Flashing-yellow-arrow safety effects: a clarification from the research team

The following is a statement from David Hurwitz and Christopher Monsere, the researchers behind the OTREC project "Improved Pedestrian Safety at Signalized Intersections Operating the Flashing Yellow Arrow" and the resulting report:

The authors have welcomed the attention that this research has gained through multiple print and web-based sources. However, several of the media articles have misstated the context and findings of this research and incorrectly interpreted some of the findings of the simulator-based study of driver performance at protected-permissive left-turn traffic signal operations that have implemented the flashing yellow arrow (FYA) indication.

Some of these summaries have stated that our work suggests that the FYA indication display may have a negative impact on pedestrian safety. This summarization of our research is not correct, and has caused some concerns with agencies throughout the country who have correctly and successfully implemented the FYA indication in left-turn operations.

To be clear, our study started with the premise that the FYA is the best indication for communicating permissive left turns to drivers based on a long history of high-quality research. So, while our research simulated intersections operating the FYA, it is not the FYA display that results in the negative impact on pedestrian safety, but rather the permissive left-turn phase.   

It would be reasonable to expect similar negative interactions of permissive left turning drivers and pedestrians with the circular green indication or other display. Our research does not provide any support or finding for the type of permissive left-turn indication implemented.

In retrospect, we could have made it more clear that study assumed the FYA as the preferred permissive display to avoid this confusion. Oregon has been an early and successful adopter of the FYA display so there is little recent local debate about the display.

Additional Background

Our research was motivated by a local issue but has broader implications.  Washington County, Oregon replaced protected only left-turn phasing with protected-permissive left-turn phasing.  The addition of the permissive left-turn phase provides additional opportunity for left-turns to be made during the traffic signal phase, but unlike the protected left-turn phase, requires drivers to yield to both opposing traffic (opposing vehicles, pedestrians in the conflicting crosswalk, and opposing bicycles) before making the left turn.  

Historically, the circular green indication would have been used to provide the permissive left-turn message to the driver.  The flashing yellow arrow (FYA) indication has been shown to improve driver understanding and safety of permissive left-turns, and has been implemented in place of the circular green indication at many hundreds of intersection approaches locations in Washington County and elsewhere in Oregon.

Years of research has shown that drivers may also interpret the circular green indication as a protected left-turn indication, and assume left-turn right-of-way when in fact they are required to yield. The FYA has been used to replace the circular green indication because it has been shown to minimize, and in some cases eliminate, the left-turn right-of-way confusion, minimizing the safety impacts of adding a left-turn permissive phase. The addition of a permissive phase to intersection has negative safety implications but capacity improvements and must be carefully weighed on a site-specific basis.

In our review of the past research, we identified that the study of driver workload at permissive left-turns is lacking, especially with the more advanced research tools we currently have available. In addition, we found the issue of using a 4-section or 3-section display for FYA was unstudied.  With a limited research budget, we prudently elected to focus on these knowledge gaps and created a research scope to answer a specific and limited set of questions.  

Our research was conducted in an advanced driver simulator —it was not a field-based study. We recruited 27 subjects and had them complete a driving course where drivers made a total of approximately 620 individual left turns.  Our research metric was driver glance patterns, measured by a sophisticated optical tracking device.  Our protocol was designed to quantify differences in driver glance patterns with varying volumes of pedestrians (and their direction of travel), volumes of opposing vehicular through movements, or the presentation of a three or four section FYA display.   

Our finding can be summarized as:

•    As conflicting volumes of vehicles increased, drivers allocated more of the visual search task to identifying a gap in vehicular traffic.
•    As conflicting volumes of pedestrians increased, drivers allocate more of the visual search task to identifying a gap in pedestrians.
•    In approximately 4 to 7 percent of the scenarios observed in the experiment, drivers did not fixate on the pedestrians in the crosswalk.
•    There was no statistical difference in the glance patterns, as measured by average total fixation duration, in the presentation of the 4-section or 3-section FYA display.

In closing, based on our finding we have the following recommendations for pedestrians, drivers, and engineers:

•    Pedestrians – remain vigilant in the crosswalk as research shows it is entirely possible that permissive left turning vehicles are not actively searching for you in the crosswalk.
•    Drivers – a circular green or FYA permissive left turn indication requires you to search for safe gaps not only in the conflicting vehicular stream, but also the conflicting pedestrian and bicycle stream as well.
•    Traffic Engineers – consider all modes of transportation when determining which type of left turn phasing to select at a signalized intersection. It is entirely possible that the context of a particular location may not lend itself to concurrent movements of the permissive left and the conflicting crosswalk. There are control strategies that can be employed to mitigate these conflicts.

-- David Hurwitz and Christopher Monsere