A p-value is the probability of obtaining a result at least as extreme as the one you observed, assuming the null hypothesis is true. A small p-value means your data would be unusual if there were truly no effect.

What does p less than 0.05 mean?

It means that if the null hypothesis were true, data this extreme would occur less than 5% of the time. By convention this is called statistically significant and the null hypothesis is rejected at the 0.05 level.

Does a p-value tell you the size of an effect?

No. A p-value only indicates how incompatible the data are with the null hypothesis. With a large enough sample, a trivial effect can produce a very small p-value. Always report an effect size alongside the p-value.

Is the p-value the probability that the null hypothesis is true?

No. This is the most common misinterpretation. The p-value is the probability of the data given that the null hypothesis is true, not the probability that the null hypothesis is true given the data.

How do you report a p-value in APA style?

Report exact p-values to two or three decimals without a leading zero, for example p = .032. For very small values write p < .001. Include the test statistic, degrees of freedom, and an effect size.

What p-value is considered significant?

Conventionally p < 0.05, but the threshold is a convention rather than a law. Some fields use 0.01 or 0.005, and particle physics requires about 0.0000003 (five sigma).

P-Value Calculator: From z, t, Chi-Square, F & r (Free)

P-Value Calculator: From z, t, Chi-Square, F or r

Enter any test statistic and get the exact p-value instantly — one-tailed or two-tailed, with a clear significance verdict, the shaded distribution curve, critical values, the classic p-value table, and ready-to-copy APA sentences. Raw data mode runs the t-test for you.

Hypothesis TestingP-ValueSignificanceFree Online Tool

Quick answer: A p-value is the probability of getting a result at least as extreme as yours if the null hypothesis were true. Small p = your data would be surprising under "no effect". Convention: p < 0.05 is statistically significant. Key landmarks: z = 1.96 → p = 0.05, z = 2.576 → p = 0.01, z = 3.29 → p = 0.001. A one-tailed p-value is exactly half the two-tailed one.

🔑 Key Takeaways

1p = P(data this extreme | null is true) — not the probability the null is true, and not the probability your finding is a fluke. That reversal is the single most common error in statistics.
2p < 0.05 is a convention, not a law of nature. Fisher picked it as a rough rule; physics demands ~0.0000003, and many fields now argue for 0.005.
3The p-value says nothing about effect size. With n = 10,000 a meaningless difference gets p < .001. Always report an effect size (Cohen's d, η², r) beside it.
4One-tailed p = two-tailed p ÷ 2. Only use one-tailed if you predicted the direction before collecting data — switching afterwards is p-hacking.
5Chi-square and F tests are naturally one-tailed (upper tail only), because only large values indicate departure from the null.
6"Not significant" does not mean "no effect." It means insufficient evidence — often just a small sample. Check the confidence interval before concluding anything.
7Report exact p-values (p = .032, not "p < .05"), with no leading zero, and use p < .001 only for very small values — that is the APA 7 standard.

🧮 Calculate Your P-Value

z-score From a z-test or standardized value

t-score

Degrees of freedom (df) n − 1 for one sample; n₁+n₂−2 for two groups

Chi-square (χ²)

Degrees of freedom (df) (rows−1)(cols−1) for contingency tables

Chi-square tests use the upper tail only — the tail setting below is ignored.

F value

df between (df₁)

df within (df₂)

F tests use the upper tail only — the tail setting below is ignored.

Correlation r

Sample size (n)

Converted to t = r√(n−2) / √(1−r²) with df = n − 2.

Choose your test Compares the means of two independent groups assuming equal variances.

Comma-separated input is the default. Group names are editable. Paired tests and correlation require both columns to have the same number of values, entered in matching order.

Tail: Significance level (α):

🛠 How to Calculate a P-Value with This Tool

Pick your test statistic. z-score, t-score, chi-square, F, correlation r, or paste raw data for an automatic t-test — or load one of the ten sample scenarios.
Enter the value and degrees of freedom. z needs no df; t needs df; chi-square needs df; F needs both df₁ and df₂; r needs the sample size.
Choose the tail. Two-tailed is the default and correct choice unless you predicted the direction in advance. Chi-square and F ignore this — they are upper-tail tests by nature.
Set α. 0.05 is conventional; the tool compares your p-value against it and states the decision explicitly.
Read, check, report. Read the exact p-value and verdict, confirm the shaded tail area on the curve, then copy an APA sentence — remembering to add an effect size.

📑 P-Value Table: Critical Values & Landmarks

These are the critical values people look up most. A statistic larger than the critical value means p is smaller than that α.

Standard normal (z) — no degrees of freedom needed

α (two-tailed)	Critical z	α (one-tailed)	Critical z
0.10	1.645	0.10	1.282
0.05	1.960	0.05	1.645
0.01	2.576	0.01	2.326
0.001	3.291	0.001	3.090

Student's t — critical values by degrees of freedom (two-tailed)

df	α = 0.10	α = 0.05	α = 0.01	α = 0.001
5	2.015	2.571	4.032	6.869
10	1.812	2.228	3.169	4.587
20	1.725	2.086	2.845	3.850
30	1.697	2.042	2.750	3.646
60	1.671	2.000	2.660	3.460
∞ (= z)	1.645	1.960	2.576	3.291

Chi-square — critical values (upper tail)

df	α = 0.10	α = 0.05	α = 0.01
1	2.706	3.841	6.635
2	4.605	5.991	9.210
3	6.251	7.815	11.345
4	7.779	9.488	13.277
5	9.236	11.070	15.086

ƒ P-Value Formulas Used

①

Definition of the P-Value

p = P( test statistic at least as extreme | H₀ true )

H₀The null hypothesis — usually "no effect" or "no difference"

NoteIt is a tail area under the null distribution, never a probability about H₀ itself

②

P-Value from a z-Score

two-tailed: p = 2[1 − Φ(|z|)] · one-tailed: p = 1 − Φ(|z|)

ΦStandard normal cumulative distribution function

Checkz = 1.96 → p = .050 · z = 2.576 → p = .010

③

P-Value from a t-Score

p = 2[1 − F_t(|t|, df)] where F_t uses the incomplete beta function

dfn − 1 (one sample) or n₁ + n₂ − 2 (two independent groups)

Notet converges to z as df grows beyond about 100

④

P-Value from Chi-Square

p = 1 − P(df/2, χ²/2) (regularized lower incomplete gamma)

df(rows − 1)(columns − 1) for independence tests

TailUpper tail only — large χ² means poor fit to the null

⑤

P-Value from an F Statistic

p = 1 − I_x(df₁/2, df₂/2), x = df₁F / (df₁F + df₂)

I_xRegularized incomplete beta function

UseANOVA, regression overall fit, variance ratio tests

⑥

P-Value for a Correlation r

t = r√(n − 2) ÷ √(1 − r²), df = n − 2

ThenFeed t into the t-distribution formula above

⑦

Independent-Samples t (Raw Data Mode)

t = (M₁ − M₂) ÷ [ s_p√(1/n₁ + 1/n₂) ], df = n₁+n₂−2

s_pPooled SD = √[((n₁−1)s₁²+(n₂−1)s₂²)/(n₁+n₂−2)]

🔍 P-Value Interpretation — In Detail

The p-value answers one narrow question: if there were truly no effect, how often would I see data at least this extreme? A p of .03 means results like yours would arise 3% of the time under a true null. That is evidence against the null — but it is not a measure of how large, important, or reproducible your effect is.

What the number does and does not mean

A p-value is not the probability that the null hypothesis is true; that reversal (confusing P(data|H₀) with P(H₀|data)) is the single most common statistical error in published research. Nor is p = .04 "twice as significant" as p = .08 — p-values are noisy, and in replications they bounce around dramatically. Nor does p > .05 prove there is no effect; absence of evidence is not evidence of absence, and small samples routinely produce non-significant results for real effects. Finally, the p-value says nothing about whether your study design, measures, or sampling were sound — it assumes all of that was correct.

Reading the magnitude

Treat p as a continuous measure of evidence rather than a switch. Values near .05 are weak evidence; below .01 is moderately strong; below .001 is strong — always conditional on the model being right. Because the 0.05 line is arbitrary, never write "approaching significance" for p = .06 or treat .049 and .051 as categorically different. The American Statistical Association's 2016 statement makes this explicit: scientific conclusions should not be based only on whether a p-value crosses a threshold.

Always pair p with an effect size and interval

Sample size drives the p-value: with n = 20 a large effect can miss significance, while with n = 20,000 a trivial effect achieves p < .001. So the complete report is statistic + df + exact p + effect size + confidence interval. The effect size tells you whether the finding matters; the interval tells you how precisely you measured it; the p-value only tells you how surprising the data would be under the null. Our effect size calculators and confidence interval calculator produce the companions this number needs.

One-tailed versus two-tailed

A two-tailed test asks "is there a difference in either direction?" and is the default. A one-tailed test asks "is group 1 greater?" and halves the p-value, which makes it tempting — and makes switching to it after seeing the data a form of p-hacking. Choose the tail before collecting data, state the choice in your methods, and stick to it. Note chi-square and F tests are inherently upper-tail: only large values signal departure from the null, so tail selection does not apply.

✍ How to Report P-Values in APA Style

APA 7 asks for exact p-values, no leading zero (p = .032, not 0.032), two or three decimals, and always accompanied by the test statistic, degrees of freedom, and an effect size.

Template 1 — Significant t-test

"The treatment group scored significantly higher than the control group, t([30]) = [7.76], p < .001, d = [2.74], 95% CI [[1.78], [3.71]]."

Template 2 — Non-significant result

"The groups did not differ significantly, t([28]) = [1.80], p = [.083], d = [0.66], 95% CI [−[0.09], [1.40]]. The confidence interval includes zero but extends to a moderate effect, so a larger sample is needed before concluding there is no difference."

Template 3 — ANOVA (F test)

"There was a significant effect of [condition], F([2], [27]) = [5.20], p = [.012], η² = [.28], indicating a large effect."

Template 4 — Chi-square

"The association between [variable A] and [variable B] was significant, χ²([1], N = [120]) = [3.84], p = [.050], Cramér's V = [.18]."

Template 5 — Correlation

"[Variable X] correlated significantly with [variable Y], r([28]) = [.42], p = [.021], 95% CI [[.07], [.68]]."

Reporting checklist

Give the exact p to two or three decimals; use p < .001 only when it truly is; drop the leading zero; italicize p, t, F, and r; report degrees of freedom in parentheses; state whether the test was one- or two-tailed if it was one-tailed; and never report a p-value without an effect size.

🔬 Example Results

Six worked examples matching the sample scenarios in the dropdown. Load any of them and click Calculate to reproduce the numbers exactly.

Example 1 — The 0.05 Landmark: z = 1.96

z = 1.96 · two-tailed
p = 2 × [1 − Φ(1.96)] = 0.0500
Verdict: significant at α = 0.05 (exactly on the line)

The most famous number in statistics. Each red tail holds 2.5% of the area; together they make the 5% that defines the conventional significance threshold.

Example 2 — Typical t-Test: t = 2.50, df = 20

t = 2.50 · df = 20 · two-tailed
critical t = 2.086 · p = 0.0212
2.50 > 2.086 → reject H₀ at α = 0.05

A routine significant result. Report it as t(20) = 2.50, p = .021 — and add Cohen's d, because this p-value alone says nothing about how large the difference actually is.

Example 3 — Borderline: t = 1.80, df = 28 (p = 0.083)

t = 1.80 · df = 28 · two-tailed
critical t = 2.048 · p = 0.0827
1.80 < 2.048 → fail to reject H₀

Do not call this "approaching significance." The honest reading: the data are compatible with the null, but also with a moderate real effect — report the confidence interval and treat the study as inconclusive.

Example 4 — Chi-Square: χ² = 3.841, df = 1

χ² = 3.841 · df = 1
p = 0.0500 (upper tail only)
The classic df = 1 critical value

Chi-square is right-skewed and one-tailed by nature: only large values indicate the observed frequencies depart from expectation. Note 3.841 = 1.96² — the same 5% threshold seen through a different distribution.

Example 5 — ANOVA F: F = 5.20, df = 2, 27

F = 5.20 · df₁ = 2, df₂ = 27
critical F = 3.354 · p = 0.0123
Report: F(2, 27) = 5.20, p = .012, η² = .28

A significant one-way ANOVA. The F distribution is also upper-tail only. Pair it with η² or ω² — our effect size calculators compute both from exactly these inputs.

Example 6 — How the P-Value Falls as the Statistic Grows

z = 1.00 → p = .317
z = 1.65 → p = .099
z = 1.96 → p = .050
z = 2.58 → p = .010
z = 3.29 → p = .001

The relationship is steep and non-linear. Small changes in the statistic near the threshold swing the p-value a lot, which is exactly why treating .049 and .051 as categorically different is indefensible.

🧭 When to Use This Tool

Use this p-value calculator whenever you have a test statistic and need the exact probability — reading a paper that reports only t or F, checking software output, doing homework or exam prep, converting between one- and two-tailed values, teaching hypothesis testing, or running a quick t-test from raw data. It suits students, researchers, reviewers, analysts, and anyone who needs a p-value without opening SPSS or R.

⚠ Assumptions & Limitations

Every p-value assumes its test's conditions hold: random sampling, independent observations, and the relevant distributional assumptions (normality for t and z tests, expected cell counts of at least 5 for chi-square, homogeneity of variance for the pooled t-test and ANOVA). If those fail, the p-value is not merely imprecise — it is answering a question about a model that does not describe your data. The raw-data mode here uses Student's pooled t-test; with clearly unequal variances Welch's version is preferable. P-values also assume a single pre-specified analysis: running many tests, peeking at data as it accumulates, or choosing the tail after seeing the result all inflate false positives (multiplicity and p-hacking). Correct for multiple comparisons when appropriate. Finally, p-values are famously unstable across replications, so a single p near .05 is weak evidence — effect sizes, intervals, and replication carry more information.

🏁 Conclusion

A p-value is a tail area: the probability of data at least as extreme as yours, computed under the assumption that the null hypothesis is true. Mechanically it is simple — find where your statistic falls on the null distribution and measure the area beyond it. For z that is the normal curve, for t the Student distribution with its degrees of freedom, for chi-square and F the upper tail only, and for a correlation you convert r to t first.

The difficulty is never the arithmetic; it is the interpretation. The p-value does not give the probability that the null is true, does not measure effect size, does not indicate importance, and does not become meaningful simply by crossing 0.05. It is one piece of evidence about compatibility between data and a model. Treat the threshold as a convention, report the exact value, and resist the urge to bin results into "significant" and "not significant."

The habit that fixes most p-value misuse is simple: never report one alone. Statistic, degrees of freedom, exact p, effect size, and confidence interval together tell a complete story — how surprising the data are, how big the effect is, and how precisely you measured it. This calculator gives you the exact p-value from any common test statistic, with plots that show the tail area you are actually computing, plus APA-ready sentences. Pair it with our effect size and confidence interval calculators, and your results section will satisfy any reviewer.

❓ Frequently Asked Questions

What is a p-value in simple terms?

The probability of getting results at least as extreme as yours if the null hypothesis (no real effect) were true. A small p-value means your data would be surprising in a world with no effect.

What does p < 0.05 mean?

If the null were true, data this extreme would occur less than 5% of the time. By convention that counts as statistically significant and the null is rejected at the .05 level — but it does not prove the effect is real or important.

How do you calculate a p-value from a z-score?

Two-tailed: p = 2[1 − Φ(|z|)], where Φ is the standard normal CDF. One-tailed: p = 1 − Φ(|z|). For example z = 1.96 gives p = .050 and z = 2.576 gives p = .010.

How do you calculate a p-value from a t-score?

Use the t distribution with your degrees of freedom rather than the normal curve: p = 2[1 − F_t(|t|, df)]. Because t has heavier tails, its p-value is always slightly larger than the z-based one for the same statistic.

What is the difference between one-tailed and two-tailed?

Two-tailed tests look for a difference in either direction; one-tailed tests look in one direction only and give exactly half the p-value. Choose the tail before seeing the data — switching afterwards is p-hacking.

Is the p-value the probability the null hypothesis is true?

No — this is the most common misinterpretation. It is P(data | null true), not P(null true | data). Answering the second question requires Bayesian methods and a prior.

Does a smaller p-value mean a bigger effect?

No. The p-value depends on both effect size and sample size. A tiny effect measured in 50,000 people yields a minuscule p; a large effect in 12 people may not reach significance. Only an effect size measures magnitude.

What does a non-significant p-value mean?

Insufficient evidence to reject the null — not proof the null is true. Absence of evidence is not evidence of absence. Check the confidence interval: if it spans from trivial to large effects, the study was simply inconclusive.

Why is 0.05 the standard threshold?

Ronald Fisher suggested it in the 1920s as a convenient rule of thumb, and convention hardened around it. There is nothing special about it mathematically — particle physics uses ~0.0000003, and many methodologists now advocate 0.005.

Can a p-value be exactly zero?

Not mathematically — the tail area is always positive. Software showing "p = .000" means the value is smaller than the displayed precision, which should be reported as p < .001.

How do I report p-values in APA style?

Exact values to two or three decimals with no leading zero: p = .032. Use p < .001 for very small values. Italicize p, include the test statistic and df, and always add an effect size.

What is p-hacking?

Manipulating analysis until p drops below .05 — trying different tests, dropping outliers selectively, adding participants until it works, or switching to one-tailed after the fact. It dramatically inflates false positives and is a major driver of the replication crisis.

What p-value do chi-square and F tests use?

The upper tail only. Both distributions are right-skewed and only large values indicate departure from the null, so there is no two-tailed option for them.

How do degrees of freedom affect the p-value?

Lower df means heavier tails and a larger p-value for the same statistic. t = 2.5 gives p = .067 at df = 5 but p = .021 at df = 20 and p = .012 at df = 100 — small samples are penalized appropriately.

How do I get a p-value for a correlation?

Convert r to t: t = r√(n−2)/√(1−r²) with df = n − 2, then read the t-distribution p-value. This tool does it automatically in the correlation tab.

Should I correct p-values for multiple comparisons?

Yes, when running many tests on the same data. Twenty independent tests at α = .05 give about a 64% chance of at least one false positive. Bonferroni (α/k) is simplest; Benjamini-Hochberg FDR is less conservative.

What is the difference between p-value and alpha?

Alpha is the threshold you set before the study (usually .05) — your tolerated false-positive rate. The p-value is what you compute from the data. You compare p to alpha to make a decision.

Can I compare two p-values?

Not directly. p = .04 is not "twice as significant" as p = .08, and a significant result in one study plus a non-significant one in another does not mean the studies disagree — compare effect sizes and their intervals instead.

What does the ASA say about p-values?

The American Statistical Association's 2016 statement warns that p-values do not measure effect size or the probability the hypothesis is true, and that scientific conclusions should not rest on whether p crosses a threshold.

What should I report alongside the p-value?

The test statistic, degrees of freedom, exact p, an effect size (Cohen's d, η², r), and a confidence interval. That combination answers how surprising, how large, and how precise — a p-value alone answers only the first.

📚 References

Wasserstein, R. L., & Lazar, N. A. (2016). The ASA statement on p-values: Context, process, and purpose. The American Statistician, 70(2), 129–133. https://doi.org/10.1080/00031305.2016.1154108
Wasserstein, R. L., Schirm, A. L., & Lazar, N. A. (2019). Moving to a world beyond "p < 0.05". The American Statistician, 73(sup1), 1–19. https://doi.org/10.1080/00031305.2019.1583913
Greenland, S., Senn, S. J., Rothman, K. J., Carlin, J. B., Poole, C., Goodman, S. N., & Altman, D. G. (2016). Statistical tests, P values, confidence intervals, and power: A guide to misinterpretations. European Journal of Epidemiology, 31(4), 337–350. https://doi.org/10.1007/s10654-016-0149-3
Fisher, R. A. (1925). Statistical Methods for Research Workers. Oliver & Boyd. Internet Archive
Goodman, S. (2008). A dirty dozen: Twelve p-value misconceptions. Seminars in Hematology, 45(3), 135–140. https://doi.org/10.1053/j.seminhematol.2008.04.003
Nuzzo, R. (2014). Scientific method: Statistical errors. Nature, 506(7487), 150–152. https://doi.org/10.1038/506150a
Simmons, J. P., Nelson, L. D., & Simonsohn, U. (2011). False-positive psychology. Psychological Science, 22(11), 1359–1366. https://doi.org/10.1177/0956797611417632
Benjamin, D. J., et al. (2018). Redefine statistical significance. Nature Human Behaviour, 2(1), 6–10. https://doi.org/10.1038/s41562-017-0189-z
Cumming, G. (2008). Replication and p intervals. Perspectives on Psychological Science, 3(4), 286–300. https://doi.org/10.1111/j.1745-6924.2008.00079.x
Halsey, L. G., Curran-Everett, D., Vowler, S. L., & Drummond, G. B. (2015). The fickle P value generates irreproducible results. Nature Methods, 12(3), 179–185. https://doi.org/10.1038/nmeth.3288
Ioannidis, J. P. A. (2005). Why most published research findings are false. PLoS Medicine, 2(8), e124. https://doi.org/10.1371/journal.pmed.0020124
Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate. Journal of the Royal Statistical Society: Series B, 57(1), 289–300. https://doi.org/10.1111/j.2517-6161.1995.tb02031.x
Sterne, J. A. C., & Davey Smith, G. (2001). Sifting the evidence — what's wrong with significance tests? BMJ, 322(7280), 226–231. https://doi.org/10.1136/bmj.322.7280.226
Lakens, D. (2021). The practical alternative to the p value is the correctly used p value. Perspectives on Psychological Science, 16(3), 639–648. https://doi.org/10.1177/1745691620958012
Gigerenzer, G. (2004). Mindless statistics. The Journal of Socio-Economics, 33(5), 587–606. https://doi.org/10.1016/j.socec.2004.09.033
Neyman, J., & Pearson, E. S. (1933). On the problem of the most efficient tests of statistical hypotheses. Philosophical Transactions of the Royal Society A, 231, 289–337. https://doi.org/10.1098/rsta.1933.0009
Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery, B. P. (2007). Numerical Recipes (3rd ed.). Cambridge University Press. numerical.recipes
NIST/SEMATECH. (2012). Critical values and p-values. In e-Handbook of Statistical Methods. https://www.itl.nist.gov/div898/handbook/prc/section1/prc131.htm
American Psychological Association. (2020). Publication Manual of the American Psychological Association (7th ed.). apastyle.apa.org
Moore, D. S., McCabe, G. P., & Craig, B. A. (2021). Introduction to the Practice of Statistics (10th ed.). W. H. Freeman. Publisher page