The assumption that facilitation is necessary to ensure to the quality of deliberation in a mini-public is widespread (Beauvais and Baechtiger 2016; Curato et al. 2019; Elstub and Escobar 2019; Fung 2007). However, scholarly work on the internal workings of deliberative processes regarding especially the role of process design and facilitation remains rather limited. While practitioners of facilitation regularly consider which approaches to facilitation might best help achieve particular goals (Büro für Zukunftsfragen 2018; IAP2 2006; White et al. 2022), the scholarly literature does not yet provide a clear consensus on what the expectations on facilitation are, and how these are to be achieved (Escobar 2019).

There are, however, a few notable exceptions. Mansbridge et al. (2006) find that facilitators consider group atmosphere and progress toward the goal of deliberation as indicators of successful facilitation. Moore (2012) points out the need for a critical reflection on facilitation practices because they inevitably exist in a unique tension between ensuring and dominating deliberation. Dillard (2013) supports this, positing that facilitation methods vary along a scale of passive to involved, and that the role of the facilitator should not be seen as neutral because they help determine where deliberations go.

Asenbaum (2016) examines the characteristics of one particular facilitation method, dynamic facilitation, documenting how these facilitators address exclusionary tendencies and enable diversity while steering toward consensus. Molinengo and Stasiak (2020) argue that for facilitation, the planned interaction modes and material ‘artefacts’ used by facilitators can play a significant role in achieving (or impairing) collaborative advantage. As regards online deliberation, Trénel (2009) found that higher quality facilitation could minimize the effects of internal exclusion. Meanwhile, Wyss and Beste (2017) found that online forums using artificial intelligence-based facilitation can realize benefits in deliberative goals, improving deliberative interaction in numerous respects, other goals such as improvements in learning throughout the process were reached only for certain types of participants, with others experiencing undesirable effects.

Citing Dryzek (2000), Moore points out a set of expectations for facilitation: to make deliberative mini-publics ‘more inclusive, more comprehensive, more careful to avoid deception, suppression and coercion’ (Moore 2012: 148). The task of facilitation becomes yet more complex when these expectations are paired with different deliberative goals. The literature provides some insight as to facilitation applied in various cases, but the work of comparing facilitation methods systematically has only just begun.

Strandberg et al. (2019) compared facilitated and non-facilitated discussion among like-minded individuals, finding that facilitation—described as interaction rules upheld by a trained facilitator—reduces polarization in discussions among like-minded individuals. While Setälä et al. (2010) do not explicitly refer to facilitation, they compare two process designs in which facilitators led groups in different methods of decision-making. This is relevant here because process design reflects the particular goal that facilitators aim to achieve. The researchers find that while the different deliberative formats did not systematically impact the development of participant opinions, participants’ knowledge-increase as well as their consideration of the facts was higher with the consensus-oriented decision-making facilitation than with the format that decided via secret ballot voting.

The literature shows various expectations from facilitation, but for the purposes of this article, we follow Escobar (2019), who identified inclusion, interaction, and impact as three expectations from facilitation (Escobar 2019: 183).

Inclusion refers to participation in deliberation and can be separated into two parts: internal and external inclusion. External inclusion is about who is invited or allowed to take part in a deliberation. There is significant attention to this in the literature, especially regarding representation and recruitment (Curato et al. 2021; Devillers et al. 2021; Dryzek and Niemeyer 2008; Elstub and Escobar 2019; Harris 2019).

Internal inclusion refers to the participation of all participants within a deliberation. When certain participants are overly dominant, the literature speaks of internal exclusion (Young 2002). Seen this way, the expectation is that ‘facilitation not only ensures that all participants have an equal chance to speak (achieving a more concrete equality of opportunity for participation), it also ensures that different sides of the debate are heard’ (Beauvais and Baechtiger 2016: 6).

Scholars expect facilitation to minimize internal exclusion by ensuring that all participants take part in deliberation (Dillard 2013; Escobar 2019; Moore 2012). One example of structural internal exclusion recognized by scholars that impacts internal inclusion is the gender divide (Einstein et al. 2019; Landwehr 2014; Young 2002). Scholars have documented ways that this form of exclusion manifests itself, for example that males speak longer and interrupt more often (Karpowitz and Mendelberg 2014; Mendelberg et al. 2014). Facilitation is expected to address this and other types of internal exclusion by designing and moderating interaction.

Interaction refers to the inter-group dynamics during the deliberation. Facilitators can affect this by establishing an interactive order through communicating rules, roles, or other norms and expectations. In order to better include all relevant arguments in a deliberation, facilitation in deliberative mini-publics can serve to counteract the [structural] power inequalities that otherwise threaten to be reproduced within the deliberative forum, for example by ‘encouraging and protecting’ participants (Landwehr 2014: 81). For example, a facilitator suggests an interactive order in which all participants are heard once before any are heard twice, or prompting quiet participants to contribute, protecting their contributions by physically positioning themselves between participants or ‘holding the space’ by prompting, focusing on, turning toward, looking at, and/or gesturing for a participant to speak while maintaining silence.

A facilitator might carry out none, some, or all of these interventions, and to varying degrees, depending on the goal and the approach. Dillard (2013) proposes a continuum of facilitations, from passive to moderate to involved. Landwehr (2014) also suggests a continuum of facilitation roles (or as she calls them: intermediary roles), suggesting various degrees of intensity in the tasks and roles. These range from organizing deliberation to enforcing interactive rules to interpreting and aggregating, resulting in different interactive orders.

Impact refers to what facilitation is able to achieve. Mansbridge et al. (2006) suggest that facilitation can impact upon group atmosphere, consistent progress toward the goal, and also increase participant satisfaction. Participant satisfaction is relevant, as some have argued that positive experiences deliberating effect factors important to acceptance and readiness to partake in deliberative democratic processes (Dienel 2002a; Landemore 2020). Boulianne et al. (2020) suggest that facilitated deliberation leading to the perception of a higher quality deliberation can increase civic and political engagement more than when deliberative standards are lower.

There is relative consensus in the literature that facilitation should be impartial, regardless of whether it is more passive or involved in enforcing interactive orders (Dillard 2013; Escobar 2019; Landwehr 2014). However, impartiality can conflict with expectations of inclusion, interaction, and impact, if facilitators use their position as an impartial party to promote more inclusive processes. Doerr (2018) describes how the position as a ‘disruptive third’ has been used on behalf of facilitators committed to the goal of inclusion and how this unique position can be effective in better achieving this goal.

There are numerous guidelines, toolkits, outlined approaches, and processes available to facilitators. These range from foundational underpinnings based on theory and experience and elaborated with examples (for example Kahane 2021; Scharmer 2009) to dedicated works offering detailed practical and methodological orientation. Amongst the latter are guidelines for specific facilitation methods; for example, Dienel (2002b) on the planning cell, Zubizarreta (2014) on dynamic facilitation, or White et al. (2022) on variations of mini-publics. There are also toolkits and guidelines that include various facilitation methods according to the context and goal of the deliberation (Büro für Zukunftsfragen 2014; IAP2 2006). The IAP2’s Public Participation Toolbox clusters its tools in different categories, suggesting different goals. Listed are techniques to: share information, compile and provide feedback, and bring people together (IAP2 2006).

Practice-oriented materials can be found that cover the range in the spectrums of facilitation roles (Dillard 2013; Landwehr 2014), some methods and approaches demanding more resources than others. These resources can include time (the duration of the process and preparation and postprocessing), training of the facilitators, and material requirements. Facilitation methods also vary in terms of goals and expected outcomes. The practice-oriented literature clearly seeks to match facilitation approaches with goals of (mini-public) deliberation, while scholarly literature has yet to systematically address this relationship. This points to the gap in scholarly literature that needs to be addressed and to which we aim to contribute to with this article.

The expectations projected upon facilitation are many, and they have been more thoroughly documented than the extent to which facilitation can meet them. The practice-oriented approaches, methods, and materials set different emphases. The question is: can facilitation achieve all the deliberative goals at once? Encouraged by the practice experience and practice-oriented literature, we argue that this is not the case.

‘Clearly, the role and tasks of the intermediary will differ depending on the precise set-up of the deliberative forum’ (Landwehr 2014: 89). Following Landwehr (2014) and other colleagues who have acknowledged the variance between facilitations, we expect that to achieve specific facilitation goals, the purpose of a mini-public’s format must be appropriately matched with a facilitation design. As illustrated above, comparative work on facilitation remains scarce. Although interest in comparative facilitation work is growing (Escobar 2019), obstacles for scholars pursuing this work also remain significant (Landwehr 2014).

Seventeen citizens took part in the deliberative mini-publics. They were recruited through invitations sent to a random sample of 702 citizens from the official city registry distributed equally according to age and gender, of which 13 participated in the deliberation. A further four participants were recruited using an intercept method, approaching citizens at random along a pre-defined path in Magdeburg’s city center. (For a more detailed description of the recruitment, see Molinengo and Stasiak 2020: 6.) The mini-publics were carried out in in Magdeburg’s city hall in June of 2019 in accordance with German data protection and privacy regulations. All participants consented to the mini-publics being recorded and analyzed by the researchers.

We worked with a media team from the regional state television station MDR to make video recordings of the deliberations. We analyzed the recordings to determine the speech time of each participant, and to identify the individual speech acts, qualifying whether the speech act was an interruption and whether the speech act took place in the plenary or in smaller groups. To supplement the video analysis, we collected data from the participants using post-deliberation surveys. Thus, we aim to generate preliminary insight regarding the different functions of deliberative processes as regards differences in facilitation.

Due to the real-world nature of the mini-publics, constraints regarding participants, facilitation design, and the limited time-frame of three hours for each process, the generalizability of the insights is limited. Our exploratory approach instead provides initial insight and a point of departure for more systematic comparison of facilitation approaches.

The participants were divided into three groups: the self-organized (SO) group with five participants, the multi-method (MM) group with seven participants, and the dynamic facilitation (DF) group with five participants. The uneven distribution of participants between the groups was due to last-minute cancellations. The task of each group was to generate recommendations in response to the question: ‘How can the city center of Magdeburg be made more pedestrian friendly?’ (Wie kann Magdeburgs Innenstadt für Fußgänger attraktiver werden?) All mini-publics were asked the same question. They also received the same briefing at the beginning, outlining their task and the resources available to them. All were instructed that they had three hours to come up with and document their recommendations. They were also given the same working material, including templates for collecting and documenting their recommendations. All groups were told that their recommendations would be received and reviewed by the City of Magdeburg and would serve to inform the city development strategy. We designed the templates following consultation with city officials so that they could be easily understood by the city administration.

There were two different types of response-templates (see Figures 1 and 2). Each group was also provided with markers and a large map of the city center of Magdeburg (seen in Figure 2). Although each group received the same introduction and materials, how and when these were applied depended on the process. For a more detailed examination of how the materials were used in the different groups, see Molinengo and Stasiak (2020). The following sections examine the three different facilitation designs.

The process design of the self-organized (SO) mini-public (see Figure 3) was limited to the introductory briefing, which identified the task and introduced the response-templates. Participants themselves established various roles within the mini-public. Participants assumed or tried to pass on roles, but these roles were often not firmly established and maintained. It was never explicitly established when either a topic or the overarching task was fulfilled. This, coupled with a lack of maintained focus on topics for longer periods, suggests that the SO process was characterized by a lack of orientation. There was also no systematic attempt to ensure internal inclusion. Nonetheless, there was a certain deliberative quality to be observed. The participants supplied reasons for their suggestions and engaged with the contributions of others and filled out the response templates.

Lacking a facilitator, the interaction mode was left open. At the beginning, a male participant established himself in a moderating role by suggesting an approach to gathering and sharing ideas:

We noted that the body-language of other participants suggested that not all participants seemed to equally support this suggestion. We observed a frown, raised eyebrows, and leaning away from the participant. Nonetheless, this decision was not contested. This suggestion was upheld for the remainder of the process, although the participant who suggested it filled the role with less clarity and energy as time went on. This participant had the highest speech time (31.4%) of the participants in the group.

One participant assumed the role of timekeeper, who reminded the group of how much time was left. Another participant left before the group was finished (after 1 hour 30 minutes), which seemed to cue to the rest of the group that the process was nearing the end. The SO group ended the process after 1 hour 45 minutes, with a bit of uncertainty. One participant asked the others ‘Do we have to sit out the time?’ (1F4)

The participants discussed without explicitly allocating speaking turns. The categories of the response templates were intermittently referred to and served as a minimal structure for organizing the content of the discussion. While this structure was mostly in the background, it helped organize the discussion. One female participant had been designated by others as the scribe, supported by comments like ‘Women mostly have better handwriting’ (1M1), ‘That’s true’ (1M5). The scribe used the template to maintain focus on a topic. When the participant who had taken on a moderating role said, ‘Should we…em, next topic?’ (1M1), the scribe pointed with her pen to the headings on the template and said ‘well, we have to still…’ (1F3) and then read aloud the template prompt. She used the template to maintain focus on the topic until the question was answered, sometimes adding instructions so the group would provide the kind of response she felt was appropriate to write in ‘They want a bit more detail, like street names’ (1F3). While this was successful at getting participants to interact by supporting, opposing, or qualifying others’ ideas, the topic of discussion often jumped from topic to topic. This led to many topics being left open or unfinished, without an explicit indication whether this was the will of the group or happenstance.

The multi-method (MM) (see Figure 4) mini-public showed the highest participant satisfaction and the greatest readiness to engage in city development. Through meticulously planning the mini-public, the facilitator sought to create a constructive environment and constantly keep the process moving forward by including various interactive moments, safeguarding reflection time, and by using first names, establishing goals and sub-goals, giving the participants ownership of the process by having them reiterate the task with their own emphasis. The participants came up with ideas, reflected on them, shared them, and developed further ideas together. The facilitator established and maintained an orientation (‘Is it clear what we’re doing?’) and bundled smaller tasks that built upon each other to achieve the main task but helped make clear the orientation of what the participants were doing at any given time.

The mini-public began with a plenary session. The facilitator (with whom we designed the MM process) received the participants in a circle of chairs. She told participants what to expect from the mini-public. She also organized an activity to let participants get acquainted with one another. The facilitator led the participants through the process design, which oscillated between interaction in the plenary, in smaller groups, and individual tasks. This encouraged repeated moments of reflection—for participants to become consciously aware of their own positions and ideas—followed by moments of sharing and reaction. And there were phases where participants spoke on the telephone with people outside of the group or were given fictional profiles based on real individuals observed and spoken to in the Magdeburg city center to study and consider. The goal of these phases was to encourage participants to think in a way that considered the needs and perspectives of others in relation to their own. The interaction mode was thus lively but at the same time often agreeable and organized. The facilitator encouraged participants to speak with and listen to one another. All participants engaged in the process until the facilitator declared the end, after 2 hours 30 minutes.

Different from the other processes, at the beginning the facilitator asked if it would be acceptable to address one another on a first name basis, which the group agreed to and did. The facilitator had participants share their own experiences in groupwork to emphasize the importance of listening. She also gave advice on how to listen and contribute constructively, for example saying:

Thus, the facilitator created a positive feeling in the group by emphasizing the benefits of working together. While the facilitator used prompts to bring people into the deliberation, they were often embedded in the overarching plan that had been established. Rather than a typical ‘calling on’ participants to continue, the facilitator prompted people to not just say something, but to contribute to the progress of the process. ‘Ralf and Adam, what else did you guys find out?’ Or, to encourage interaction with ideas: ‘So when you listened to that, the concepts, the ideas, where did your heart warm up?’

The facilitator emphasized certain norms of interaction, making the participants a part of their establishment for the group. At rare points, she also clearly re-established the rules of interaction. For example, while gesturing between two groups: ‘Ok, there’s the one rule though: you guys get to speak, and you guys get to listen’ (MM Facilitator).

In the dynamic facilitation (DF) mini-public, the process design determined the interaction mode: speech interaction was dialogue between the facilitator and a single participant at a time, with the facilitator asking prompting questions to elicit more elaboration on ideas from the participant while the other participants listened. Relative to the other two processes, participants spoke for fewer, but much longer turns, and watched as their ideas were written down and filled the room around them.

The facilitator of the DF process began with a round of introductions, with the participants sitting in a half-circle. After the introduction, the facilitator changed the setting (see Figure 5), asking the participants to move their chairs into a line all facing the facilitator and four flip-charts, saying ‘…so that you are all not like just now in a circle speaking with each other, but speaking with me.’ She explained the mode of interaction, the rules and what participants could expect. She said:

She then read aloud the titles of the four charts (Solutions/Ideas, Concerns, Information/Perspectives, Questions/Challenges). The facilitator explained what she would do (‘I listen, and I write what I hear onto the paper’) and explained what she meant with the jumping: ‘You can imagine it as if you’re doing a puzzle together. One person is doing the sky, another person is looking for all the green pieces. And a third person tries to find out the corners. And to the extent that we over time are busying ourselves with all these pieces, the bigger picture emerges.’ (DF Facilitator)

The facilitator then announced she would now listen to each participant speak about their thoughts on the question at hand and did this one by one. Each participant spoke for as long as they wanted, their turns protected by the facilitator, who held the space by concentrating on current participant with eye-contact, focus, and hand gestures. The facilitator wrote many of their points on the four flipcharts, sometimes asking if the formulation was correct or if it was on the correct flipchart and hanging the poster-sized papers on an adjacent wall when full. Another round in the same format followed, with the facilitator emphasizing that participants should follow ideas that evoked emotions. The facilitator held the space for the participants to finish their thoughts, telling other participants that their turn would come if they began to speak out of turn. In this way, the facilitator allowed and encouraged participants to fully share, without pressure to quickly get to the point or finish. When participants voiced concerns or described problems, the facilitator would often respond with: ‘Ok. Do you have a solution or idea for that?’ and write down their responses.

Our observation of the participants’ body language suggested that it required energy to listen to the long speech contributions of the other participants. Participants who were not speaking shifted in their seats, cradled their heads in their hands, or twiddled their thumbs. The facilitator had prepared them for this: ‘That’s when your patience is needed.’ The participants in the DF process were engaged in the process the whole time, beginning and ending together, for 2 hours 45 minutes. During that time, they were together in the plenary, except for 20 minutes toward the end when they split up into 2 smaller groups as directed by the facilitator. During these 20 minutes, the participants filled out the response templates, considering what they had just heard and experienced.

Further research comparing facilitations should aim to help convenors of deliberative formats better judge what facilitation designs will serve their needs and which facilitators are best equipped to carry out that facilitation. Convenors, especially when it comes to public sector convenors commissioning deliberation as a part of democratic processes, should be making informed decisions to make the facilitation fit the task of deliberation. Without identifying pluralities and differences within facilitation design and practice, and establishing standards or guidelines, this remains a challenge.

We need more research to establish categories and standards for facilitation, and to sharpen the indicators we will use to see whether facilitations are living up to their promise and potential. Work done on identifying categories and tasks of facilitation, and the roles of facilitators, is extremely valuable (Dillard 2013; Landwehr 2014; Mansbridge et al. 2006; Moore 2012), but researchers will need to get into more detail and importantly, connect their categories and standards closely to facilitation practices. To this end, close collaboration with the emerging communities of practice that are growing around certain facilitation methods such as dynamic facilitation or approaches such as the Art of Hosting can be useful.

To sharpen the indicators, for example, we will need to develop concepts beyond speech time to measure internal equality, as has already been suggested (Escobar 2019; Landwehr 2014). Indicators for good listening will also be helpful. The quality of listening may prove difficult to capture, but Scudder (2020b, 2020a) offers a fertile point of departure. Developing a nuanced understanding of different facilitations would be a valuable step to better enabling deliberation and its practical ability to further deliberative-democratic practices.