Levene

Levene's test is a statistical procedure for assessing the equality of variances across multiple groups. It is named after a statistician who proposed the method and is widely used as a more robust alternative to Bartlett's test when data deviate from normality. The test is commonly applied as a preparatory step before analyses that assume homogeneity of variances, such as ANOVA.

The standard approach computes the absolute deviations of observations from their group centers (usually the group mean) and then performs a one-way ANOVA on these deviations. The resulting F statistic tests the null hypothesis that all group variances are equal. Because it relies on deviations rather than raw values, Levene's test is relatively insensitive to non-normal distributions.

Variants of Levene's test exist to improve robustness. The Brown-Forsythe modification uses deviations from the group median, which can better handle heavy-tailed or skewed data. In practice, software implementations often provide these options and may default to the mean-based version, with the median-based variant available as an alternative.

Beyond statistics, Levene is also a surname of English origin. Individuals bearing the name appear across various professions, including academia and industry. The term Levene, in its statistical sense, remains primarily associated with this variance-homogeneity test and its widely used derivatives in data analysis.