ANOVA Calculator
Enter data for 2 or more groups to test whether their means are statistically equal — get F-statistic, p-value, degrees of freedom, and a complete ANOVA table with between-group and within-group variance.
Enter your values above to see the results.
Tips & Notes
- ✓ANOVA only tells you that at least one group mean differs — it does not tell you which groups differ. Always follow a significant ANOVA with post-hoc tests (Tukey, Bonferroni) to identify specific differences.
- ✓Groups do not need equal sample sizes for ANOVA, but unequal sizes reduce power. Dramatically unequal sizes (1:10 ratio) can affect the reliability of variance estimates.
- ✓Check homogeneity of variances before ANOVA using Levene's test. If group variances differ substantially, use Welch's ANOVA instead — it does not assume equal variances.
- ✓ANOVA is robust to mild violations of normality when sample sizes are equal and large (n > 15 per group). For small samples with clear non-normality, use Kruskal-Wallis (non-parametric alternative).
- ✓Effect size for ANOVA is eta-squared (η²) = SS_between/SS_total. Small: 0.01, Medium: 0.06, Large: 0.14. A significant p-value alone does not tell you the practical importance of the difference.
Common Mistakes
- ✗Running multiple t-tests instead of ANOVA. With 3 groups and α=0.05, three t-tests give a 14.3% chance of at least one false positive. ANOVA controls the error rate at 5% across all comparisons simultaneously.
- ✗Stopping at ANOVA when p < 0.05 and concluding which groups differ. ANOVA only confirms that at least one mean differs — post-hoc tests are required to identify which specific group pairs are different.
- ✗Ignoring the assumption of homogeneity of variances. If one group's variance is 10× another's, the F-test is unreliable. Always check with Levene's test and use Welch's ANOVA if variances differ significantly.
- ✗Using ANOVA for a single comparison of two groups. For exactly two groups, a t-test is equivalent and simpler. ANOVA is designed for three or more groups — using it for two groups adds complexity without benefit.
- ✗Confusing statistical significance with practical significance. ANOVA with large samples (n=500 per group) can detect trivially small mean differences as statistically significant. Report effect size (η²) alongside p-value.
ANOVA Calculator Overview
ANOVA (Analysis of Variance) tests whether the means of three or more groups are equal — or whether at least one group mean differs significantly from the others. Running multiple t-tests to compare groups inflates the false positive rate; ANOVA avoids this by testing all groups simultaneously in a single F-test. ANOVA is the standard method for comparing group means in experiments, clinical trials, and research studies.
The core logic — between-group variance vs. within-group variance:
F = MS_between / MS_within = Variance explained by groups / Variance within groups
EX: Three teaching methods, student scores: Group A [70,75,80], Group B [85,90,95], Group C [60,65,70] → Group means differ substantially → large F → small p-value → reject equal meansANOVA table — step by step:
SS_between = Σnⱼ(x̄ⱼ − x̄)² | SS_within = ΣΣ(xᵢⱼ − x̄ⱼ)²
EX: SS_between = 3×(75−75)² + 3×(90−75)² + 3×(65−75)² = 0+675+300 = 975 | SS_within = (70−75)²+...= 150 | F = (975/2)/(150/6) = 487.5/25 = 19.5Complete ANOVA table structure:
| Source | SS | df | MS | F | p-value |
|---|---|---|---|---|---|
| Between groups | SS_b | k−1 | SS_b/(k−1) | MS_b/MS_w | From F-distribution |
| Within groups | SS_w | N−k | SS_w/(N−k) | — | — |
| Total | SS_tot | N−1 | — | — | — |
Frequently Asked Questions
ANOVA (Analysis of Variance) tests whether the means of three or more groups are statistically equal. Use it when: you have one categorical independent variable (group), one continuous outcome variable, and three or more groups to compare. Examples: comparing test scores across three teaching methods, blood pressure across four drug treatments, or sales across five regions. For exactly two groups, a t-test is the simpler equivalent.
Running multiple t-tests inflates the false positive rate. With 3 groups and α=0.05, you need 3 t-tests — the probability of at least one false positive rises to 1−(0.95)³ = 14.3%. With 5 groups (10 t-tests), the false positive rate reaches 40%. ANOVA tests all groups simultaneously in one F-test, keeping the error rate at the chosen α throughout.
F = MS_between / MS_within. It compares the variance explained by group differences (between-group) to the variance within groups (random noise). F=1 means group differences are no larger than random within-group variation — consistent with equal means. Large F (>>1) means groups explain far more variation than expected by chance — evidence that at least one mean differs. The p-value converts F into a probability.
Run post-hoc tests to identify which specific groups differ. Common options: Tukey's HSD (balanced designs, controls family-wise error), Bonferroni correction (conservative, any number of comparisons), or Scheffé (most conservative, any contrast). Do not simply declare 'groups differ' — stakeholders need to know which group pairs are different and by how much.
Three main assumptions: (1) Independence — observations are independent within and across groups. (2) Normality — values within each group follow approximately a normal distribution (less critical for n > 30 per group). (3) Homogeneity of variances — group variances are approximately equal (test with Levene's test). If variances differ significantly, use Welch's ANOVA. If data is clearly non-normal with small samples, use Kruskal-Wallis.
Report: F(df_between, df_within) = F-value, p = p-value, η² = effect size. Example: 'A one-way ANOVA revealed a significant effect of teaching method on test scores, F(2, 24) = 19.5, p < 0.001, η² = 0.62.' Then report which groups differed from post-hoc tests. Always include group means and standard deviations so readers can evaluate the practical size of the differences.