These lessons are adapted from my graduate methods course at Texas A&M. The rule that organizes them: no term appears before the thing it depends on. Concrete examples first, formal language last, every load-bearing word grounded the moment it shows up.
New lessons are added weekly. They're free, and they're written for anyone — researchers, product teams, applicants, people who run evals and suspect the numbers are lying to them.
Three criteria, three ways the arrow fools you, and why the pattern being real doesn't make your explanation true.
The first translation problem in all of statistics — and a two-step diagnostic that settles any variable in ten seconds.
NEXT IN THE SERIES — Validity & Reliability · Sampling Logic · Missing Data (MCAR / MAR / MNAR)