ANOVA (One-Way)
Compare three or more group means: the F statistic, both degrees of freedom, an upper-tail p and eta squared.
What this page computes
Put one group per line with its values comma-separated. The page forms SS between = Σn_g(x̄_g − x̄)² and SS within = ΣΣ(x − x̄_g)², divides each by its degrees of freedom (k − 1 and N − k), and takes F as their ratio. The p-value is the upper tail of F(k − 1, N − k), computed through the incomplete beta function rather than read from a table, and the page states that it is the upper tail. Eta squared is SS between over SS total. Figures print to six significant figures.
A significant F says the group means are not all equal; it does not say which pair differs, and the page runs no post-hoc comparison. A group with fewer than two values is refused, because its within-group variance would be zero, and so is a set of groups where every value is identical.
The groups never leave the tab, so reloading and repasting gives the same F and p.
The page runs a one-way analysis of variance from grouped data: SS between = Σ n_g(x̄_g − x̄)², SS within = ΣΣ(x − x̄_g)², F = (SS_b/(k − 1))/(SS_w/(N − k)) and a p-value from the upper tail of F with those two degrees of freedom. Three groups of three give F = 13 on (2, 6) degrees of freedom and p = 0.0065918, with η² = 0.8125. Each group needs at least two values, because within-group variance would otherwise be undefined.
Compare the means of several groups with a one-way ANOVA. The p-value is the upper tail of F and both degrees of freedom are printed.
How to use it
Paste one group per line, with that group’s values comma-separated — this page reads groups, not a single column. It needs at least two groups, and every group needs at least two values; a group of one is refused by its line number. The table prints the sums of squares, both mean squares and η² beside the F and p, so the arithmetic can be followed.
The upper tail, the two degrees of freedom and η²
The p-value is the upper tail of F(k − 1, N − k) and both degrees of freedom are printed as a pair, because an F test has no second tail and the two numbers together identify the distribution. η² = SS_between/SS_total is the share of the total variation the grouping explains, which is the effect size the F statistic itself does not give. A significant F says the group means are not all equal; it does not say which pair differs, and the page does not run a post-hoc comparison.
What to Expect
Best for
- Comparing three or more group means in one test
What this tool keeps
- Nothing leaves this tab: the arithmetic runs in the page you are reading, and a reload discards both the input and the result.
Sample input
1, 2, 3 2, 3, 4 5, 6, 7
Sample output
F = 13, df = (2, 6), p = 0.0065918 (upper tail, H₁: not all group means are equal) SS between = 26, SS within = 6, SS total = 32 MS between = 13, MS within = 1 η² = 0.8125 (share of variance explained by the group) k = 3 groups, N = 9 values
Frequently asked questions
Which tail is the p-value from, and what are the degrees of freedom?
The upper tail of F with (k − 1, N − k) degrees of freedom, both printed beside the statistic. For three groups of three that is F = 13 on (2, 6) and p = 0.0065918.
How is the data entered?
One group per line with its values comma-separated. Every group needs at least two values, and a group of one is refused by its line number rather than being dropped.
Does a significant result say which group differs?
No. The test says the group means are not all equal. The page prints η² as the share of variation explained and does not run a pairwise post-hoc comparison.
Build custom tools for your store
Join the SDK program to create useful ecommerce utilities for your store.