Problem 1
State the question and data structure, apply examining subgroup performance, measurement bias, feedback loops, and deployment risk, check assumptions or design conditions, quantify uncertainty, and interpret the result in context.
Faculty solution guide
Fairness, Responsible AI and Statistical Harm
Accept equivalent statistically correct reasoning. Require context, design, assumptions, uncertainty, and limits.
Problem 2
A common error is to report a calculation without identifying the population, design, assumptions, uncertainty, context, or limits on causal and generalizing claims.
Problem 3
A complete response identifies population, cases, variables, parameter or estimand, study design, and conditions for Fairness, Responsible AI and Statistical Harm.
Problem 4
Repair the earliest invalid assumption, display, calculation, or inference; then rerun the interpretation and state how the correction changes the claim.
Problem 5
A strong application defines context, data source, design, analysis plan, uncertainty, and limits on generalization or causation.
Problem 6
Use at least two of: independent recomputation, alternative graph or model, residual/diagnostic check, resampling, sensitivity analysis, subgroup analysis, simulation, or comparison with a known benchmark.