Comparison of computerized and examiner-administered neurobehavioral testing techniques

To better understand the relationship of newly developed computerized neurobehavioral tests to their examiner-administered counterparts, performance on four pairs of tests was compared among 185 solvent-exposed workers in the paint manufacturing industry. Two approaches to the comparison of the test methods (computer or examiner) were employed: (1) the multitrait-multimethod correlation matrix and (2) linear regression modeling of the difference scores between pairs of tests (same functional domain, different test formats). The data revealed that the validity criteria of the multitrait-multimethod technique were met for the tests of simple visual reaction time and attention-concentration (digit symbol substitution). Linear regression analyses demonstrated that the digit symbol substitution and visual memory test pairs were differentially affected by age, vocabulary, solvent exposure, or manual dexterity. Overall, this latter approach holds promise as a means of understanding test format similarities and differences.