[¦meth·əd əv ′sem·ē‚av·rij·əz] (statistics) A method for providing a quick estimate of a linear regression line, in which data are divided into two equal sets and the means of the two sets or two other points representative of each set are determined and a straight line drawn through them.