About
About these calculators
A companion to unspurious.com, the library of statistical illusions. The main site explains how correct numbers mislead; this corner gives you the tools to compute them correctly — and read them honestly.
The principle
Most statistics calculators online give you a number and nothing else: no formula, no assumptions, no interpretation, and a thicket of ads. That's how a confidence interval gets mistaken for a probability, a p-value for the chance a hypothesis is false, a relative risk for an absolute one, and a positive test for a diagnosis. These calculators are built on the opposite instinct — to make the working visible and the meaning explicit.
Every calculator
…shows the formula and the plugged-in numbers; gives a plain-English reading of the result; warns about the assumptions that matter (small samples, unequal variances, sparse cells, multiple comparisons, low base rates, heterogeneity, post-hoc power); offers a journal-style "how to report it" line; and hands you copyable R and Python to reproduce the calculation. Where a result touches a known trap, it links to the relevant illusion.
The mathematics
All computation runs in your browser — nothing is sent anywhere. The distribution functions (normal, t, χ², F, gamma and beta families, and their inverses) are implemented from standard numerical methods and tested against reference values from R and SciPy; forward and inverse functions agree to around twelve decimal places. Each calculator's result is independently checked against R. If you ever find a discrepancy, it's a bug worth reporting.
Accuracy & caveats
These tools are for learning, checking and reporting — not a substitute for a statistician on a study that matters. They compute what you ask correctly; they cannot tell you whether it was the right thing to ask. For that, the illusions are the better teacher.