Estimating Bicycling Demand

Abstract

Simple and reliable tools for estimating and predicting the amount of bicycling in an area would be useful for a variety of investment and policy decisions. Previous efforts to develop such tools have typically tried to develop demand estimates from basic descriptors of the population, land use, and bicycling facilities of an area. This paper takes an alternative approach by using the idea of deriving estimates of the likely range of total bicycling demand in an area on the basis of census commute-to-work data. The paper makes three contributions. The first is a general discussion of the total amount of bicycling in the United States and how it varies across places, on the basis of a number of surveys and some original data analysis. The second is the development of an argument that predictive models based on land use and transportation factors are unlikely to ever be accurate or useful because of a number of intractable problems. Third, a simple model is developed for estimation of a range of current levels of bicycling in a given geographic area with reasonable and known accuracy and by use of easily available data. While this model stops short of predicting bicycling levels or demand on specific facilities, it is an important first step in reaching these objectives. There is such a high degree of local variation in bicycling rates in the United States that attempts to predict bicycling levels directly without accounting for current levels are unlikely to be consistently successful.