P Value T Test Calculator
A t statistic on its own answers nothing: it still has to be turned into a p-value before you know whether your result is statistically significant. This calculator runs the full test for you, one-sample, Welch two-sample, or paired, from summary statistics or pasted raw data, or converts a t statistic and its degrees of freedom entered directly into a p-value, and reports the t statistic, degrees of freedom, exact p-value, critical value, confidence interval, Cohen’s d, and a plain-English verdict.
Sample data
Quick Answer
A p-value is the probability of seeing a t statistic at least as extreme as yours if the null hypothesis were true. If it is at or below your significance level (usually 0.05), reject the null. The calculator runs the whole test: enter summary statistics or paste raw data, pick one-sample, Welch two-sample or paired, or enter a t statistic and its degrees of freedom directly, and get t, df, the exact p-value, the critical value, a confidence interval, Cohen’s d and the verdict in one table.
How to use this calculator
Choose the test type first. For a one-sample test, enter the sample mean, standard deviation, size and the hypothesized population mean. For a Welch two-sample test, fill in both samples (the sample 1 fields stay visible when you switch modes). For a paired test, paste the two matched lists, one value per line. Alternatively, paste raw data in any mode and it overrides the summary fields. Pick the tail direction and alpha, then Calculate. Results stay hidden until you click, and Reset restores the worked example below.
If you already have a t statistic, from software, a textbook or a published table, choose Direct t statistic (t + df) as the test type instead, enter the statistic and its degrees of freedom, and Calculate returns its p-value, critical value and verdict. The confidence interval and Cohen’s d rows read “Not applicable” in this mode, because both need the sample data behind the statistic.
Which test do I need?
Welch’s version is the default here rather than the classic pooled-variance two-sample test, because real groups rarely have identical variances, and Welch’s test stays valid either way. That is also why Welch degrees of freedom can be fractional: the calculator shows them to two decimals when they are not whole.
How the p-value is computed
The t statistic measures how many standard errors the observed difference sits from zero under the null. The p-value is the tail area of the t distribution beyond that statistic at the test’s degrees of freedom, doubled for a two-tailed test. This calculator evaluates the t distribution’s cumulative probability through the incomplete beta function, the same exact method statistics packages such as R use, so the p-value is exact to about six decimals rather than rounded from a table. A result smaller than 0.000001 is shown as < 0.000001, never as a misleading flat zero. Direct t entry runs through exactly the same incomplete beta calculation and the same tail handling, so a statistic entered directly returns the identical p-value the full test produces when it computes that same statistic.
Worked example
A researcher weighs 40 turtles: sample mean 300 pounds, sample SD 18.5, testing whether the species mean differs from 310 pounds (two-tailed, α = 0.05). The calculator is pre-filled with exactly these inputs:
- Standard error: 18.5 / √40 = 2.9251
- t = (300 − 310) / 2.9253 = −3.4187; df = 39
- p-value = 0.001487, well below 0.05
- Critical t (95%) = 2.0227; 95% CI = [294.08, 305.92], which excludes 310
- Cohen’s d = −0.541, a medium effect
- Verdict: Reject H₀. The data provide statistically significant evidence that the mean differs from 310 pounds.
Press Calculate and the tool returns these exact values, because they are its own outputs, not hand-rounded figures.
One-tailed or two-tailed?
Use a two-tailed test when you care about a difference in either direction. Use a one-tailed test only when the research question is directional from the start, for example testing whether a new drug lowers blood pressure. Switching to one-tailed after seeing the data is a common misuse that halves the p-value without justification, roughly doubling your false-positive rate. The calculator defaults to two-tailed, the conservative choice.
Reading the verdict
The results table gives you two independent ways to reach the same conclusion, and they always agree. The p-value route: reject H₀ when p ≤ α. The critical-value route: reject H₀ when |t| exceeds the critical value. The confidence interval adds a third view: if the interval excludes the null value, the test rejects. When the verdict says “fail to reject,” that is not proof the null is true, only that the evidence fell short. A statistical significance calculator frames the same decision in general terms, while a confidence interval calculator builds intervals on their own.
Cohen’s d: is the effect big or small?
A tiny p-value only says an effect is unlikely to be zero; with a huge sample, trivial differences turn significant. Cohen’s d standardizes the difference in standard-deviation units so you can judge practical importance: below 0.2 is negligible, 0.2 to 0.5 small, 0.5 to 0.8 medium, 0.8 and above large. The turtle example’s d of −0.541 is medium, so the 10-pound gap is both statistically significant and practically meaningful. For two samples the calculator uses the pooled standard deviation; for paired data it uses the standard deviation of the differences. Unlike p-values, effect sizes are comparable across studies, which is why journals increasingly ask for both.
Assumptions behind the tests
- One-sample: observations are a random sample and the population is approximately normal. With n of 30 or more, the central limit theorem makes the test robust to non-normality.
- Welch two-sample: the two samples are independent and each is approximately normal. Equal variances are not required, which is the point of Welch’s correction.
- Paired: the differences are approximately normal, and pairs are independent of each other. Check normality on the differences, not on the raw lists.
- If data are heavily skewed with a small sample, a nonparametric alternative such as the Wilcoxon signed-rank test is safer. A area to z-score calculator helps when you need normal-distribution tail areas directly, and a correlation coefficient calculator covers the related question of how two continuous variables move together.
What the p-value does not mean
Three misreadings refuse to die. A p-value is not the probability that the null hypothesis is true; it is the probability of data this extreme assuming the null is true. A non-significant result does not prove the null; it only says the data are compatible with it. And p = 0.049 versus p = 0.051 is not a cliff: the evidence is nearly identical on either side of the 0.05 line, which is a convention, not a law of nature.
What competing calculators miss
The top results for this keyword each cover part of the workflow:
- Penn State STAT 462: the gold-standard teaching page for t-based p-values, but it walks through Minitab dialogs rather than computing anything in the browser.
- Pearson: a polished statistic-to-p-value tool for z, t, χ² and F with visualizations, but it needs a precomputed test statistic and runs no test from data.
- StatGarden: a full one- and two-sample t-test with p-values, but the two-sample side pools variances and there is no paired mode and no raw-data paste.
- Conyso and The Magic Codex: clean t-score plus degrees-of-freedom to p-value converters with plain-English verdicts, but again statistic-entry only.
- QuadraticHQ: a template covering one, two and paired tests, but as a downloadable sheet rather than an instant web tool.
- The gap this page fills: one-sample, Welch two-sample and paired tests on one page, summary statistics or pasted raw data, exact p-values by the incomplete beta method, critical value, confidence interval, Cohen’s d with interpretation, and a reject/fail-to-reject verdict in a single results table with copy and share, plus a direct t + df entry mode for a statistic you already have.
Takeaways
- A p-value is a tail area under the t distribution, not the probability the null is true.
- Match the test to the design: one mean vs a value, two independent groups (Welch), or matched pairs.
- Fix the tail direction before collecting data; switching afterward inflates false positives.
- Report the effect size (Cohen’s d) alongside the p-value; significance is not importance.
Frequently asked questions
It is the probability of observing a t statistic at least as extreme as yours if the null hypothesis were true. A small p-value means your data would be surprising under the null, which is evidence against it.
Use the t-test when the population standard deviation is unknown and estimated from the sample, which is the usual case. The z-test applies when the population standard deviation is known in advance, or the sample is so large the estimate is effectively exact.
Choose two-tailed unless your research question named a direction before you collected the data. Switching to one-tailed after seeing which way the data lean halves the p-value and roughly doubles the false-positive rate, which is why the calculator defaults to two-tailed.
Welch’s t-test compares two independent means without assuming equal variances. It adjusts the standard error and the degrees of freedom (which is why df can be fractional). Use it as the default two-sample test unless you have a solid reason to believe the variances are equal.
Yes. In one-sample mode, paste numbers into the raw-data box and it overrides the mean, SD and n fields. In two-sample mode, paste both samples’ raw data. In paired mode, raw data is the only input: paste the before and after lists, one value per line.
It means the evidence fell short of your significance threshold, not that the null hypothesis is true. A larger sample might still detect a real effect, and the confidence interval shows which values remain plausible.
Cohen’s d expresses the difference in standard-deviation units: below 0.2 is negligible, 0.2 to 0.5 is small, 0.5 to 0.8 is medium, and 0.8 or more is large. The calculator labels each result with its interpretation automatically.
The Welch-Satterthwaite equation blends the two samples’ variances and sizes into a df that is rarely a whole number. It is not a rounding error: the t distribution accepts fractional df, and the calculator displays them to two decimals.
Yes. Choose Direct t statistic (t + df) as the test type, enter the statistic (keeping its sign, which matters for one-tailed tests) and its degrees of freedom, then pick the tail and alpha. It returns the p-value, critical value and verdict through the same calculation as the full tests. The confidence interval and Cohen’s d are not shown, because they need the sample data behind the statistic.
Add This Calculator to Your Website
Embed this calculator on your own site with the code below. Pick a size to preview it live, then copy the code.
Powered by Calculator Zilo
References
- Penn State Eberly College of Science, STAT 462. Find a t-based P-value. Accessed October 23, 2026.
- Price, P. C., Jhangiani, R., and Chiang, I. Research Methods in Psychology: Some Basic Null Hypothesis Tests. Accessed October 23, 2026.
- UCLA Institute for Digital Research and Education. SPSS Class Notes: Analyzing Data (t-tests). Accessed October 23, 2026.
- UCLA Institute for Digital Research and Education. What Statistical Analysis Should I Use? Statistical Analyses Using SPSS. Accessed October 23, 2026.
- BCcampus Open Education, Beginner Statistics for Psychology. Effect Size: Cohen’s d. Accessed October 23, 2026.