One-Way ANOVA Calculator

Understanding the one-way ANOVA — concept and principle

One-way analysis of variance (ANOVA) tests whether the means of three or more independent groups are all equal, using a single F test instead of many pairwise t-tests. It is used to compare teaching methods, fertiliser types, production lines or any factor with several levels. Enter each group's values on its own row, separated by semicolons; the calculator returns F, the p-value, both degrees of freedom and η².

How it is calculated

1. Compute each group mean and the grand mean of all N values.
2. SS_between = Σ nⱼ (group meanⱼ − grand mean)², the spread of group means.
3. SS_within = Σ Σ (x − group mean)², the spread inside groups.
4. Mean squares: MS_between = SS_between ÷ (k − 1) and MS_within = SS_within ÷ (N − k), where k is the number of groups.
5. F = MS_between ÷ MS_within; the p-value is the upper tail of the F distribution with (k − 1, N − k) df.
6. η² = SS_between ÷ (SS_between + SS_within).

The principle

Total variation splits exactly into a part explained by group membership and a part left inside groups. MS_within estimates the random noise variance whatever the truth. MS_between estimates the same noise variance only if all population means are equal; if they differ, it is inflated. Their ratio is therefore close to 1 under the null hypothesis and large when the means really differ. With normal data and equal variances, the ratio follows the F distribution introduced by R. A. Fisher.

Worked example

1. Groups: {5, 6, 7, 8}, {7, 8, 9, 10}, {6, 7, 6, 8}; k = 3, N = 12.
2. Group means 6.5, 8.5, 6.75; grand mean 87 ÷ 12 = 7.25.
3. SS_between = 4(0.5625 + 1.5625 + 0.25) = 9.5.
4. SS_within = 5 + 5 + 2.75 = 12.75.
5. MS_between = 9.5 ÷ 2 = 4.75; MS_within = 12.75 ÷ 9 ≈ 1.4167.
6. F ≈ 3.3529, df 2 / 9, p ≈ 0.0816.
7. η² = 9.5 ÷ 22.25 ≈ 0.4270.

How to read the result

With p ≈ 0.082 > 0.05 the means are not significantly different at the 5% level; the 5% critical F for (2, 9) df is about 4.26. Yet η² ≈ 0.43 says group membership accounts for 43% of the observed variation, a large share; the sample is simply too small to be conclusive. A significant F only says that at least one mean differs; a post-hoc test such as Tukey's HSD is needed to find which.

Common mistakes

• Running many t-tests instead of ANOVA, which inflates the false-positive rate.
• Using ANOVA for repeated measurements on the same subjects; that needs repeated-measures ANOVA.
• Reporting η² as if it were unbiased; in small samples it overstates the effect (ω² is less biased).

Frequently asked questions

Can I use ANOVA with only two groups?

Yes. With two groups F equals the square of the pooled two-sample t statistic and gives the same p-value.

Do the groups need equal sizes?

No, but equal sizes make the test more robust to unequal variances.

What counts as a large η²?

Common rough benchmarks are 0.01 small, 0.06 medium and 0.14 large, but context matters more.

Sources: Fisher, R. A. Statistical Methods for Research Workers (1925). · Montgomery, D. C. Design and Analysis of Experiments (Wiley).