Subtypes of reading disability.

Subtypes of reading disability were studied through the application of cluster analysis. A battery of 23 measures of reading, language, auditory and visual processing, memory, perceptual organization, and visual-motor coordination was administered to 65 reading-disabled children. The primary clustering technique was the K-means iterative partitioning method which yielded three clusters that were subsequently found to be stable and reproducible. The clusters were characterized by the following: (1) a visual processing deficit; (2) a generalized language disorder; and (3) a minimal deficit subtype. These subgroups were consistent with findings from previous investigations of subtypes of reading disability. The quality of the clusters was assessed through a series of evaluation techniques developed specifically for this purpose and through a cluster analysis of randomly generated data drawn from a normal distribution with the same covariance matrix as the actual data. These analyses indicated that the clusters were relatively heterogeneous, and thus this typology may have limited clinical utility. It is possible that future investigations with more severely disabled readers may produce more homogeneous clusters.