Methodological issues and learning disabilities diagnosis in clinical populations

Previous research suggests that the diagnosis of a comorbid learning disability is dependent on the method used for making the LD diagnosis. This study investigated that proposition by studying the effects of using three approaches to the assessment of learning disabilities in a sample of 177 six- to thirteen-year-old boys referred to outpatient mental health clinics for behavior problems. The use of these three procedures to diagnose comorbid learning problems produced significantly different results. All methods identified significant numbers of children in the clinical population as learning disabled; however, each method identified children with differing characteristics. Consistent with predictions from measurement theory, the commonly used simple standard score discrepancy method was more likely to identify children with above-average IQs as learning disabled, whereas a regression approach identified learning disabilities more consistently across the ability range. These results were interpreted as supporting the use of regression approaches to diagnose co-occurring learning disabilities, as that method is less likely to be biased by the child's intelligence test score. The implications of the use of each method in research investigations is also discussed.