Paper

The Structure and Meaning of a Computerized Neurocognitive Test Battery


Authors:
C. Thomas Gualtieri; Aaron S. Hervey
Abstract
Background: There exists a prevailing assumption that a neurocognitive test administered by a computer is equivalent to the same test administered conventionally by a human being. There is reason to question this assumption. Method & subjects: A computerized neurocognitive test (the CNT) was subjected to critical analysis. Study 1 addressed the appropriate standardization method and was conducted in 3420 normal subjects between the ages of 4 and 94. Study 2 examined the factor structure of the CNT in the same group. Study 3 examined the discriminant powers of the CNT in comparison with 3295 normal subjects compared to 4084 subjects with attention deficit hyperactivity disorder (ADHD), 694 patients with severe traumatic brain injuries and 90 patients with mild cognitive impairment (MCI) or early dementia. Study 4 was a comparison of the CNT with The Wechsler Adult Intelligence Scale (WAIS) in a clinical sample of 179 patients. Analysis & results: Study 1: Regression analysis indicated that age and years of education contributed significantly to a subject’s performance on the CNT but race and gender did not. Study 2: Exploratory and confirmatory factor analyses indicated that the seven tests of the CNT generated three factors: memory, processing speed and attention. Processing speed was the latent variable underlying performance on the test battery. Study 3: Neither stepwise discriminant function analysis nor logistic regression indicated a specific pattern to performance for the three clinical groups relative to each other. Study 4: Performance on the CNT was most highly correlated with the most highly g-loaded WAIS subtests. Two tests of the CNT - visual memory and shifting attention - were predictive of Full Scale Intelligence Quotient (FSIQ). Conclusion: These four studies raise questions about the continued use of computerized tests in clinical practice and research. The latent variable underlying test performance is processing speed. On the CNT, tests of executive function, memory and attention are, in the main, measuring the processing speed component of those functions. A critical examination of one CNT suggests that it may be doing less than it was designed to do.
Keywords
Computerized Test; Test Standardization; Processing Speed; General Mental Ability
StartPage
11
EndPage
21
Doi
Download | Back to Issue| Archive