t-Test Calculator — Welch two-sample & one-sample
Run a Welch two-sample or one-sample t-test: means, standard deviations, t statistic, degrees of freedom, exact two-tailed p-value and Cohen’s d.
🔒 Runs entirely in your browser. Your input and output are never sent to a server or stored — fully private.
Welch’s t-test with the Welch–Satterthwaite df; p-values come from the exact t-distribution (regularized incomplete beta). Assumes roughly normal data — for very small or heavily skewed samples consider a nonparametric test (e.g. Mann–Whitney U). Educational aid, not a substitute for statistical advice. Click a result to copy.
About this tool
This t-test calculator checks whether the means of two groups really differ, or whether one group's mean differs from a target value. Paste the numbers for group A and group B and it runs Welch's t-test (which does not assume equal variances), reporting the means, standard deviations, the t statistic, degrees of freedom, an exact two-tailed p-value and Cohen's d effect size. A one-sample mode compares a single set against a hypothesised mean. It is a quick significance check for experiments, lab data or coursework — computed entirely in your browser.
Frequently asked questions
Why Welch's t-test and not Student's?
Student's t-test assumes both groups have the same variance, which is often false and, when violated, inflates false positives. Welch's version does not make that assumption and is nearly identical to Student's when variances happen to match, so it is the safer default and is what modern statisticians recommend by default. This tool uses Welch's for the two-sample case; if you specifically need Student's for a course, note that the two can give slightly different degrees of freedom and p-values.
What does Cohen's d add on top of the p-value?
The p-value tells you whether a difference is real; Cohen's d tells you how big it is, in standard-deviation units. Roughly, 0.2 is a small effect, 0.5 medium and 0.8 large. This matters because with a big sample even a trivial difference becomes statistically significant, so a tiny p with a d of 0.05 signals a real-but-negligible effect. Report both: significance without effect size can make an unimportant result sound impressive.
Is my data processed locally?
Yes. The means, variances, t statistic and exact p-value are all computed by JavaScript in your browser, so the numbers you paste never leave your device and it works offline once loaded — useful for unpublished experimental data. One caveat: the t-test assumes roughly normal, independent data; for very small, heavily skewed samples, consider a nonparametric test like Mann–Whitney instead, and treat this as an educational aid rather than a substitute for statistical review.