Fifteen shapes data takes
Part One is a story. Part Two is a different register — scannable, no narrative, built for the moment you're staring at a histogram and need a name for what you're looking at. This is one page from its Field Reference.
"Recognising a distribution on sight is most of statistical fluency. The name is not the point — knowing what it implies about the process behind it is."
Six shapes two variables can take together
What the cloud of points is telling you before any model touches it. Four of six read live off the book's ActiveEdge membership file; two stay illustrative because no real column pair in this dataset happens to fan out that way — said plainly on the card, not glossed over.
"Correlation has a sign, a strength, and a shape. A single number throws two of those three away."
A seventh shape belongs here too — Anscombe's Quartet, four datasets with identical r and a wildly different picture each time. See it right below.
"Mean, SD, and r describe the data." Four datasets below — pick one. Every number in the panel is computed live from its eleven points, and it barely moves.
Eleven questions, eleven tools
Knowing every tool is not the same as knowing which one this question needs. Pick the question — the shape of what you're actually asking usually says more than the method name does. The union of this book's own model table and the platform's own question list — merged into one.
"The question comes first. The question determines the method. The method is only as good as the reasoning that chose it."
Ten assumptions, quietly load-bearing
Every method in this book, and on the platform, assumes something about the data before it can be trusted. This is the spread print space hurts most — here it's the full table.
"An assumption you haven't checked is a bet you didn't know you made."
| assumption | where it matters | how you'd check | notation | the business risk | see it break |
|---|
Statistical terms, plainly
The book's own vocabulary (21 terms) merged with every acronym the platform shows in a warning or a summary without explaining (66 more) — one list, searchable.
"The name is not the point — knowing what it implies about the process behind it is."
What the business says, what the analyst hears
The same idea, in the two vocabularies this book has been translating between. Nine of the thirteen rows have a demo on this page that shows exactly the gap.
"A number is not a sentence. Someone still has to say what it means, and that's where the two vocabularies split."
A few more from the book
Worked examples the spine doesn't have room for — the full interactive version of each lives in its chapter.
Old Faithful
Eruption durations cluster around two very different lengths — the dataset every statistics course reaches for.
Price Elasticity
Why price-vs-demand is exponential, not linear — and what elasticity actually means for revenue. On the platform.
Process Control
When a process drifts out of control — and why the Western Electric rules catch it before a threshold breach does.
Full story: "A Rule Built for Telephone Relays…" — coming soon