Detects differential variability for individual CpG sites in methylation data. DiffVar employs an empirical Bayes model framework that can take into account any experimental design and is robust to outliers. DiffVar’s approach is inspired by Levene’s z-test. It is a simple and computationally efficient test that is robust against non-normality and outliers. A major advantage of this method is that it is suitable for any experimental design; it is not limited to a two-group scenario. DiffVar can also be applied to DNA methylation sequencing data or any set of β or M values from CpG sites or regions. It will transform a matrix of β values into M values by applying a logit transformation. In general, the DiffVar method for testing differential variability can be applied to any omics data that use the limma pipeline for analysis. DiffVar is available as a function in the missMethyl Bioconductor R package, and depends on the limma framework.