Psychometric analysis of a computer-based measure of hidden maze learning

Date of Completion

January 2007


Psychology, Psychometrics




Neuropsychology's foundation lies in the ability to measure brain-behavior relationships in a psychometrically sound manner. Clinicians and researchers can only make meaningful interpretation of behavioral results if neuropsychological instruments provide reliable and valid data. Therefore, it is of critical importance that all measures used have the ability to provide data that are stable across time (i.e., reliability across multiple assessments) and that show validity in reference to meaningful theoretical constructs. ^ The Groton Maze Learning Test© (GMLT) is a computer-based measure of spatial learning, memory, and executive function currently undergoing this rigorous validation process. This test was initially developed for use in clinical trials with the anticipation that it would exhibit sound psychometric properties to allow for repeated administration, and yet, with the sensitivity to identify change following administration of various pharmacological challenges. Based on older tests of visual learning and memory, this modern instrument allows for greater precision of measurement because of its computer administration (e.g., response times recorded in milliseconds) and unique design features (e.g., pseudo-random selection of matched alternate forms). ^ The principal objective of the following set of studies was to evaluate several key psychometric properties of the GMLT. Two paradigms known to perturb cognition were employed to assess response during GMLT task completion: that of acute alcohol intoxication and 24-hour sustained wakefulness. This allowed for the estimation of reliability (test-retest, alternate forms, internal consistency) and validity (construct and ecological). Secondarily, these studies permitted a closer examination of cognitive changes across alcohol's dose response curve, as previous studies have shown mixed results due to limitations in instrumentation. The GMLTs ability to replicate or refute a postulated consequence of acute alcohol intoxication—the Mellanby Effect, or acute tolerance—was examined. Lastly, we included two additional measures during the alcohol intoxication study for further analyses: (1) a subjective rating of intoxication measure (visual analogue scale (VAS)), to exploit alcohol's strong expectancy in responding, and (2) an Internet-based visuomotor integration task to allow for measurement of procedural learning, ecological and construct validation, and additional examination of data analysis techniques used across the dose-response curve. ^