace produced a substantial generation with strongly developed political skills. H1 Capability hypothesis: Interest groups have above-average politically capable citizens as members In the next sections we link variation in membership structure to several organizational characteristics: the level of political orientation, the ability to translate membership demands to political activity, and the type of issues groups focus on. In each case we explore whether the assumption of Schattschneider (1960) that organization is the mobilization of bias is valid. In other words, we explore whether initial discrepancies between societal groups and organizational 8

10 membership are reverberated into the organizational capacity of interest groups and the goals they set for the political arena. About the first, political orientation, Salisbury (1969) notes that interest group entrepreneurs invest capital in their organization in order to be able to offer prospective members an attractive set of benefits. By doing so, interest groups mobilize the political resources of individuals (McCarthy and Zald, 1977; Brady et al 1995, 290fn3), and do this in a highly strategic manner, making use of advanced marketing and targeting techniques (Jordan and Maloney, 1998). Organizational leaders strategically target specific groups in society and the set of benefits they offer to prospective members largely depends on the political capabilities of their target group. First, most importantly, target groups with relatively few political capabilities are likely to be offered a set of membership benefits with a relatively strong solidary component. Solidary benefits include such rewards as socializing, congeniality, the sense of group membership and identification, and status resulting from membership, fun and conviviality, the maintenance of social distinctions and so on (Clark and Wilson, 1961, ). Hobby, sports and cultural clubs are typical examples of groups in which solidary benefits motivate individuals to become and stay member. These groups sometimes, as a by-product (Olson, 1965), represent collective interests in relation to public policy, such as on sports subsidies, tax deductibility of charity donations and animal rights regulations. At the same time, as noted by Clark and Wilson (1961, 135), these benefits are, in contrast to the material or expressive incentives, largely independent of the goals of the organization because these benefits refer to the process of coming together rather than the substantive interests represented. But, these goals must first and foremost be non-controversial (...) and never relate to an issue (the conflict of ends) because organizations dominantly 9

11 relying on solidary benefits seek to avoid conflict with other associations as well as avoid it internally (Clark and Wilson, 1961, 142). In other words, club-like associations will not recruit political resources because of the limited availability in their target-groups and because they do not have the management capacity to build external coalitions or internal consensus on concrete policy issues beyond their immediate self-interest (De Bruycker et al, 2017). Second, related, leaders of interest groups with a strong political orientation, such as groups engaging in policy negotiations with government officials, are likely to be in especially strong need of political resources such as membership discipline or mobilization capacity (e.g. Streeck and Kenworty, 2005). Leaders of policy interested groups are therefore, even more than others, in need of political resources and able to offer expressive incentives to their prospective members. They are consequently likely to especially target politically capable potential constituencies rather than individuals with relatively few political capabilities. H2 Organizational orientation hypothesis: Strongly politically oriented groups have more politically capable members than weakly politically oriented groups The capacities of interest groups to translate membership demands into political activity potentially create, at least theoretically, additional inequalities in the representation of interests of subsections of the population. More to the point, the weaker social groups will also be relatively weakly organized. This should both be the case in terms of the strength of individual organizations and in terms of the organizational niches their organizations are part of. First, this follows from the population-ecological idea that organizations need resources for their survival and that (potential) members to a large extent supply these resources (the area term in the 10

12 Energy-Stability-Area model) (Gray and Lowery, 1996). When a lot of people care about international development, and when these people have resources to devote to development charities, then we should observe large numbers of such organizations. The numbers of organizations in a given niche will continue to grow until the carrying capacity of that niche is realized (Gray and Lowery, 1998). In order to avoid direct competition, organizations are likely to partition resources and specialize in certain target groups, tactics or policies. The competitive pressures in dense niches also make sure that only the most professional and resourceful associations survive. Denser niches are more specialized and are situated in parts of the interest group system where there are more abundant resources than elsewhere. In other words, politically capable subgroups in the population form the breeding ground for such niches of relatively specialized, professionalized and resourceful associations. Second, similarly but in more empirical and normative terms, this is also observed by Schattschneider (1960, 34) when he notes that scope and bias are aspects of the same tendency. He characterizes the interest group system as consisting of few very large organizations (the churches, organized labor, farm organizations, and veteran organizations) representing a relatively broad social base and being relatively broad in scope, and a large number of narrowlyscoped special-interest organizations of small numbers of individuals who are acutely aware of their exclusive interests representing a small segment of the population. This leads biases in participation to be exacerbated in the process of political organization because social groups with weaker or less politically capable members will associate in a low number of large organizations rather than a large number of specialized groups. 11

13 H3 Cumulative inequality hypothesis: More resourceful, specialized and professionalized groups have more politically capable members than less specialized groups Last, we can distill from the literature that variation in membership structure affects the particular issues interest groups focus on. The exchange relationship between leaders and members occurs in a public policy context. Distinct groups in society have preferences and material interests on actual policy issues, and policy issues are part of the broader and intereststructuring public policy-making process (Hacker and Pierson, 2014). Old people have a stronger interest in retirement benefits than young people, highly educated are more interested in university policies than lower educated, and professionals have a stronger interest in professional accreditation policies than unemployed. These interests are triggered, or disturbed in a Trumanian sense (1951), to a different degree and in distinct ways depending on the specific policy controversy. First, the extent to which interests may be triggered is likely to depend on the extent to which government is involved in handling certain activities. This translates into government demand (Leech et al 2005; Coen ) for interest representation or, in a less actorcentered formulation, in mobilisation energy (Gray and Lowery, 1996). More government regulation, spending and other activity should lead us to observe higher numbers and activities of interest groups. Second, as regard the type of interest disturbance, the most common approach (e.g. Rasmussen and Carroll, 2014) to theorize about the implications of such policy-induced interest disturbances, is to differentiate between the regulatory or redistributive nature of the policy. This Lowi-Wilson 2x2 classification of the nature of politics relates to the expectations of people regarding the concentrated or distributed structure of costs and benefits of a given policy 12

14 programme (majoritarian, entrepreneurial, client and interest group politics) (Lowi, 1972; Wilson, 1974). In cases where people expect concentrated costs and concentrated benefits one should expect conflictual and substantial interest group mobilization. This classification tells us to expect substantial variation across policy areas in the type and extent of interest group politics, but does not predict the types of social bases mobilized. Therefore, we rely on a more ideological differentiation of policy topics. For this we differentiate among policy areas in terms of the values that policy programmes seem to aim to realize. We hereby rely on Inglehart (1997) who has identified a modernization shift in the decades around the Seventies among groups of people in Western societies towards 'post-material issues. Among other manifestations, this is exemplified by the causes such as peace, gender equality and a clean environment on which the New Social Movements in the Seventies mobilized (Kriesi et al, 1995). More to the point, we expect that politically capable citizens mobilize on relatively post-material policy areas. H4 Policy area hypothesis: Groups working on post-material public policies have politically more capable members than groups working on material public policies Research design We use data of the Dutch part of the Comparative Interest Group Survey (Beyers et al. 2016). We rely on data collected by the Dutch survey team because this is the only team that asked specific questions about the nature of the membership. The Dutch case serves as a great case to test the hypotheses developed in the United States. The United States serves as a most likely case for several reasons. First, the welfare state is not as developed as in most Western- European countries which has led to much more societal inequality and poverty. Moreover, the 13

15 government does not provide much assistance to interest groups as they do in Europe. Combined, this means that we should see a much stronger divide between classes in society. Also the costs of collective action are much higher. This means collective action problems are less severe than in the US, hence the chance of more equal representation should be higher. The Netherlandsserves as a least likely case given that the welfare state is highly developed, the funding system for NGOs by the government is among the most generous in the world, and the interest group community is considered as one of the densest systems on the globe. All this should lead to less inequality in society and less obstacles to organize. If we find similar trends in the Netherlands as we do in the United States, this is a strong indication of a more general trend across countries. To identify the population of interest groups in the Netherlands, two sources were combined. First, organisations were included from attendance lists of public hearings of the Dutch House of Representatives between 2012 and 2014 (Hoorzittingen en Ronde Tafel Gesprekken van (commissies van) de Tweede Kamer). The organisation type of each organisation in these lists was coded manually based on its website. Only membership organisations were included in the sampling frame, excluding for instance firms, (semi)public institutions, and experts. Organisations that recently participated in Dutch public hearings are likely not to be representative of the entire population of interest groups. In such a top-down approach, business associations and organizations with higher levels of political activity or involvement can be expected to be overrepresented (Berkhout et al, 2016). To account for this, second, organisations were included from the Dutch Pyttersen s Almanak This is a register containing Dutch non-profit organisations and institutions that mainly represent or serve interests of a supralocal level. Due to these registration requirements of the Pyttersen s Almanak, no further 14

16 selection was needed. Such a bottom-up approach is relatively more likely to include citizen groups and organizations that are less involved in politics (Berkhout et al, 2016). The combination of these two approaches provides us with a sample in which different types of groups 459 organisations were included from the attendance lists of public hearings organisations were included from the Pyttersen s Almanak, amounting to a total sample of 3329 (113 organizations were included in both lists). To survey our population, for each organisation we searched for contact details of up to two high-level representatives (often including the director, president, chair, secretary general and/or director of public affairs). Out of the 3329 organizations, we were able to find such details of 2479 organizations (in 86 percent of the cases from the public hearings lists and in 74 percent of the cases from the Pyttersen s Almanak, as this also includes much smaller organizations). During the Spring of 2016, 150 invitations were sent out every month to organizations to participate in our survey. A week later a reminder was sent by , which was followed by a phone call another week later. A final reminder notified the respondent by of the closing date of the survey. If by this time the organization had not responded, the second contact person was contacted. This led to a response rate of 37,2 percent. The key dependent variables in our analysis are derived from answers to a question list on the types of members of association responding. We asked: Looking at the following social characteristics, how would you describe the membership base of your organization?. The answer-categories were provided on ten point scales. These items include: mainly youth versus mainly elderly, women versus men, lower educated versus higher educated, people with a left political preference versus people with a right political preference and Dutch citizens with a migrant background versus Dutch citizens without a migrant background. In the descriptive 15

17 analysis, the scales are used unaltered. In the statistical analysis, scores are standardized, representing relative membership bases compared to that of all other organisations. In the descriptive analysis (to test H1), we use population data from several sources to compare the members of the interest groups to the population. In the data on educational attainment, derived from the Dutch Central Bureau for Statistics (2016a), five categories are distinguished: primary, lower secondary, upper secondary, post-secondary and tertiary education. Ethnic background, gender and age are also retrieved from data offered by the Central Bureau for Statistics (2016b). Data of self-placement on a left/right scale of the Dutch population is based on the most recent European Social Survey (Norwegian Centre for Research Data, 2014) and is measured on an 11 point scale. In order to compare the data from the CIG project with the population data we made the scales comparable. For instance, we calculated based on the answers of the respondents the percentage of women that was a member and the percentage of man that is a member. Or, in the case of age, we calculated the average age of the membership of each interest group in our dataset. This makes it easy to compare the overall distribution of age in our sample; with the population demographics in the Netherlands regarding age. Only for the political orientation scale we were unable to normalize the scale because in our survey we asked about the political orientation on a scale of 1-10, whereas the ESS asked about self-placement of citizens on a scale of This means that there is no middle category in our scale, whereas there is a middle category in the data of the CBS. To make sure this doesn t affect our results we provide the original scales of both surveys in our comparison. [INSERT TABLE 1 AROUND HERE] 16

18 In the second section of our analyses we explore how membership structure influences several organizational goals and objectives. We use answers to a number of our survey-questions for this paper of the analysis. First, in order to assess hypothesis H2, we measure the political orientation of an organization by combining three variables. Two are derived from the following question: Looking at the list below: To what extent is your organization involved in the following activities? The items lobbying and representing interests are included, both measured on a 3 point scale (ranging from not involved to very involved). The third variable is based on the question: Organizations like yours have to spend resources on many things. Can you tell us what percentage of your total resources is spent on influencing public policy? This question is measured using 6 categories. These three variables are standardized with a mean of zero and standard deviation of one, after which they are added up, and used as an indicator for political orientation. Second, in order to evaluate hypothesis H3, we measure the policy diversity (the inverse of specialization) of an organization using the following question: How involved is your organization in the following policy areas?. For each organization, the number of fields that it is very active on are summed up. Also in relation to H3, the professionalization of an organization is analyzed by combining items on several topics: decision making, hiring policy, training of staff and career prospects of employees (see appendix 1). Also related to H3 we use the financial resources of an organization, measured as the annual budget of Finally, in order to study hypothesis H4 on policy areas, we asked organizations in which policy area they are most active, such as the environment, social policy, economic policy, migration, etc. (see appendix 2 for the full list). We then coded each policy field as more materialistic, more post materialistic or in between (see appendix for coding tables A3 and A4). Finally, we 17

19 control for the resources an organization has (measured as the turnover in the previous year) and the expenses of the Dutch government in the field that the organization is most active on. Results We analyse the survey data in two ways: first, we evaluate the plausibility of our capability hypothesis related to mobilization bias by comparing the social base of the interest groups surveyed with distributions of the categories studied in the actual population. Second, we examine our organizational and policy-related hypotheses (H2-4) by means of several multivariate OLS regression models on the assumed political capability of the socio-economic membership profile of the organizations studied. We start with our analysis on mobilization bias (H1). Figures 2 to 6 present the frequency distributions of the answers provided by groups with citizen members in our survey and distributions of the categories studied in the actual population. We discuss the results below. For starters, Figure 2 compares the distributions among organisations and within the population in terms of education. The results show that the part of the public that is identified by group leaders as their members is higher educated than the population as a whole. To be more precise, 28 percent of the population belongs in the upper categories of educational attainment, while 61 percent of our respondents said that their membership dominantly includes higher educated (see two highest bars). For the lowest educated citizens the reverse is true. While 33 percent of the population is considered lowly educated, among the interest group population this is only 10 percent. These findings are in line with much research focusing on the political capabilities of citizens with varying levels of education. Moreover, it fits with the representational bias observed in parliaments towards highly educated citizens (Bovens and Wille, 2011; Hakhverdian 18

20 and Schakel, 2017). Apparently, interest groups have a similar bias towards the representation of highly educated members. This confirms our expectations. In Figure 3 we present the distributions by gender. In Figure 3 one can clearly see that males are overrepresented in interest group membership. In the population 49,5 percent are men, whereas among the membership of interest groups this percentage is 57,5 percent (following a chi-squared test, the difference is not significant with a P-value of.09). To indicate how the distribution is precisely shaped, in appendix 4 we present the specific answers respondents gave. Some interesting, additional, observations can be made. That is, thirty three percent of our respondents presumably interpreted category five as the middle category and indicated that their group equally mobilises men and women. Sixty five percent note that their membership is more or less balanced in terms of gender (categories 4 to 7). The interesting difference is between the number of respondents who said that their organizations dominantly had a male (>8) or female (<3) membership. Twenty three percent of the organizations tend to mobilise men and twelve percent has a largely female membership. In other words, organizations of men are more common than organizations of women. This, again, is in line with the literature on political participation by gender (e.g. Marien et al (2010, 193) and it also confirms our expectation. Figure 4 shows the distributions by age. In Figure 4 one can clearly see that in the lower half of the figure (referring to younger citizens), membership is underrepresented compared to the population, whereas in the upper half of the figure (indicating older citizens) the members are overrepresented (a chi-squared test indicates that the difference is statistically significant). To be more precise, 64 percent of our respondents choose category seven or higher and in that way indicate that they tend to have relatively old members. This part of the population only constitutes for 25 percent, a vast difference. If we look at the bottom three categories (i.e. 19

21 younger than 30 years old), there are hardly any members from this age group (8 percent), whereas in the population 36 percent is 30 years or younger. Clearly young people are not well represented in the membership of interest groups, whereas older people are overrepresented. We hypothesized that older people are more politically experienced and capable than younger people and this confirm our expectations in this regard. Figure 5 presents the proportions of individuals with a migration background. The results show a mismatch between the population and interest group membership, but in the opposite direction. We see relatively more members with a migrant background than in the population (22 percent versus 30 percent). Again, to be more specific, we present the exact distribution within the interest group population in appendix 5. Here the upper and lower categories are of interest: That is, twenty one percent of our respondents said that their membership largely consists of people with a migration background (>8), and sixty percent identified their members as largely non-migrants (<3), and in only a minority of cases (20 percent) a diversity of membership was noted (4-7). On aggregate terms this is reflective of the population as a whole (a chi-squared test indicates that the difference between the two is not statistically significant, with a P-value of.08). But it seems that people with a migration background inhibit a separate island in the land of interest groups. We expected people with a migration background to have relatively few political capabilities and therefore relatively low presence, both as members of diverse groups, and in own organizations. This is not supported by the survey respondents. Figure 6 shows the distribution by political identification. The results indicate a strong peak with rather weak shoulders (categories 2-3, 7-8). Among the interest groups no less than 50 percent have members that are neither left nor right, whereas this is only 30 percent in the population. In contrast, extreme score are almost absent in the interest group population, whereas 20

22 these are still substantial categories in the population (a chi-squared test indicates that the difference between the population and the members of interest groups is statistically significant). This means that interest groups are more politically moderate than citizens. Formulated differently, citizens with political positions somewhat outside the political centre will find it difficult to find an interest organization with members with similar positions, and consequently do not find their views well represented via groups. It seems that interest groups moderate the positions of members, or, at least, the leaders like to think of their members in such as manner. Another notable observation is that the membership of interest groups is more right than left. This is somewhat surprising as membership with citizen groups tents to be associated with left leaning citizens. Our results indicate the opposite. Overall, we find some support for the assertion that more moderate citizens are better represented than citizens on the outside of the political spectrum. In sum, we find support for our first hypothesis that interest groups tend to mobilise relatively politically capable groups in society. This is especially the case for highly educated people and higher age groups, but also for political orientation. We did not find that people with a migration background are disproportionately absent from interest group membership. [INSERT FIGURE 2-6 AROUND HERE] We now turn to the second part of our analyses and examine whether these inequalities are exacerbated or reduced in the process of political organization. Most notably, we assess whether political orientation (H2), cumulative inequality (H3) and the policy focus of the organizations (H4) favours some constituent groups over others. We test these hypotheses by 21

23 means of lobby specialization, organizational resources, organizational professionalization, and level of government expenditure in and value -nature of the policy area of concern. Table 2 presents the results of nine OLS regression models on each of the membership answer dimensions in our survey. We present separate models for the different categories rather than constructing clusters of strong or weak subgroups. The categories hardly correlate and therefore form only incoherent clusters (see correlation matrix in appendix). [INSERT TABLE 2 AROUND HERE] What are the findings? As can easily be seen in the R-squared scores and the limited amount of statistically significant results all models have a very low explained variance. This means that in very general terms our hypotheses 2-4 find limited support in our survey findings. One finding does stand out, namely that groups with more migrant members are less resourceful than groups with less migrant members. This is notable because this was the one area in which we found no bias in terms of mobilization. This means that these members are more adequately represented than other marginalized groups in society, but apparently by weaker groups. Overall, the result thus show that the discrepancy we found between key demographic groups and interest group membership does not lead to, as Schattschneider (1960) called it, the organization of bias. In contrast, we do not find any indication that membership bias lead to weaker organizations or different organizational goals. We discuss the implications of these findings in the conclusion. But first, we do not think that this is a result from the relatively simple modelling strategy used. That is, the scales could be transformed in other ways, the linear nature of the regression could be adapted or separate analysis of distinct sets of independent variables may be 22

24 used. However, the strength (or weakness) of the findings indicates that this is unlikely to fundamentally affect these convincing null-findings. Conclusion Schattschneider s depiction of the interest group universe as the heavenly chorus that sings with an upper-class accent is arguably one of the most used quotes in the interest group literature. Yet, despite its allure, what Schattschneider was truly referring to has not been studied as often. That is, with this quote Schattschneider, (1960, 32) highlighted that marginalized groups in society, such as people with low socio-economic status, women and those living in rural areas are hardly ever members of interest associations. Moreover, as Schattschneider (1960, 34), assumes, such biases in participation will eventually be exacerbated in the process of political organization. In this paper we have modestly tried to revive this debate in an EU context. Do interest groups indeed ignore marginalized groups in society and mainly represent highly educated and other privileged groups in society? Moreover, do groups representing more marginalized constituents face steeper obstacles in organizing politically? Our findings provide mixed result. We found a clear mobilization bias towards stronger and more politically capable, privileged groups in society. Citizen groups have disproportionally older, highly educated, and politically moderate citizens as members. Yet, surprisingly, this doesn t seem to translate into organizational bias. We found no evidence that groups with members with relatively few political resources have weaker political organizations than associations with more privileged members. This means that from an equal representation perspective we have good news and bad news. We must be pessimistic about our observation that certain groups are poorly represented by interest groups. We are more optimistic about our 23

25 null findings. The inequalities in participation are not substantially amplified by the unequal organizational development of some interest associations compared to others. Obviously, this is not the end of the story. In our analyses we merely showed that there is no difference in the organizational structure or capability of interest groups in relation to different types of members. Yet, this doesn t mean that the influence targets of interest groups such as members of parliament, government ministers or other policy makers, may provide some groups a more favourably hearing than others. Indeed, the relatively closed and specialised nature of policy circles seems to favour interest groups that rely on relatively politically capable individuals. Moreover, as Strolovitch (2006) shows, interest groups might still favour stronger members over weaker or deprived members in setting organizational goals. This all falls beyond the scope of this paper, but future studies can productively relate the distinct socio-economic membership profiles to the selective nature of agenda setting, policy access and influence. The results also speak to the broader debate on whether or not some demographic groups (based on ethnicity, education, etc.), and their issues, are represented proportionately in politics. Interest groups are one channel to which citizens can enter the political arena. We show that the patterns of bias found in studies of political parties and social movements are also present in interest group politics. As such they cannot be seen as a cure to inequalities in political agenda-setting and decision-making. Rather they largely reflect the unequal distribution of political capabilities in society and reproduce these in the political arena. 24

26 References Almond, Gabriel Abraham (1960) The American People and Foreign Policy, Praeger. Bernhagen, Patrick & Mitchell, Neil J. (2009) The Determinants of Direct Corporate Lobbying in the European Union. European Union Politics, 10(2): Bovens, M. A. P., & Wille, A. (2011). Diplomademocratie: Over de spanning tussen meritocratie en democratie. Amsterdam: Bert Bakker. Brady, Henry E., Verba, Sidney & Schlozman, Kay Lehman (1995) Beyond SES: A Resource Model of Political Participation. American Political Science Review, 89(02): Burns, Nancy, Schlozman, K.L. & Verba, S. (2001) The Private Roots of Public Action, Harvard University Press. Central Bureau for Statistics (2015). Statistics Netherlands: Werkzame beroepsbevolking Central Bureau for Statistics (2016a). Statistics Netherlands: Hoogst behaald opleidingsniveau. Central Bureau for Statistics (2016b). Statistics Netherlands: Population; key figures. Clark, Peter B. & Wilson, James Q. (1961) Incentive Systems: A Theory of Organizations. Administrative Science Quarterly, : Cobb, Roger W. & Elder, C.D. (1983) Participation in American Politics: The Dynamics of Agenda-Building, Sec. edn, Johns Hopkins University Press, Baltimore. De Bruycker, I., Berkhout, J. and Hanegraaff, M. (2017) The Paradox of Collective Action., Paper prepared for presentation at the EUSA Annual Conference Miami. Edwards, B. & McCarthy, J.D. (2004) Resources and Social Movement Mobilisation. Snow, D.A., S.A. Soule and H. Kriesi (eds) The Blackwell Companion to Social Movements., Blackwell, Malden: Eurostat (2015). General government expenditure by function (COFOG). Gilens, Martin & Page, Benjamin I. (2014) Testing Theories of American Politics: Elites, Interest Groups, and Average Citizens. Perspectives on Politics, 12(03): Grier, Kevin B., Munger, Michael C. & Roberts, Brian E. (1994) The Determinants of Industry Political Activity, American Political Science Review, 88(04): Hakhverdian, A., & Schakel, W. (2017). Nepparlement?: een pleidooi voor politiek hokjesdenken. Amsterdam University Press. Inglehart, Ronald (1997) Modernization and Postmodernization: Cultural, Economic, and Political Change in 43 Societies, Princeton University Press. 25