Assessing the impacts of bicycle infrastructure on the willingness to commute via bicycle

Abstract

Past research has been conducted into factors that guide the travel mode choice of individuals through surveys, usership data, simulations, etc. Studies which focus on mode choice factors have often used surveys. Due to these survey-based methods, there are many sources of potential bias that may inflate the importance of factors and/or the willingness of respondents to change existing behaviors. The following study aims to reduce potential bias by using publicly available Public Use Microdata Samples (PUMS) data as well as a variety of spatial/temporal data. PUMS data is used to determine the effects of demographic factors on the probability of biking, while spatial/temporal data is used to determine the effects of population density, employment density, median property value, median housing cost, median income, crime rate, and bike accessibility. The bike accessibility measure is based on 20-minute isochrones with impedance correlating to the proportion of the population willing to use a particular type of infrastructure. The data used within the study represents 17 unique Public Use Microdata Areas (PUMAs) within Chicago, Illinois over the course of eight years (2012-2019). This study produces a logistic regression model that estimates the probability of choosing to bicycle to work given a variety of demographic and environmental factors. Based on the model’s output, it was shown that one-point bikeability index increase correlates with a 0.33-percent increase in the likelihood of an individual choosing to bike. Men were to be 2.61 times more likely to bike to work than women. Contrary to prior research, black/African American and Hispanic persons were found to be less likely to bike to work than the rest of the population, possibly due to local variation in the effects of demographics or unequal distribution of infrastructure. Age had a negative correlation with likelihood of biking to work and education had a positive correlation.