*WINNER* Evaluating the Statistical Significance of Pedestrian Crossings on Superstreets using D-values instead of p-values

Abstract

"Restricted Crossing U-turn Intersections, RCUTs, also known as Superstreets, are an innovative interchange system that eliminate the need for left turn and through movements from the minor streets of a roadway by turning them into U-turns instead. A study by Hummer (2014) looked to integrate pedestrian crossings into Superstreets. The study considered four pedestrian crossings, and within these crossings, different signal offsets, cycle lengths and splits, and length to U-turn were considered. A total of sixteen scenarios were considered, and the “best” pedestrian crossing was chosen based on the p-value found when comparing three mean measurements of efficiency for each scenario.
The current practice for measuring statistical significance between means is to calculate a p-value. Recently, there has been a lot of research into the misuse of the p-value. The questions regarding the p-value stem from the fact that the p-value is often driven by the sample size of the study, and it is well known that large sample sizes result in p-values that suggest statistical significance. A recent study by Demidenko (2016) developed a D-value of measuring statistical significance that eliminates population size as one of the variables considered within the model. This D-value allows for evaluation of significance on an individual level rather than a group one. This paper looks to reevaluate the level of significance from the Superstreet project by considering the D-value of each of the crossing types instead of the p-value."