Estimating the amount people drive based on accessibility measures

By Logan Dredske

How does the built environment influence the amount people drive? Research by SSTI’s Logan Dredske worked to answer this very question. The focus of his research was to create a framework for estimating vehicle miles traveled based on conditions of the built environment. His goal was to use measures of accessibility as the principal proxy for the built environment. The research also converted vehicle miles traveled into greenhouse gas emissions and evaluated the ability of transportation projects to reduce emissions. Dredske’s research was completed as a joint effort between SSTI and the Department of Planning and Landscape Architecture at the University of Wisconsin–Madison.

Traditionally, the built environment has been measured using the D-variables. Unlike traditional methods, using measures of accessibility as a proxy for the built environment has the potential to be used in project evaluation. As an example, imagine a new overpass constructed to assist pedestrians with the crossing of a busy roadway. Traditional measures such as intersection density or street width would not be able to account for this improvement. However, accessibility measures account for the physical connections of a transportation network, while also accounting for all the other aspects of the D-variables.

Dredske utilized the National Household Travel Survey’s weekday household vehicle miles traveled data in his research. This data allowed him to estimate daily travel behavior for a specific area. He then was able to overlay travel behavior with built environment measures such as the D-variables from the EPA’s Smart Location Database and accessibility measures from Sugar Access, a GIS add-on. Demographic information was also included in the model.

Results suggested that the most useful predictors of vehicle miles traveled are:

  • Access to non-work destinations (e.g., restaurants, stores, medical centers, leisure activities) by walking
  • Access to jobs by automobile
  • Ratio of jobs accessible by transit compared to by automobile
  • Total household income
  • Number of adults per household
  • Ratio of children to adults in the household
  • Number of vehicles per household

When comparing the usefulness of accessibility measures to the D-variables, Dredske’s models suggested that knowing only a few accessibility measures (the three indicated above) allows you to estimate vehicle miles traveled just as well as using many built-environment D-variables. Another major takeaway from the research was that accessibility measures are important predictors of vehicle miles traveled, but household characteristics are the dominant predictors. However, vehicle miles traveled cannot be predicted with household characteristics alone and must be associated with measures of the built environment.

One example of how Dredske’s research can be applied is by estimating the ability of a transportation improvement project to reduce vehicle miles traveled, hence reducing greenhouse gas emissions. He applied his model to a relatively new transit line in Madison, WI, to evaluate the line’s impact (Figure 1).

Figure 1. Percent reduction in VMT and GHG emissions because of Madison Metro’s Route 31 transit line in Madison, WI.

The added transit line provided neighborhoods on Madison’s southeast side with transit service that connected them to a job center on the city’s east side. Based on this research, the added transit line is estimated to reduce vehicle miles traveled and greenhouse gas emissions by up to 14 percent for some households.

Logan Dredske is a Project Assistant at SSTI.