(1) A statistical procedure that allows subdivision of the total variation among experimental units into known sources of variation and provides a measure for each source. (2) The statistical analysis that tests the significance of variable sources.