Examining how different measurement approaches impact safety outcomes in child pedestrian research: Implications for research and prevention

Abstract

There has been a great deal of research aimed at understanding the causes of child pedestrian injury. Many different methods have been employed with the goal of designing simulations that produce rigorous assessment of children’s behaviors without putting children at risk of actual pedestrian injury. Most research has assessed children’s pre-crossing decision making and extrapolated crossing outcome measures from estimates of mean walking speed. This study explores the nature and extent of measurement bias that is introduced when average walking speed is used to produce estimates of outcomes versus measuring actual in-road behavior directly. Using a within-subjects design and a fully immersive virtual reality pedestrian simulator, both measures were taken. Comparisons based on regression models revealed the extent of differences in results produced by measurement bias. Results indicated that measurement bias is produced when average walking speed is used such that hits and high risk crossings are overestimated and missed opportunities are underestimated, resulting in an overall overestimate of children’s risk for pedestrian injury. The discussion highlights how these two measurement approaches emphasize different underlying processes as determinants of child pedestrian injury risk.