Abstract
This paper is concerned with the synthesis and analysis of S-P sub-tables for learning diagnosis. Two methods are presented for partitioning a given S-P table into several S-P sub-tables so that they have highly reproducible Guttman scales. Using these S-P subtables, some applications for learning diagnosis and evaluation are discussed. The relations between a criterion in the methods and several coefficients of S-P sub-tables are also presented.
