The Thanksgiving Table: A Role-Play Activity for Statistical Communication

Students practice explaining one study to a relative

for professors
curriculum
statistical communication
Author
Published

November 20, 2026

I normally write blog posts after I implement things in my class. This one is different. I will run this activity on Monday in STAT 100, and I am sharing it in advance in case other instructors would like to implement it next week.

Thanksgiving week

Thanksgiving week is always difficult to teach. We do not want to lose a class on the schedule, but students also travel. I normally teach on Mondays and Wednesdays, so during Thanksgiving week I only have a Monday class.

My class is normally in-person. Just for this class, I gave students the option to come in-person or join on Zoom. That meant I had to design an activity that works in a classroom and in breakout rooms at the same time.

I also wanted the class to be worth attending. By this point in the semester students have seen sampling, study design, p-values, and intervals. On Thursday, many of them will sit at a table where someone asks about a study they saw in the news. So the topic for Monday is statistical communication, and the activity is called the Thanksgiving Table.

The slides

Here are the slides. They include links to all of the materials that students use: a guide, four study cards, and five relative cards.

Before the role-play

I start with a short introduction on why statistical communication matters. Communicating a result is part of doing statistics, not an extra step at the end.

The main example is the 1995 pill scare in the UK, as described by Gigerenzer et al. (2007). Women were told that some newer birth control pills doubled the chance of dangerous blood clots. In counts of people, that was a change from about 1 in 7,000 to about 2 in 7,000. It is the same number told as two different stories.

Then we talk about why facts alone rarely change minds. People also want to know whether you understand their worry, whether you respect them, and whether you are honest about what you do not know. This leads to four qualities of good statistical communication: accurate, clear, honest about uncertainty, and respectful.

The activity

The goal is stated on the slides: you will not win Thanksgiving. Nobody does. The goal is to leave the table with a relative who understands one study a little better and still wants to talk to you.

Roles

Students work in small groups.

  • Explainer explains one real study using a study card.
  • Relative plays a family member with a real reason for their view, using a secret relative card that the explainer does not see.
  • Cousin Sam joins groups of three or four. Cousin Sam is on the fence, a bit lost, and asks “What does that mean?”

Students rotate roles each round, so everyone gets to explain and everyone gets to be a relative.

Study cards and relative cards

Each study card summarizes one published study: who was sampled, the design, the results as counts of people, the limitations, and plain-language swaps for jargon.

Card Study Design Relative
A MMR vaccine and autism (Denmark) Observational Uncle Ray
B Pfizer–BioNTech COVID-19 vaccine Randomized experiment Aunt Dee
C Mediterranean diet (Spain) Randomized, with problems Grandma Lou
D Quitting Facebook (US adults) Randomized experiment Mom or Dad

I chose the studies so that the statistics lesson is different on each card. Card A is observational. Card B is a clean randomized experiment. Card C is a randomized experiment that was retracted and republished after problems with how people were assigned to diets, so students have to talk about science correcting itself in public. Card D has a statistically significant result with a small effect, so the explainer has to tell a parent that the evidence is weaker than the parent hoped.

The relatives are not caricatures. Uncle Ray, for example, watched his best friend’s son get diagnosed with autism not long after his toddler shots, and he does not trust drug companies because of what happened with opioids. He cares about protecting kids and not being lied to. Each relative also has a question about a p-value or an interval that they must ask. Uncle Ray asks whether “not significant” just means the study could not prove harm. He is partly right, and the explainer has to say so.

The relative cards tell students when to soften and when to dig in. Relatives soften when they feel heard, when they hear honest uncertainty, or when they hear an example from their own life. They dig in when they hear unexplained jargon, “just trust the science,” or a textbook definition recited at them. They are told to move only as much as a real person would.

Five courses

The conversation has five courses.

  1. Appetizer: listen first. The explainer may not explain anything until they ask a question, reflect back what they heard, and name a shared value.
  2. Who and how. Who was in the sample? Was it an experiment or an observational study?
  3. What they found and how sure. One count of people, the p-value or interval in plain words, and one thing it is not.
  4. Who it applies to. What generalizes, and one real limitation.
  5. Dessert: one brick, door open. One study is one brick. Summarize, then ask “What’s still bugging you?”

The first course is the one I care about the most. Students who have just learned what a p-value is want to start with the p-value. The rule forces them to start with the other person.

Feedback

After each round the relative gives feedback first. Did the explainer ask about my view before explaining? Did I understand the p-value or interval explanation? Did they name something the study cannot tell us? Then one sentence that made them feel respected and one word they did not understand.

At the end students step out of the role out loud: “I’m not Uncle Ray. I’m ___.”

Making it work on Zoom and in-person

Since students can attend either way, everything they need is a web page. There is nothing to print and nothing to hand out. A student in the classroom and a student in a breakout room open the same links from the same slide.

The groups are flexible for the same reason. I will not know how many students are in the room and how many are on Zoom until class starts. The activity works with two, three, or four people, and the slides say what to do with each group size.

A few things are specific to Zoom.

  • Students rename themselves to their character when they play a relative, and rename back afterward.
  • Explainers and relatives keep cameras on, because faces are part of listening.
  • Nods are hard to see on video, so explainers check for understanding more often.
  • Anyone who needs a pause to look at their notes says “Pass the gravy.”
  • Breakout rooms are not recorded.

After the role-play

We debrief as a class. What made relatives soften or dig in? Which sentence worked best for explaining a p-value or an interval? Did admitting a limitation make the explainer look weaker or stronger?

The exit ticket is a three-sentence text message to the relative. It summarizes the study, gives one number in plain language, and names one thing the study cannot tell us.

A note on role-play

This is my first time doing role-play in class, so I am not really the expert to write on it. I was inspired by Mario Davidson’s roundtable, Implementing Role-Play in Statistical Courses, at JSM this past August, with co-authors Julia Sharp, Alexandra Hanlon, Ryan Peterson, and Emily Griffith. The idea for this activity came to me there and then.