Selecting Components for a Screening Test to Identify Three-Year-Olds at Risk for Mild Learning Problems

Abstract

Thirty-four (Study 1) and 37 (Study 2) pairs of three-year-old children with, and without, mild learning problems, were matched on chronological age, gender, race/ethnicity and testing language. Children were presented a battery of five (Study 1) or six (Study 2) simple cognitive tasks. The children without mild learning problems performed at a significantly higher level than their matched peers with mild learning problems on 17 of the 18 measures. A subset of three measures in Study 1 and three measures in Study 2 were associated with classification accuracy levels for the children with learning problems of 91% (Study 1) and 84% (Study 2). Similarly, 85% (Study 1) and 86% (Study 2) of the children without learning problems were correctly classified. The measures were differentially effective in correctly identifying children classified as learning disabled (LD), educable mentally handicapped (EMH) or developmentally delayed (DD). Gender was not significantly related to performance, nor was testing language, although children who needed to be tested in their native language (Spanish) had lower mean scores. Race/ethnicity was significantly related to performance in Study 2 but not in Study 1.