Skip to main content

The Thanksgiving Table: Guide

The goal

You will not win Thanksgiving. Nobody does.

The goal is to leave the table with your relative understanding one study a little better and still wanting to talk to you. A relative who moves a little is a success.

The roles

The three roles
Role Your job
Explainer You explain one real study to a relative. Use your study card.
Relative You play a family member with a real reason for their view. Use your secret relative card. Don’t show it to the explainer.
Cousin Sam (groups of 3 or 4) You’re on the fence and a little lost. Ask “Wait, what does that mean?” at least once. Use the Cousin card.

The conversation: five courses

Every explainer follows the same five courses, in this order.

1. Appetizer: listen first

You may not explain anything until you have done all three:

  • Ask an open question: “What worries you about it?” or “Where did you hear that?”
  • Reflect back what you heard: “So it sounds like you’re worried that…”
  • Name a value you share: protecting kids, not being lied to, staying healthy, spending money wisely.

2. Who and how: sampling and design

  • Who was studied? How many people, and how were they found? (Volunteers? Everyone in a national registry? Patients at certain clinics?)
  • What did the researchers do? Was it an experiment (researchers decided who got what) or an observational study (researchers watched what people already did)?
  • If it was an experiment: what does random assignment buy you? Try: “It’s like shuffling a deck and dealing the cards: the two groups end up alike on average, except for the one thing we changed.”

3. What they found and how sure we are

  • Give one number as a count of people: “Out of every 1,000 people in the placebo group, about 9 got sick.” A sample proportion of 0.009 is the same information, but “9 out of 1,000” is much easier to picture.
  • Explain the p-value or the interval in plain language (see the box below).
  • Say one thing it is not (see below).

4. Who this applies to

  • Generalization: Who can we reasonably extend this to? Who was not in the study?
  • Cause or association? Can this study say A causes B, or only A goes with B?
  • Name one real limitation without being asked. Admitting a limitation builds trust.

5. Dessert: one brick, and leave the door open

  • One study is one brick. We trust an idea when many studies, by different people with different methods, point the same way. Science also corrects itself, sometimes in public.
  • Summarize in one sentence.
  • Check, don’t demand: “Does that make sense? What’s still bugging you?”

Explaining a p-value

“Suppose [the vaccine / the diet / quitting Facebook] did nothing at all. How surprising would results like these be, just by chance? The p-value measures that surprise. A small p-value means: this would be pretty weird if nothing were going on.”

In class terms: the p-value is the probability of a sample statistic as extreme as ours, or more extreme, if the null hypothesis were true. The sentence above says the same thing without the words “null hypothesis.”

Three things a p-value is NOT (say at least one):

  1. It is not the chance that the result is true or that the treatment works. (It is not the probability that the null hypothesis is true.)
  2. “Not significant” does not mean “proven: no effect.” It means results like these would not be unusual if there were no effect.
  3. “Significant” does not mean big or important. A tiny effect can be statistically significant in a big study. That’s the difference between statistical and practical significance.

Two analogies you can borrow:

  • Coin flips. If your cousin flips 9 heads out of 10, you start to suspect the coin. A p-value tells you how suspicious to be.
  • Courtroom. “Not guilty” isn’t the same as “innocent.” It means the evidence wasn’t strong enough.

Explaining an interval

“Here’s the range of effects that fit the data reasonably well.”

Then ask two questions out loud:

  • Does the range include “no difference”? (For a slope or a difference, “no difference” is 0. Two of the study cards compare groups with a ratio; for a ratio, “no difference” is 1.)
  • Does the range rule out a big effect? Sometimes this is the more convincing part.

Two kinds of intervals appear on the study cards:

  • A confidence interval (frequentist): if we repeated the study many times, 95% of intervals made this way would capture the true value.
  • A credible interval (Bayesian): the middle 95% of the posterior. Given the data and the model, there’s a 95% probability the true value is in this range.

Plain-language tips

  • One key message. If your relative remembers one sentence, what should it be?
  • Swap the jargon. See the table below. Your study card has more swaps.
  • Use their words and examples from their life.
  • Short sentences. Then pause.
  • No “just trust the science” and no “that’s been debunked.” Explain instead.

Words from STAT 100, and how to say them at the table

You know these terms. Your relative doesn’t. Use the idea, not the word.

Class terms and plain-language versions
Class term At the table
population and sample “who they wanted to learn about” and “who was actually in the study”
representative sample “the people in the study look like the people we care about”
observational study “they watched what people already did”
experiment with random assignment “a coin flip decided who got what”
confounding variable “something else that’s different between the two groups”
placebo “a fake treatment, like a shot of salt water”
null hypothesis “suppose it did nothing at all”
p-value “how surprising the results would be if it did nothing”
statistically significant “would be unusual if there really were no effect”
practical significance “big enough to matter in real life”
confidence or credible interval “the range of answers that fit the data”
holding other variables constant “comparing people who are alike in other ways”

The papers also use a few statistics we have not covered in class. Your study card rewrites them as proportions, counts of people, and ratios. You don’t need any term that isn’t on your card.

After each round: relative feedback

The relative (and Cousin Sam, if you have one) tells the explainer:

  1. Did they ask about my view before explaining? (yes / sort of / no)
  2. Did I understand the p-value or interval explanation? (yes / sort of / no)
  3. Did they name something the study can’t tell us? (yes / no)
  4. One sentence that made me feel respected. And one word I didn’t understand.

Relative goes first: *“Here’s what moved me, and here’s what made me dig in.”