What is an ANOVA Calculator?
This ANOVA calculator runs a one-way analysis of variance: it tests whether the means of two or more independent groups are all equal. It splits the total variation into a between-groups part and a within-groups part and compares them with F = MSB / MSW. A large F, with a small p-value from the F distribution on k − 1 and N − k degrees of freedom, means at least one group mean differs.
What the ANOVA Calculator does
The calculator handles a one-way (single-factor) ANOVA for 2 to 10 groups, from raw data or from group summaries (n, mean, SD). It fills in the complete ANOVA table (SS, df, MS, F and p for between and within groups), reports the critical F and the decision, and gives two effect sizes, η² and ω². Two follow-ups that most free tools leave out are built in: Tukey HSD post-hoc comparisons with adjusted p-values and confidence intervals, and Welch’s ANOVA for groups with unequal variances. Every number is worked through step by step, so it also works as an ANOVA solver for checking homework.
Use it when one numeric outcome is compared across the levels of a single factor: crop yield under several fertilizers, test scores for three teaching methods, waiting times at four branches. With two groups, ANOVA gives the same p-value as the pooled t-test (F = t²). It does not run a two-way, 2×2 or repeated-measures ANOVA; for two factors, read the Two-Way ANOVA: Main Effects, Interaction and a Worked Example guide and one-way vs two-way ANOVA. The One-Way ANOVA: Definition, Formulas and a Worked Example by Hand guide explains the ideas behind the table.
How to use the ANOVA Calculator
- Set the significance level α (0.05 is the usual choice).
- Choose Raw data and type or paste each group’s values into its own box (use Add group for more), or choose Summary statistics and enter each group’s name, n, mean and SD.
- Read F, the p-value and η², then check the ANOVA table and the Group summary.
- If F is significant, use the Tukey HSD post-hoc comparisons table to see which pairs of groups differ.
- If the group SDs are very different, report the Welch’s F and p-value instead of the classic F.
How to fill in an ANOVA table
Homework problems often give a one-way ANOVA table with blanks. Every cell follows from a few relationships, so you can fill it in from any two or three known values. SS and df add up down the columns; each mean square is its SS divided by its df; F is MSB divided by MSW.
- Find the df first: k − 1 between, N − k within, N − 1 total. If you know two of them, the third is their sum or difference.
- Get any missing SS from SST = SSB + SSW, or from MS × df.
- Divide each SS by its df to get MSB and MSW, then F = MSB / MSW.
- Look up the p-value for F with (k − 1, N − k) df, or compare F with the critical value from the F-table.
- Example: with k = 3 groups, N = 18, SSB = 67.4444 and SST = 105.7778, you get SSW = 38.3333, df = 2 and 15, MSB = 33.7222, MSW = 2.5556 and F = 13.1957.
| Source | SS | df | MS | F |
|---|---|---|---|---|
| Between groups | SSB | k − 1 | MSB = SSB / (k − 1) | F = MSB / MSW |
| Within groups (error) | SSW | N − k | MSW = SSW / (N − k) | |
| Total | SST = SSB + SSW | N − 1 |
ANOVA formula
| Symbol | Meaning |
|---|---|
| k | number of groups |
| N | total number of observations |
| nᵢ, x̄ᵢ, sᵢ | size, mean and SD of group i |
| x̄ | grand mean of all observations |
| SST | total sum of squares, SSB + SSW |
η² tends to overstate the effect in small samples; ω² adjusts for this and can be slightly negative when the true effect is near zero (read it as 0). Rough guides: 0.01 small, 0.06 medium, 0.14 large.
Group SDs use the sample (n − 1) divisor. The F test, Tukey HSD and Welch results were checked against SciPy (f_oneway, tukey_hsd) to at least four decimals.
Worked example
Worked example: three fertilizers
Problem. Yields (kg per plot) for six plots per fertilizer: A = 20, 22, 19, 24, 21, 23; B = 25, 27, 24, 26, 28, 25; C = 22, 21, 23, 20, 24, 22. Do mean yields differ at α = 0.05?
- 01Group means: 21.5, 25.8333 and 22; grand mean = 23.1111 (N = 18).
- 02SSB = 6(21.5 − 23.1111)² + 6(25.8333 − 23.1111)² + 6(22 − 23.1111)² = 67.4444, df = 2, MSB = 33.7222.
- 03SSW = 5(1.8708²) + 5(1.4720²) + 5(1.4142²) = 17.5 + 10.8333 + 10 = 38.3333, df = 15, MSW = 2.5556.
- 04F = 33.7222 / 2.5556 = 13.1957; critical F(2, 15) = 3.6823.
- 05p = P(F₂,₁₅ ≥ 13.1957) = 0.0005; η² = 67.4444 / 105.7778 = 0.6376, ω² = 0.5754.
- 06Tukey HSD (groups 1, 2, 3 = A, B, C): B − A = 4.3333, p = 0.0008; C − B = −3.8333, p = 0.0023; C − A = 0.5, p = 0.8522. B differs from both; A and C do not differ.
Answer. Reject H₀ (p < 0.001). Fertilizer B gives a higher mean yield than A and C, which do not differ from each other. Fertilizer explains about 64% of the variation in yield. These are the calculator’s default inputs, so you can check each step above.
Tukey HSD post-hoc test and Welch’s ANOVA
A significant F only says that the group means are not all equal. Tukey’s HSD (honestly significant difference) compares every pair of means while holding the family-wise error rate at α. For each pair it divides the difference in means by SE = √(MSW / 2 × (1/nᵢ + 1/nⱼ)) and refers the result, q, to the studentized range distribution with k groups and N − k df. This is the Tukey–Kramer form, which also works when group sizes differ.
Differences are listed as later group minus earlier group, as in R’s TukeyHSD. In the fertilizer example below (groups 1, 2 and 3 are A, B and C), B − A = 4.3333 (q = 6.6398, adjusted p = 0.0008) and C − B = −3.8333 (q = 5.8737, p = 0.0023) are significant, while C − A = 0.5 (q = 0.7661, p = 0.8522) is not. The critical q for k = 3 and df = 15 at α = 0.05 is 3.6734, so any two of these means must differ by at least 3.6734 × 0.6526 = 2.3974 kg.
Welch’s ANOVA drops the equal-variance assumption by weighting each group by nᵢ / sᵢ² and adjusting the error df. Prefer it when the largest SD is more than about twice the smallest, especially with unequal group sizes. For the example it gives F = 13.3004 on (2, 9.8757) df, p = 0.0016, which agrees with the classic test.
Common use cases
- Comparing outcomes across three or more treatments, doses or teaching methods.
- Testing whether average output or defect counts differ between several machines, shifts or sites.
- Comparing customer satisfaction scores across regions or product versions.
- Checking a textbook or homework ANOVA table, or one from Excel, SPSS or a TI-84, step by step.
- Analysing designed experiments with one factor before moving to two-way ANOVA.
Assumptions and limitations
- Independence: observations within and between groups must be independent. Repeated measures on the same subjects need repeated-measures ANOVA, which this calculator does not run.
- Normality: residuals should be roughly normal within each group; ANOVA tolerates moderate departures with balanced groups of 15 or more. For skewed small samples consider the Kruskal–Wallis test, or check each group with the Shapiro–Wilk test.
- Equal variances: group variances should be similar (largest SD under about twice the smallest). Check with Levene’s test; when they are not, especially with unequal group sizes, use the Welch’s ANOVA result.
- One factor only: two-way, factorial (2×2) and mixed designs need a different model. See Two-Way ANOVA: Main Effects, Interaction and a Worked Example.
- Running many t-tests instead of a post-hoc test inflates false positives; the Tukey HSD p-values here are already adjusted.
Frequently asked questions
What does the F value in ANOVA mean?
How do you calculate degrees of freedom in ANOVA?
What are the assumptions of one-way ANOVA?
What should I do after a significant ANOVA?
Can this calculator do a two-way ANOVA?
How do I run a one-way ANOVA on a TI-84?
How do I report a one-way ANOVA?
Cite this page
Statculator. (2026, November 14). ANOVA Calculator. Retrieved November 14, 2026, from https://statculator.com/hypothesis-tests/anova-calculator/