Levene's test checks one of the assumptions of the t-test and the ANOVA: that the groups have similar variances (homogeneity of variance). It is reported briefly, usually in one clause before the main test, but it decides which version of the main test you report. This guide gives you the APA 7 format for both outcomes.
What you need before you write a single word
- Levene's F statistic.
- Its two degrees of freedom: number of groups minus 1, and total N minus the number of groups.
- The exact p value.
- Which version of Levene's test your software ran (based on means or on medians, the Brown-Forsythe variant), if it gives a choice.
The APA 7 format template
When the assumption holds:
Levene's test indicated that the assumption of equal variances was met, F(df1, df2) = X.XX, p = .XXX.
All 53 templates, ready to paste with the italics set: the Rescue Vault, $27 →
When it is violated:
Levene's test indicated unequal variances, F(df1, df2) = X.XX, p = .XXX, so [Welch's t-test / Welch's ANOVA] was used.
Formatting rules reviewers actually check:
- Italicize F and p; give both degrees of freedom.
- Keep the leading zero on F (it can exceed 1): F = 0.13, not .13.
- Say what the result means for the analysis, not only the numbers.
Worked examples
Equal variances. Comparing weight loss across three diets (n = 25 each):
Levene's test indicated that the assumption of equal variances was met, F(2, 72) = 0.13, p = .876, so a one-way ANOVA with Tukey's HSD post hoc tests was used.
Unequal variances. Comparing exam scores of 28 online and 40 in-person students:
Levene's test indicated unequal variances, F(1, 66) = 20.53, p < .001, so Welch's t-test was used.
The second example is written up in full in how to report a Welch's t-test.
Mistakes reviewers catch
- Reading the result backwards. A significant Levene's test (p < .05) means the variances are not equal.
- Reporting the test and then ignoring it. If it is significant, the main analysis must change (Welch's test, or Games-Howell post hoc tests).
- A leading zero dropped from F. F can exceed 1, so write 0.13.
- Leaning on it in very large samples. With hundreds of cases it flags trivial differences; look at the ratio of the largest to the smallest variance too.
Let KyroStat do the write-up for you
KyroStat runs Levene's test before every comparison, switches to Welch's version when the variances differ, and writes both sentences from your data.
Frequently asked questions
Where does Levene's test go in the Results? Right before the main test, or in a short assumptions paragraph. See how to check statistical test assumptions.
Mean-based or median-based Levene's test? The median-based version (Brown-Forsythe) is more robust to non-normal data. Say which one you used.
What if Levene's test is significant in an ANOVA? Report Welch's ANOVA with Games-Howell post hoc tests.
Do I need Levene's test for a paired t-test? No. A paired test works on the differences within each pair, so there are no separate group variances to compare.