Pre-Destination-Choice Walk Mode Choice Modeling
Corresponding Author: Norman Marshall, Smart Mobility Inc.
Presented By: Norman Marshall, Smart Mobility Inc
Walk trips are a very significant and growing portion of travel. In recent household travel surveys, walk mode shares have increased from previous surveys of the same region. In the National Household Travel Survey, the walk mode share increased from 8.6% in 2001 to 10.4% in 2009. In the 2010-2012 California Household Travel Survey, the statewide walk share almost doubled since 2000 to 16.6%.
Reinforcing trends likely will increase walk mode share in the future including: lower licensure and auto ownership by younger adults, shift in residential preference towards mixed-use walkable areas, investments in walk and walk safety infrastructure, shared economy technologies including car share and Uber/Lyft services, interest and initiatives in health and fitness, and greenhouse gas emission reduction policies.
As walk mode share continues to increase, accurate walk trip forecasting will become increasingly important. Currently, most travel demand models (trip-based or activity-based) model walk trips using post-destination-choice mode choice models. These models generally poorly match the locations of walk trips, because they do not model actual behavior. People do not look at all destinations, pick one, and then realize they cannot walk to it. The walk mode decision is made either simultaneously with the destination choice decision or prior to it.
Pre-destination-choice walk mode choice models have been found to be more accurate in models in several different regions. For example, the conventional Austin regional model only matches walk mode shares for work trips at the Census Tract level at a correlation of 0.38 In sharp contrast, a simple pre-destination-choice walk mode choice model achieves a correlation of 0.89.
Modelers sometimes say that walk trips do not really matter because walk trips only substitute for short vehicle trips, or that walk trips do not matter because walk trips are mostly intrazonal trips. These observations are true of the models, but not true of the real world. In the real world, walk trips often substitute for longer auto trips that otherwise would have been made. Switching to pre-destination-choice walk mode choice modeling is not difficult, and is strongly recommended.