Reliability of traditional neurological sensory and motor tests

Tests of sensory and motor function are widely used by clinicians and researchers in neurology, psychiatry, and neuropsychology. Less than perfect performance on many of these tests may be considered pathognomonic of central nervous system dysfunction. Unfortunately, differences across practitioners in specific test selection, administration procedures, and scoring criteria have resulted in inconsistencies which confound attempts to study the incidence and patterns of deficits (Adams & Victor, 1993; Glick, 1993). Although as a group psychologists favor standardized, quantitative instruments, the field has lacked a comprehensive, standardized sensory and motor battery. That psychologists have not developed such a battery may reflect the notion that pathognomonic signs are ambiguous and unstable (Buchanan & Heinrichs, 1989) and the fact that some traditional test development procedures are often inappropriate.For the Dean-Woodcock Sensory and Motor Battery, Dean and Woodcock (1994) selected measures representative of those included in the traditional neurological examination.Measures of subcortical function, unavailable in the major neuropsychological batteries, were included to allow differentiation with right hemisphere impairment.Using standard procedures for administration and scoring, this study gathered preliminary data regarding the incidence of pathognomonic signs in a normal adult population, identified items with difficulty levels likely to result in overidentification of abnormality, and estimated the interrater agreement and interrater reliability for items and tests most vulnerable to subjective interpretation. Data analysis reflected consideration of Franzen's (1989) argument that reliability can be better understood through use of multiple estimation strategies and Cicchetti's argument that data needs to be considered at "finer levels of molecular analysis" (p.621). In addition to investigating interrater agreement, this study applied generalizability theory which allows for simultaneous estimation of the relative proportion of variance contributed by multiple sources and their interactions.Results suggested adequate to excellent rater agreement and reliability (i.e., generalizability). Also, with minor modification of specific items, generalizability of items may be expected to increase. Future studies should sample from a more heterogenous general population and specific clinical populations.Department of Educational LeadershipThesis (Ph. D.