My thesis completes a visualization and bivariate, multivariable statistical analysis of 1.5 million trips being made using Washington D.C.’s Capital Bikeshare system. These trips span from April 1 to September 30, 2013. My thesis explores the relationship between bikeshare trips, separated cycling lanes and the Frequent Transit Network. I have explored other aspects of my thesis findings such as the evidence of trip behaviors for road pricing here, the relationship between trips and transit here, the effects of bikeshare pricing structures here.
There is a wealth of research that shows separated cycling lanes are a critical ingredient for getting the 2/3 of the population that are “interested but concerned” in cycling. According to a study by Kay Teschke, and Meghan Winters one of the main deterrents to getting more people to cycle is safety and the perception of safety. Physically separating two-tonne vehicles from people on a 30-pound bike is one key aspect for creating a safe and comfortable riding area.
While there is plenty of research showing this connection between separated cycling lanes and increased rates of general cycling, most of this research relies on surveys and does not track actual trips. This is one of the benefits of bikeshare systems, due to modern technology the origin and destination of each trip can be tracked and quantified. While there are still limitations with not knowing the exact paths taken between bikeshare stations, using bikeshare trip data of actual trips adds another piece of the puzzle and can compliment survey studies.
While there could be hundreds of variables influencing the rates of cycling, this study focuses on controlling for the strongest known variables influencing the rates of cycling based on a literature review. This results in 14 of the known strongest socio-economic demographic and built environment variables affecting the rates of cycling.
When you take these variables into account along with the separated cycling infrastructure in a statistical multivariable linear regression models. These models take into account the infrastructure, land use, and demographic context around each bikeshare station in order to determine if there is a statistical relationship with certain characteristics surrounding the stations. This study also created these multivariable linear regression models based on different characteristics of the trips, such as the starting and ending time of the day, weekend vs weekday, the trip duration and the locations.
Separated Bike Lanes
The results from this study show a clear relationship between having a high volume of separated bike lanes within walking distance from bikeshare stations. This results held true for all trips overall, but also trips that were longer than 10 minutes, and trips taken in the peak time.
The membership type did not make a difference, 24 hour, 3 day, monthly, and annual pass holders all had a positive statistically significant relationship with separated cycling lanes. I mentioned this in an earlier post that the “casual members” (24 hour or 3 day passes) present the highest revenue generation potential under the Capital Bikeshare pricing structure. Therefore by extending the potential for safe, separated cycling you can maximize revenues from tourists using the bikeshare system.
The results are pretty conclusive, people are using bikeshare stations that are in close proximity to separated cycling infrastructure.
The following tables below summarize my research statistical findings. Green indicates a statistically significant relationship, red indicates a negative, while no color indicates no statistical relationship.