Development and Evaluation of Driver Response Time Predictors Based upon Meta Analysis 2003-01-0885

The goal of this research was to develop mathematical equations that would estimate the response times of drivers in various situations. This research involved two studies.

The first study involved the development of a series of equations that predict driver response times [DRTs]. Compiling a database of over 130 studies that measured DRT and coding for over 20 methodology and substantive variables was the source from which the equations were developed. Multiple Stepwise Linear Regression analysis was performed on the database. The analysis produced an empirical equation that revealed which variables and methods were statistically significant predictors of DRTs. The analysis showed that when all research data was analyzed together an accurate predictor could not be developed. However, when the database was divided into smaller sets based upon where the target emerged, empirical equations were developed. Each of six equations reached statistical significance.

The second study compared the prediction results of the equations from the first study, to the results obtained from time/position analysis of the video record of actual traffic incidents at an intersection in Kentucky. The meaning, limitations and significance of the analysis results are discussed and compared to previously published prediction models.

The equations predicted response times within ½ of a second 74 percent of the time, and within ¾ of a second 90 percent of the time.