Homogeneity of Variance (Homoscedasticity)
You'll learn how to verify that all groups in your ANOVA have similar variances using Levene's and Bartlett's tests.
The Core Idea
ANOVA assumes that each group you're comparing has roughly the same amount of variability—like comparing three basketball teams where players' heights vary by about the same amount in each team. If one team has heights ranging from 5'8" to 6'2" but another ranges from 5'2" to 7'0", that unequal spread (variance) violates the assumption of homoscedasticity (equal variances).
When variances differ dramatically across groups, the F-statistic in ANOVA becomes unreliable, potentially leading to incorrect conclusions about whether group means differ.
Testing Homogeneity of Variance
Two statistical tests help you check this assumption:
Levene's Test
- More robust to departures from normality
- Preferred in most practical situations
- Tests whether the absolute deviations from group medians (or means) are equal
Bartlett's Test
- More powerful when data truly are normal
- Very sensitive to non-normality—can reject even when variance equality is fine
- Best reserved for cases where you're confident about normality
Both tests have:
- Null hypothesis (H₀): All group variances are equal
- Alternative (H₁): At least one group has different variance
If the p-value > 0.05, you fail to reject H₀ and can proceed with standard ANOVA. If p-value ≤ 0.05, variances differ significantly, and you may need Welch's ANOVA (a modified version that doesn't assume equal variances).
Key Takeaway: Always check homoscedasticity before trusting ANOVA results; use Levene's test as your go-to option, and consider Welch's ANOVA if variances differ substantially.