The Art of Statistics
About this book
Spiegelhalter frames statistics as the practice of turning a real question into data and back into an answer, using a problem solving cycle as the spine of the book. Topics run from describing data and communicating risk through to probability, confidence intervals, significance testing, and the reproducibility problems that have hit published research. Every idea is worked through a real case, including survival rates for heart surgery, the Harold Shipman murders and national estimates of sexual behaviour. A recurring thread shows how the same number can be presented in ways that mislead without being technically wrong. Read it if you sign off on analysis rather than produce it.
Description via Product Digest.
Topics
Related concepts
- LLM-as-a-Judge Evaluation: Rubrics, Bias and CalibrationUse an LLM judge to score open ended responses with a clear rubric, then measure agreement, false passes and position bias against human labels.
- Segmentation in analyticsCutting a population so an average stops describing nobody, the cuts that usually pay, and a worked case of Simpson's paradox where a signup rate improves on every device and the combined figure falls from 14.8 per cent to 12.9 per cent.
- Alternatives to a controlled experimentThe changes that cannot be randomised into two groups, the four designs available when a split is impossible, what each one assumes and how much smaller the claim becomes once the randomisation has gone.
- The peeking problemWhat repeated checking does to the false positive rate of a running test, the published figures for one check against ten, why the habit is invisible from inside a team and the two legitimate ways to stop a test early.
- Multiple comparisonsWhy examining many measures at the five per cent threshold produces winners out of nothing, the arithmetic for twenty measures and for ninety six segment cuts, the two corrections available and the practice that avoids needing either.