Validation of a Battery of Preschool Screening Tests for Predicting Special Education Placement

Abstract

The accuracy and bias of a kindergarten screening battery for predicting educational placement was examined. The battery contained four well-validated commercial instruments. Subjects were 519 White and 183 American Indian children who had been tested prior to entering kindergarten. The children were monitored and 98 of the original 702 (14%) were eventually placed in special education resource classes. The battery predicted educational placement equally for both groups. However, the hit rate for both groups combined was only slightly over chance. While the battery had good accuracy for identifying normals (97%) it correctly identified only 39% of the special education students. A predictive value statistic was used to determine that the battery would have flagged 59 children as at-risk, 38 of which would have qualified for special education (or 64% of those identified). Inasmuch as the overall hit rate masks the low rate of identification of true positives, sensitivity and predictive value are argued as being more meaningful in evaluating the accuracy of a screening battery.