Skip to content
Statculator
P-Values & Critical Values

P-Value from F Calculator

Free p-value from F calculator. Enter F, df₁ and df₂ to get the right-, left- or two-tailed p-value for ANOVA or regression, the decision and every step.

Free, no sign-up Runs in your browserBy Hira Noor, BS MathematicsUpdated

Try

ANOVA: k − 1. Regression: number of predictors.

ANOVA: N − k. Regression: n − k − 1.

Tail

ANOVA and regression F-tests are right-tailed. Two-tailed is for comparing two variances.

Leave blank to get the p-value without a decision.

Results

p-value (right-tailed)
0.013
p = P(F ≥ 4.12), df = (3, 36)
Critical F at α = 0.05
2.8663

Significant at α = 0.05: reject H₀

p = 0.013 ≤ 0.05. Reject H₀ if F ≥ 2.8663, and the observed F is 4.12.

Right tail, P(F ≥ x)
0.013
Left tail, P(F ≤ x)
0.987
Two-tailed, 2 × smaller tail
0.0261
Degrees of freedom (df₁, df₂)
(3, 36)
Reciprocal 1/F on (df₂, df₁)
0.2427
Implied η² or R² = F·df₁ / (F·df₁ + df₂)
0.2556

Interpretation: If H₀ is true, an F statistic of 4.12 or more would turn up about once in every 77 repetitions with df = (3, 36). If F comes from a one-way ANOVA or an overall regression test, the effect explains about 25.56% of the variation.

f distribution curve with the shaded probability region, markers at F = 4.12F = 4.120.00.51.01.52.02.53.03.54.04.5F distribution, df₁ = 3, df₂ = 36
Shaded area = p-value = 0.013. Dashed line: your F statistic.

Decision at common significance levels (right-tailed, df = (3, 36))

αCritical FReject H₀?
0.12.2426Yes
0.052.8663Yes
0.0253.5047Yes
0.014.3771No
0.0055.0616No
0.0016.7436No

p = 0.013. “Yes” means p ≤ α, the same as F falling in the rejection region.

Step-by-step solution

  1. 1

    Write the p-value as a probability

    A large F means the numerator mean square is big relative to the denominator, which is evidence against H₀. The p-value is the area to the right of F under the F(3, 36) curve.

    p = P(F ≥ 4.12), F ~ F(3, 36)
  2. 2

    Convert F to the incomplete-beta argument

    The upper tail of the F distribution is a regularized incomplete beta function evaluated at w = df₂ / (df₂ + df₁·F).

    w = 36 / (36 + 3 × 4.12) = 0.744417
  3. 3

    Find the upper-tail area

    P(F ≥ 4.12) = I_w(df₂/2, df₁/2) = I(0.744417; 18, 1.5) = 0.01305
    P(F ≤ 4.12) = 1 − 0.01305 = 0.98695
  4. 4

    Read off the p-value

    p = 0.013
  5. 5

    Decision: reject H₀

    p = 0.013 ≤ α = 0.05
    Reject H₀ if F ≥ 2.8663; observed F = 4.12

What is a P-Value from F Calculator?

A p-value from F calculator turns an F statistic and its two degrees of freedom (numerator df₁ and denominator df₂) into a p-value. For ANOVA and regression the p-value is the area to the right of F under the F distribution, since only large ratios of mean squares count against H₀. For example, F = 4.12 with df = (3, 36) gives p = 0.0130, which is significant at α = 0.05. A two-tailed option covers tests that compare two variances.

What the P-Value from F Calculator does

Enter an F value with df₁ and df₂ and the calculator returns the right-tail, left-tail and two-tailed p-values, the critical F at your α and the decision, and shades the matching area on the F curve. It also reports the η² (or R²) implied by F and its df, a quick effect-size check for one-way ANOVA and overall regression tests. It works as an F-test p-value calculator, an F statistic to p-value converter or an F score to p-value tool: whatever the name, the input is F with its two degrees of freedom, and the F stat p-value comes out.

F statistics appear in one-way ANOVA, the overall F-test of a multiple regression, comparisons of nested models, and the F-test for two variances. Use this page when you have F and the degrees of freedom but no p-value, or when a printed F-table only lists a few α levels.

How to use the P-Value from F Calculator

  1. Enter the F statistic, the ratio of two mean squares or two variances. It cannot be negative.
  2. Enter the Numerator df (df₁) and Denominator df (df₂). Order matters: F(3, 36) is a different distribution from F(36, 3). The Degrees of Freedom Calculator works them out if you are unsure.
  3. Keep Tail on Right for ANOVA and regression. Choose Two-tailed or Left only for a variance-ratio test with that alternative.
  4. Optionally enter Significance level (α) and read the p-value, the decision and the shaded curve. The steps show the incomplete-beta calculation.

How to calculate the p-value from an F statistic

To get the p-value from an F-test you need three numbers: the F statistic (sometimes called the F score or F value), the numerator df and the denominator df. The p-value of the F-test is the probability that an F(df₁, df₂) variable is at least as large as the F you observed, assuming H₀ is true.

  1. Compute F. It is a ratio of two mean squares: in one-way ANOVA, F = MS between / MS within. For example, MS between = 24.6 and MS within = 5.97 give F = 24.6 / 5.97 = 4.12.
  2. Find df₁ and df₂. One-way ANOVA: df₁ = k − 1 and df₂ = N − k. Overall regression: df₁ = number of predictors and df₂ = n − k − 1. Two variances: n₁ − 1 and n₂ − 1.
  3. Find the right-tail area. p = P(F ≥ your value) = 1 − CDF. A calculator or software gives it exactly. With a printed F-table, look up your (df₁, df₂) cell in the α = 0.05 and 0.01 tables; if F lies between the two critical values, 0.01 < p < 0.05.
  4. By hand for small df₁. When df₁ = 1, F = t², so the p-value is the two-tailed t p-value for t = √F with df₂ degrees of freedom. When df₁ = 2 there is a closed form: p = (1 + 2F / df₂)^(−df₂/2). For F = 3.5 with (2, 20), p = 1.35^(−10) = 0.0497.
  5. Compare with α. If p ≤ α, reject H₀. If p > α, fail to reject H₀.

F-test p-value table for 1 numerator degree of freedom

Right-tail p-values for F with df₁ = 1, the case of a two-group ANOVA, a single regression coefficient or a 1-df contrast. Read across from your denominator df to the F nearest your statistic; for an exact value, or for other df₁, enter it in the calculator above. The last column is the critical F at α = 0.05.

Right-tail p-value, P(F ≥ x), for df₁ = 1
df₂F = 1F = 2F = 3F = 4F = 5F = 7.5F = 10F = 15Critical F (α = 0.05)
50.36320.21640.14380.10190.07560.04090.02500.01176.608
100.34090.18770.11390.07340.04930.02090.01010.00314.965
150.33320.17770.10380.06390.04100.01520.00640.00154.543
200.32930.17270.09870.05930.03690.01270.00490.00094.351
300.32530.16760.09350.05460.03290.01030.00360.00054.171
600.32130.16250.08840.05000.02910.00810.00250.00034.001
1200.31930.15990.08580.04780.02720.00710.00200.00023.920
∞0.31730.15730.08330.04550.02530.00620.00160.00013.841

F-test p-value formula

The F distribution’s tail area has a closed form in terms of the regularized incomplete beta function Iₓ(a, b).

Right-tailed p-value (ANOVA, regression)
p = P(F ≥ x) = I_w(df₂/2, df₁/2), w = df₂ / (df₂ + df₁·x)
Left-tailed p-value
p = P(F ≤ x) = 1 − P(F ≥ x)
Two-tailed p-value (two variances)
p = 2 × min{ P(F ≤ x), P(F ≥ x) }
Reciprocal property
P(F(df₁, df₂) ≤ x) = P(F(df₂, df₁) ≥ 1/x)
Formula symbols
SymbolMeaning
xobserved F statistic
df₁numerator degrees of freedom
df₂denominator degrees of freedom
I_w(a, b)regularized incomplete beta function

The reciprocal property means a left-tail area can always be found as a right-tail area by flipping F to 1/F and swapping the two df. Printed F-tables rely on it because they list only upper-tail values.

The distribution follows the NIST/SEMATECH e-Handbook definition of the F distribution. Critical values here agree with its table of F critical values, and the p-values match R’s pf and SciPy’s f.sf.

Worked example

Worked example: four teaching methods

Problem. A one-way ANOVA compares exam scores under four teaching methods with 10 students each. The ANOVA table gives F = 4.12. Find the p-value and decide at α = 0.05.

  1. 01Degrees of freedom: df₁ = k − 1 = 4 − 1 = 3 and df₂ = N − k = 40 − 4 = 36.
  2. 02ANOVA is right-tailed: p = P(F ≥ 4.12) with F ~ F(3, 36).
  3. 03Incomplete-beta argument: w = 36 / (36 + 3 × 4.12) = 36 / 48.36 = 0.744417.
  4. 04Upper tail: p = I(0.744417; 18, 1.5) = 0.0130.
  5. 05The critical value for α = 0.05 is F(3, 36) = 2.8663, and 4.12 > 2.8663.

Answer. p ≈ 0.0130 < 0.05, so reject H₀: at least one teaching method has a different mean score. The implied η² = (4.12 × 3) / (4.12 × 3 + 36) ≈ 0.2556, so about a quarter of the score variation is explained by method.

Where F statistics and their df come from

Each F statistic is a ratio of two independent variance estimates. The numerator df belongs to the top of the ratio and the denominator df to the bottom, so reading them from the right places is the most important step.

TestF statisticdf₁df₂Tail
One-way ANOVA (k groups, N values)MS between / MS withink − 1N − kRight
Overall regression F (k predictors, n cases)MS regression / MS residualkn − k − 1Right
Nested models (q extra parameters)[(SSE_reduced − SSE_full)/q] / MSE_fullqdf of full-model errorRight
Two variancess₁² / s₂²n₁ − 1n₂ − 1Two-tailed, left or right

What the p-value of an F-test means

The F-test p-value is the probability of an F statistic at least as large as yours if H₀ were true. In a one-way ANOVA, H₀ says all group means are equal; in an overall regression test, that every slope is zero; in a nested-model test, that the extra terms add nothing.

A p-value of 0.05 or less means a ratio of mean squares this large would turn up in at most 5% of samples under H₀, so you reject H₀. With (3, 36) df that happens once F reaches 2.866; with (1, 30) df, once it reaches 4.171. An F near 1 is what H₀ predicts, because then the two mean squares estimate the same variance.

A small p-value says the means (or slopes) are not all equal, not which ones differ or by how much. Report the F with both df and the p-value, for example F(3, 36) = 4.12, p = .013, and add an effect size such as η² or R².

F-test p-value in Excel, R, Python and on a TI-84

Every package computes the same right-tail area. The examples use F = 4.12 with df₁ = 3 and df₂ = 36, which all return 0.0130.

ToolRight-tailed p-valueLeft-tailed p-value
Excel / Google Sheets=F.DIST.RT(4.12, 3, 36)=F.DIST(4.12, 3, 36, TRUE)
Rpf(4.12, 3, 36, lower.tail = FALSE)pf(4.12, 3, 36)
Python (SciPy)f.sf(4.12, 3, 36)f.cdf(4.12, 3, 36)
TI-83 / TI-84Fcdf(4.12, 1E99, 3, 36)Fcdf(0, 4.12, 3, 36)

Common use cases

  • Getting the p-value for a one-way or factorial ANOVA from the F in an ANOVA table.
  • Testing the overall significance of a regression model, or whether a block of extra predictors improves it.
  • Comparing two population variances with the variance-ratio F-test, one- or two-sided.
  • Checking F(df₁, df₂) values reported in papers, such as “F(2, 57) = 3.41, p = .040”.

Assumptions and limitations

  • The p-value assumes the statistic really follows an F distribution under H₀: independent observations, normal errors and, for ANOVA, equal group variances. Welch’s ANOVA or a nonparametric test is safer when variances differ.
  • The variance-ratio F-test is highly sensitive to non-normality; Levene’s or the Brown–Forsythe test is more robust for checking equal variances.
  • A significant ANOVA F says the means are not all equal, not which ones differ. Follow it with post-hoc comparisons.
  • The implied η² is exact for one-way ANOVA and for the overall regression F (where it equals R²), but for a single effect in a factorial design it is a partial η².

Frequently asked questions

How do you calculate the p-value from an F score?

Find the area to the right of your F statistic under the F distribution with the numerator and denominator degrees of freedom: p = 1 − CDF(F; df₁, df₂). Enter F, df₁ and df₂ in the calculator, or use =F.DIST.RT(F, df₁, df₂) in Excel or pf(F, df₁, df₂, lower.tail = FALSE) in R. For example, F = 4.12 with (3, 36) df gives p = 0.0130.

How do you calculate an F score?

Divide one mean square by another. In one-way ANOVA, F = MS between / MS within, where MS between = SS between / (k − 1) and MS within = SS within / (N − k). For two variances, F = s₁² / s₂². The ANOVA Calculator and F-Test Calculator compute F from raw data.

Can you calculate a p-value from a z-score?

Yes, with the P-Value from Z Calculator, which uses the normal curve instead of the F distribution. The two are linked: a z-score squared follows F(1, ∞), so z = 1.96 and F = 3.8416 with (1, ∞) df both give p ≈ 0.05.

Is the F-test one-tailed or two-tailed?

ANOVA and regression F-tests are right-tailed, because only large F values indicate that the effect explains more variation than chance. The F-test comparing two variances can be two-tailed, in which case the p-value is twice the smaller tail area.

Which df goes first in the F distribution?

The numerator df comes first and the denominator df second. In a one-way ANOVA these are k − 1 (between groups) and N − k (within groups). Swapping them gives a different distribution and a wrong p-value.

What does an F value below 1 mean?

F below 1 means the numerator variance estimate is smaller than the denominator one. In ANOVA it means the group means are closer together than random noise alone would usually produce, and the right-tailed p-value will be large, typically above 0.3.

How is the F p-value related to the t-test p-value?

When df₁ = 1, F equals t² for the matching t-test, and the right-tailed F p-value equals the two-tailed t p-value. For example, t = 2.31 with df = 14 and F = 5.3361 with (1, 14) df both give p ≈ 0.0366.

Can I use non-integer degrees of freedom?

Yes. Welch’s ANOVA and some corrections, such as Greenhouse–Geisser in repeated-measures ANOVA, produce fractional df. The F distribution is defined for any positive df, and this calculator uses them exactly.

Cite this page

Statculator. (2026, November 14). P-Value from F Calculator. Retrieved November 14, 2026, from https://statculator.com/p-values-critical-values/p-value-from-f-calculator/