Assessing Tools for E-Democracy: Comparative Analysis of the Case Studies

Korthagen and van Keulen compare in this chapter the 22 case studies of digital tools discussed in part II of the book. They use Qualitative Comparative Analysis (csQCA) to study which conditions lead to actual impact of the tools on policy, (1) decision-making or (2) agenda-setting. Sixteen conditions identi ﬁ ed from the literature review are compared. Ultimately, the most important conditions for successful e-participation identi ﬁ ed by the authors are as follows: a close and clear link of e-participation processes to a concrete formal decision-making process should be available; the participatory process and the contribution of its outputs to the overall decision-making process have to be clari ﬁ ed to the participants from the start; feedback to the participants about what has been done with their contributions is an indispensable feature of the process; a participative process should not be limited to one event but should be imbedded in an institutional ‘ culture of participation ’ ; and, ﬁ nally, e-participation must be accompanied by an effective mobilisation and engagement strategy, involving communication instruments tailored for different target groups.


Introduction
In this chapter, the 22 case studies of digital tools discussed in part II of the book are compared. In the comparison, we analyse which conditions lead to impact on decisions or agenda-setting. The case studies were compared in a crisp-set Qualitative Comparative Analysis (csQCA). This was also the approach that Pratchett et al. (2009) used to compare different cases of e-participation (i.e. e-fora and e-petitions) in relation to the empowerment of communities influencing local decision-making. The foundation of csQCA lies in Boolean algebra. Hence, the scores of the cases on the different conditions and the outcome are dichotomised in the course of the procedure.
The number of conditions included in the QCA needs to be relatively low because the number of possible logical set-combinations quickly exceeds the number of cases (Berg-Schlosser and De Meur 2009). In other words, the empirically observed cases will occupy only a tiny proportion of the potential 'logical space' (3 conditions result in 8 possible logical combinations, 6 conditions lead to 64 possible combinations). Moreover, the fewer the number of 'causes' which are needed to explain an outcome, the closer we come to the core elements of causal mechanisms. A large number of conditions tend to individualise each case, which makes it difficult to find any regularity or synthetic explanation of the outcome across the cases. In our intermediate-N research design, working with a number of 4-7 conditions is advised in the literature (Berg-Schlosser and De Meur 2009: 28). The ideal balance between the number of cases and the number of conditions is found through trial and error.
The list of conditions we researched in our case studies is larger than the list of conditions that formed the configurations later on. The final conditions and outcome 'variables' are formulated in the last stage of the research, during which the variation of conditions and the outcome amongst the different cases is determined. At least one-third of the scores on a condition must be a one (1) or a zero (0); when conditions or outcome scores do not show enough variation amongst the cases, they are excluded or adjusted (Berg-Schlosser and De Meur 2009: 45). When a condition or outcome is difficult to score in too many cases (e.g. because of a lack of information, contradictory statements in the literature or from two interviewees in one case), it also cannot be used for further analysis. We based the scoring of the conditions and outcomes on the case studies. The data for the case studies was collected from (grey) the literature about the case, with both a standardised questionnaire and a semi-structured interview with two respondents per case. According to the design of the study, the two respondents are usually (1) a professional who is involved in the process of the case (i.e. organiser) and (2) an expert who studied the case (i.e. academic researcher).
The comparative analysis leads to two types of findings. First, a comparison of the conditions and outcomes of the cases. In Sect. 12.2 and 12.3, we introduce the conceptualisation of each condition-based on the literature review in Chap. 4-and provide a short analysis of how the condition is scored amongst the 22 cases. Where possible, we explain some of the underlying mechanisms of the conditions: why is the condition relevant to digital participation trajectories? This second part of the analysis identifies the conditions under which digital tools can successfully facilitate different forms of citizen involvement in decision-making processes, which answers the main research question. Success means that citizen involvement has led to either impact on decisions or impact on political or policy agendas. In Sect. 12.4, we analyse the configurations leading to impact on decisions as well as configurations that lead to agenda-setting. We distract a 'minimisation formula' as it is called. We conclude both sections by reflecting on what can be learned from the descriptive formulas and how impactful e-democracy on the European scale can be organised.

Assessment of the Different Conditions
Our complete study compares the 22 cases on 16 conditions and 2 outcomes (Korthagen et al. 2018). However, in this book we only report the results for the conditions that are part of the final configurations. These conditions concern whether cases use a combination of online and offline participation, a link is created with the formal decision-making process, the tool is sustainable, the participation process was clear from the start, a mobilisation and engagement strategy was in place, feedback was provided, voting was possible and interaction possibilities existed. The assessment of other conditions, such as the user-friendliness of the tool, moderation and whether the initiative is a governmental initiative ('invited space') or not ('invented space'), was less related to the outcome of impactful e-participation. In this section we discuss the nine conditions, their relevance to the outcomes, and identify the cases that score positively on each of the conditions.

Combination of Online and Offline Participation
This condition evaluates whether the participation process offers the opportunity to participate not only online but offline as well ('hybrid or blended format'). This condition can be expected to have an effect on the outcomes, since offering both online and offline possibilities encourages the inclusion of citizens. For example, a combination of online and offline channels to maximise inclusiveness is now state of the art in German participatory budgeting projects (Heidelberger 2009). Kies and Nanz (2013) evaluated different EU participation tools and recommend a combination of online and offline activities-an open online phase carefully connected with a phase of face-to-face consultations-to improve EU citizens' deliberation activities. By offering offline opportunities, everybody should be able to participate, even if they do not have online access or do not have sufficient digital skills. Or, as one of the interviewees of the participatory budgeting case in Berlin-Lichtenberg said: 'Because not everyone is comfortable with just one way [of participating]'. Another consideration for combining online with offline activities is that deliberation works better offline than online. The founder of petities.nl stated: 'The moral of the [online] medium is that you can endlessly 'fork' as we call it. If you do not agree, you move on to another website, another Whatsapp group, etc. Online there is no scarcity of space'. Kersting (2013: 278-279) is an advocate for a 'blended democracy combining online and offline instruments' because online spaces can lead to self-affirmation and in-group bonding. And yet another argument for a combination is that online activities which build on existing offline networks are more effective in mobilising 'real world' participation (Gibson and McAllister 2013: 21). However, online and offline participatory activities do not always have to complement each other. In cases of petitioning (petities.nl) or contacting politicians (theyworkforyou.com), the activities can substitute for one another (see Gibson and Cantijoch 2013).
In 14/22 cases, participants had the possibility to participate online and/or offline. The case of Wiki Melbourne, the crowdsourcing of a new constitution in Iceland and also the case of the European Citizens' Consultation (ECC) are classic examples of how digital instruments can contribute to democratic processes alongside offline participatory events. Those have been extensive and long-lasting participation processes consisting of different online and offline phases. For example, in the Wiki Melbourne case, existing offline networks have been used (and created) and perfectly integrated into the online process. Firstly, meetings with different stakeholders were organised to draw up a draft plan. This draft was then published as a wiki webpage, to which changes could be made online (although not yet by the public). A stakeholder consultation of 2 weeks was held in which changes in the document were made by specific stakeholders. A few weeks later, a month of public consultation was organised, in which the wiki was open for anyone to edit. Various meetings and events were organised to gather input for the document, making the project an actual combination of on-and off-line community activity. In other cases, like Futurium and Berlin-Lichtenberg, offline meetings such as workshops, public events, community meetings, etc., feed the online discussion, and vice versa. For political parties such as Podemos, the German Pirate Party and the Five Star Movement, offline meetings also play a vital role in the decision-making processes. Additionally, in several cases it was possible to vote online as well as offline (Participatory Budgeting (PB) in Paris, e-voting in Switzerland and Estonia), or to sign a proposal online or offline (European Citizens' Initiative, voting in Estonia, voting in Switzerland, Open Ministry-at least, for the Finnish Citizens' Initiative).

Link to the Formal Policy or Political Process
This condition reflects the official status of the digital tool and the participation process. It concerns the embeddedness of the tool in the formal processes of decision-making is organised. By designing such a link, the participation process and its outputs are clearly connected to what politicians or policy-makers are addressing at that time. A link to the formal decision-making process might therefore be of vital importance for the impact of the participatory input. For example, Font et al. (2016) looked at proposals resulting from participatory processes in Spain by studying 611 proposals from 39 different processes. They found that '[. . .] the odds that a proposal emerging from a participatory budget or other permanent mechanisms (e.g. citizen councils) is fully implemented double those of proposals coming out from a case of strategic planning or other temporary processes' (Font et al. 2016: 18).
A link to a formal process can be designed in several ways. It might be a policy or legal framework that prescribes what the participation process is about, its preconditions and how its results should be handled. An example of a prescription about how results should be handled is an obligation to provide adequate feedback on participative input. Such an obligation also creates pressure on the decisions to be taken, as it entails the acknowledgment of the participants as legitimate political actors (Badouard 2010). In Slovenia, proposals in Predlagam have to receive an official response from the competent authority of the government of the Republic of Slovenia, if at least 3% of the users active in the previous 30 days voted in favour of the proposal, and if there are more votes in favour than opposed. Official responses are also required in the cases of the Finnish and European Citizens' Initiatives.
Almost all cases had some sort of formal link to the decision-making process. We therefore demarcated the condition 'link to the formal policy or political process', so that it must be clear which formal decision-making process it concerns: the link has to connect the participation process to, e.g. a specific policy document, specific public funds, a specific internal democratic party process or an election. This means that although the Finnish Citizen Initiative is legally embedded in the parliamentary decision-making process by an amendment to the Constitution which allows citizens' initiatives to be submitted to Parliament, the participatory tool is meant to generate ideas for new policies and bills. Therefore, the tool does not link the citizens' input to a specific existing formal decision-making process on a certain topic.
A carefully designed link facilitates the political uptake of a proposal, request, or any other input from citizens. A link to the formal decision-making process does not necessarily signify that the outcomes of the e-participation initiative are legally binding. In fact, generally this is not the case. In the Berlin-Lichtenberg case, it is written in the 'Rahmenkonzeption' that citizens can suggest and discuss how public money should be spent, but that politicians will decide which suggestion will be included in the budget plan (Bezirksamt Lichtenberg von Berlin 2008).
We discerned two types of links, on the basis of their different roles in the policy cycle: (a) Link to a formal currently existing agenda-setting process (10 of the 22 cases score positively); (b) Link to a formal currently existing policy or political decision-making process (15 of the 22 cases score positively).
The cases score positively on either of these two conditions when the link facilitates the input of the participants to be taken up in one or both of these phases of the policy cycle. The cases that score negatively on condition (a) (agenda-setting process) are cases like the Dutch e-petition case, or the Slovenian Predlagam case, or the European cases ECI and ECC09. Positive scores are assigned to participatory budgeting cases (established link to existing political discussions about budgets) and political parties' cases (established link to the political agenda of the parties in question). More cases score positively on condition b than a. This is because some of the tools are just not agenda-setting tools, such as the e-voting cases and Belo Horizonte.

Sustainability
The sustainability of a digital participation tool was characterised by whether or not provisions for the future, like maintenance and improvement or expansion of the tool, are taken. For example, were user experiences used to improve the tool? This condition is taken from a study by Panopoulou et al. (2014) which attempted to determine the success factors for e-participatory projects, based on reviewed literature on e-government and e-participation success and on a survey of practitioners across Europe. Badouard (2010) studied different EU participatory tools and concludes that important conditions for the sustainability of participative instruments are their official status and a legal framework on their position in the decision-making process. Sustainability was seen as a success factor in the literature as well as by the practitioners. There are different reasons why the sustainability of a tool is important for success, an important one being the attempt to improve the userfriendliness of the tool. Or, as one of the developers of the Betri Reykjavik tool said: 'We are always working on simplifying the process, in terms of how to participate. And that, I think, is in general a weakness of participatory processes, that they can be too complicated'. In other cases, improvements have been made over time to increase positive responses from government authorities. This is highlighted by the case of Predlagam, which introduced a lower limit of endorsing five or six proposals on a monthly basis to the competent authorities, so they now carry more weight and are more likely to succeed.
The majority of the cases (14/22) have a positive score for this condition: abgeordenetenwatch.de, theyworkforyou.com, PB Paris, PB Berlin-Lichtenberg, PB Belo Horizonte, Betri Reykjavik, E-voting Estonia, E-voting Switzerland, Five Star Movement, Your Voice in Europe, Futurium, Podemos, Dutch e-petitions and Predlagam. Cases of tools which have not been used repeatedly scored negatively, for instance, the European Citizen's Consultation (ECC09) or the Iceland constitutional crowdsourcing case. One interviewee is quite critical of the lack of sustainability of the ECC: 'They [European Union] are aware that we need to try to find new ways of involving citizens. So that's why they have been spending all this money. But then they are doing a one-shot experiment and they don't include it into the decision-making process. That is a problem. They don't think of a long-term solution for implementing citizen participation at the EU level. So it cannot work. Then it's better to do nothing'.
Sometimes, tools which have existed for a longer period of time were not, or only marginally, improved and scored negatively on the sustainability condition as well. Such an example is the Dutch e-consultation website internetconsultatie.nl or-at least until more recent times, as evidenced by the proposal for revisions from September 2018-the European Citizens' Initiative. This might be explained by a lack of political urgency or willingness. Other reasons for a lack of sustainability can be a lack of funding, as in the case of the Open Ministry in Finland, which caused the downfall of the Open Ministry as a crowdsourcing service platform.

Communication or Engagement Strategy
This condition reflects on the communication or engagement strategies used to mobilise participants. Questions raised here are: Has the possibility to participate been effectively communicated to the target group? Have different strategies been used to attract different target groups? Has the strategy succeeded in mobilising different groups of citizens to use the tool? In Panopoulou et al. (2014), a 'promotion plan' was mentioned as a success factor for designing e-participation initiatives, defined in terms of utilising the most appropriate promotional activities for each stakeholder group. The engagement and communication strategy can thus be very significant in predicting the outcomes of the e-participation process. A lack of diversity amongst participants, and/or low representativeness of the participants, can result in decreasing interest from policy-and decision-makers in the input, and therefore in lower impact.
Mobilisation has proven to be one of the great challenges of participatory projects in general. One of the explanations is that citizens have low confidence that their input in such projects will have any real weight in the decision-making processes. When it comes to e-participation at the EU level, this scepticism appears to be wellfounded, as is made clear in the literature review (see Chap. 4). Deliberative civic engagement tends not to be embedded in political decision-making, often being short-lived and temporary and focused on single issues. There can also be a lack of support and engagement from decision-makers. Other barriers preventing mobilisation are language problems and a low interest in European-level matters.
Some of the tools have facilitated different e-participation trajectories, such as the Dutch e-consultation website, Futurium, Your Voice in Europe and the ECI. In these cases, there is quite a lot of variation between the different trajectories. In order to assess the score for the communication and engagement strategy of these tools, we therefore took into account to what extent the tool/platform itself is well-known.
In half of the cases (11/22), an effective communication or engagement strategy was in place. The mass media are important mediators in several cases, and the attention of the mass media for the tool and the participatory process is generally important for mobilising participants. As in the case of abgeordnetwatch.de, the annual report of the monitoring website states media partners serve as important crowd-pullers, with one-third of visitors finding the platform through media. This can be seen in other cases, including Predlagam, German Pirate Party, Podemos and the Five Star Movement. Two of the participatory budgeting cases also received a lot of media attention (Belo Horizonte and Paris). However, in these 6/22 cases, media attention has not been constant. After the first buzz around the launch of the initiative, the attention of the media regressed.
Different target groups require different engagement strategies. In order to reach a high diversity of participants, it can be important to have an offline communication strategy as well. This might be easier to organise for local initiatives, like the participatory budgeting case of Berlin-Lichtenberg. The researcher and administrator interviewed for this case stated that decentralised meetings in community centres were an important way for community workers to reach new people every year and to get them involved in the participatory budgeting for the district.
In the other half of the cases (11/22), the general public appears not to be familiar with the tool, and lay citizens were not mobilised. This was the case for Predlagam, the Dutch e-consultation, theyworkforyou.com, Wiki Melbourne and Open Ministry (related to the Finnish CI). In the Dutch e-consultation case, the researcher interviewed noted that some civil servants did not have a problem with the tool being unknown to the general public; they did not want too many responses in the consultations and only wanted a few people who knew the ins and outs to react. Remarkably, all European-level cases also score low on their engagement strategy: the Green Primary, Futurium, ECI, Your Voice in Europe and ECC. In the European-level cases, not much effort has been invested in gaining a broader reputation amongst target groups other than the usual suspects (civil society organisations at European level).
Sometimes an active large-scale engagement strategy is not required to mobilise participants. The Dutch e-petition site gets about 2 million visitors per month without having to spend one euro on it. It gets its name and fame mostly through a snowball effect via social media, and more importantly-according to the foundere-mail as well. The low threshold of participation in this tool-sign a petition by entering your name and e-mail address-plays an important role here, as well as easy ways to share e-petitions via social media and e-mail.

Clarity on the Process
This condition reflects how clearly the participation process has been organised (for participants) and to what extent expectations about the process are managed properly. Is it clear to participants from the outset what the goals of the process are? How far does their influence reach? What will be done with their input? Is it clear to participants which actors have responsibilities in the decision-making process? For example, an analysis of the ECC by Karlsson (2011) shows that members of the European Parliament (MEPs) as well as participating citizens have been disappointed in the participation tool. Karlsson found the design of the ECC project, at least in part, as being responsible for the failure. The procedure suffered from a lack of clarity over what inputs are desired by the MEPs and which inputs are expected from the citizens.
In 15/22 cases, clarity for participants had been adequately delivered on the participatory process: abgeordnetwatch.de, theyworkforyou.com, PB Paris, PB Berlin-Lichtenberg, PB Belo Horizonte, Betri Reykjavik, E-voting Estonia, E-voting Switzerland, Five Star Movement, German Pirate Party, Your Voice in Europe, Wiki Melbourne, Green Primary, Constitution Iceland and Open Ministry (Finnish Citizen Initiative).
The City of Paris provides extensive information about the participatory budgeting process. Firstly, the website provides infographics, FAQs and information, which explain the process of Budget Participative and how to participate. In the proposal submission phase, information regarding the legal framework and support on financial aspects is provided by the administration to the submitting participant(s). Also in the case of Melbourne, as well as in the case of participatory budgeting in Berlin, the expectations on the process were well-managed online as well as offline. The organisers in Melbourne were clear that: 'There is no guarantee that all suggestions can be incorporated into the Future Melbourne draft plan. A number of the recommendations fall outside the City of Melbourne's areas of responsibility'. This kind of transparency did not seem to discourage participants.
In other tools, clarity was particularly lacking with regard to the decision-making process and how the input of participants is part of that process. One of the interviewees on Predlagam argued: 'The policy process is very complex. And citizens should be aware how complex it is. I don't think that they should be fooled. And in this case, in the case of this tool, I think they are being fooled, because there are still a lot of proposals and they are just going into a blackbox where nothing happens with them'. For the European Citizens' Consultation 2009, the argument was that: 'So the process in itself was clearly presented and well communicated but the organisers were unable to say what would be the impact'. And there are more cases like this. In the European Citizens' Initiative and Podemos, the official steps in the participation process are clear, but almost no proposal reaches the final stage. Politicians from Podemos claim to incorporate input from the online discussions in their considerations, but it is not clear how this indirect influence of participants actually works in practice.
Clarity on the process is supposed to encourage and empower participants, and ultimately it should prevent participants being disappointed. However, disappointment can be found in several cases. Beside the Podemos case, the digital budgeting case in Belo Horizonte is the most striking. The winning project in 2008 has not been finished because there is a problem in terms of land use and land ownership. After this disappointment, participation has fallen significantly: from 124,320 citizens in 2008, to 25,378 in 2011 and 8900 in 2013. Trust is hard to gain but easy to lose. In the case of the crowdsourcing of the constitution in Iceland, the transparency of the participation process seems to have created a lot of public appreciation and even a sense of co-ownership with the participants, according to one of the interviewees. Impact on decision-making is easier to achieve if it is clear beforehand exactly how the participatory process will contribute to the final decisions.

Possibility to Interact with Other Participants
We were also interested in how the diversity of views is managed within the different tools. Does the tool offer the possibility to deliberate? Deliberation is broadly defined here as the opportunity for participants to exchange views within the digital tool(s) available in the case.
The need for deliberative possibilities in e-participation projects is debated in the literature. Deliberation is supposed to enhance input quality when it comes to e-consultation (Albrecht 2012), and Albrecht advocates a model of deliberative e-consultations, which not only consists of collecting comments on a policy proposal but also allows for discussions on these amongst the participants and with representatives of the EU institutions concerned (see next condition). Organ (2014) points out that even if no legal outcomes of e-participation are achieved, the legitimacy of the policy agenda can be increased through the act of deliberation. However, deliberative civic engagement seems to be of a temporary nature, being employed for single issues and spanning only a short amount of time (Leighninger 2012).
Kersting criticises the quality of online deliberative instruments, which appear to be '[. . .] more oriented towards the construction of identity and community building than towards political dialogue and deliberation' (Kersting 2013: 270). He also observes that web forums on the internet are low in deliberative quality, meaning that '[. . .] they are not argumentatively-respectful and consensus-oriented, but are often pure monologues and frequently aggressive' (Kersting 2013: 277). Another interesting argument against deliberation, but pro voting or signing, was made by a researcher who studied petities.nl: 'You can only sign or not sign. You cannot co-edit a text for example. At the same time, your voice is not lost as happens often in deliberative settings where a participant can take part in a discussion but where in the end it is difficult to ascertain where and how one's input has been used. With petitions, your voice just counts'. The added value of participation in a digital tool thus seems to depend on how the deliberation is organised and the extent to which people use the options provided.
In 13/22 cases, it was possible to interact with other participants in the online tool. Where crowdsourcing was used to co-create a proposal, the tools facilitated deliberation between participants: Open Ministry related to the Citizens' Initiative in Finland, the constitutional crowdsourcing process in Iceland, Wiki Melbourne and Predlagam. Registered users of the political parties who are also aiming for collaborative decision-making (German Pirate Party, Podemos and Five Star Movement) have several tools at their disposal to debate issues. These include the European Citizens' Consultation 09 and the Futurium.
The four participatory budgeting tools include the possibility to comment on proposals to spend the municipal budget. This worked particularly well in the case of Betri Reykjavik, where the most popular arguments against the proposal were presented next to the most popular arguments in favour of it. One of the interviewees mentioned that by structuring the debate in this way, views are exchanged strictly by arguing for or against proposals, which helped to improve its quality: 'What we tried to do was to split the screen in two so people who support the idea can write points for it on the left side of the screen (. . .), and on the right side of the screen, people who are against the idea can put their points. . . And almost overnight (. . .) the quality level of the debate increased a lot'. This approach minimises the extent to which a comment can refer to another comment rather than the proposal itself: 'If you see a point you don't agree with, there's no way to comment on it. You have to write a counterpoint'.
In the case studies of Wiki Melbourne and the German Pirate Party, the exchange of ideas was seen as stimulating a more constructive mindset amongst participants rather than just approving or disapproving of ideas. However, the possibility to interact does not equal deliberative quality. In some cases in which interaction between participants was facilitated, like the PB in Berlin-Lichtenberg, the diversity of views on the different proposals appeared to be limited: only a few reactions can be found online. In the case of ECC09, the online deliberation varied widely between countries.

Possibility to Interact with Decision-Makers
This condition reflects whether the tool offers the possibility to deliberate with decision-makers. As with the former condition, deliberation in this context means the opportunity to ask questions and/or exchange views. Decision-makers can be administrators as well as politicians. Do they participate in the online tool? Barrett et al. (2012) mention that in order for deliberative civic engagement processes to be successful, one needs the engagement of public officials and politicians. Another example showing the same is OurSpace, an international project dedicated to improving the engagement of the youth of Europe with European decision-makers through the combination of ICT use, information and motivation to participate. The engagement of decision-makers was in the end an important factor in the success of the project (Parycek et al. 2014).
Interactions between decision-makers and participants contribute to a better match between the needs of decision-makers and citizens' input and to the quality of the input. Research on the case of ECC09 brought to light that politicians criticised participants for not understanding political reality and therefore recommends a 'meet and greet' between politicians-in this case MEPs-and participants at an early stage (Karlsson 2010). In this way they can exchange perspectives and knowledge before the content of the proposals is decided upon. The interaction between participants and decision-makers would thus improve the quality of the output (i.e. closer to political reality), and therefore probably the impact of the participatory input on political agendas or final decisions.
This same argument was made by the interviewed researcher who studied Predlagam, who claimed the tool was too open and recommended it should provide more information on what kind of input the government wants from citizens and should provide more such structures in its design. Furthermore, the initiator of Open Ministry proposed an improvement of the participatory process around citizens' initiatives, whereby citizens would work together with the parliamentary committee. The hope was that it would stimulate a discussion between citizens and politicians on the content of the proposal to increase mutual understanding which, in the end, might help to improve the legal quality of the law proposal.
In 8/22 cases, there is some form of interaction between the participants and the decision-makers. In 5/22 cases, this interaction takes place between participants and politicians, including in all four cases of the political parties, where public servants are involved (Five Star Movement, Podemos, German Pirate Party and-only in incidental Facebook chats-in the online Green Primary), as well as in the case of abgeordenetenwatch.de, where Q&As between politicians and citizens are moderated. In Wiki Melbourne a team of city officers answered questions by participants, corrected factual errors made in edits, linked citizens to relevant documents and updated participants on events and developments concerning the project. In the participatory budgeting cases of Berlin-Lichtenberg and Paris, policy officers also interacted with citizens about their proposals.

Quantitative Aggregation
Quantitative aggregation is easily done online, and the numbers provide an indication of the level of support for a proposal. This indication is relevant for decisionmakers in considering the proposal. When a proposal is supported by many people, this might increase the chances for political uptake of these ideas. However, political willingness is also necessary. The crowdsourced constitution in Iceland gained the support of 67% of voters during a referendum (voter turnout was 49%), but still the constitution was not voted upon by the parliament due to political unwillingness. At the same time, one should be careful of giving too much weight and meaning to voting results in digital tools, taking into consideration that the representativeness of the participants could be low.
A total of 17/22 cases use some form of quantitative aggregation. To be able to make an appropriate comparison, we have further specified this condition by distinguishing between: • Voting on (or signing for) proposals with the aim to reach a certain threshold ( The first type of votes, often in the form of signatures, is collected in the agendasetting phase. An example can be found in the Predlagam case, where at least 3% of users that were active in the previous 30 days need to have voted in favour of the proposal. The cases of the political parties of Podemos and the German Pirate Party also show comparable procedures for individual ideas, which need to reach a certain level of support before the proposals are taken into further consideration. Other examples are the Finnish Citizens' Initiative or the European Citizens' Initiative, where 50,000 and 10,00,000 signatures are needed, respectively. When these thresholds are met, the Finnish parliament is obliged to discuss the proposal and vote on it, and the European Commission must examine the proposal for legislation and decide whether or not the initiative warrants taking legislative steps. The second type of voting takes place in a later phase of the decision-making process. These are votes for specific proposals in order to prioritise the range of proposals, or votes in elections and referenda. An example of this second kind of voting is the participatory budgeting case in Berlin-Lichtenberg, where different budget proposals are voted upon by participants online and via surveys, resulting in a top ten. In the participatory budgeting case in Paris, the online and offline votes on specific proposals in the final phase of the process determine which projects receive the estimated budgets. Another example is the ECC09, where 88 recommendations from the national consultations were presented on each national website; the 1635 participants were asked to vote (online or by mail) for 15 recommendations that they wanted to become the final result of the ECC.

Feedback to Participants
This condition reflects the extent to which participants receive feedback from the organisers and/or the addressees, such as administrators or politicians, on (a) their contributions and (b) the final decisions (i.e. do they get informed about the way their contributions have been used?).
Feedback is significant because it relates to the trust participants have in the process and the political system. The interviewed organiser of Wiki Melbourne put it as follows: 'It is almost like you extend the respect to people as if they were sitting in a room talking to you. You would expect to have to respond to them. Otherwise it's just plain rude, right? [. . .] If you take that mind-set, you just leave a comment: 'I just moved this over to this section, because it seemed more appropriate over here' or 'Sorry, that point, we're not legally able to change that part of the law, so I had to delete it. But I'll point you to the state government body who is responsible for that.' It is those types of contributions and changes that maintain the trust during the process'. The organiser of Betri Reykjavik who was interviewed is also insistent about the importance of proper feedback, in terms of common courtesy. 'And obviously at the end, when the idea is agreed on or rejected, then everybody gets an email as well. It's super important [. . .]. Otherwise, you're really not respecting people's time'. This is confirmed in a survey amongst participants of the Dutch e-consultation case, where participants indicated that participation should be rewarded more, for example by ensuring that responses are published on the site without delay.
Feedback, even if the message is that the participants' input is not going to be used, can increase the democratic value of the tool: 'It is more about participating in a democratic process. To me, a petition is also a success when the answer of a recipient is: 'sorry, that is not going to happen, for this and this reason.' After which the signatories might even agree', according to the initiator of petitie.nl. In the case of Predlagam, it turns out that despite the high amount of negative responses, users appreciate the feedback the ministry provides, as it shows it is giving adequate consideration to their suggestions. In contrast, in cases where participants perceive responses to be standardised, cynicism increases. Badouard (2010) argues in his study on EU participation tools that obligations to provide adequate feedback also create some pressure on the decisions to be taken and the recognition as a policy instrument, together with institutional accountability, and it brings the Commission to acknowledge the participants as legitimate political actors.
When the organisation is able to provide feedback to participants, it is a sign of a well-organised participation process. Feedback implies that the organisation knows how it can and will use the input of participants, or why it can't or won't. The impact on decision-making thus gets deliberated in the process. Such feedback was given in 14/22 cases. These were: abgeordenetewatch.de, Predlagam, Open Ministry and the Finnish Citizens' Initiative, constitutional crowdsourcing in Iceland, Wiki Melbourne, Berlin-Lichtenberg, Futurium, Five Star Movement, PB Belo Horizonte, PB Paris, Betri Reykjavik and the three e-voting cases.
The extent to which feedback was given differs amongst these cases. Some of them can be considered best practices when it comes to providing feedback. Digital tools can be very supportive in providing transparency about the participatory outcomes and final decisions. For example, the wiki tool used by the municipality in Melbourne to open up the vision document for input was an instrument to maximise transparency. All contributions throughout the process and the outcomes of offline activities were fed back into this wiki by City of Melbourne officers. The wiki tool manages revisions and shows participants what has happened with their contributions. Also in the case of Betri Reykjavik, the website forum, the municipality website and emails are used to inform citizens about developments in the decisionmaking process, as well as implementation and later developments (Bjarnason and Grimsson 2016): 'If there's an idea that is going into processing, people can track it on the website (. . .) and each time there's a status update, you know, it goes into a committee and is discussed and there are meeting notes, they are sent to all the participants' (Interview 39, organiser).
When we look at the cases which score negatively on providing feedback to participants, it is striking that it is especially the tools at the EU level that often fail to provide proper feedback: ECI, Your Voice in Europe and ECC09. However, from the literature review, we know that the European Parliament Petitions Portal has been improved regarding this point. In November 2014 a new petitions web portal was introduced, possessing more feedback features on the status of petitions (alongside more information on the Parliament's areas of competence). In the case of ECI, the information supplied by the website itself is generally very good, with exceptions in the area of result feedback, where there is a lack of clear organiser feedback to supporting citizens due to a gap in the existing online collection system. The recent proposal for revision of the ECI addresses this by allowing organisers or the European Commission to collect email addresses to improve communication efforts. With Your Voice in Europe, a synopsis report on the outcomes of an e-consultation is required, but in many cases, it is not provided (yet). And in the case of ECC09, no feedback was given on the final outcomes of the process.

Assessment of the Outcomes
A common critique of e-participation practices at the EU level is that they are a successful civic instrument but not a convincing policy instrument (as Kies and Nanz 2013: 24, with regard to ECC). It seems to be an ongoing theme that e-participative projects might provide added personal value for participants and community capacity, but suffer from a lack of direct, or even indirect, political impact. Impact on the policy or political agenda, or on the final decisions made, have therefore been the focus of this study.
This study identified two key outcome factors defining a positive result for the different e-participation tools: • Impact on the final decisions; • Impact on policy or political agenda-setting.

Impact on Final Decisions
The outcome measure 'Impact on final decisions' reflects the extent to which the results of e-participation initiatives were taken up by the policy-makers and/or politicians and actually influenced their final decisions. Van Dijk (2012) calls the outcome 'influence on political decisions': 'The decisive touchstone of eParticipation in terms of democracy'. The most relevant question we considered was as follows: Is the majority of the input suggested by the participants recognisably incorporated in law proposals, policy documents such as EU Communications, political party programmes or election results and/or implemented in municipal budgets, etc.? Did the participatory input have a substantive and/or repeated impact on decisions made?
In some cases, the participatory input entailed many different proposals/consultations, such as Predlagam, Open Ministry (Finnish Citizen Initiative), the Dutch e-consultation and Your Voice in Europe. In these cases, we scored whether the majority of the input had an identifiable impact. On the basis of desk research, questionnaires and interviews, it was assessed that there was a substantive impact on the final decision in 11/22 cases. That half of the cases show an impact on decisionmaking is a positive result, since in the literature it is generally concluded that few decisions of government, political representatives or civil servants have changed on account of the input of citizens through e-participation. Van Dijk (2010) concluded that 'scarcely any influence of eParticipation on institutional policy and politics can be observed yet' (Van Dijk 2010). Millard et al. (2008: 76) wrote: 'Most administrations do not (yet) have mechanisms and capacities in place to cope with a significant increase in participation'. This share of positive outcomes within the cases examined might be explained by the case selection. Many cases were individually requested in the project specifications, defined by STOA, in order to learn how to strengthen participatory and direct democracy.
The 11/22 cases which score positively on 'Actual impact on final decisions' are Wiki Melbourne, PB Berlin-Lichtenberg, Your Voice in Europe, German Pirate Party, Five Star Movement, PB Belo Horizonte, PB Paris, PB Betri Reykjavik, the Green Primary and e-voting in Estonia and Switzerland. It is interesting to note that all three of the e-voting cases and the four participatory budgeting cases have an impact on the final decisions. For the e-voting cases, this may not be that surprising, since voting is a legal right with direct impact. And the literature review predicted that when it comes to influencing decision-making, the area of e-budgeting has produced some of the strongest results.

Impact on Policy or Political Agenda-Setting
This outcome factor is related to the outcome factor of 'Impact on final decisions' but focuses on an earlier phase within the policy cycle: the agenda-setting phase. We assessed for each of the cases if the input to the online participation process has had a substantive and/or repeated effect on the policy or political agenda. Impact on the policy or political agenda concerns the effects of the contributions from e-participation on the political or policy debate, without necessarily influencing the decision-making process per se. For instance, in the case of the Finnish Citizen Initiative, 15 legislative proposals by citizens reached the threshold of 50.000 signatures to be debated in parliament. These proposals were handled properly: initiators are heard by committees, and these committee hearings were open to all MPs and to the media (which was a novelty). However, only one of these citizens' initiatives has led to changes in the law: the gender-neutral marriage legislation. The input of citizens in the form of legislative proposals did have a significant and repeated impact on the political agenda, but the impact on final decisions lags behind. The Iceland case also scores positively on 'agenda-setting' while not having an impact on the final decisions. The Constitutional Council of 25 citizens presented its draft constitution to Althingi, the House of Representatives in Iceland, where it was discussed. However, the draft met resistance from politicians which led to troubled parliamentary deliberations. A referendum on the draft constitution followed, with a majority in favour of its adoption. However, the impact on the decision-making process remained zero, since in the end parliament never took up the proposed constitution, it was never brought to vote, and it never went into effect.
A total of 11/22 cases score positively on the outcome factor 'Impact on policy or political agenda-setting'. Two positive cases have already been mentioned: the Finnish Citizen Initiative (with the involvement of Open Ministry) and Iceland constitutional crowdsourcing. The other cases are The EC tool Your voice in Europe, the Dutch e-consultation, Wiki Melbourne, Futurium, participatory budgeting in Berlin, Paris and Reykjavik and the collective decision-making tools of the German Pirate Party and Five Star Movement.

Analysis of Configurations
Qualitative Comparative Analysis enables systematic analysis of the conditions that are necessary and/or sufficient to produce an outcome. In the previous section, the data collected was explored, scores assigned and the cases compared for the nine conditions. In this section the data is minimised by grouping the cases that have the same scores on relevant conditions and the outcome. The resulting tables, in which the cases that show similar configurations are clustered, are called truth tables. Through these steps, similarities and differences between cases on the conditions and outcome values come to light systematically. The different paths towards the outcomes 'Impact on final decision' and 'Impact on political or policy agenda' are assessed in the final steps of the csQCA.

Impact on Decision-Making
Two cases are eliminated from the analysis of configurations for the outcome impact on final decisions. The two monitoring websites do not aim to have an impact on final decisions and are therefore not included in this truth table. The six conditions included in the truth table appeared to have a stronger connection with the outcome than the other conditions we assessed in the case studies; these six conditions showed frequent presence in combination with the positive outcome (and non-presence in relation to the negative outcome) (Table 12.1).
Out of the 20 cases in this truth table, 12 show significant impact on final decisions. Seven of these twelve cases (7/12) score positively on all six conditions Participatory Budgeting (PB) in Paris, PB in Berlin-Lichtenberg, PB in Belo Horizonte, Betri Reykjavik, e-voting in Estonia, e-voting in Switzerland and the Five Star Movement. The findings suggest that having impact on final decisions involves: 1. Creating a link to formal decision-making (in these cases via embeddedness in the policy process, elections/referenda and official political representation) 2. A digital tool that has existed for a while and where several alterations have been made to improve the participatory process (sustainability) 3. An active mobilisation and engagement strategy 4. Clarity on the participatory process and its contribution to the overall decisionmaking process from the start (for the participants) 5. Providing feedback to participants 6. Including an option where participants can vote to decide via prioritising proposals or elections/referenda The other five (5/12) cases show that not all six conditions are necessary to produce the outcome. The Pirate Party in Germany is positively rated on the link to formal decision-making (1), the mobilisation strategy (3), clarification of the  participatory process (4) and the possibility to vote to (co-)decide (6). But the political opinion formation and decision-making software called Liquid Feedback-the backbone of online democratic processes within the German Pirate Party-appeared not to be sustainable, and insufficient feedback to participants was provided. Your Voice in Europe has positive scores on the link to formal decision-making (1), the sustainability of the tool (2) and clarity for participants on the participatory process (4), but not on the other three conditions. Futurium is linked to formal decision-making (1), is sustainable as a tool (2) and also provides feedback to participants (5). The path of Wiki Melbourne also includes a link to the formal decision-making process (1) and has a clearly communicated participatory process (4), and feedback is provided to participants (5).
One of the configurations is inconsistent: the combination of (1) a link to formal decision-making; (4) a clearly communicated participation process; and (6) the possibility to vote, corresponds with a positive as well as a negative outcome. These conditions are positively scored in the Green Primary case as well as in the case of the Iceland constitution. The contradictory configurations can be explained by a difference in the type of links to formal decision-making. Although the cases both have a link to the formal decision-making process, in the Iceland constitution case, the link still leaves a lot of room to the decision-makers in the Icelandic Parliament. The link entails a first parliamentary constitutional committee, which initiated the Constitutional Council with 25 member citizens from Iceland. Subsequently another governmental committee was established to prepare further decision-making about the new constitution. This committee published a provisional report in the spring of 2014, which identified the Constitutional Council's draft as one of several possible alternatives for a new constitution, thereby leaving the draft constitution on ice (negative impact on final decision-making). In the Green Primary the online voting result is directly translated into the election of two 'Spitzenkandidaten' (top-ranked candidates), which leaves no room to make a different decision (positive impact on final decision-making). This comparison of cases makes clear that there are different paths to impact on the final decision; different combinations of conditions can lead to the same outcome. The path with six positive conditions shows consistency and explains seven cases, which makes it an empirically stronger result than the five individual paths in which two or three of the conditions are lacking and where one path is inconsistent.
The minimisation of the configurations, without logical remainders (unobserved cases), leads to the following formula ( Fig. 12.1): The link to formal decision-making is present in all configurations with a positive outcome. The minimisation formula thus clearly shows that it is necessary to establish a link to the formal decision-making process that organises the potential uptake of the participatory input. Eleven of the twelve figurations also include the condition that the participatory process and its aims are sufficiently clarified from the start. Strictly speaking this is thus not a necessary condition since it is not present in all configurations. The importance of the condition of a clear process is however supported by the fact that none of the cases that have a negative outcome score positively on both of the conditions 'link to formal decision-making' and 'participatory process clarified' (excluding the case of the Iceland constitution discussed above). To emphasise its importance, only one other case with a negative outcome-Finnish Citizens' Initiative via Open Ministry-clarified the participatory process and its aims.
The link to formal decision-making, even in combination with the clarification of the participatory process, is however not sufficient to produce the outcome. To create impact on final decisions, it also helps to have a sustainable tool, which has been improved over time (9/12 cases), to have an active mobilisation and engagement strategy (8/12 cases), to provide feedback to participants (10/12 cases) and to include a possibility to vote (9/12 cases).

Conclusions: How to Organise Impact on Decisions at a European Level
What do we learn from this descriptive formula? The six conditions included in the configurations for impact on decision-making mostly have to do with a clearly organised participation process in which the expectations of participants and decision-makers are well-managed from the beginning. Interesting to note is that three of the six conditions are in fact easier to meet with online participation tools. For example, digital tools are very useful in creating transparency and accountability, providing (a) clarity on the participatory process and (b) feedback on the results. Furthermore, it is an advantage of online participation practices to combine deliberative processes with (c) voting processes. This study is aimed at drawing lessons from the comparative analysis for the EU level. Are the observed conditions that contribute to impact on decision-making present in all the studied participation tools at the EU level? If not, could the factors be realised at the EU level, or do particular challenges arise at the EU level? We discuss the different factors below, ranked via frequency.
• Starting with what we identified as the most significant necessary condition: a link to a specific formal decision-making process (present in all configurations with a positive outcome). Two observed EU-level cases-Futurium and Your Voice in Europe-have such links established. When we look at the tool of Futurium, we see that the more recent consultations of the tool, such as 'eGovernment4EU' and 'Digital4Science' are linked respectively to the 'eGovernment Action Plan (2016-2020)' and the 'Future Horizon 2020 Work Programme (2018-20)'. The earlier consultation 'Digital Futures' was linked to the European Strategy and Policy Analysis System (ESPAS) and Horizon 2020's strategic programming exercise 2016-18. When we take a look at Your Voice of Europe, we see that participants are asked to deliver input through questionnaires for specific policy proposals, which are regularly influenced by the online input. In contrast, the Europeans' Citizens Initiative case scored negatively, since the ECI facilitates new ideas to be raised by participants, which do not necessarily relate to a specific existing formal policy. The European Citizens Consultation (ECC) case was also very broad in scope and not linked to a specific policy or political process, which makes it more difficult to create actual impact in policy or politics. Proposals that are too general do not match the needs of decision-makers, as we saw in the case of the ECC, as well as in the cases of Predlagam and the Finnish CI (Open Ministry). Moreover, without a link to a formal decision-making process, it is not clear who is responsible for processing the input in the decision-making process. • The second most important condition is clarity on the participatory process and its contribution to the overall decision-making process from the start, particularly from the perspective of participants (present in almost all configurations with a positive outcome). This indicates that the participation process should be wellembedded in the decision-making process, and participants and decision-makers have to know what to expect. At the EU level, it can be challenging to offer clarity on the overall decision-making process since it can be very complex, involving many different actors. Moreover, European citizens are less knowledgeable about EU decision-making processes than they are about national or local processes. From the four observed EU-level cases, only Your Voice in Europe scores positively on this condition. Futurium has recently improved this clarity for its more recent consultations; however, the process of the 'Digital Futures' consultation that we were able to monitor from beginning to end did not show this clarity. Your Voice in Europe clearly has made an effort to explain the consultation process and its aims in an accessible way. Accessibility, however, does not imply that the process should be oversimplified. Oversimplification can be counter-effective, as in the case of Predlagam. Because Predlagam does not pay attention to the overall, complex decision-making processes, it might-in the words of one of our respondents-'fool' people and will inevitably lead to disappointment of participants. In contrast, Wiki Melbourne and Berlin-Lichtenberg are best practices in being clear about the expectations participants can have. These platforms are also explicit in that there is no guarantee that each proposal will be implemented. The case of participatory budgeting in Paris shows how the use of infographics can help to clarify the participation process and its contribution to the final decisions. • A third important condition in the observed cases that succeeded in having impact on policy or political decisions is providing feedback to participants. Providing feedback is a sign of a well-organised process in which it is clear how exactly the participatory input has contributed to the decisions made. In addition, feedback to participants is a form of accountability. To make participation processes rewarding for citizens, their proposals should be given adequate consideration. Otherwise, it will lead to distrust not only in the participatory process itself but also in the political system as a whole. This is thus all the more important in the European context, given the democratic deficit and the negative public discourse around the EU. Yet, of the four observed EU-level cases, only Futurium provides considerable feedback to participants: Every participant in the Futurium consultation 'Digital Futures' received an email with the final report. The 'eGovernment4EU' consultation that is now running on the Futurium platform will not only provide information about its progress and results on the platform, but participants (i.e. proposers) will also be notified about the decisions on the platform and how the actions will be implemented. The other three EU-level cases in this study lack feedback mechanisms to the participants (ECI, ECC09, and Your Voice in Europe). One solution at the EU level might be-as is the case in Predlagam for example-the obligation for government or political authorities to provide feedback. This can create pressure on the actions to be taken and acknowledges participants as legitimate political actors. However, procedures alone are not enough, as the Your Voice in Europe case and the Dutch e-consultation case illustrate. And such an obligation should not result in standardised responses to citizens about their contributions and their impact. Time investment is required to make an accessible report or to create another form of feedback, and it might help to implement feedback options in the design of the tool. An inspiring example is the Participatory Budgeting site of Berlin-Lichtenberg, in which decisions on proposals are motivated in short messages in a 'traffic light-format' (green for accepted proposals, orange for proposals in process and red for rejected proposals).
• A fourth condition that contributes to impact on final decisions is sustainability. It takes time to organise a digital participation process to run smoothly, which often implies adjustment over time. In Futurium the tool was made more user-friendly on the basis of their experiences during the first project, Digital Futures. DG CONNECT organised three public workshops in 2015 to collect best practices, ideas and feedback on how to engage with stakeholders online, especially through Futurium. This can be seen as best practice: the users'/citizens' perspective is greatly involved in the evaluation of the tool and broader process. With regard to the Your Voice in Europe tool, the aim is to unify the separate consultation pages to improve the process. This leads to central management of the page internally. In addition, YVIE strives for simpler visual guidance and explanation of where a particular initiative currently is in the policy-making process. In contrast, the European Citizens' Initiative has not made much improvement, although several evaluations have made suggestions. The one-time experiment of the European Citizen Consultation is problematic; it was not well-implemented in existing decision-making processes, and this could not be revised in time. Time to learn and improve the digital tool is important in order to create impactful participation. Experiments are riskier and have less chance of success. • The possibility to vote was present in 9 of the 12 cases in which an impact on decisions made was detected. It is a particular advantage of online participation practices that votes can be easily collected and even combined with deliberative processes. The advantage of the combination of deliberation and voting is that it can show if the participative input is supported broadly, or not. The European Citizens' Consultation included such an option. The national consultations had resulted in 88 recommendations. Subsequently, participants were asked to vote for 15 recommendations that they wanted to be part of the final result of the ECC. However, because there was no link to a specific policy or political process, the results barely had any impact. The European Citizens' Initiative does include the option to sign a proposal, which also indicates the support for a proposal. But this sort of quantitative aggregation seems to have no significant impact, as other cases in our study illustrate (the Finnish Citizens' Initiative via Open Ministry, and the Dutch e-petitions case). This lack of impact can probably be explained by the link to the decision-making process: the signatures are collected in order to put a proposal on the agenda, but they leave all further interpretation to decisionmakers. • The sixth condition, an effective mobilisation and engagement strategy (3), is probably one of the greatest challenges of e-participation, especially for the EU institutions. All the cases on the EU level, including the Green Primary, score low on the condition of an effective mobilisation and engagement strategy. The challenge to mobilise and engage EU citizens is even larger than it is to mobilise citizens on the national or local level, since: -EU citizens form a very large and diverse group of people, who generally do not share a sense of European citizenship -Mass media form an important mediator in mobilising the general public on the national and local level, but they cannot be expected to play a comparable role at the EU level (for instance, due to the negative discourse about the EU and the different national foci on EU decision-making, related to national interests) A lesson from the primaries of the European Green Party is that it might help to create commitment of partners at the national level, who can help mobilise the national publics. Another important point to note here is that different target groups require various mobilisation and engagement strategies, for which serious investments are needed. The possibilities that transnational social media offer could also be further explored in this regard.

Agenda-Setting
Only 16/22 cases are included in the truth table on agenda-setting. The websites abgeordenetenwatch.de and theyworkforyou.com are aimed at monitoring politics; the e-voting cases (including the Green Primary) and PB Belo Horizonte are aimed at making final decisions. These six cases (6/22) are therefore excluded from this analysis.
In the analysis of configurations for agenda-setting, five conditions are included. It appeared that these five conditions have a stronger connection with agenda-setting than the other conditions measured in this study (Table 12.2).
Eleven cases scored positively on the outcome agenda-setting. For five of these cases (5/11)-Participatory Budgeting in Paris, Participatory Budgeting in Berlin-Lichtenberg, the Five Star Movement, the German Pirate Party and Wiki Melbourne-the path towards agenda-setting involved: 1. A link to a specific existing formal agenda in policy or politics 2. Clarity on the participatory process and its goals from the start (for the participants) 3. The possibility to participate offline as well as online 4. The possibility within the tool to interact with other participants 5. The possibility within the tool to interact with decision-makers Six cases that succeeded in setting the agenda (6/11) did not check all these boxes. The case of Betri Reykjavik, Your Voice in Europe and the Dutch ministerial e-consultation did not include the possibility to participate offline. In Betri Reykjavik and the Finnish CI (via Open Ministry), new ideas are raised by participants that do not necessarily relate to a specific existing formal agenda in policy or politics. In Your Voice in Europe and the Dutch e-consultation cases, interaction between participants is not facilitated by the tool. The tools of Open Ministry and the Finnish Citizens' Initiative, the Iceland constitution process, Betri Reykjavik, Futurium and the Dutch e-consultation do not offer the possibility to interact with decision-makers online. In the case of Futurium and the e-consultation in the Netherlands, it is not made sufficiently clear in the tool how the participation works and/or how the participatory input contributes to the decision-making process. Information on the participatory process and its aims for participants is lacking in these cases.
As this comparison demonstrates, in the truth table more unique pathways are identified for impact on agenda-setting processes than for impact on decisionmaking processes. The observed cases showed more variety in the paths towards political agenda-setting. The case of the Dutch e-consultation deserves attention  particularly, because this case only scores positively on the link to the formal decision-making process. Official policy around the ministerial e-consultation is that unless there is a valid reason why e-consultation does not suit the legislative process, it must be applied. Procedures also prescribe a report on the results of the e-consultation, which naturally facilitates an agenda-setting effect. However, this has only a modest impact on the policy agenda, which is not significant in all e-consultations. In some cases no input is collected. And in many cases, civil servants acknowledge they do not have much room to manoeuvre, as they are barely able to deviate from the law proposal that is already negotiated. Yet in other instances, knowledge from specialists or tacit knowledge is very valuable for policy-makers to improve the legislative proposal. Civil servants argue that in the majority of cases e-consultation improves the quality of legislative proposals that are subsequently discussed in parliamentary debate. Both interviewees in the case study on the Dutch e-consultation argue that as e-consultation is an obligatory step, and many civil servants are just doing their duty; this explains why the link to the formal policy agenda is such a decisive condition in this case. However, other non-observed conditions might also contribute to the agenda-setting effect of e-consultation, such as the available knowledge at the ministry on the subject and the quality of the contributions. The minimisation of the configurations, without logical remainders, leads to the following formulas for agenda-setting ( Fig. 12.2): The minimisation results in two different formulas. The first formula represents nine cases (9/11) and has therefore a stronger empirical basis than the second formula which stands for two cases (2/11). The first formula indicates that it is necessary to create a link to a specific existing formal policy or political agenda. The links have different forms in the cases. Links are established that connect the digital participatory input through official municipal budgeting processes (PB Paris, PB Berlin-Lichtenberg, PB Betri Reykjavik). Other links are created through official political representation (German Pirate Party, Five Star Movement, PB Betri Reykjavik), or via a law on Citizens' Initiatives to parliamentary debate (Finnish CI via Open Ministry) and via consultation in official policy-making processes (Wiki Melbourne, Dutch e-consultation, Your Voice in Europe, Futurium). However, in most cases more conditions need to be met in order to set the agenda. This is also clear from the cases with a negative outcome; in four of these five cases, a link to the formal decision-making process is there, but is not enough to substantially or repeatedly affect the political or policy agenda (European Citizen Consultation, Predlagam, Dutch e-petitions, European Citizens' Initiative).
The second formula describes the combination of interaction between participants and a clarified participative process as necessary conditions to succeed in setting the agenda. This combination of conditions is not present in any of the five cases with a negative outcome.
In nine of the eleven (9/11) cases in which an agenda-setting effect is measured, interactions between participants are facilitated in the online tools. Deliberation might increase the quality of proposals and/or shows the social support for proposals. Clear goals and procedures contributed to the substantial or repeated effect on the political or policy agenda in nine of the eleven (9/11) observed cases, as well as a combination of online and offline participation tools (8/11 cases). A final interesting finding is that interactions between decision-makers and participants within the digital participatory process are facilitated in 5 of the 11 cases (5/11) with an agenda-setting effect, while this was only facilitated in 1 case without an agenda-setting effect.

Conclusions: How to Organise Impact on European Agendas
The outcome of the csQCA about agenda-setting is less clear-cut than the outcome of the csQCA about 'Impact on final decisions', but what can we learn from this descriptive formula? The five conditions included in the configurations that show an effect on the policy or political agenda mostly have to do with how the participation process is organised and the type of participation that is facilitated. Just as with the outcome 'Impact on final decisions', a link to an existing policy or political process, as well as the clarity of the participation process are important factors for 'Impact on agenda-setting'. Three further factors appear to contribute to an impact on agenda-setting. In the section below, we therefore address the following questions: Are these three factors present in the EU-level cases? And if not, could the factors be realised at EU level or do particular challenges arise at the EU level?
• First, the possibility to participate offline as well as online. A combination of online and offline participation improves the inclusiveness of the tool. Any method to improve inclusiveness is important at the EU level, since most EU citizens feel detached from the EU and generally there is a low level of interest in matters at European level. Combining online and offline participation is also important to overcome the digital divide, which is present in various EU countries, although with regard to access to basic broadband this no longer maps onto the underlying divides between richer and poorer regions of Europe (Negreiro 2015). Three of the EU tools in the cases already provide offline participation possibilities: Futurium encompasses many 'engagement activities', including offline meetings or workshops that feed the online discussions, and vice versa; ECC started off with an online phase to collect as many proposals as possible, which in their turn formed input to the national offline consultations; ECI offers the opportunity to sign an initiative offline (alongside the online collection system certified by national authorities in the Member States). • Second, the possibility within the tool to interact with other participants. In the literature there is an unresolved debate about the need for deliberation in online participatory processes. In this QCA configuration, the possibility to interact with other participants seems to contribute to an effect on the political or policy agenda. Deliberation could enhance the quality of the input, and better proposals might more easily find their way to the political or policy agenda, but it does not seem to be a decisive condition. Deliberation between participants is facilitated in both Futurium and in the European Citizens' Consultation, but not in Your Voice in Europe or ECI. The different consultations on the Futurium platform has an interactive design, and participants can react to one another's input; according to the interviewed developer, it resembles a social network. During the online first phase of the ECC, participants had the opportunity to discuss one another's contributions, but how much deliberation actually took place differed per national website. Deliberation between participants from different EU countries was not possible, which is an often-heard critique of the ECC. This draws attention to a challenge that arises at the EU level: deliberation between participants from different EU countries is difficult to organise, particularly because of language barriers. Technological measures, like translation software, are not yet able to overcome this barrier. Futurium uses English as the common language, which suits a professional target group, but excludes many European citizens. • A third condition is the possibility within the tool to interact with decisionmakers. A good connection between the input from participants and the political reality would increase the chance of citizens' input being incorporated in the political or policy agenda. Interaction between participants and decision-makers can enhance this connection and can be realised online. Good practices in this regard are the Wiki Melbourne case, in which a vision document was co-created between citizens and officials, and the PB Paris case. In the phase of project assessment, there is room for the Paris municipality to combine, pair and interpret the proposals of citizens. This phase was opened up for citizen involvement in 2016: 'We made a big effort to involve citizens into the merging phase and really encouraging them to go together and defend the project together. First of all, to have less projects to deal with but also to have more comprehensive projects and finally because we need people to get more involved in the campaign phase', the interviewed organiser explained. The four observed digital participation tools at the EU level lack the possibility to interact with decision-makers. This might be less problematic in the e-consultations of Your Voice in Europe or Futurium, where participants' input is interpreted by officials related to the specific policy. However, in relation to the more open tools of ECI and ECC, the interaction between participants and decision-makers would contribute to a better match between citizens' input and the reality and practices at the political and policy level. And last but not least, online interaction can also contribute to an increase in transparency and accountability.

In Sum: Digital Participation Is Not a Quick-Fix
A long-standing and continuing democratic deficit of the European Union is detected in both public and scholarly debate. This democratic deficit is explained by the complex and mutually reinforcing mix of the institutional design features of the EU and is held to contribute to the lack of a sense of European citizenship and the negative and national-oriented public discourse around the EU. It is still believed by many that the perceived democratic deficit of the European Union indicates the need to foster a European public sphere as a space of debate across national public spheres. Moreover, there is a consensus that the new modes of political communication and participation via the internet can play a role in this respect. Redressing the democratic deficit is obviously a daunting task which cannot be accomplished through the introduction of e-participation tools alone. Far-reaching expectations of a fundamental reform of modern democracy through the application of online participatory tools are vanishing after two decades of e-democracy. However, if properly designed and implemented, e-participation has the potential to contribute to accountability and transparency, the transnationalisation and politicisation of public debates, and the improvement of exchanges and interactions between EU decision-making and European citizens.
A common critique of e-participation practices at the EU-level is that they are a successful civic instrument but not a convincing policy instrument. It seems to be an ongoing theme that e-participative projects might provide added personal value for participants and community building, but suffer from a lack of direct, or even indirect, political or policy impact. In our comparative case study, we therefore focused on the factors within e-participation practices that contribute to impact on the political or policy agenda or on the final decisions made. The most important factors for successful e-participation identified in the report are: • A close and clear link between e-participation processes and a concrete formal decision-making process • Clarity of the participatory process and the contribution of its outputs to the overall decision-making process for participants from the start • Feedback to the participants about what has been done with their contributions • E-participation that must be accompanied by an effective mobilisation and engagement strategy, involving communication instruments tailored for different target groups • Participative processes that should not be limited to one event but should be embedded in an institutional 'culture of participation' To realize these conditions in practice requires serious investments (in time and costs) and the commitment of all actors involved; digital participation is not a quickfix.