Key Enhancements to the WFRC/MAG Four-Step Travel Demand Model

Reid Ewing, University of Utah



Conventional four-step travel demand modeling is overdue for a major update. The latest NITC report from the University of Utah offers planners better predictive accuracy through an improved model, allowing for much greater sensitivity to new variables that affect travel behavior. Specifically, it accounts for varying rates of vehicle ownership, intrazonal travel, and multimodal mode choices.

Used by nearly all metropolitan planning organizations (MPOs), state departments of transportation, and local planning agencies in the United States, the importance of travel demand modeling for project selection cannot be overstated: They are the basis for forecasting future travel patterns and developing long-range regional plans.

Led by Reid Ewing, Director of the Metropolitan Research Center at the University of Utah, the research team includes University of Utah doctoral student Sadegh Sabouri, and UU alumni Keunhyun Park (now at Utah State), Guang Tian (now at the University of New Orleans) and former NITC fellow Torrey Lyons (now at the University of North Carolina at Chapel Hill). The enhanced model was developed in partnership with two Utah MPO's with plans for implementation: Wasatch Front Regional Council (WFRC) and Mountainland Association of Governments (MAG).

The Traditional Four-Step Travel Demand Model
Trip Generation tells us the number of trips generated in each traffic analysis zone (TAZ), usually based on some prediction of vehicle ownership.

Trip Distribution tells us where the trips go, matching trip productions to trip attractions. Particularly tricky are predictions of trips that remain within the same zone. 

Mode Choice tells us which mode of travel is used for these trips. 

Route Assignment tells us what routes are taken, assigning trips to networks that are specific to each mode. 

A major flaw of the four-step model is its relative insensitivity to the so-called D variables: characteristics of the built environment that are known to affect travel behavior. The D variables are:

Development density
Land use diversity
Street network design
Destination accessibility
Distance to transit
What Does the New Model Accomplish?
This report introduces a vehicle ownership model, an intrazonal travel model, and a mode choice model that considers all of the D variables based on household travel surveys and built environmental data. These models were calibrated with data from the University of Utah's 32-region household travel database, the largest household travel database of its sort ever assembled. This database has been linked to built environmental data as well.

Vehicle ownership is often treated as a function of sociodemographic variables only. But in reality, a phenomenon known as "car shedding" means that vehicle ownership rates go down as the built environment becomes denser. Researchers pooled regional household travel survey data from 32 diverse regions of the United States and generated consistent measures for all regions, then modeled vehicle ownership (see Section 2 of the final report for more details). Results suggest that areas with high population and employment density, good street connections, great transit service, and high accessibility allow direct substitution of transit, walk, and bike travel for automobile travel.

Intrazonal travel is hard to predict accurately using conventional models. Researchers offer a new method which accounts for important built-environment related measures like activity density, street connectivity, and mixed land uses and how they impact intrazonal trip making. They also use discrete choice modeling, a significant improvement over standard intrazonal modeling efforts, since it more accurately represents the behavioral aspects inherent in individual travel decision making. See section 3 of the final report for more details. 

Multimodal mode choice is also more accurately predicted by the new model. Many traditional models focus exclusively on vehicle trips. Bicycling, in particular, is seldom treated as a separate transportation mode. Compared to the traditional walk/bike mode choice model which only controls for the trip distance, in this study researchers were able to control for most of the critical sociodemographic and built environment variables. Results confirm that in all models, some D variables will reduce the share of vehicle trips and will encourage travelers to use non-motorized modes of travel, as well as transit.

Implementation of the New Model
Researchers calibrated the new model and validated its results by comparing its predicted trips to actual travel survey data. The new model was found to consistently outperform and offer far better predictive accuracy than WFRC and MAG's current models. Going forward, both MPOs will incorporate this into their existing four-step modeling process.

This model will also be made available to MPOs across the nation, the vast majority of whom still use four-step models. All MPOs will be sent copies of the NITC final report and all peer-reviewed publications to arise from this project, in an effort to reduce barriers and actively promote innovation to enhance the performance of the nation’s transportation system.

More NITC Travel Demand Modeling Research
This project parallels another NITC effort, led by Kelly Clifton of Portland State University, which addresses a knowledge gap in trip generation by making travel models more sensitive to the built environment and contextual factors, allowing for better accuracy in predicting bicycle and pedestrian trips. Read more about Clifton's trip generation research.

This research was funded by the National Institute for Transportation and Communities, with additional support from the University of Utah, the Utah Department of Transportation, Utah Transit Authority, Wasatch Front Regional Council, and Mountainland Association of Governments.

Project Details

Project Type:
Project Status:
End Date:
October 31,2019
UTC Grant Cycle:
NITC 16 Initial Projects
UTC Funding:

Other Products

  • Park, K., Ewing, R., Sabouri, S., Choi, D. A., Hamidi, S., & Tian, G. Guidelines for a Polycentric Region to Reduce Vehicle Use and Increase Walking and Transit Use. Journal of the American Planning Association, 14. doi:10.1080/01944363.2019.1692690 (PUBLICATION)
  • Park, K., Ewing, R., Scheer, B. C., & Khan, S. S. A. (2018). Travel Behavior in TODs vs. Non-TODs: Using Cluster Analysis and Propensity Score Matching. Transportation Research Record, 2672(6), 31-39. doi:10.1177/0361198118774159 (PUBLICATION)