An investigation of cyclist behavior at intersections

Abstract

The designing of bicycle lanes can be a complicated issue. Some bicycle lane treatments are meant to encourage cyclists to merge with traffic before the intersection, while some treatments channelize cyclists to merge with traffic in a predetermined location. Two intersections (Bizzell/George Bush Drive and Throckmorton/George Bush Drive) at the Texas A&M University were chosen to observe cyclist behavior. Both intersection approaches contained an island separating right turning traffic from straight and left turning traffic. However, the right turn lane at Bizzell begins at the island, while at Throckmorton the right turn lane is present throughout the approach to the intersection. Video cameras were used to record cyclists as they approached and traveled into the intersections. The distance from the intersection where a cyclist merged with motor vehicle traffic, the cyclist maneuvers as they entered the intersection, and the cyclist traffic control compliance was recorded. The motor vehicle volumes were also recorded to determine if they influenced cyclists' decisions to merge with traffic. Using the Chi Square test of independence, the collected behaviors were evaluated to determine if the behaviors were dependent upon the intersection location. Linear regression was used to determine the influence of motor vehicle volumes on cyclists and the test of proportions was used to decide if the proportions of behaviors were equal at both intersections. It was found that the intersection approach geometry, striping and lane configuration does influence cyclists to merge with traffic at different locations and motor vehicle volumes did not influence cyclists' decisions to merge with traffic. The proportions of inappropriate behaviors were different between the locations, while the traffic compliance rates were the same.