U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
|This report is an archived publication and may contain dated technical, contact, and link information|
Publication Number: FHWA-RD-98-165
Date: July 1999
Guidebook on Methods to Estimate Non-Motorized Travel: Overview of Methods
4.0 Conclusions and Future Needs
A bicycle or pedestrian planner wishing to estimate future levels of non-motorized travel has a number of options. These include comparisons of proposed projects with usage on similar existing projects, calculations based on census and other available local data and assumptions, aggregate and disaggregate behavior models to predict travel choices, and inclusion of bicycle and pedestrian factors in existing regional travel models. Alternatively, the planner may choose to look at measures of the potential market for bicycling or walking, rather than explicitly forecasting demand. The planner may also use these measures in conjunction with measures of the quality of facilities supplied to prioritize improvements where they are most needed. Finally, these methods can be enhanced by tools and techniques such as GIS and preference surveys of travelers.
In addition, planners may develop combinations of existing and new approaches. Bicycle and pedestrian travel demand forecasting is an evolving field, and creative thought is needed by those who are confronted with planning needs in the real world. The best approach for any particular situation will depend on available knowledge, data, financial, and technical resources, as well as the specific purpose for which the demand forecasts are being developed.
Finally, planners should be aware of the limitations as well as the advantages of existing methods, and should supplement quantitative forecasts with the judgment of local practitioners and advocates when planning projects. Despite limitations, however, the methods discussed in this guidebook can provide valuable information, both for estimating the benefits of proposed projects and for prioritizing projects and improvements to achieve the greatest benefits to users.
4.2 Future Needs
As a result of developing this guidebook, a number of areas have been identified in which additional research and methodological development could be particularly useful. These suggestions are presented so that users of this guidebook can consider the limitations of existing knowledge when developing their own methods, collecting data, and conducting research. Recommended future efforts include:
Development of a Manual for Bicycle and Pedestrian Sketch-Planning
In the absence of better methods, practitioners who need to estimate usage on a non-motorized facility generally resort to back-of-the-envelope calculations based on readily available data and rules of thumb on travel behavior. These methods are somewhat crude and generally have not been tested for accuracy, but nevertheless may be the best that is possible given limitations on data, resources, and expertise. Development of a sketch-planning manual for bicycle and pedestrian forecasting would improve the state of practice in this area and could be widely used by bicycle and pedestrian planners. Such a manual would include methods and supporting data for developing local estimates of demand. Specific elements of the manual might include:
The sketch-planning techniques could, at a minimum, draw from techniques that already have been developed by practitioners and identified in this guidebook. Ideally, such techniques would be further developed and tested in practice to ensure that they are applicable to a variety of areas and that they give reasonable results.
Additional research useful for this type of guidebook might include further analysis of data sources, such as trail user counts and surveys in conjunction with other trail-related data, to look for patterns in facility usage and to provide information useful for the planning of comparable facilities.
Research on Factors Influencing Non-Motorized Travel Behavior
Along with the short-term documentation of planning methods and data for practitioners, more fundamental research is needed into the factors influencing non-motorized travel behavior and how these factors can be modeled to support demand forecasting. Particular attention should be given to identifying factors that are both of significance in predicting non-motorized travel behavior and that can be collected or created with relative ease from existing data sources or future survey efforts. Factors should be investigated that can be useful in a variety of forecasting methodologies ranging from sketch-planning techniques to travel demand and network modeling. Focusing on the individual traveler as the unit of analysis, rather than on aggregate-level studies, will provide richer information that will be useful not only for improvements to current efforts but to future modeling efforts such as activity-based analysis and microsimulation.
Facility design characteristics. Significant research has focused on developing quantitative measures of the quality or compatibility of facilities for bicyclists and pedestrians. The next step is to integrate these measures into methods of forecasting travel demand. Research is needed into how to aggregate facility-level compatibility measures, such as the Bicycle Compatibility Index, into an overall route or network compatibility measure, including facilities of varying quality as well as intersections and other discontinuities. Ultimately, the overall route or set of route options, rather than just individual facility characteristics, determines whether or not the bicyclist or pedestrian makes the trip.
Environment factors. Area-level environment factors that describe, or act as a proxy for, the relative attractiveness of bicycling or walking at an area/zonal level are potentially useful and should be further developed and tested. Pedestrian environment factors should be further refined and tested to verify their predictive capability. (Efforts in this area should build on recent research relating neighborhood design factors to levels of walking.) Bicycle environment factors also should be developed and tested for predictive capability. Other possibilities include the quality or impedance of alternative modes (traffic speeds, LOS, cost of parking, etc.) and the potential demand based on trip-end characteristics (population, employment, special generators, etc.). These factors should be useful both in sketch-planning techniques and in regional travel models where the scale of resolution is too coarse to model every facility in the network.
Attitudinal and perceptual factors. The relative importance of attitudinal and perceptual factors in the choice to walk or bicycle, as well as their potential uses in modeling, should be investigated. While gathering such data requires additional collection efforts, factors of this type have been found to be highly significant in determining travel behavior. Research in this area should focus on (1) which factors are most important; (2) how they can best be described/standardized; (3) what level of resources are required to collect these data on an ongoing basis; (4) how the factors may change over time; (5) how they can most effectively be influenced; and (6) how they can be integrated into modeling/forecasting techniques to predict the impacts of various policies. Research in this area can build on behavioral research from the public health field, as well as on existing studies of attitudes and perceptions regarding bicycling and walking.
Factors influencing recreational travel. None of the methods discussed in this guidebook make an explicit distinction between recreational and utilitarian travel. Many aggregate-level methods consider both types implicitly by looking at overall travel on a facility, while others such as regional travel models consider only utilitarian trip-making. Forecasting recreational travel at the individual or disaggregate level requires a different analysis framework, involving lifestyle and activity patterns, than is generally used in transportation modeling. Approaches from the public health arena that model the decision to exercise as a function of various personal/attitudinal characteristics and social factors should be helpful for incorporating recreational travel in transportation modeling.
Figure 4.1: Models Need to be Capable of Modeling Both Utilitarian and Recreational Travel.
Market research. Marketers in competitive industries have long recognized that marketing success depends on targeting the right customer with the right product. State-of-the-art techniques from the field of market research can be used to better identify the "market segments" for non-motorized travel, the travel characteristics of each market segment, and the facility design factors that are important in attracting increased usage from each segment. The trip and personal characteristics of recreational travelers, for example, should be differentiated from those of utilitarian travelers, while utilitarian users may be further distinguished as necessity vs. discretionary, commute vs. non-commute, etc. While some research has been conducted in defining non-motorized market segments, planners have not adequately identified the differences in techniques required for identifying the needs and predicting the behavior of these various groups.
Integration of facility/environment, policy, and personal/attitudinal variables into an overall modeling framework. Insights from the public health and social marketing fields suggest that personal attitudes and beliefs interact strongly with environmental and policy variables to influence travel behavior and mode choice, particularly for bicycling. Accurate forecasting of bicycle travel will require integrating these variables into a modeling framework which can include personal/attitudinal variables, and which can account for the fact that the effects of facility/environmental improvements will depend on (as well as influence) the levels of these other variables.
Integration of Bicycle and Pedestrian Considerations into Mainstream Transportation Models and Planning
As a final recommendation, further development of modeling techniques and data sources is needed to better integrate bicycle and pedestrian travel into mainstream transportation models and planning activities. Regional travel models have the unique advantage of representing an integrated framework for predicting travel decisions, considering all trips and modal options, as well as personal and household characteristics, within the spatial structure of the surrounding area. Furthermore, they are widely used and accepted as demand forecasting methods for automobile and transit planning. Improvements to existing models should significantly increase their usefulness for analyzing non-motorized policies and facility improvements. Specific near-term and long-term improvements might include:
Data collection on bicycle and pedestrian travel. A general need for all types of bicycle and pedestrian planning is better data on trip and personal characteristics of travelers. Household travel surveys performed for modeling purposes are a potentially effective means of collecting these data. While data on non-motorized trips are increasingly being collected in these surveys, surveys must be designed carefully to ensure that all non-motorized trips are reported. Also, since there are generally few reported bicycle trips, additional means of collecting data on bicycle trips, such as supplemental stated preference surveys, may be required. The potential for non-motorized data collection using emerging ITS information technologies should also be investigated.
Spatial scales of models. The scale at which travel is modeled should be refined to be more relevant to the short distances involved in bicycle and pedestrian travel. Improvements in computational power and in data management tools will make it easier to analyze smaller-scale networks of bicycle and pedestrian facilities rather than just major roadways.
Facility design factors. For travel models in which bicycle and pedestrian networks can be accurately represented, the most important design variables for predicting mode and route choice should be identified and included in the network link characteristics in the model. This will require quantifying tradeoffs between these variables and link travel time or distance. Travel time penalties also need to be developed for major intersections or other discontinuities in the network. The validity of aggregating link-level factors across routes and networks to produce an overall "utility" or "compatibility" should be tested. In addition, the potential for transferring preferences for facility design from studies conducted in one area to other areas, to avoid the need for locally-specific surveys, should be investigated.
Environment factors. For regional models in which zones are too large to model local non-motorized networks, further development and testing of zone-level environment factors are needed to validate the usefulness of these models for analyzing non-motorized travel. These efforts can build on the outcomes of basic research into these factors and can also utilize GIS data bases and analysis techniques to develop better factors. In addition, environment factors should be developed for bicycles as well as pedestrians.
Other environmental and policy variables critical to non-motorized modeling. Factors such as the presence of bicycle parking and workplace showers and lockers may be just as important as facility and network design factors in determining the decision to walk and particularly to bicycle. Methods should be investigated for collecting data on these factors; describing them in a way in which they can be included in travel models; and verifying the relationship of the identified factors with levels of non-motorized travel.
Modeling behavioral change in multiple stages. Methods and data requirements for modeling bicycle use in multiple stages should be investigated. Multi-stage behavior models may improve forecasting efforts because the individual must first decide to even consider bicycling or walking as a viable travel option. Only when bicycling or walking is regarded as a viable option does the question of whether to bicycle or walk for a particular trip become relevant. These methods should be tested for improving the sensitivity and predictive power of travel models. The results of research into attitudinal and perceptual factors, as well as modeling approaches from the public health and market research areas, can inform this process.
Inclusion of recreational travel. To be useful for modeling non-motorized travel particularly on separate facilities, travel models will need to be capable of modeling recreational as well as utilitarian travel. Advances in activity-based modeling, which looks at personal and household activity patterns rather than simply trip-making patterns, may be useful in this effort. Research and methods from the public health arena are also relevant to modeling recreational travel.