Compare the means of three or more groups in one test. Paste each group's data, get the full ANOVA table, F-statistic, p-value, η² effect size, and per-group 95% confidence intervals — all client-side.
Paste your groups (one value per line in each box) and press Calculate to run one-way ANOVA.
A buyer receives the same component from three suppliers and wants to know whether their average measurement really differs — or whether the gaps are just normal sampling noise. Enter the lab to see how ANOVA settles it.
Enter the measurements for each group (supplier, machine, line, treatment) in its own box, one value per line. The groups can have different sample sizes — they do not need to be equal.
The calculator measures two things: how different the group averages are from each other, and how much everyday variation there is within each group. Those two numbers are the key to everything that follows.
It divides one number by the other. If the group differences are big compared to everyday noise, that ratio comes out high — suggesting the groups probably aren't all the same. The result is a p-value (real difference, or just luck?) and a strength-of-effect score (what percentage of the total variation comes from which group a reading belongs to). Turn on Tukey's test in Advanced options to find exactly which pairs of groups differ.
One-way ANOVA is the right tool when you need to compare the means of three or more groups at once. Instead of running many separate t-tests — which inflates the false-positive risk — ANOVA gives a single, controlled test of whether at least one group genuinely differs.
One-way ANOVA (analysis of variance) tests whether three or more independent groups share a common mean. It compares the spread of the group means around the grand mean (between-group variation) with the natural noise inside each group (within-group variation). If the between-group signal is large relative to the within-group noise, the F-statistic is large and you conclude at least one group differs.
ANOVA compares two kinds of variation. First it measures how far each group's average sits from the overall average — that is the 'between-groups' signal. Then it measures how much the readings bounce around inside each group — that is the 'within-group' noise. Dividing the signal by the noise gives the F-ratio: a large F means the gaps between groups are too big to be chance. The p-value turns that F into a clear yes/no, and the effect size shows what share of all the variation is explained by which group a reading came from.
A buyer tests three suppliers by measuring the yield from five batches each. Supplier A averages 82%, Supplier B averages 85%, and Supplier C averages 79%. Just looking at the numbers, B seems best — but are those differences real, or just normal batch-to-batch variation?
Running the ANOVA gives a p-value of 0.0002 — well below 0.05. That means yes, at least one supplier is genuinely different. This isn't noise.
What to do with this: Switch to Supplier B. You now have statistical evidence to back the business case, not just a hunch.
Comparing suppliers, machines, shifts, or treatment settings nearly always involves more than two groups. ANOVA gives a defensible, single-shot answer to "are these groups really different?" — replacing eyeballing the means with statistical evidence and an effect size that quantifies how much of the variation the grouping explains.
Use one-way ANOVA in the Analyse phase of DMAIC to confirm that a categorical factor (supplier, line, operator) drives a continuous response. If ANOVA is significant, follow up with a post-hoc test (such as Tukey) to identify exactly which groups differ before you act on a specific source.
ANOVA is a core hypothesis test in the Six Sigma toolkit and the foundation of designed experiments. It tells you whether a factor is a genuine source of variation, giving you both the significance (p-value) and the practical importance (η² and group means) to report at a DMAIC tollgate.
If ANOVA is significant, run a post-hoc test (Tukey's HSD) to pinpoint which specific groups differ, then act on the responsible source — switch supplier, retune a machine, or standardise the best method. Confirm the assumptions with a normality check and an F-test for equal variance. If the result is not significant, the groups are statistically indistinguishable; pool them or collect more data before drawing conclusions.
Combine one-way ANOVA with the F-test for equal variances, run charts, and a DMAIC project structure to find and act on sources of variation.
One-way ANOVA (analysis of variance) tests whether the means of three or more independent groups are all equal. It does this by comparing the variation between the group means with the variation within the groups. If the between-group variation is large relative to the within-group variation, the F-statistic is large and at least one group mean is judged to differ. ANOVA gives a single overall test rather than many separate comparisons.
Running many t-tests inflates the false-positive (Type I) error rate. With three groups you need three t-tests, with four groups six, and each carries its own 5% risk, so the chance of a false significant result climbs quickly. ANOVA controls this by testing all groups together in one analysis at your chosen α. If ANOVA is significant, you then use a post-hoc test (such as Tukey) to find which specific pairs differ.
The F-statistic is the ratio of the mean square between groups (MSB) to the mean square within groups (MSW): F = MSB/MSW. MSB measures how far the group means spread around the grand mean; MSW measures the average variability inside the groups. When the groups truly have the same mean, F is around 1. A large F means the group means differ by more than chance alone would explain, and the p-value from the F-distribution tells you how unlikely that result is.
One-way ANOVA assumes: (1) observations are independent within and between groups; (2) the data in each group are approximately normally distributed — check with a normality test, though ANOVA is fairly robust for larger samples; (3) the groups have roughly equal variances (homogeneity of variance) — check with an F-test or Levene's test. If variances are very unequal or the data are heavily non-normal, use Welch's ANOVA or the nonparametric Kruskal-Wallis test instead.
Tukey's Honestly Significant Difference (HSD) test compares every pair of group means after a significant ANOVA while holding the overall (family-wise) false-positive rate at your chosen α. For each pair it builds a confidence interval for the difference using the studentized range distribution: the margin is q × √(MSW/2 × (1/nᵢ + 1/nⱼ)), where q is the studentized-range critical value for k groups and the within-groups degrees of freedom. Any pair whose interval excludes zero differs significantly. Enable it under Advanced options to see the full pairwise table.
A significant ANOVA tells you at least one group differs. Structured training shows you how to pinpoint the source and act on it.
More calculators for statistics, quality, and improvement.
Want to compare groups and find sources of variation inside a full DMAIC project? The Black Belt course covers advanced statistics, hypothesis testing, and the full analytical toolkit.
View Black Belt →