Evaluating the diagnostic utility of attention-deficit/hyperactivity measures using discriminant function analysis
Abstract
Discriminant function analysis was used to evaluate the diagnostic utility of five measures commonly used in ADHD assessment: the Gordon Diagnostic System, the Restricted Academic Observation, the Tower of London, the Child Behavior Checklist, and the Teacher Report Form. The 109 children with ADHD performed more poorly than 68 children with other clinical disorders (Learning Disabilities, psychiatric disorders, and Disruptive Behavior Disorders other than ADHD) on 4 out of 5 measures. Various combinations of these variables were entered into the discriminant function analysis, with the function comprised of the Tower of London, Restricted Academic Observation Off-Task and Fidgeting, and Gordon Diagnostic System Commissions resulting in the best classification (87.2% of the participants with ADHD and 44.8% of the clinical comparison group). Positive predictive power for the measures was satisfactory, while negative predictive power was quite poor. Sensitivity was particularly problematic, and specificity was complicated by the dimensional nature of ADHD symptoms. An optimal diagnostic battery is suggested and the limitations of ADHD instrumentation are discussed.
This paper has been withdrawn.