P-Value Calculator
A p-value answers one question: if the null hypothesis were true — if there were really no effect — how surprising would your result be? A p-value of 0.03 means a result at least this extreme would turn up by chance about 3% of the time. Small p-values cast doubt on the null; large ones do not. The conventional cut-offs are 0.05 (significant) and 0.01 (highly significant), though a p-value is a continuous measure of evidence, not a verdict.
Enter your test statistic — a z-score, a t-score with its degrees of freedom, or a chi-square statistic — choose one-tailed or two-tailed, and the calculator gives you the p-value plus a plain-English reading of what it means at the 5% and 1% levels.
P-value
How It Works
The p-value is the probability, under the null hypothesis, of getting a test statistic at least as extreme as the one you observed. How "extreme" is counted depends on the test:
- Two-tailed (z or t): both tails count — p = 2 × P(X ≥ |statistic|). A z-score of 1.96 gives p = 0.0500, and a z-score of −1.96 gives exactly the same p-value, because the normal distribution is symmetric.
- One-tailed right: only the right tail counts — p = P(X ≥ statistic). A z-score of 1.645 gives p = 0.0500. Use this only when you predicted the direction before collecting the data.
- One-tailed left: only the left tail counts — p = P(X ≤ statistic).
- Chi-square: always right-tailed — p = P(X ≥ statistic). The chi-square statistic measures how far the data sits from what the null predicts, so only large values count against the null, and the statistic can never be negative.
The t distribution needs the degrees of freedom as well as the statistic: with fewer degrees of freedom the tails are heavier, so the same t-score gives a larger p-value. A t of 2.5 with 10 degrees of freedom gives p = 0.0314 (two-tailed); the same 2.5 as a z-score gives p = 0.0124, because the normal distribution has thinner tails.
The numbers behind the famous thresholds are worth knowing. For a two-tailed z-test, 1.96 is the 5% critical value and 2.576 the 1% value. For a t-test with 9 degrees of freedom, 2.262 is the 5% value. For chi-square, 3.841 (df = 1) and 5.991 (df = 2) are the 5% values. Each of these gives a p-value of 0.0500 — try them here.
Worked Example
A z-test gives z = 1.96. Two-tailed, the p-value is 0.0500 — right on the conventional 5% line. Push the statistic to z = 2.576 and the two-tailed p-value falls to 0.0100, the 1% threshold. A one-tailed test with z = 1.645 also gives 0.0500, because one tail at 1.645 holds the same area as two tails at 1.96.
A t-test gives t = 2.5 with 10 degrees of freedom. Two-tailed, the p-value is 0.0314 — significant at 5%, but not at 1%. With t = 2.262 and 9 degrees of freedom the two-tailed p-value is 0.0500, the 5% critical value for that t distribution. For chi-square, χ² = 3.841 with 1 degree of freedom gives 0.0500, and χ² = 5.991 with 2 degrees of freedom also gives 0.0500.
Assumptions
- The test statistic must be computed correctly first. This calculator turns a statistic into a p-value; it does not check whether the statistic itself is right, or whether the test's assumptions (normality, independence, expected counts) hold.
- One-tailed or two-tailed is decided before seeing the data. Picking a one-tailed test after noticing which way the result went doubles the false-positive rate. When in doubt, use two-tailed.
- Chi-square is right-tailed only. The statistic is a sum of squared values, so it cannot be negative, and only large values argue against the null hypothesis.
- Degrees of freedom are whole numbers from 1 to 1,000. Above 1,000 the t distribution is practically the normal distribution, so the z-score mode gives the same answer.
- P-values are numerical approximations, accurate to about 4 decimal places for typical ranges. Extremely small p-values (below 0.0001) are shown in scientific notation in the details table.
- A p-value is not an effect size. A tiny p-value can come from a trivial effect measured on a huge sample, and a large p-value can hide a real effect measured on a small one. Report the effect and its confidence interval alongside the p-value.
Sources: No external data is used on this page — every figure is calculated from the values you enter. The normal CDF uses the error function evaluated through the regularized incomplete gamma function; the t CDF uses the regularized incomplete beta function (Numerical Recipes betacf); the chi-square survival function uses the regularized incomplete gamma function. All three were checked against published critical values (z = 1.96, t = 2.262 at df 9, χ² = 3.841 at df 1 all give p = 0.0500). Last reviewed: 15 November 2026.
Built and tested by SumKitTool — see how we check our numbers.
How to Use This Calculator
Pick the test statistic type, enter the statistic (and the degrees of freedom for t and chi-square), and choose the tail. The headline result is the p-value to four decimal places; the table shows the exact value, whether the result is significant at the 5% and 1% levels, and a plain-English interpretation.
Try the classic values: z = 1.96 two-tailed gives 0.0500, t = 2.5 with 10 degrees of freedom gives 0.0314, and chi-square 5.991 with 2 degrees of freedom gives 0.0500. Switch the z-test to one-tailed right with z = 1.645 to see 0.0500 from a single tail.
Tips and Practical Advice
Do not treat 0.05 as a cliff. A p-value of 0.049 and one of 0.051 are nearly identical evidence, even though one is "significant" and the other is not. Report the p-value itself and let readers judge, rather than reducing it to a yes/no.
Watch out for multiple testing. If you run 20 tests on data with no real effects, you should expect about one p-value below 0.05 by chance alone. Testing many hypotheses without adjusting for it manufactures false discoveries.
A p-value never proves the null hypothesis is false — and a large p-value never proves it is true. "Not significant" means the data did not provide enough evidence against the null, which is often a statement about the sample size rather than about reality.
Frequently Asked Questions
What does a p-value of 0.05 mean?
It means that if the null hypothesis were true — if there were really no effect — a result at least this extreme would occur by chance about 5% of the time. It is the conventional threshold for statistical significance, but it is not a magic line: p = 0.049 and p = 0.051 are nearly identical evidence. A z-score of 1.96 (two-tailed) is the classic example that gives p = 0.0500.
When should I use a one-tailed test?
Only when you predicted the direction of the effect before collecting the data — for example, that a new drug can only help, never harm. A one-tailed test puts the whole 5% in one tail, so the critical value is lower (z = 1.645 instead of 1.96). Choosing one-tailed after seeing which way the result went doubles the false-positive rate, so when in doubt use two-tailed.
Why is the chi-square test only right-tailed?
Because the chi-square statistic measures how far the observed data sits from what the null hypothesis predicts — it is a sum of squared differences, so it can never be negative, and only large values argue against the null. A small chi-square value means the data fits the null well. There is no meaningful left tail to test, so the p-value is always the area to the right of the statistic.
Is a small p-value proof that my hypothesis is true?
No. A p-value measures how surprising the data would be if the null hypothesis were true — it is not the probability that the null is true, and it is not the probability that your hypothesis is true. It also says nothing about the size or importance of the effect: a tiny p-value can come from a trivial effect on a huge sample. Always report the effect size and a confidence interval alongside the p-value.
Last updated: