# Slide Outline: Bayesian Statistics Workshop

## 1. Workshop Goal

- Turn one uncertain decision into a Bayesian model criticism report.
- End with a generative story, candidate models, priors, checks, comparison, and recommendation.

## 2. Models Are Checked Assumption Sets

- Priors imply fake data before seeing observations.
- Computation can fail.
- Posterior predictive checks should match the decision.

## 3. Write The Generative Story

- Units
- Outcome
- Group structure
- Noise
- Parameters
- Decision-relevant variation

## 4. Compare Models

- Complete pooling
- No pooling
- Partial pooling
- What each model assumes away

## 5. Team Exercise

- Draft one prior predictive expectation.
- Choose one posterior predictive check.
- State what result would change the decision.

## 6. Peer Critique

- Are the priors plausible on the outcome scale?
- Which check matters for rollout?
- Where might pooling borrow too much information?

## 7. Close

- Share one model revision triggered by criticism.