Free, accurate web tools — built to be genuinely useful.
A complete, free statistics toolkit: 73 calculators with step-by-step working, plain-English guides, reference tables, and interactive simulators — computed on a real Python/SciPy backend, so results match SPSS, R, and Stata. No sign-up.
t-tests, ANOVA, regression, confidence intervals, chi-square, and more — with interpretation, not just numbers.
Interactive games and simulators: guess-the-correlation, a Central Limit Theorem demo, a regression sandbox, and more.
Plain-English explanations of p-values, confidence intervals, effect size, hypothesis testing, and choosing the right test.
Notes on building practical AI systems and statistics tools. Six posts so far: what makes statistics tools reliable, interpreting effect size, an interactive Central Limit Theorem walkthrough, and lessons from building an offline meeting-minutes AI.
Stat-Tools is organised around what you are trying to do. Every calculator shows its working and explains what the result means, so you can check the method and report it correctly.
Enter or paste your data and get the test statistic, p-value, effect size and a plain-English conclusion.
Printable references for homework, exams and quick checks during analysis.
Short guides that explain the idea first and the formula second, with worked examples.
Statistics for comparing models honestly: uncertainty on benchmark scores, paired comparisons and agreement between raters.
Some ideas are easier to see than to read about. The playground has simulators and games where you change the inputs and watch the statistics respond.
Hypothesis tests run on a Python backend using SciPy, the same numerical library used by researchers and by many statistics packages, so a t-test or ANOVA here gives the same statistic and p-value you would get in R or SPSS for the same method. Interactive tools that run in your browser are checked against reference results from SciPy and statsmodels before they are published. Each calculator states which method it uses (for example, Welch's t-test by default rather than the pooled version), and the methodology page lists the formulas and references. If you find a result that disagrees with another package, email [email protected] and it will be investigated and corrected.
Students checking homework and preparing for exams, researchers and thesis writers who need to report a test correctly, analysts who want a quick second opinion on a result, and machine-learning engineers who need confidence intervals and significance tests for model evaluations. Everything is free, works on a phone and needs no account. Data you enter is used to compute the result and is not saved, unless you create an account and choose to save a result to your history.
Questions or feedback? [email protected] · About · Methodology