Cluster analysis was used in the attempt to derive a meaningful taxonomy of academic competence in 117 adults with temporal lobe epilepsy. Reading, spelling and arithmetic achievement scores from the Wide Range Achievement Test-Revised were subjected to hierarchical complete linkage cluster analysis, and the stability of the obtained solution was examined by the application of two additional clustering procedures. A six cluster solution was obtained, and tests of internal validity showed membership in four of the clusters to be extremely stable (Moderate and Marked Reading/Spelling Underachievement, Above Average Achievement, Reading/Arithmetic Underachievement), while the remaining two clusters showed some exchange of cases (Arithmetic Underachievement, Average Achievement). Modest evidence of external validity was found through the comparison of specific clusters to the Average Achievement group on measures of neuropsychological function, neurological and demographic characteristics. It is concluded that cluster analysis may be helpful in developing taxonomies of psychosocial impairment in epilepsy, and may facilitate an understanding of the determinants of variable quality of life outcomes associated with epilepsy.
Read full abstract