, Volume 8, Issue 2, pp 121-132

A partitioning technique for defining instructional groups

Abstract

A technique is presented for partitioning N students into K groups of fixed sizes using a given measure of proximity for all student pairs. The measure of proximity is typically calculated from a set of variables, such as completed curriculum units or learning style, and constitutes the data needed for a criterion of partition “fit”. This latter index is explicitly defined by the sum of within-group proximities and when used in conjunction with the optimization procedure discussed, homogeneous groups can be obtained that satisfy externally imposed size requirements. Finally, a simple generalization is suggested for the related task of grouping students to meet upper limit size constraints only.