Does Democratic Innovation Reduce Bias? The G1000 as a New Form of Local Citizen Participation

ABSTRACT

Since less well educated citizens are underrepresented in many forms of citizen participation, deliberative ‘mini-publics’ have been introduced as a democratic innovation to redress some of this inequality. Using data from eleven Dutch deliberative mini-publics (G1000s), this study shows that despite attempts to broaden the appeal of the mini-public, a clear educational bias in the output of the deliberations remains; the output largely reflects the wishes and preferences of the more highly educated. Themes like crime and safety, immigration and integration, which are important concerns for many citizens in their local communities, rarely made it onto the G1000 ‘Agenda for the City’. From a perspective of democratic legitimacy, this finding is worrisome. The paper concludes with some suggestions that combine the strength of a deliberative setting with equality and external democratic legitimacy to overcome this problem.

KEYWORDS:

• Innovation
• deliberative democracy
• citizen participation
• inequality

Introduction

Across Europe, satisfaction with the institutions and processes of representative democracy is declining, as many citizens feel they are inadequately represented by politicians (Dalton, 2008; Den Ridder & Dekker, 2014; Van Dijk & Coffé, 2011). In response, politicians, citizens and scholars have called for democratic innovations based on more direct citizen involvement. These innovations can take various forms. Some are mainly technological, such as online consultation, online voting or e-petitions. Other forms require a physical meeting space, like a mini-public or a citizens’ conference. All of these innovations share an emphasis on interaction and the direct involvement of a larger number of citizens than in traditional forms of political decision making.

Previous research, both in the United States and Europe, indicates that participation in non-institutional forms is unequal. It consistently shows an overrepresentation of highly educated individuals (e.g., Mariën et al., 2010; Schäfer, 2012). Even with more innovative forms of participation, such as e-participation and citizen governance, little has changed: the level of educational attainment still appears to be a decisive factor (Bovens & Wille, 2017; John, 2009; Stolle & Hooghe, 2011). This could lead to serious problems of democratic legitimacy if this bias in participation leads to the neglect or exclusion of the interests and preferences of the less well-educated groups.

Deliberative fora or mini-publics are a relatively new democratic innovation, which aim to attract a broadly inclusive and representative sub-group of an affected population (Ryan & Smith, 2014). Sortition is often used as a selection mechanism, to give each individual an equal chance to participate. Therefore, it is argued that this type of citizen engagement might redress some of the imbalance in participation. Leydet (2019) refers to the potential advantage of a mini-public in terms of both the equality of presence and the equality of voice. Yet, as is argued in this paper, the highly educated tend to be overrepresented in deliberative fora as well (cf. Jacquet, 2017; Ryfe & Stalsburg, 2012), leading to a similar participation bias.

Concerns, preferences and opinions differ between groups of citizens with lower and higher levels of educational attainment (Bovens & Wille, 2017; Kriesi, 2010). Therefore, a lack of diversity among participants may have substantial consequences for the topics that are on the table during mini-publics and the proposals that they produce. A lack of representativeness then leads to a lack of inclusiveness. This is precisely the issue that this paper will tackle, through the analysis of a number of mini-publics in the Netherlands – the G1000s – that were organized between 2014 and 2017. The main question is: To what extent do the outcomes of the G1000s display a bias towards specific topics that mainly reflect the concerns and preferences of higher educated citizens?

The paper first outlines how an education bias can be witnessed both in traditional and innovative forms of citizen participation. It then argues how a bias in participation may lead to a bias in the outcome of the deliberations. The second section introduces the context and specifics of the mini-publics in the Netherlands. A section on methods and data is followed by a section about the outcomes of the deliberations. In the subsequent sections, the findings are discussed and conclusions are drawn about the potential of mini-publics to mitigate inequality.

Participation, education, and patterns of exclusion

Participation of citizens can take many forms, ranging from autonomous decision making in local communities to nationwide societal dialogues or internet consultation. There is a long tradition of participation in many countries, which often precedes the current use of terms like participation or deliberation. Almond and Verba (1963) were among the first to stress the central importance of education in various forms of political participation. Participation is ‘costly behaviour’ (Filetti & Janmaat, 2018, p. 329), which requires various intellectual resources and has to be weighed against other possible activities. Recent research about non-institutional forms of participation, both inside and outside the political arena, points to education as one of the main drivers for participation (Bovens & Wille, 2017; Filetti & Janmaat, 2018; John, 2009; Schäfer, 2012).

Both in e-participation and interactive policy-making, it is again the well-educated citizens who are overrepresented, even if e-participation is more likely to mobilize young people (Christensen et al., 2017; Stolle & Hooghe, 2011; Thijssen & Van Dooren, 2016). The same holds for citizen governance, which involves citizens in directly providing and shaping the delivery of public services through decision-making forums. A study based on data from England and Wales, looking at initiatives in the fields of regeneration, crime, education, and health, found that education was the major predictor for participation (John, 2009). Interestingly, unlike many other forms of civic participation, citizen governance was characterised by a greater involvement of both young people and ethnic minorities.

Similar results can be found in a study from Mariën et al. (2010), that compares participation in 25 countries: the higher the education level, the more likely an individual is to participate in various political acts. Although these non-institutionalized forms are more inclusive in the sense that women and young people are represented better (Mariën et al., 2010, p. 205), they do not diminish the inequalities between participants in terms of education level. Likewise, Van der Meer and Van der Kolk (2016) find a difference of 22% in electoral turnout at local elections in the Netherlands between lower and higher educated, and they show that higher educated are twice as active in various forms of citizen participation.

Many people feel increasingly alienated from our democratic institutions and politicians, which leads to widespread support for giving citizens a more direct say in political decision making, complementing existing forms of representative democracy (Inglehart & Welzel, 2005). New and innovative forms of participation have been introduced in many countries, including forms of participatory or interactive policy making, citizen governance, popular assemblies, and e-participation (Pogrebinschi & Ryan, 2018; Smith, 2009). However, stark differences in the extent of participation and the type of (policy) proposals remain, related to education level, as will be outlined below.

Participation bias

The overrepresentation of the higher educated in various forms of participation has a number of consequences. First of all, these participants differ from those with lower educational attainment in attitudes towards political institutions and democracy. The higher educated have a greater trust in politics and political institutions, they show a greater interest in politics, they are more satisfied with representative democracy and vote more frequently (Armingeon & Schädel, 2015; Bovens & Wille, 2017; Coffé & Michels, 2014).

Not only do the higher educated and the lower educated differ in attitudes, they also differ in policy preferences. Research into the differences in opinions between citizens with lower and citizens with higher levels of educational attainment, points at different concerns and preferences (Bovens & Wille, 2017; Lesschaeve, 2017). The higher educated have a more globalist, multiculturalist and environmentalist outlook compared to the lower educated. This translates into a new socio-cultural cleavage (cf. Kriesi, 2010; Marks et al., 2006) which is linked to fundamental values rooted in the developments of globalization, denationalization and the opening up of national borders. These processes cause a division between the losers and winners of globalization and modernization, with the higher educated on the ‘winning end’.

According to Kriesi, citizens are becoming increasingly segmented into social groups divided by value orientations (Kriesi, 2010, p. 678). Public debates are increasingly fought out in cultural terms with a strong focus on issues such as EU integration, the environment, crime, and immigration. Different terminologies are used to characterize this new cleavage: Flanagan and Lee (2003) refer to it as the libertarian-authoritarian cleavage, Inglehart and Welzel (2005) use the labels post-materialist and materialist. This socio-cultural dimension also affects the policy positions taken up by political elites: as Lesschaeve (2017) shows, they are biased towards the interests of the privileged groups (highly educated, high income). Similarly, Filetti and Janmaat (2018) argue that the well-off are more effective in advancing their own interests, while preventing other interests and issues from reaching the policy agenda.

The potential benefits of deliberative mini-publics

Mini-publics aim at being more or less representative of the population and they follow the rules of deliberation to enable a dialogue between participants. Even if the process of (self)selection would lead to an overrepresentation of the higher educated, the deliberative design should ensure that a broad range of policy preferences is considered and discussed during the mini-public. The concept of deliberative mini-publics is used for participation in many shapes and forms (Ryan & Smith, 2014), examples of which are citizens’ assembly or summits, planning cells, consensus conferences, and citizens’ juries. Although definitions and usages of the concept of mini-publics vary, they do share some common characteristics (cf. Leydet, 2019).

One of the main characteristics of mini-publics is structured deliberation, often enabled by independent facilitators. Mini-publics are designed with the aim of being deliberative, which means that the focus is on following the ideal deliberative procedures. Opinion formation and the exchange of arguments are more important than decision-making, i.e. talking instead of voting (Chambers, 2003). A possible consequence, and perhaps a disadvantage, is that mini-publics are less connected to the decision-making process, as the link with the political and institutional context often remains implicit (Meijer et al., 2017; Michels, 2011; Riedy & Kent, 2017).

A second key characteristic of the mini-public is the participation of “ … a broadly inclusive and representative sub-group of an affected population” (Ryan & Smith, 2014, p. 20). Often, some form of sortition is used for this purpose. As a consequence, deliberative mini-publics are less open to all citizens as compared to other innovative forms of participation. Sortition replaces the equality of opportunity by the equality of chance or probability (Leydet, 2019), since no member of the population is excluded beforehand. Conversely, this means that if someone wishes to participate, he or she will not be permitted if not selected by lot.

Although a key element of mini-publics is that they try to attract a diverse crowd, evidence regarding deliberative fora shows that those who decide to attend are usually well educated, older, politically active and interested in politics (O’Flynn & Sood, 2014; Smith, 2009; Stadelmann-Steffen & Dermont, 2015). In other words, mini-publics appear to be subject to participation biases as well. Stratified sampling, often used in smaller mini-publics, serves as an instrument to ensure that citizens from salient groups are selected and that the forum better reflects socio-demographic variations (Ryan & Smith, 2014). However, even then, a bias towards the higher educated, the more politically knowledgeable and the people who are less cynical about politics remains (Curato & Böker, 2016; Smith, 2009).

The reason for this bias is that citizens often do not take up opportunities to participate, because there is both no obligation to do so and, more importantly, quite some uncertainty about what will be done with the outcomes (Jacquet, 2017). In particular, citizens with lower education are more likely to decline the invitation, since this practical effect of participation is of more importance to them than to the higher educated. This means that the process of self-selection creates a participation bias.

Why do mini-publics not produce a more diverse and balanced group of participants? This is partly due to the fact that more than in other forms of participation, mini-publics have a strong focus on deliberation and the exchange of arguments. The level of education affects the ability to collect and analyse information, to express views and opinions and to translate these into policies. Higher educated people are more likely to have the cognitive skills, better access to political information and higher levels of political interest and efficacy. Therefore, they will be better able to participate in this type of citizen involvement (cf. Christensen et al., 2017). Moreover, the higher educated may dominate the conversation and be less open or even anaphoric to attitudes and views that diverge from what is familiar and accepted among higher educated (Lee et al., 2015). This implies that although sortition provides equal chances, it does not remove the substantial post-selection barriers to participation.

The policy agenda proposed by citizens

For a long time, the literature on deliberative democracy was mainly focused on the internal dynamics: who are the participants and how do they interact with each other? Which of the conditions for genuine deliberation are met and how can this be improved? The basic rationale for the mini-public approach is that if the random sample that is gathered to deliberate is representative of the population, and if it deliberates under good conditions, then “ … its considered judgments after deliberation should represent what the larger population would think if somehow those citizens could engage in similarly good conditions for considering the issue” (Fishkin & Mansbridge, 2017, p. 9). Recently, attention has shifted to this micro-macro link: how does the mini-public relate to the broader community (Olsen & Trenz, 2016)? This is part of a broader trend among deliberative democracy scholars to focus on the impact of deliberation (Pogrebinschi & Ryan, 2018; Strandberg, 2015), i.e. how political actors respond to the ideas and proposals that are developed during mini-publics.

The potential equalising effect of the selection method (sortition) could be largely undone by the dispersing effect of actual participation. Mini-publics, like other forms of citizen participation, are characterised by an overrepresentation of participants with higher levels of education. The design of the mini-public does not compensate for this lack of diversity among the participants, but with its focus on language, deliberation and arguments potentially enhances the dominance of the higher educated. This will in turn have an impact on the issues and preferences put forward – and their uptake by politicians and civil servants.

Therefore, we expect a bias in the output of mini-publics along the socio-cultural dimension. That is, a bias towards the topics or issues that are of concern mainly to the citizens with higher education: a strong focus on green and libertarian topics as opposed to immigration and law-and-order topics. The next section specifies on which issues this bias is expected to be most prominently reflected.

Context of the G1000s in the Netherlands

The first G1000 citizens’ summit was held in Belgium in 2011, organized by a group of citizens (Caluwaerts & Kavadias, 2014; Caluwaerts & Reuchamps, 2015). This G1000 is often seen as a response to the political instability that Belgium was experiencing in these days. The aim of the summit was to empower citizens and to reach agreement, where politicians had failed to do so. Its name is a wink to the G7 or G8 summits of world leaders, but this time with citizens in charge instead of politicians. The number refers to the 1,000 participants the organizers hoped to mobilize. Participants were randomly selected, to ensure that those taking part constituted a representative group. About 600 people took part in the deliberations, which were held over the course of just one day in the Belgian capital of Brussels. While the summit itself received much media attention, there was hardly any political follow-up.

The Belgian experience with a large-scale citizens’ summit has inspired several groups of citizens, politicians and public officials in the Netherlands to organize similar types of events. As in many other Western countries, declining levels of satisfaction with the institutions and processes of representative democracy have sparked discussion about ways to improve citizen involvement or citizen self-governance. From the 1980s onwards, various ideas have been put forward, but none of the structural reforms (e.g., a new electoral system or the direct election of mayors) have managed to receive the required support in Dutch parliament. Likewise, there have been several attempts to introduce a national binding referendum. These efforts, too, have failed to generate a parliamentary majority, while the Advisory Referendum Act, which took force in 2015, was repealed only three years later.

Given the lack of change in democratic repertoire at the national level, attention has gradually shifted to the level of the municipalities. Particularly at this level, citizens and stakeholders participate in projects aimed at improving the liveability and safety in a neighbourhood, the redevelopment of a particular area, or the founding of a local cooperation for renewable energy. Since the 1990s, the Netherlands has developed broad experience with processes of co-production or collaborative governance in which governments, citizens and (quasi)non-governmental organizations attempt to join forces in policy-making (Hendriks & Michels, 2011). Over the last five to ten years, a wide variety of citizen participation initiatives have emerged, of which the G1000 is one example.

Amersfoort was the first city to host a G1000 in the Netherlands in March 2014, a couple of days after the municipal elections. This example was followed by Uden in October 2014 and by Groningen in June 2015. Another series of G1000s took place in 2016: Schiedam (February), Nijmegen (March), Amersfoort (2^nd time, April), Gemert-Bakel (June), Eindhoven (June) and Coevorden (October). In the first half of 2017, G1000s were organized in Schiedam (2^nd time, March) and Borne (May).

The G1000s that are discussed and analysed in this paper share a number of features, with some couleur locale added by the organizers in a particular municipality. The independent platform ‘G1000.nu’, a citizens’ initiative, is the organization behind most of the G1000s which promotes a specific format:

1. It is organized from the bottom up, by a group of citizens. The role of the local authorities is confined to facilitating and promoting the event, e.g., through a financial contribution and the use of local media outlets. The only exception is Uden where council members and the council secretary took the initiative for the G1000.

2. There is no role for experts, as the relevant expertise is to be provided and shared by all participants during the day.

3. The selection of participants takes place through sortition to give everyone an equal chance to take part, except in Uden where local regulation prevented this.

4. The topics to be discussed are not set in advance: the ‘Agenda for the city’ is developed in a sequence of dialogues, which finishes with a voting round. Two central questions were put to the participants during the G1000: (a) What ideas would you wish to see achieved in your city during the next 4 years? and (b) What will be your contribution to achieving these ideas?

5. Participants talk with each other in small groups (6 to 10 people at each table), with an emphasis on dialogue and deliberation, supported by trained facilitators.

Method and data

This study investigates the outcomes (‘Agenda for the city’) of eleven G1000s organized between 2014 and 2017. Table 1 lists all cases that are included in this study, ordered according to population size. As can be seen from this table, G1000s have been organized both in larger cities with more than 150,000 inhabitants, and in smaller municipalities. The table shows that the voter turnout at local elections was close to the national average of 54% in most of the municipalities. There are two exceptions though: Eindhoven (44.7%) and Schiedam (42.6%) witnessed substantially lower turnout rates. Turnout was highest in the smallest municipality in the case selection: Borne (57.3%).

Table 1. G1000s in Dutch municipalities, political and educational background

	Inhabitants (1 January 2016)	turnout 2014 municipal elections	% of highly educated*
Borne	22,343	57.3%	31.3%
Gemert-Bakel	29,647	55.8%	17.4%
Coevorden	35,381	54.0%	19.2%
Uden	41,247	55.3%	22.6%
Schiedam	77,108	42.6%	22.4%
Amersfoort	153,602	56.3%	38.1%
Nijmegen	172,064	54.6%	47.4%
Groningen	200,952	54.5%	39.2%
Eindhoven	224,755	44.7%	37.1%
Netherlands	16,980,049	54.0%	32.4%

* data on educational background are from Statistics Netherlands (CBS)

Nearly one third of the Dutch population holds an academic degree or a degree from a university of applied sciences. This is the group classified as ‘higher educated’ in this paper. Eindhoven, Groningen and Nijmegen all have universities, which explains the higher-than-average percentage of highly-educated individuals found in these cities (almost half of the inhabitants in Nijmegen). In Amersfoort, too, nearly four out of ten inhabitants were highly educated. On the other hand, Gemert-Bakel (17.4%), Coevorden (19.2%), Schiedam (22.4%) and Uden (22.6%) all have fewer inhabitants with high levels of educational attainment. With respect to the educational background of the G1000 participants, survey data are available for the G1000s of Amersfoort, Uden, and Groningen. These data show an overrepresentation of participants with high educational qualifications (Binnema & Michels, 2016; Michels, 2019). For Groningen and Amersfoort it was found that more than 70% of the participants had a high level of educational attainment (graduates from higher vocational colleges or university) and for Uden the percentage was a little less than 60%. This indicates a participation bias towards the higher educated, which is in line with previous research on mini-publics.

Policies

The final result of each G1000 was a top-10 list of proposals, the ‘Agenda for the city’ which were ranked in a concluding voting round. As explained earlier, these proposals are developed during the day, going from rather abstract ideas to more tangible projects on which citizens can act themselves. These proposals are presented through Prezi or on a poster, and participants walk through the meeting room to evaluate them. The Agendas have been collected from the web sites and newsletters of the local G1000s.

The issues on the Agendas for the city can be compared to the results of the Dutch Local Election Study 2016 (Van der Meer & Van der Kolk, 2016). This survey was conducted among 2,643 citizens approximately two years after the local elections of 2014. It covers a wide range of topics regarding local democracy, like support and legitimacy, political views, political knowledge and voting behaviour. It also has a part on the problems that citizens deem important in their municipality. The sample is representative for many characteristics of the Dutch population, including level of educational attainment.

We already referred to the difference in electoral participation between higher and lower educated. The data from the Local Election Study also show that higher educated are more active in contacting politicians or civil servants, signing petitions, starting local action groups, and contacting media or using social media. The higher educated are less supportive of local referenda, while they emphasise the role of the municipal council in making decisions. In other words, this is a group of citizens that knows how to find their way in local politics and administration, with a focus on indirect rather than direct influence.

The survey included an open-ended question asking the respondents what they thought were the most important problems in their municipality. In other words, they were asked to identify these problems in general, not whether these affected them personally. This question was answered by 40% of the respondents, whereas the higher educated among them were more active in answering this question. Moreover, their answers were longer and they listed more problems than the lower educated did. The answers to this open question were subsequently coded by the research team of the Local Election Study into 12 categories (cf. Van der Waal et al., 2017). The categories were adapted from a list used in other surveys which queried the most important problems at the national level. Table 2 provides the ranking of the problems and the relative share of each of these.

Table 2. Most important problems in Dutch municipalities, local election study 2016

Topic	Share (%)
Infrastructure and public works	21.8
Politics and governance	19.6
Crime and safety	10.7
Immigration and integration	10.6
Health care, care for the elderly	8.1
Income and economy	5.0
Youth and family	2.4
Community, norms and values	2.3
Nature and environment	2.3
Employment	1.9
Social security	1.7
Education, innovation, arts, culture	0.8
Other/none/don’t know	12.7

The problems mentioned most often concerned transport and public works, covering a range of topics from parking problems in the inner cities and the reliability of public transport, to social housing and the demise of shopping streets. Second most common were problems relating to politics and governance. These mainly focused on the integrity of local authorities, the quality of the people involved (politicians) and the procedures (political decision making). Notice that employment and social security score a low 1.9 and 1.7%, although an important change in local politics in the Netherlands has been the decentralisation of a number of policy tasks from the national and regional level. In January 2015, municipalities assumed responsibility for youth care, care for the elderly and labour market activation. Nevertheless, the figures suggest that people still consider these national problems (to be solved at the national level), or that their economic situation was unproblematic.

With respect to issues on the socio-cultural dimension, the table shows that crime and safety and immigration and integration are more or less tied for third place, with almost one out of nine respondents reporting either of these as the most important problem in their municipality. On the other side of the socio-cultural dimension, nature and environment (2.3%) and education, innovation, arts, and culture (0.8%) are considered of minor importance at the local level.

Based on the overrepresentation of higher educated citizens in the G1000s, it is expected that the policy agenda resulting from the G1000 predominantly reflects the interests and preferences of the more highly educated groups: green and libertarian issues. Topics such as crime and safety, and immigration and integration are expected to be less prominent, with more emphasis being placed on art and culture, nature and environment.

The output of the G1000

In the final round of a G1000, all participants are invited to vote for their favourite proposals (save their own proposal), resulting in a top-10 which then forms the ‘Agenda for the City’. All 110 proposals have been assigned to one of the categories from the Local Election Study – as can be seen in Table 3. There were some instances where various topics were combined, e.g., becoming acquainted with the neighbours and greening the neighbourhood. In those cases the proposal was assigned to the category that covered the central element of the initiative. Yet, a number of proposals from Schiedam dealing with the image of the city could not be assigned to any of the categories. Next, one of the proposals from Eindhoven did not fit into the topics identified, because it covered nearly all local issues, from education and liveability to social interaction and public transport, with no clear prioritization of one aspect. Altogether, five proposals were assigned to the ‘other’ category.

Table 3. Topics on the Agenda (top-10) of the G1000s, 2014–2017*

	Amersfoort I	Uden	Groningen	Schiedam	Nijmegen	Amersfoort II	Gemert-Bakel	Eindhoven	Coevorden	Schiedam II	Borne	Total (+%)	Local Election Study
	Mar 14	Oct 14	Jun 15	Feb 16	Mar 16	Apr 16	Jun 16	Jun 16	Oct 16	Mar 17	May 17	2014–2017
Infrastructure and public works	1	2	2		1		1		2	1	3	13 (11.8)	21.8
Politics and governance	2	2	2	1	2	1	4	2	1	2	1	20 (18.2)	19.6
Crime and safety	1					1		1				3 (2.7)	10.7
Immigration and integration					1			1			1	3 (2.7)	10.6
Health care/care for the elderly	1	2										3 (2.7)	8.1
Income and economy			1									1 (0.9)	5.0
Youth and family							1		1			2 (1.8)	2.4
Community, norms and values	3	1	2	4	5	4	3	3	4	5	2	36 (32.7)	2.3
Nature and environment	2	1	2	1	1	4	1	3	1		3	19 (17.3)	2.3
Employment									1			1 (0.9)	1.9
Social security												0	1.7
Education, innovation, arts, culture		2	1	1								4 (3.6)	0.8
Other				3						2		5 (4.5)	12.7

* The proposals were collected from the reports and updates on the web sites of the local G1000 organizers

Overall importance: community and politics

Two topics made it onto the Agenda of all the G1000s: 1) community, norms and values (32.7%), and 2) politics and governance (18.2%). The recurrence of the first topic on the Agendas stems from the growing concern about the individualization and fragmentation of society, leading to a call for initiatives that help restore social cohesion and important values of Dutch society. It is a theme that is shared by many participants, and the attractiveness of these proposals is that the implementation of such initiatives will have to take place in the neighbourhoods where people live. By contrast, this theme occupies a far less prominent position on the list of important problems derived from the Local Election Study, where it ranked eighth (2.3%).

The second topic concerns politics and governance. Many proposals pertain to a new balance between the responsibilities of citizens and of the government. They entail new ways of participation and decision making, and focusing on neighbourhood governance without too much interference from local politics. The initiatives proposed mainly focus on ‘thinking’ instead of ‘doing’: giving input at the start of the policy process, increasing coordination and dialogue between citizens and civil servants, developing new, more interactive procedures of decision making, and the like. They require some insight into the local policymaking structure to have an impact. While in the light of their high position on that list, the importance attached to these issues might be said to be mirrored in the Local Election Study, the tone is quite different. In the municipalities where this was considered an important local problem, this was mainly in connection with the lack of integrity on the part of politicians and the lack of transparency in decision-making. In contrast, the G1000 proposals have a more optimistic tone, focusing on cooperation between citizens and the government and improving policies together.

Education cleavage: nature, environment, crime, immigration

As argued before, topics such as crime and safety, and immigration and integration are expected to be less prominent, with more emphasis being placed on art and culture, nature and environment, issues that were identified as being more prevalent among the higher educated. Most of the findings are clearly in line with this expectation. First, nature and environment occupies the ninth place on the list of problems and priorities from the Local Election Study, yet it is a prominent topic on the Agenda of the G1000s. Various proposals relate to the greening of the neighbourhood or providing local solutions to climate change. This reflects the difference of opinion regarding the scope of the climate problem and also the difference in the financial means required to switch to renewable energy sources.

Crime and safety, which was seen as the third most important local problem in the LES, made it to the Agenda of the G1000 in only three out of eleven occasions: Amersfoort (twice) and Eindhoven. What is more, the tone is quite different. To the extent that it reached the Agenda of the G1000s, the nature of the proposals is rather positive – a similar mirror image as observed earlier on for the topic of ‘politics and governance’. The focus of the G1000 proposals is not ‘law and order’ or ‘tough on crime’ but rather on keeping an eye on each other and feeling safe in the neighbourhood. This also fits the more libertarian end of the socio-cultural dimension.

Expectations are confirmed with respect to the issue of immigration and integration as well. This was ranked fourth in the Local Election Study, but it reached the Agenda in only three instances: Nijmegen, Eindhoven and Borne. And again, the tone is more positive. To the extent that these topics made it to the Agenda for the city, the proposals were not geared towards stopping immigration or the downsides of the multicultural society, but they concerned initiatives to welcome refugees or to organize more interaction between different ethnic groups. It should be added that this also had to do with the different moments at which the G1000s were held: the refugee crisis was most urgent in 2015, but since then gradually faded from the headlines. Moreover, there are large differences between the municipalities in size of the non-western foreign-born population. Particularly in Schiedam, more attention for this topic might have been expected given that almost three out of ten inhabitants have a non-western immigrant background.

Discussion of the findings

Using data from eleven G1000s that were held in the Netherlands between 2014 and 2017, a clear bias can be observed in the output of the G1000s, the ‘Agenda for the city’. As the analysis shows, crime and safety, and immigration and integration, two local problems that citizens find important according to the Local Election Study (Van der Meer & Van der Kolk, 2016), rarely make it onto the Agendas. In contrast, an issue ranking a mere ninth in the LES, nature and environment, occurs on all but one of the eleven Agendas, adding up to a total of 19 proposals. This confirms the expectation that the output of the G1000s largely reflects the wishes and preferences of the more highly educated.

In addition to this and based on the findings, there are a number of other interesting observations. Most of all, the findings show that where the topics of crime and safety, and immigration and integration reach the top-10, the connotation is predominantly positive and ‘soft’ on crime and safety, immigration and integration. Part of this may be explained by the deliberative setting of face to face deliberations. From previous research on the G1000 in Belgium it is known that even in strongly divided groups, deliberation led to high quality deliberations; it created openness towards others by confronting people with different views. But, the predominantly positive tone might also be explained by the overrepresentation of higher educated citizens who in their daily lives usually are less confronted with crime and the problematic aspects of immigration and integration.

Some other observations are related to the design and characteristics of the Dutch local G1000s. First, the design of a G1000 and the type of questions that people reflect on during the day, may lead to a focus on the immediate environment and on initiatives in which participants themselves can play a role. For that reason, it comes as no surprise that the topics of community, and nature and the environment, figure prominently on the Agendas. This may also explain why there are no striking differences between the municipalities; not between larger cities and smaller municipalities, and even not between the G1000 in Uden and other municipalities, although Uden had a slightly different format.

Second, these mini-publics are held at the local level, which could mean that participants also take into account whether local government has a certain influence or authority on these topics (or the perception thereof). This could explain why ‘income and economy’ or ‘health care’ hardly receive a mention, even now that many of the policy competences have been decentralised to local government.

Finally, two other design characteristics of the Dutch G1000s may have contributed to a dominance of the voice of the higher educated during the deliberations: the absent role of experts and the open agenda. The choice not to give experts a role benefits those who already have knowledge about policies and policy processes. Working with an open agenda, finally, may have discouraged people who did not feel a sense of urgency to participate and may have benefited those who are better at expressing ideas and convincing others. The ‘open agenda’ means that the topics of the deliberation are not set in advance, which causes uncertainty about what will be discussed during the day and what will be the effect on local government. This seems to be an important factor to understand why such a high percentage of the invitees (90 to 95%) decided not to take part. Previous research into participation also indicates that citizens often do not take up opportunities to participate, when there is no obligation to participate or no reimbursement (cf. Jacquet, 2017). The deliberative design of the mini-public (throughput) does not compensate for the input bias.

Conclusion

Even if dissatisfaction with the institutions of representative democracy is growing, elections still remain a popular form of political participation. Elections are also, if only for the sheer number of participants, among the most equal forms: although turnout is somewhat higher among higher educated, many lower educated citizens vote as well. Alternative forms of participation, and in particular those that can be headed under democratic innovation tend to be more unequal, as people with higher levels of educational attainment are overrepresented.

This bias in participation based on education level has at least two consequences that deserve further discussion, First, as this paper shows, it may lead to the neglect of the wishes and interests of the lower educated groups in society, which might constitute a serious problem of legitimacy. Individuals or groups within the community may not support the policies that are developed, or even be opposed to them. Moreover, the debate between the two groups is increasingly framed in socio-cultural terms with a strong focus on issues such as EU integration, the environment, crime, and immigration. Previous research on political inequalities showed that lower educated citizens have other values, concerns and preferences than higher educated. Yet, this group is less successful in raising awareness for these issues and in putting them on the policy agenda.

This paper considered whether the democratic innovations known as mini-publics perform better in this respect. Do they reduce the bias stemming from differences in educational background or do they lead to the same kind of bias? The background to this question is that a key element of mini-publics is their aim to attract a broadly inclusive and representative sub-group of the population. Often, sortition is used to (randomly) select participants, hoping to avoid the bias of self selection. If indeed mini-publics would be characterised by such inclusiveness and representativeness, a more balanced outcome of the deliberations might be expected. That is, the ideas developed during the mini-public would be a reflection of the interests and preferences of the community as a whole and not mainly of a highly educated subgroup. However, as this paper shows for eleven G1000s in the Netherlands, despite attempts to broaden the appeal of the mini-public, a clear educational bias in the outcome of the deliberations remains.

All in all, the conclusion about the success of the G1000s as a democratic innovation to mitigate inequality and the biases in the policy agenda that stem from this, is rather pessimistic. As this paper has shown, mini-publics again give the more highly educated groups a strong voice, comparable to interactive policy making or e-participation. From a perspective of democratic legitimacy this is worrisome, even more because the higher educated are already more politically active and more capable of connecting with local policy makers. Therefore, the G1000s rather strengthen existing patterns of inequality instead of setting a new and more equal pattern of participation.

On the other hand, a strong element of the deliberative setting of mini-publics is its discursive power and the creation of a respectful dialogue, which is reflected in a softening of tone about otherwise contested issues about crime, immigration and integration. This leads to a more effective and more substantial consideration of such issues than during a community meeting or hearing. The main challenge for mini-publics therefore is to improve the selection of participants, to make the forum more diverse and more representative. If this were achieved, the bias in the selection of topics during the mini-public is likely to be decreased which, in turn, could increase its legitimacy among the local community.

To live up to these expectations about the mini-public implies, first, an increased effort to foster participation, for example, by paying a fee to the participants or by combining sortition with targeting specific groups through formal and informal networks. Second, it implies that participants need to be aware beforehand what will be done with the outcome of the deliberations, e.g., a debate in the local council or a local referendum. The latter option relates to a growing strand in the literature on deliberative democracy, which emphasizes the importance of a mini-public’s embeddedness in the broader deliberative system. The combination of deliberations with political or public ratification not only enhances the quality of deliberation, but also increases equality and democratic legitimacy by connecting the outcome of the mini-public to the broader political sphere.