Neuropsychological assessment of relatively intact individuals: pychometric lessons from an HIV+ sample

The assessment of relatively intact individuals poses special problems for neuropsychologists, because fine discriminations are often needed to identify subtle deficits or gradual declines in performance. Many neuropsychological tests, however, are not well-suited for making such exact discriminations. 117 HIV+ individuals were administered 26 different neuropsychological tests that produce 48 scores. Measures of skewness and kurtosis were used to identify nonnormal sampling distributions. While many tests showed good sampling distributions, several demonstrated ceiling effects and other restrictions of range. This included some tests, such as Boston Naming and Lafayette Grooved Pegboard, that are not ordinarily considered screening instruments. Such nonnormal distributions distort the interpretation of clinical and research data, and indicate a need to use tests that are suited to the abilities of the population being assessed.