Methodology & accuracy

Every result on Statistics Calculator is produced by a single, open statistics library that runs directly in your browser. This page explains how those calculations are made and how we keep them accurate. Last reviewed August 2026.

Calculations run in your browser

The numbers you enter are never sent to a server. Each calculator calls the same shared TypeScript library of pure, deterministic functions — the mean, variance, distribution CDFs, regression, and hypothesis tests are all computed locally and instantly.

Tested against reference values

The library is covered by an automated test suite that checks each function against known reference values — worked textbook examples, standard statistical tables, and the results returned by established tools such as Excel, R, and SciPy. For example, the Student’s t and F critical values shown in our tables are generated by the same functions that power the calculators and are verified against published critical-value tables.

Formulas and conventions

• Sample statistics use Bessel’s correction (dividing by n − 1); population statistics divide by N. You choose which applies to your data.
• Skewness and kurtosis use the bias-adjusted sample formulas (matching Excel’s SKEW and KURT), with the population versions also shown.
• Continuous distributions (normal, Student’s t, F) are evaluated with well-established numerical methods for the error function and the regularized incomplete beta function.

Found a discrepancy?

We aim to match authoritative references exactly. If a result looks wrong, please double check the inputs and the sample-versus-population setting first — and see our calculators for the step-by-step working that shows exactly how each answer is derived.