Accounting for drivers' bicycling frequency and familiarity with bicycle infrastructure treatments when evaluating safety.

Municipalities in the United States often encourage bicycling for the health, economic, and environmental benefits by implementing new and innovative bicycle infrastructure treatments. Unfortunately, many treatments are unfamiliar to or misunderstood by drivers, especially when lacking explicit rules (e.g., shared lanes). To date, research has largely investigated bicycle infrastructure from a bicyclist's perspective, but with little research from the driver's perspective. The objective of this research is to utilize a driving simulator to investigate driver behavior towards different bicycle infrastructure treatments when driver behavior is not provoked by an interaction with bicyclists. More specifically, this research intends to investigate the impact of bicycling frequency and treatment familiarity, as well as the combined effect of the two, on driver behavior at each treatment type. The treatments investigated are shared lane markings called "sharrows", standard bike lanes, bike boxes, and merge lanes. The results show that bicycling frequency significantly affects the proportion of drivers making eye glances at treatments. In addition, drivers more familiar with bike boxes stopped significantly further back from bike boxes, and drivers more familiar with merge lanes performed the merge maneuver significantly earlier. Furthermore, driver speed and lane positioning at bike lanes was significantly affected by the combination of bike lane familiarity and bicycling frequency, but not individually. This research is a first step towards understanding driver behavior and expectation of bicyclists; an essential understanding for infrastructure treatments that do not provide physical barriers between bicycles and automobiles, and instead rely on driver behavior for safety.

A mixed methods investigation of bicycle exposure in crash rates.

Crash rates are an essential tool enabling researchers and practitioners to assess whether a location is truly more dangerous, or simply serves a higher volume of vehicles. Unfortunately, this simple crash rate is far more difficult to calculate for bicycles due to data challenges and the fact that they are uniquely exposed to both bicycle and automobile volumes on shared roadways. Bicycle count data, though increasingly more available, still represents a fraction of the available count data for automobiles. Further compounding on this, bicycle demand estimation methods often require more data than automobiles to account for the high variability that bicycle demand is subject to. This paper uses a combination of mixed methods to overcome these challenges and to perform an investigation of crash rates and exposure to different traffic volumes.