The Logic of Sample Selection: Who Represents What?

Abstract

This article focuses on sampling problems related to the use of system-identified subjects in special education research. Differences in school district criteria and procedures for selection of pupils into special education programs for the mildly handicapped have resulted in limited generalizability of research findings across studies. Overlap between samples identified as LD and EMR is particularly noticeable. Suggested approaches to deal with these sampling problems are identified. These include the use of rationally defined subgroups, the statistical procedure of cluster analysis, and marker variables.