Jude Wallis

AP Statistics is being rebuilt. The course drops from nine units to five in Fall 2026, and the first exam on the new course is May 2027. That timing is the reason this site exists in the form it does: almost every study resource a student can find right now describes the old course, and a resource that is out of date about which topics are examinable is not slightly less useful, it is actively misleading.

StatsLearn was written against the new course structure from the first page. Every unit, every topic code, and every claim about the exam format traces to the official College Board course and exam description rather than to a memory of how the course used to work.

Everything is free, and nothing needs an account: 148 guides that answer one question each, 16 calculators that show every step instead of only the final number, 55 topic pages covering the full CED, 224 glossary entries, 53 practice sets with 424 worked problems, and 8 interactives for the ideas that static pages handle badly, like sampling distributions, confidence interval coverage, and influential points.

A wrong standard error on a statistics site is a bug, not a typo, so accuracy here is a process rather than an intention. Every page is written, then handed to a separate adversarial review whose explicit job is to find defects: it recomputes each number from the page’s own stated inputs, solves every practice problem from the question alone before reading the provided solution, and tries to build a counterexample to any sentence containing the words always or never. The findings are applied, and the page is re-checked end to end afterwards, because a revision pass breaks what it touches.

Automated checks then run before anything can ship. They fail the build on a KaTeX expression that does not compile, an internal link that does not resolve, a page with no inbound links, structured data that references something that is not there, and statistical claims this site refuses to make: no page may tell you to accept the null hypothesis, describe a p-value as the probability that the null is true, or say a 95% confidence interval has a 95% chance of containing the parameter.

That process is not a guarantee. It is a reason to believe a correction will be made quickly when something is wrong.

If you find a mistake, please report it. Wrong statistics content teaches something false to whoever reads it next, so corrections take priority over new pages. Reach Jude on LinkedIn. Point at the page and quote the sentence; that is enough.

More on how this site is built, what it refuses to do, and the other subjects in the network: about StatsLearn.