Which test is commonly used to assess equality of variances across groups?

• A. Shapiro-Wilk test
• B. Levene's test
• C. Durbin-Watson test
• D. Chi-square test

Answer: (B)

Equality of variances across groups means the spread of scores is similar from one group to another, an assumption often needed for ANOVA and similar tests. Levene's test is designed to check that assumption. It works by measuring how far each observation is from its group center (usually the mean, though using the median makes it more robust). You then perform an ANOVA on those absolute deviations across groups. If the variances are truly equal, those deviations should be about the same on average across groups, leading to a non-significant result. A significant result suggests that the group spreads differ, indicating heterogeneity of variances. Levene's test is favored in many cases because it remains reliable under non-normal data, whereas tests like Bartlett's are more sensitive to departures from normality. Other tests address different questions: normality, autocorrelation, or categorical frequencies, which is why they don’t assess equality of variances.