(ANOVA) a statistical procedure by which (a) total variance is separated into components attributable to defined sources (e.g., treatments, sites, families), and (b) mean square ratios are calculated and compared to determine the probability that differences among treatments or populations are too large to be due to chance —see generalized linear models This definition last updated 10/23/2008