Using educational placement in third grade to select and validate a preschool screening measure

The early detection of children with mild learning problems remains a problem. New screening tests are being developed that will be more effective in identifying children who fall into this "difficult to identify" group. This study examined the predictive validity of one potential screening test. The scores of 268 children on each task of a nine-task preschool cognitive battery were evaluated in terms of the accuracy with which they predicted the classification of those same children into regular education (n = 254) or special education (n = 14) 4 years later when the children were in third grade. A classification accuracy level of 79% was achieved for the exceptional children, and a classification level of 70% for the normally achieving children. Racial/ethnic status, testing language, and socioeconomic status all impacted the children's screening performance. Lower screening scores and a greater proportion of the sample with scores below the cut were associated with minority status, not being tested in English, and lower socioeconomic status. Preschool testing in Spanish or both English and Spanish was associated with poorer achievement 4 years later. Many of the children not in special education, but with a screening score below 13, had serious academic problems in third grade that were reflected in their low percentile scores on one or more subtests of the Stanford Achievement Test.