The first day of school: The predictive validity of early school screening

Abstract

This study assessed the predictive validity of a multidomain early school screening procedure. Two cohorts of participants (n = 427, n = 352) were followed through the first three years of school. Outcome variables included retention, special education services, teacher nominations of behavior and emotional problems, and performance on standardized achievement tests. Discriminant function analyses were conducted with Cohort 1 and cross-validated in Cohort 2 using predictor variables for which significant group differences existed between "problem" and "nonproblem" groups. The rate of correct classifications for predicting problem outcome status ranged from 74% to 89%. Classification rates were similar in both cohorts. Fine-motor skills, general cognitive level, ethnicity, and maternal education were predictors of a variety of school problems. The results are discussed in relation to screening for school readiness and developmental constraints on the prediction of school outcomes.