Abstract

Previous studies have documented validity evidence of neuropsychological measures in the assessment of dementia. However, known cognitive test measures were mostly validated as separate measurements rather than asconstituents of a whole battery. In this study, the neuropsychological battery (CERAD-K) and neuropsychiatric measures were acquired in older adults with Mild Cognitive Impairment, dementia of Alzheimer’s Disease (AD), and Vascular dementia (VD). The assessment measures and demographic information were used to predict two validity criteria: dementia severity (CDR) and dementia type (AD or VD). A correlation between a single test measure and the target criteria indicated univariate validity, whereas relative impor- tance among multiple regression models indicated the multivariate validity of a single measure as a constituent of the battery. We identified that test measures including the Boston Naming Test, Trail Making Test, and Word List Recall were predictive of the clinical outcome criteria as univariate validity; however, this strength of association did not remain consistent when evalu- ated in terms of multivariate validity. Regarding the multivariate validity, measures including Word List recognition, and neu- ropsychiatric impairment showed robust validity. This contrasting validity indices between univariate and multivariate frame- works may be owing to shared information between other measures, which can distort the conclusions of validity evidence. The findings suggest that the validity of a neuropsychological test differs as a function of the target criteria and whether ad- ministered as a whole battery. The findings suggest that the validity of a neuropsychological measure differs as a function of the criteria of clinical context and whether tested under a comprehensive battery.