Should I Stay or Should I Go? Uncovering the Factors of Red Light Runnings in a Field Study

Should I Stay or Should I Go? Uncovering the Factors of Red Light Runnings in a Field Study


L Carnis, University Paris-Est and IFSTTAR-DEST; R Dik, E Kemel, CETE de l'Ouest, FR


This field study identifies several factors involved in red light running behaviours at the individual level.The findings show that users seem to weight the expected benefits and risks of violations, but that context also matters.


Traffic light runnings are typical examples of risky driving behaviours and have been found to represent a substantial threat to traffic safety in urban contexts. They constitute a clear violation of road safety rules and may involve several types of users in asymmetrical situations such as car versus motorcycle, cycle or pedestrians. The so-called vulnerable users are placed in a more vulnerable situation in such situations.

A better identification of the factors driving red light running behaviour could help to design efficient road safety policies for reducing the number of cases and related road fatalities.

Most of the available literature on red light runnings relies on stated preference surveys or on aggregated field data. It is thus difficult to assess both the characteristics of the violators and those of the situation. The goal of this contribution is to identify several factors involved in running behaviours at the individual level and through field observations.

Using recently developed counting technologies, data have been collected for four days on three different sites equipped with traffic light in the Nantes area. Over 45,000 crossings were considered, including over 400 red light runnings. For each crossing, counters recorded precise information allowing to describe the case and its context. Among others, the data provided information about the type of vehicle involved, its speed and the time spent since the light turned red. Variables describing the context of the cases dealt with traffic conditions, durations of green and led light periods and whether or not other vehicles ran the red light in the same period.

The statistic treatment of this rich data set allowed to capture clear-cut and significant behavioural patterns. The findings suggest that users weight the expected benefits and risks when deciding to respect or not a red light. For instance, more than 90% of runnings happened during the first two seconds of the red light signal, that it to say when crash risk is minimal. Regarding the benefits, we observed that running proportions significantly increased with the duration of red lights, as if users accounted the waiting time that can be saved by running the traffic light.

These observations are consistent with the rational behaviour framework. Other trends, however suggest that context also impact these decisions. We observed that more than 20% of red light runners actually followed another car that crossed the orange light, suggesting the presence of interaction between users and imitation effects. Moreover, we observed that running behaviours vary significantly according to the sites and the type of user involved.

Our results show that numerous dimensions are at play in the decision to run a red light and that several directions should thus be explored to address these behaviours. If users weight the benefits and costs of their behaviours, they should be sensitive to changes in deterrence policy. However, the context and the user type have also been found to play a significant role for understanding red light running behaviour. They are a part of the complex picture describing these behaviours. For these aspects, significant improvements may be expected from behavioural based policies, focusing on perception, error committed by the driver, and user interaction and group identification.


Association for European Transport