| Over the past several years, driving performance research has undergone significant changes. Many of the changes have been the result of advances in computer and engineering technology. High-fidelity, motion-based simulators are available for research that was never before considered possible. These technological innovations have changed the elements of driving performance assessment. When simulation studies were gaining popularity a generation ago, Leonard and Wierwille (1975) recognized that the evaluation of the psychometric properties of the simulator data was one of the greatest challenges to researchers utilizing such tools. The examination of measurement methods applied in a novel situation, such as a simulation environment, was the primary motivation of this research. This research examined various characteristics of data collected by the Iowa Driving Simulator (IDS) and applied internal consistency methods to determine the reliability of the driving performance of speed and steering control. Of particular interest was the effect of time on the magnitude of the reliability coefficient and the validity of psychometric methods for estimating how time affects reliability. Woodruff's ANOVA for estimating conditional standard errors of prediction and Thorndike's difference method for estimating conditional standard errors of measurement were considered to estimate the standard error of measurement of an individual's driving behavior. The extent to which these two methods for estimating a standard error of measurement for an individual accurately assessed that error when compared to the actual variability of individual measurements was examined. |
