Definition

Many testing procedures such as the Analysis of variance for several Means, require that the Variances across the groups be equal, or homogenous. Bartlett's test can be used to evaluate this assumption, and therefore is called a test of the 'homogeneity of variances'.

The Bartlett test requires that the data be continuous and normally distributed and is not valid if the data are not normal. The Levene's and modified Levene's test (also called the Brown-Forsythe test) is another test of homogeneity that is more robust to violations of the normality assumption.

See Also

Homogeneity of Variances

Levene Test

External Links

Bartlett's test from NIST: - http://www.itl.nist.gov/div898/handbook/eda/section3/eda357.htm