Direct comparison
t-test vs ANOVA
A t-test compares the means of two groups, while ANOVA (analysis of variance) compares the means of three or more groups simultaneously, controlling the overall Type I error rate that repeated t-tests would inflate.
The step most authors miss
Doing CRediT right? Don’t stop at the statement.
A CRediT statement credits you inside one paper. The recognition CRediT was built for happens when those roles are tied to you, persistently. Sign in with your ORCID — free — and claim your CRediT contributions on casrai.org, the home of the standard. They become a verified, portable part of your identity, not a line that disappears into one PDF.
Free: claim your contributions, then export a journal-ready CRediT statement, schema.org structured data, JATS XML, CSV or BibTeX — and preview your public profile. A membership publishes that profile publicly and verifies the journals you serve.
Side-by-side comparison
| Dimension | t-test | ANOVA |
|---|---|---|
| Number of groups | Compares two means: two independent groups, one group against a reference value, or two paired conditions. | Compares three or more group means simultaneously in a single test (one-way ANOVA and its extensions). |
| Test statistic | Produces a t-statistic, referred to Student’s t-distribution with its degrees of freedom. | Produces an F-statistic, the ratio of between-group to within-group variance, referred to the F-distribution. |
| What it tells you | Whether the two means differ significantly, and (with the sign of t) in which direction. | Whether at least one group mean differs from the others — an omnibus result that does not, by itself, name which. |
| Error-rate control | Controls Type I error at α for a single two-group comparison; running many t-tests inflates the family-wise α. | Holds the family-wise Type I error at α across all groups using one omnibus test, avoiding that inflation. |
| Post-hoc tests | Not needed — with only two groups a significant result already identifies the difference. | Needed after a significant omnibus F to locate which pairs of groups differ (e.g. Tukey HSD, Bonferroni-adjusted comparisons). |
| Relationship | A two-group special case of ANOVA: for two groups the two tests are equivalent and F = t². | The general method that subsumes the t-test; with exactly two groups it reduces to the (squared) t-test. |
| Assumptions | Parametric: approximately normal data and (for the standard form) homogeneity of variance; independent observations. | Parametric: approximate normality of residuals, homogeneity of variance across groups, and independent observations. |
| When to use | Use when the design has exactly two means to compare. | Use when comparing three or more groups, or with factorial designs involving more than one categorical factor. |
| Example | Comparing mean recovery time between a treatment group and a placebo group. | Comparing mean recovery time across three doses — low, medium and high — in one analysis. |
Common questions
FAQ
Why not just run multiple t-tests instead of an ANOVA?+
Each t-test carries its own chance of a false positive, so running several across the same groups inflates the family-wise Type I error rate well above the nominal α. ANOVA tests all groups in one omnibus F-test that holds the overall error rate at α, which is why it is preferred for three or more groups.
Is a t-test just a special case of ANOVA?+
Yes. For a comparison of exactly two groups, a one-way ANOVA and an independent t-test give the same p-value, and the statistics are related exactly by F = t². ANOVA generalises the same logic to three or more groups.
My ANOVA is significant — which groups actually differ?+
A significant ANOVA is an omnibus result: it tells you that at least one mean differs, but not which. To locate the specific differences you run post-hoc pairwise tests, such as Tukey’s HSD or Bonferroni-adjusted comparisons, which keep the error rate controlled across the multiple comparisons.
Going deeper
