Homework 5

Practical Bayes

Author

Term 2 - HW 5

Goal

Conduct a practical Bayesian analysis that connects a real decision to a transparent model, prior, posterior, and posterior predictive check.

Requirements

Use a public or approved research dataset and state one decision-relevant estimand. Fit a Bayesian model with at least two defensible prior specifications (including a sensitivity prior). Report posterior summaries as probabilities or decision quantities—not only credible intervals—and conduct posterior predictive checks and convergence diagnostics where applicable.

Compare the conclusion under the prior choices, name one model limitation, and provide a 2-page report plus reproducible analysis. Include code, data-access instructions, environment information, and a README.md.

AI can help explain software or draft code, but document prompts in AI_USE.md and independently check model specification, priors, and interpretation. Do not claim causal effects without a design that supports them.