Consensus conference

STS 201 students debating state policies on AI and the workforce, Fall 2017

In my introductory STS class, I run a mock consensus conference on a science policy issue of the students’ choosing. I wrote up these notes on the mechanics of the conference for fellow instructors who might be interested in doing a similar exercise. Please feel free to use/adapt these materials for your own teaching.

Prep work

I’ve run the conference in two formats: one where all students take on the role of citizens (version one) and another where they could choose from additional role options (version two). Here’s roughly what’s entailed in each.

In both versions, I have students choose the topic the conference will focus on, and I give them a lot of latitude there (so long as the topic is in some way related to science, technology, or medicine). I show students a few ideas from previous conferences and give them time to brainstorm their own ideas in class. Then I take 6-8 ideas from the brainstorming session that seem viable to me, write a short description of each, and have them vote on those selected topics in class. Here’s a sample list of topics.

Version one

A few weeks before the conference, I devote one class period to a discussion of what it means to represent the population of Wisconsin. We discuss questions such as: in what ways do the people of Wisconsin vary? Is it more important to attend to some types of variation over others for the conference? Should we aim to create a representative sample, or do we want to oversample minority groups or people who are particularly affected by the issues at hand? Following this discussion I have them do some research in small groups on the variation they thought was most relevant (eg one group might work on finding demographic data on age, or on employment by sector in Wisconsin). The class then works as a group to choose individual character profiles in a way that ensures we have a population sample that meets their definition of “representative.”

After that session, each student will have a rough outline of their character role (eg. a 55-year-old white man from a rural area in Wisconsin, employed in the service industry) and can begin doing research for their background document. The background document is essentially a character sketch that prepares students for their role in the conference, and makes some educated guesses about how their character is likely to respond to the issue at hand (sample here). I emphasize that these documents must be research based, not drawn from students’ imagination/stereotypes of what it would be like to be someone else. I encourage them to think broadly about what research means—most students will start out with some survey data on social and/or science and technology policy issues, and will not initially view sources such as letters to the editor, reddit threads, blogs, etc as legitimate sources because they are not “academic.” In addition to their character roles, I ask a small subset of students to volunteer as peer facilitators for the small group discussions that take place during the conference, and give them a little bit of training in listening/facilitation skills.

This version of the conference also involves some of prep work for me, in the form of seeking out a few local “experts” to give short (5-10 min) presentations and answer “citizen” questions on the day of the conference. I also prepare a short statement to frame the purpose of the conference, and get together some supplies such as nametags, attendance sheets, etc.

Version two

In this version of the conference, students can choose from one of four roles: citizen, scientific expert, lay expert, or social scientist. Rather than inviting outside experts to come present, the students themselves research the issues at hand and present as “experts.” And rather than me organizing and facilitating the conference, the “social scientists” research best practices for running a consensus conference, decide on the details of the conference format, and run the conference themselves. Allocating roles is best done in class, since in my experience students will want some clarification on the roles and the instructor will want to manage the balance between roles (I often have many students who want to be scientific experts, but too many experts slows up the flow of the conference and decenters the goal of having citizen input).

I devote a class period to preparation in this model as well, but break the students up into three groups according to their roles. The citizens decide on what it will mean to constitute a representative sample of the Wisconsin population, as described above. The scientific and lay experts discuss the range of expert viewpoints that might be relevant for the topic, what information citizens might want to know about the topic (sometimes they consult with the citizens on this question), and allocate roles collectively to ensure that a variety of viewpoints are represented and critical information is covered. I work with the social scientists on developing facilitation and listening skills, and the social scientists plan the agenda for the conference, the framing remarks they will deliver at the beginning of the conference, and the mechanisms they will use to gather citizen input and develop it into a consensus statement. I typically spend most of my time with the social scientists because having a good agenda is really crucial for the success of the conference. In future iterations I might develop a short document for them that gives a loose agenda and highlights various decisions they will need to make (eg. should the experts be “on call” to answer citizen questions during the discussion, or should they be embedded in the small groups?).

The citizens write a background document for the conference as described above, and the experts write both a research brief and a short (5 minute) presentation based on their research brief. Again, it’s helpful to emphasize that non-traditional sources can be really useful here, especially for the lay experts. For example, in one conference on intergenerational organ transplantation I had a student who took on the role of a woman who had donated a kidney to her aging father, and used primarily patient blogs/forum posts to craft her position on organ donation. Unlike the rest of the class, the social scientists submit their written work after the conference. They research best practices in running consensus conferences beforehand, and then produce a report after the conference that both describes the rationale for the choices that they made and reflects on how well those techniques worked in the actual conference itself.


In both versions, I’ve found it necessary to devote two class periods to running the conference itself. Day one begins with a short presentation that frames the objectives of the conference (either by me or by the social scientists), followed by expert presentations, and then questions/responses from the citizens. Keeping a close watch on timing for the expert presentations is critical, because they can very easily expand to eat up all of the time.

Day two begins with small group sessions where the citizens begin discussing the issues at hand and developing suggestions for a consensus statement. The small groups are facilitated either by a peer facilitator (version one) or a social scientist (version two). The experts can either be absent for day two (version one), or on call to answer citizen questions (version two), or embedded in the small groups (version two). I’ve found that when the experts are embedded in the small groups they tend to overtake the discussion, but rather than fighting too hard against this I’ve used this as a way to discuss the limitations of the consensus conference model (see the reflection section below).

Halfway through day two, the citizens take a short break and the peer facilitators report a summary of their discussion to me (version one)/the social scientists confer with each other (version two). We then move into a large group discussion aimed at crafting a short policy statement. When the social scientists are running the conference I leave the task of deciding how best to do this to them; when I am running the conference I put up a google doc with some draft text that I have created, and then ask the citizens to join me in editing it either by responding verbally to the text or by making comments/suggestions in the google doc. The outcome of the conference is typically a one paragraph statement of some kind, directed at a relevant authority (e.g. the UW Health system, in the case of inter-generational organ transplantation).


I devote about half a class period to reflecting on the strengths and weaknesses of the consensus conference as a model for science policy. One of the issues that almost always comes up is that the citizens feel under-prepared to comment on science policy issues, and end up deferring to the experts. Another is the drawbacks of aiming for consensus—students whose characters were in the minority often report that they withdrew from the conversation when it became clear that the consensus statement was moving in a direction with which they disagreed. I typically assign this article for reading on that day so that they can see that the issues they encountered mirror those encountered in real-world consensus conferences. We also discuss ways that we could modify the conference format to better address these issues in future conferences.

Pros and cons of the two formats

One of the biggest differences between the two formats is that the version two gives students more options for roles and allows them to participate in a way that more closely aligns with their own interests. But, some students feel that the workload in this model is distributed unfairly. For example, the experts are required to do a formal oral presentation, while the citizens participate orally in a much less structured way during the conference. In version one all of the students have the same role and are graded using the same rubric, so I get few/if any complaints about fairness.

Both versions one and two are workable in a small class, while version one would probably be harder to do in a large class. There’s a hard limit on the number of experts you can have before the conference schedule becomes unmanageable, and since many students want to play the role of experts it might be trickier to fairly allocate those roles and ensure that everyone ends up with a role that they’re happy with. In a class with sections, it would also be more difficult to have the experts/social scientists coordinate with each other across sections, unless you do the prep work in lecture time. For these reasons I’ve tended to use version one in the 80+ student version of my intro STS class, and version two in the 30 student version. For the instructor, the main differences are that version one entails more work in finding the experts and running the conference itself, whereas version two entails more work in distributing the roles, making sure everyone understands their roles, and working with the different grading schemes to evaluate each of these roles.