Observing the observation of (vulnerable) road user behavior and safety: a scoping review into current practices

Abstract

The main strength of behavioral observation studies is that naturalistic data is collected without the road users’ knowledge that they are being observed for research purposes. It enables the observation and identification of behavioral and situational processes that contribute to unsafe traffic events, while minimizing the influence of behavioral adaptation effects. Such studies have already been reported since the 1930s, but an overview of the current extent, range and nature of this type of research is lacking. Therefore, a scoping review evaluating all peer-reviewed journal articles published in English was conducted in order to a) investigate their purposes, b) identify common topics and behavioral indicators, c) examine study characteristics and d) prevent duplicate research efforts. In total, 600 journal articles found in three major databases were included in this review. The majority of the studies collected data of car drivers (81%), while pedestrians, cyclists and motorcyclists were found in 23%, 9% and 5% of the studies respectively. The main goal of behavioral observation is to observe what happens at selected locations (51%), followed by the evaluation of a specific safety improving treatment (38%) and the development of behavioral models (10%). Most common research topics were crossing, yielding and red-light running behavior for pedestrians and cyclists and yielding and speeding for car drivers. Speed and red-light-running were two of the indicators that have been measured most. With regard to study characteristics, four main research designs were identified: single observation, before-after studies, with-without and crosssectional setups. Some difficulties in determining truck-involvement, characteristics of the observation period and sample sizes were encountered. Finally, because the use of video cameras to capture behavioral observations has become the major data collection technique in recent years, the current efforts to improve and further develop automated video-analysis software tools can prove to be a valuable asset in behavioral observation studies and traffic safety evaluation in general.