Safety Effects of Flashing Yellow Arrows Used in Protected Permitted Phasing: Comparison of Full Bayes and Empirical Bayes Results. Issue 21 (December 2018)

• Record Type: Journal Article
• Authors: Appiah, Justice; King, F. Adam; Fontaine, Michael D.; Cottrell, Benjamin H.
• Is Part Of: Transportation research record. Volume 2672:Issue 21(2018)
• Page Start: 20
• Page End: 29
• Publication Date: 2018-12
• Subjects: Transportation -- Periodicals; Roads; Transport -- Périodiques; Routes -- Périodiques; Routes -- Conception et construction -- Périodiques; Roads; Transportation; 388.05
• DOI: 10.1177/0361198118794289
• Languages: English
• ISSNs: 0361-1981

Abstract

Using the flashing yellow arrow (FYA) signal indication for the permissive portion of protected-permissive left-turn (PPLT) phasing has become an increasingly popular treatment for left-turn signals as drivers are believed to understand the FYA better than the traditional circular green indication. A before-and-after safety evaluation of deploying FYA at PPLT signals at 28 intersections in Virginia was conducted. Each of the study intersections had FYA for the permitted portion of the phase on at least one left-turn approach. The focus was on left-turns that operated in the protected-permissive mode (with circular green indication for the permissive portion) before being converted to PPLT operations with the FYA indication for the permissive portion (PPLT-FYA). Crash records from before and after the activation of FYA were compared using both the full Bayes and empirical Bayes approaches. The results indicate that using the FYA signal indication instead of the circular green indication had a statistically significant effect in reducing overall frequency and severity of crashes. For the intersections studied in this research, total crashes reduced by 12% following conversion from PPLT to PPLT–FYA. The results also indicated that the full Bayes approach to safety effectiveness evaluation can, at a minimum, provide similar results to the well-established empirical Bayes approach. The 95% credible intervals for the expected crash reduction rates estimated with the full Bayes method were generally narrow, suggesting a good degree of confidence in the estimates.