The use of clinical neuropsychological data in learning disability research

Abstract

This study addressed the issues of using clinical archival data for research with a sample of children with learning disabilities who were referred for neuropsychological evaluation. Clinical samples are not representative samples, and this must be acknowledged when conducting research. One bias is the increased rate of comorbidity among clinical samples, and they can be used to study the mechanisms of comorbidity. When testing the hypothesis of attention deficits in comorbid reading disability (RD) and attention deficit hyperactivity disorder (ADHD) being secondary to RD (phenocopy hypothesis), it was found that comorbid RD/ ADHD group showed deficits in abilities related to both RD and ADHD. The results were not supportive for the phenocopy hypothesis. One frequent problem in clinical datasets is missing data. The study tested several data imputation techniques, their accuracy in predicting missing values, and their subsequent effects on the parameter estimates. Clinical neuropsychological datasets consist of measures covering all domains of neuropsychological functioning and are especially suitable for investigating the properties of neuropsychological measures. Rapid Serial Naming (RSN) tasks require to speedily name serially presented familiar visual stimuli. RSN tasks comprising different stimuli were found to measure partly different skills. RSN performance was related to phonological skills, motor dexterity, verbal fluency, and processing speed. The Trail Making Test-Part B (TMT-B), requires to speedily follow and alternate between numerical and alphabetical sequence. Children with RD were slower than controls on TMT-B. These differences were accounted for by differences in following the alphabetical sequence. Clinical assessment and research differ in several ways, resulting in problems and limitations, but also possibilities, to the application of clinically collected data in research.