Diagnostic Accuracy of Two Tests of Preschool Language

The Clinical Evaluation of Language Fundamentals-Preschool (CELF-P) and the Structured Photographic Expressive Language Test-Preschool (SPELT-P) were administered to 20 children developing normally and 20 children with specific language impairment in order to assess the tests' diagnostic accuracy. For both tests, the group with normal language scored near the mean of the normative group. The group with specific language impairment scored a standard deviation or more below the normal group mean for each test. Discriminant analyses resulted in statistically significant group differences for both tests. In addition, the discriminant analysis resulted in fair to good (80% or better) diagnostic accuracy in differentiating between the subjects with normal language and the subjects with language impairment. However, the cutoff scores that maximized diagnostic accuracy differed greatly between tests. The findings argue against the common clinical practice of applying one cutoff score across tests to identify an impairment. The data illustrate that the use of a predetermined cutoff score inevitably leads to poorer diagnostic accuracy than one that is empirically derived.