The hidden multiplicity in a single reported test
An analyst insists no multiple-testing correction is needed because "we only ran one significance test." But along the way they chose how to define the outcome, which outliers to drop, which subgroup to focus on, and which covariates to include, all after glancing at the data.
Explain the "garden of forking paths" and why a single reported test can still have an inflated false-positive rate.
Your answer
This one is open-ended. Work it through, then check your reasoning against the full solution.