Analysis of Network and Non-Network Factors on Traveler Choice Toward Improving Modeling Accuracy for Better Transportation Decisionmaking

CHAPTER 5. ATDM CASE STUDY

INTRODUCTION

The traveler choice focus area targets a traveler’s higher-level predictive strategic choices influenced by a range of variables such as travel time reliability, congestion (recurrent and non-recurrent), weather, pricing, availability of transit services and parking, and sidewalks. Traveler choices in network and non-network conditions can be influenced by dynamic factors (e.g., travel time, level of congestion, and weather that travelers will encounter on trips) and static factors (e.g., availability of transit services, parking, sidewalks, and bike routes). The capability of existing transportation analysis tools to accurately model and simulate traveler choices is limited due to the lack of adequate methodologies and reliable data. Consequently, it is critical to understand choices made by travelers under various circumstances and the impact of these choices on the transportation system.

Within the area of ATDM, the study focused on identifying information and data that can inform the development of factors that are critical to bicycle riders’ travel decisionmaking. The examination included a review of information and data collected by local areas in regional case studies on active transportation demand and supply travel information and identify available data and information that may inform the choices that travelers make about under what circumstances bicycle trips occur. This work builds on previous efforts related to bicycle travel data in the Metropolitan Washington region conducted by the Metropolitan Washington Council of Governments (MWCOG) as well as the Southern California region.

CURRENT AGENCY PRACTICES

The study reviewed the current practice and availability of active transportation planning data by region by conducting a literature search of publicly available documents, reports, and data. The Web sites of the significant transportation planning agencies within each region, including the MPO, large cities, and urban counties were reviewed. The study included telephone interviews with appropriate transportation planning staff of local and regional transportation planning agencies. The goal of the interview was to collect information on current practices in planning for active transportation and identify available data that the agencies have collected or used for this purpose. Some local bicycle travel data were requested and collected from government transportation planning agencies as examples of the state of the professional practice and the state of the art are for bicycle planning data collection efforts.

The following four urban metropolitan regions in the United States were examined for this effort:

• Washington, DC, metropolitan region.
  
  
• Southern California metropolitan region (Southern California Association of Governments (SCAG) region).
  
  
• San Francisco Bay area region.
  
  
• Cleveland region.

The initial review of existing local and regional transportation planning agency reports found that bicycle travel data were being collected and used to some degree for descriptive purposes. Data collected from the four metropolitan regions confirm that bicycle travel is increasing both as an active transportation mode and as a means of travel demand management. Bicycle travel supply and demand variables collected from local agencies varied in quality and robustness. Some agencies are beginning to integrate bicycle use data into travel forecasting, but simple trend extrapolation is the most common use. Changes in road capacity usage, particularly in larger urban areas, has the potential to impact automobile travel capacity and travel speeds along key urban street corridors.

The use of bicycle travel data for travel demand forecasting purposes was more limited. Leading edge travel demand modeling agencies are incorporating bicycle travel into overall regional travel forecasting, but significant data gaps limit the completeness and robustness of locally collected bicycle data for these purposes.

Information about data collection efforts in the San Francisco Bay Area Rapid Transit (BART) Agency includes the following:

• Station profile studies.
  
  
• Customer surveys.
  
  
• Passenger online surveys.
  
  
• Bike station surveys.
  
  
• Bicycle parking inventories.

Regional household travel surveys, such as the one conducted by MWCOG in 2007–2008, included data collection on the bicycle mode of travel. Some examples of data that were collected or estimated based upon the household travel surveys include the following:

• Number of bike trips.
  
  
• Bike trip distance.
  
  
• Percentage of population for whom biking is the primary travel mode.
  
  
• Percentage of daily trips by biking.
  
  
• Percentage of work trips by biking.
  
  
• Percentage of school trips by biking.
  
  
• Percentage of employers providing bike/pedestrian facilities or services.
  
  
• Type of bikeway mostly used (off-road bike trails versus on-road lanes).
  
  
• Percentage of daily trips by walking.

Other more specialized regional transportation planning studies collected data on bicycle modes as part of examinations of accessibility of neighborhoods and of transit stations. For example, in the Washington, DC, region, an internal study of the transportation and land use interaction included consideration of access to the bicycle mode and the potential use of the bicycle mode under different scenarios for transportation and land use in the region. Another internal study in the Washington, DC, region for the regional rail transit agency (Metrorail) collected and estimated data on the following:

• Bicycle infrastructure availability.
  
  
• Mode share of biking for Metrorail access by station type.
  
  
• Mode share of walking for Metrorail access by station type.

FRAMEWORK FOR EVALUATION

As discussed in chapter 4, the policies of interest in the realm of ATDM call for a deeper understanding of mode use in the context of individuals and household activity engagement decisions and behaviors. As a result, some of the same factors discussed in chapter 4 in relation to transit mode use and the influence of network and non-network factors, particularly land use, urban design, and safety/security perceptions are applicable in this case study, as well. Some of the main differences arise from the particular characteristics of active modes, especially bicycles, including their perceived (and actual) safety when running unprotected along with vehicular traffic, the physical effort that needs to be exerted in connection with frequent stop-start patterns, and the relative lack of protection vis-à-vis inclement weather. However, the basic structure of the modeling frameworks remains the same, with activity-based models providing an appropriate construct and tools to examine bicycle mode use as part of auto demand management strategies.

In the short term, it is expected that travelers’ perceptions, prior experience, and built environment factors will influence mode choice in general and bicycle use in particular. However, in the medium- to long-term horizons, policy will indirectly influence further user decisions. This is illustrated in figure 23.

Figure 23. Illustration. Modeling framework for analyzing ATDM policies.

BEHAVIORAL DIMENSIONS AND IDEAL DATA

Relevant Bicycle Use Patterns

There is a considerable need for additional data to be collected within individual metropolitan regions regarding patterns of bicycle travel in order to increase the understanding and modeling of bicycle travel behavior. The overall sense obtained from the review of existing data, studies, and current planning practices is that within the past decade, cities and regions around the United States have begun collecting some bicycle data that are locally specific but are constrained by funding, resource, and expertise limitations. One example of this is in the San Francisco Bay area, where data on bicycle travel speeds were not available for modeling purposes. As a result, the staff from the county planning agency reported that individual staff members would spend time on weekends and other non-work hours riding their own bicycles on different streets in order to estimate bicycle travel speeds for these links that could then be used in a travel demand modeling framework.

In addition, bicycle trips may constitute only one segment or link of a longer multimodal trip. As a result, there is a need for more data and understanding about metropolitan travel (e.g., commuting) where a bicycle may provide one trip segment of a multimodal trip.

Previous work has identified that bicycle travel includes different types of trips with different characteristics. Good active transportation data should include robust and reliable information on different types of bicycle trips, including the following:

• Commuting trips that are completely on bicycle mode.
  
  
• Non-commuting bicycle use for short trips.
  
  
• Longer distance bicycle on public transportation trips.

Examples of the travel data that this project sought to identify and analyze included the following:

• Commuter trips bike mode share.
  
  
• All trips bike mode share.
  
  
• Bike trips for the region disaggregated on a county by county basis.
  
  
• Trip distances, including access segment, bicycle segment, and linked line-haul public transportation segments.
  
  
• Mode shift from pedestrian and automobile to bicycle mode.
  
  
• Baseline mode splits for active transportation modes and induced (new trips) on bicycle mode.
  
  
• Bicycle trip counts at specific locations, station areas, and cordons.
  
  
• Trip purposes for different bicycle trips.
  
  
• Bicycle travelers access to motor vehicles.

The findings of this study are that while basic bicycle use data elements are being collected to an increasing degree by local transportation planning agencies, these data are of limited use for travel demand forecasting. Since mode shift data and data on other traveler attributes are usually not collected, it is difficult to measure and model a trend in mode shift or forecast forward in time based on historical trends. As a result, to the degree that travel demand forecasting is occurring for the bicycle mode, they are often being forecasted or modeled using assumptions about future mode shares and based on simple trend extension.

Since most local governments around the United States rely on federally designated and federally funded MPOs for travel demand forecasting, it is often the case that mode-based travel forecasts are not available at a city or county level unless they are obtained from regional MPO forecasts. Local governments appear to be focused on developing plans for infrastructure improvements related to bicycle travel but do not always link them to the travel demand forecasting and data. This appears to be somewhat different for cities and counties that have greater in-house travel demand analysis and forecasting capabilities. San Francisco is an example of this, where the central city and county maintains its own advanced travel demand forecasting capabilities.

This review of existing practice suggests that even amongst the more progressive MPOs, inclusion of bicycling and active transportation options as integral parts of activity-based model systems remains in the very early stages.

Data Needs for Incorporating Bicycle Use into Travel Demand Forecasting

Some counties and metropolitan regions that conduct travel forecasting and modeling incorporate limited bicycle capacity and bicycle travel usage data into their forecasting on a limited basis (e.g., general mode shares). Without sufficient data to calibrate or corroborate the forecasting or modeling data, these travel forecasts may be limited in usefulness.

The study identified one example by an MPO where bicycle travel was attempted to be forecast forward in time. Discussion of this example highlights the data needs and the data collection gaps that exist in order for more robust travel demand forecasting to occur with respect to bicycle travel. The Washington, DC, MPO MWCOG staff developed a spreadsheet model to estimate the benefits and costs of bicycle travel using a planned bikeshare system. The model was developed and used as part of a process to apply for Federal grant support for the system.

Bicycle sharing systems are increasingly popular and diverse. A number of bicycles are made available for shared use by individuals who do not own bicycles. Public bicycles are a mobility service, mainly useful in urban environments for proximity travels.

It has been estimated that as of 2010, there were more than 200 such schemes operating worldwide. The early attempts at unregulated bikeshare programs encountered numerous problems such as theft and vandalism. In 1993 in Cambridge, UK, the majority of the fleet of 300 bicycles was stolen in one program, and the program was abandoned.

The latest generation of this program includes bicycles that are kept at self-service terminals throughout the city. Individuals registered with the program identify themselves with their membership card (or a smart card, cell phone, etc.) at any of the hubs to check out a bicycle for a short period of time, usually less than 2 h. In many schemes, the first half hour is free, such as the Capital Bikeshare program in Washington, DC.⁽¹¹⁷⁾ Additionally, many of the membership programs are being operated through public-private partnerships. Several European cities, including the French cities of Lyon and Paris as well as London, Barcelona, Stockholm and Oslo, have signed contracts with private advertising agencies that supply the city with thousands of bicycles free of charge (or for a minor fee). In return, the agencies are allowed to advertise both on the bikes themselves and in other select locations in the city.

The spreadsheet model estimates the regional bikeshare use on the then planned (now active) bikeshare system in the Washington, DC, metro region. Some examples of basic data elements that were estimated or assumed in the spreadsheet model include the following:⁽¹¹⁷⁾

• Number of bikes.
  
  
• Number of bikes added each year.
  
  
• Number of riders per bike.
  
  
• Daily bicycle riders.
  
  
• Number of trips per day.
  
  
• Average bike trip length (miles).
  
  
• Bicycle miles traveled per day.

It is evident by comparing the data from the spreadsheet analysis tool with the actual data that are being collected in the Washington, DC, region that there is a significant gap between the bicycle travel data that has been found to be collected and the data needs for even a relatively simplified spreadsheet estimation tool for bicycle travel.

The review of existing data and studies found some promising data sources that potentially could be adapted to be incorporated into a more robust travel demand forecasting framework. Two examples of this are specialized travel surveys and specialized service operator data from both public transportation agencies and bicycle service providers (e.g., bikesharing operators). Specialized travel surveys ask respondents a set of standardized questions and collect detailed data (e.g., time of day, location, etc.) that could be used to develop and calibrate models on general relationships between bicycle travel and other variables (e.g., demographic, trip purpose, time of day, etc.). Specialized service operator data (e.g., Capital Bikeshare and Cleveland Transit) incorporate origins and destinations and, as a result, can be linked with geographic and other datasets (e.g., weather) and used to estimate travel speeds. In the absence of such specialized data, travel speeds for bicycle travel must be collected on a link-by-link basis and is extremely time intensive (e.g., Santa Clara County in the San Francisco Bay area). Discussion and examples of the different data types and elements found are presented in the following section.

CHARACTERIZATION OF BICYCLE USE PATTERNS BASED ON AVAILABLE DATA

Data Sources and Elements

Some local governments collect bicycle count data for intersections and corridors of interest. The bicycle count data are often limited in usefulness since they are not linked to mode shift and other data attributes of travelers. In addition to a review of regional transportation planning data and studies, local government data collection and studies were also reviewed. Many local governments have been developing a bicycle master plan that focuses on the development of additional bicycle infrastructure. However, these plans are not strongly linked with historical bicycle use data and bicycle travel forecasting. For example, in Southern California, all six counties in the MPO region either have bicycle master plans or non-motorized transportation plans that include the bicycle mode.⁽¹¹⁸⁾ Transportation planning for the bicycle mode has often been incorporated into broader studies by local governments related to TDM, in which bicycle use is included as one element of a larger portfolio of TDM strategies and efforts. In these cases, since the bicycle mode is only one element of a larger set of topics, specific detailed historical data on bicycle travel were not usually collected, nor were specific travel demand forecasts made for the bicycle mode.

In spite of the limited amount of data collection and robust travel forecasting with respect to bicycle traveling, local and regional transportation planning agencies often incorporate planning processes and committees with respect to bicycle, and increasingly, pedestrian travel. As a result, while the political and organizational will to conduct bicycle planning appears to exist at the local and regional level, a commensurate degree of data collection and data analysis does not seem to be arising, perhaps due to limited resources for these efforts.

Selected niche data collection is occurring for specific segments of bicycle travel of key interest to organizations. Examples of this include bikesharing system usage data collected by the operating agency, bicycle on rail system data collected by the public transportation agency in specialized studies, and bicycle on bus data collected as part of normal bus transit operations.⁽¹¹⁹⁾ Some examples of promising and successful local and regional bicycle travel data collections efforts identified include the following:

• Los Angeles County Metropolitan Transportation Authority (LAC MTA) study collected data of last mile bicycle trips that are linked with transit trips.⁽¹²⁰⁾ This focused study collected data from intercept surveys of bicyclists at a subset of Metrorail stations. Counts and surveys were conducted during weekday morning commute and weekday evening commute periods. Volunteers collected 605 usable surveys and counted 2,305 bicyclists at the 19 transit stations that were part of the study.
  
  
• Data on bicycle sharing in the Washington, DC, region, which is one of the first regions of the country to have an operational bicycle sharing system.⁽¹¹⁷⁾ Capital Bikeshare in Washington, DC, began operations in 2010 with 1,100 bicycles at 114 stations throughout Arlington, VA, and Washington, DC. The system has expanded to 1,670 bicycles at 175 stations. Membership has reached 19,000 persons, making it the largest bikesharing organization in the United States as measured by membership.

Since bikesharing relies to a large degree on technology, there is potential for significant data to be available for planning purposes based on the actual database of use by members. In addition, if confidentiality issues can be overcome, the trip patterns of individual members may also be linked with other attributes of the bikeshare users. Capital Bikeshare makes data about usage patterns available on its Web site and also conducts regular surveys of its members.

Bikesharing is expanding rapidly across the United States. Plans are being implemented for operation of bikesharing systems in both the San Francisco Bay area and over a larger portion of Los Angeles, as well as other parts of the United States.

Data collected by the San Francisco County Transportation Authority (SFCTA) and the San Francisco Bicycle Coalition on the GPS-data related to routes taken for bicycle commuter trips. With funding from a California Department of Transportation (Caltrans) planning grant, SFCTA was able to work with San Francisco Bicycle Coalition member volunteers who agreed to travel on their regular bicycle commuting routes with a GPS device. With the information from these GPS tracking of bicycle commuters, SFCTA was able to obtain and incorporate more and better information on bicycle commuting route choice and time of day patterns than they had available before the study.

The ideal robust data collection for bicycle travel for a regional or local transportation planning effort is one where at least three types of data elements could be identified and collected. The following data needs to be local and specific in order to be incorporated into travel demand forecasting methods:

• Bicycle use basic data elements.
  
  
• Bicycle mode shift table data elements.
  
  
• Bicycle travel linked to other traveler attributes.

The first type, count and overall usage, is discussed in the following subsection using counts in the Washington, DC, metropolitan area, including Arlington, VA, as well as counts in the Los Angeles, CA, area. The second type, data that could support mode shift analyses, is discussed in connection with survey data of bike on rail users in both the Los Angeles and San Francisco, CA, areas. The third type, explicitly linking traveler and service attributes to the likelihood of bicycle rail usage, is illustrated using data from Cleveland, OH.

Bicycle Use Count Data: Washington, DC, and Southern California

The bicycle mode still contributes a relatively small share of overall trips and travel distances at a national scale and also at smaller geographies within the four individual metropolitan regions examined. There are some cities that have relatively higher mode shares for bicycle commuting, but bicycle use is small overall. At the same time, bicycle use has been growing in many cities in recent years, and cities have been planning and implementing dedicated bicycle infrastructure.

Local governments, including both cities and counties, are increasingly collecting basic bicycle travel use data. One example of basic bicycle use data are intersection bicycle count data from Arlington County, an inner-ring suburb in the Washington, DC, metropolitan region. Another example is corridor-based bicycle counts relative to time of day in Washington, DC.

Within Washington, DC, the local council government approved a complete streets policy in October 2010. There are 40 locations where bicycle counts are collected by the Washington, DC, Department of Transportation, including 9 or so bridge locations and other strategic locations. The data are used to assess general trends and to help increase understanding of usage of the bicycle and road facilities. The data are not yet incorporated into travel demand forecasting. Some specific corridor studies have included bicycle use data. The bicycle counts program has been funded through technical assistance funding from MWCOG, the regional MPO. MWCOG also keeps the regional travel demand model that is the general forecasting model for the region. Table 7 shows an example of some of the results of basic bicycle and pedestrian use data collected within Arlington County at a specific intersection during a morning peak hour period.

Table 7. Basic non-motorized use data element: intersection non-motorized counts in
Arlington County, VA.

Selected summary results and data from Washington, DC, bicycle counts data collection are shown in figure 24 through figure 26. These graphs generally reveal the increasing trends in bicycle use and trips.

 
Data for this figure were provided by DDOT staff and is available to the public (public domain).
Figure 24. Graph. Increase in hourly bicycle counts on specific street corridor from
2004–2012 in Washington, DC.

Data for this figure were provided by DDOT staff and is available to the public (public domain).
Figure 25. Graph. Increase in peak hour bicycle counts on specific street corridor from
2004–2012 in Washington, DC.

 
Data for this figure were provided by DDOT staff and is available to the public (public domain).
Figure 26. Graph. Average peak hour bicycle counts on specific street corridor per mile of
bicycle lanes from 2004–2012 in Washington, DC.

Another example of data being collected for bicycle travel consists of specialized trip data collected by major public transportation agencies in both northern and southern California.

The data examples show data collection from a LAC MTA study of last mile bicycle trips that are linked with transit trips.⁽¹²¹⁾ These data provide a good example of the existing data that have been collected at a local level within one region of the country.

Table 8 shows data collected from LAC MTA of the number of Metrorail stations where bicyclists boarded or alighted.

Table 8. Number of Metrorail stations where bicyclists boarded or alighted, by rail line;
LAC MTA.

The unpublished study estimated that bicyclists made up 1.3 percent of all annual Metrorail trips. In terms of mode shift, the study found that 27 percent of bicycle-rail trips replaced a motor vehicle trip. Additionally, 13 percent of bicyclists report that they would not take the trip if they did not have the option of making the combined bicycle-rail trip. On average, bicyclists traveled 2.2 mi to access the Metrorail stations. As a result, LAC MTA was able to obtain more information on the bikeshed of Metrorail stations.

Only to a limited degree was bicycle use data collected by local governments linked with other traveler attributes. In specialized cases of public transportation agencies, some effort has been made in leading practice examples to collect additional traveler attribute data to link to bicycle mode use. For example, in the case of the Los Angeles County Bicycle Transit study, the trip purposes of bicycle mode users were included as part of the survey data collected.

Table 9. Bicycle boardings data at transit stations: LAC MTA blue line.

Mode Shift Analysis: Los Angeles and San Francisco Bike-on-Rail User Surveys

The unpublished LAC MTA study is a good example of more robust data collection, since some information was collected that could be used for mode shift analysis. Table 10 and table 11 show to what degree bicyclists had access to motor vehicles and the results of a question that asks bicycle riders what mode they might have used if not for the availability of the bicycle-rail trip, respectively.

There is a need to examine actual mode shift to bicycle from other modes to ascertain changes over time in order to accurately estimate, model, and better forecast future travel. Without a good basis for estimating mode shift to the bicycle mode, it is clear that the future estimates would be limited in value and robustness.

Table 10. LAC MTA bicycle transit survey responses to motor vehicle access question,
“How often do you have access to a motor vehicle?”

Table 11 . LAC MTA bicycle transit survey responses to motor vehicle access question,
“If you did not have your bike, how would you get from your origin to the first station?”

Some of the data elements related to mode shift that were assumed or estimated by MWCOG include the following:⁽¹²²⁾

• Percentage of riders shifted from auto to bike.
  
  
• Percentage of riders shifted from bus/rail to bike.
  
  
• Percentage of riders who also use transit.
  
  
• Percentage of riders with increased access to transit.
  
  
• Percentage of bike riders as new transit riders.
  
  
• Percentage of new transit trips coming from auto.
  
  
• Average transit trip length.
  
  
• Percentage of trips not made before.
  
  
• Percentage of riders shifted from walk to bike.
  
  
• Percentage of riders shifted from taxi to bike.

Data are available from an MWCOG analysis of BikePODs, and the following mode shift percentages off of baseline were assumed for one bikeshare system similar in size to that in Montreal, which contains roughly 5,000 bicycles and 400 stations:⁽¹²²⁾

• 8 percent of bike riders shifted from SOV to bike.
  
  
• 3 percent of riders shifted from taxi to bike.
  
  
• 8 percent of bike riders as new transit riders.
  
  
• 10 percent of new transit trips shifted from SOV to transit.

In 2010, MWCOG proposed to develop a regional bikesharing system for the Washington, DC, area. MWCOG conducted a benefit-cost analysis for the proposed system in application for funding from the Transportation Investment Generating Economic Recovery II Grants Program.

In developing the Washington, DC, Bicycle Sharing Spreadsheet Tool, it became clear that no applicable data would be available from cities in the United States, since the Washington, DC, region was one of the first regions in North America and the United States to implement bikesharing. As a result, the estimation process relied on adapting data from European cities’ experience with bikesharing. Example mode shift values for Paris, France, and London, United Kingdom, are presented in table 12. It is clear that the validity of this data for estimating future bikesharing in a U.S. city like Washington, DC, would be limited and that a preferable situation would be data collection of actual bikesharing use from U.S. cities.

Table 12 . Example of comparison data from other cities: mode shift to bikeshare.⁽¹²²⁾

Table 13 provides an example of bicycle use data linked to other traveler attributes, namely the trip purpose for which the bicycle is used in connection with an intermodal metro rail trip in the Los Angeles, CA, metropolitan region.

Table 13. Example of bicycle use data linked to other traveler attributes: bicycle trip
purpose data from Los Angeles County.

Another example of the specialized data collected by a public transportation agency is a study by the BART Agency entitled, BART Bicycle Plan: Modeling Access to Transit. Like the SFCTA study, the study was funded by a Caltrans planning grant.⁽¹²³⁾

An example of mode share information that is collected by a public transportation agency is data from the BART Agency related to mode shares for specific rail stations on the system.⁽¹²¹⁾ Based on data collected from two station profile studies with a 10-year interval between the data collection, BART has been able to document bicycle mode shares for individual stations for the two study years as well as interpolate a growth rate in bicycle mode access to the stations. Data collection from these studies is a promising approach to obtaining data for travel demand forecasting for bicycle-rail transit trips.

As part of a larger study, BART developed a bicycle investment tool, which is intended to help BART and other rail transit operators in the San Francisco Bay Area with estimating the effects of bicycle-related investments on bicycle access rates at individual rail stations in order to compare costs of bicycle-related investments with the cost of providing more automobile parking.⁽¹²⁴⁾ The BART system overall has established a goal of doubling the bicycle access for regional trips from approximately 4 to 8 percent by 2022, and the analysis tool is intended to aid in this effort.

Cleveland Bicycle on Bus Boardings

One of the more promising examples of modeling a subset of bicycle trips is in the Cleveland metropolitan region. An academic study conducted by researchers at Temple University estimated a regression model to identify what factors predicted bicycle-bus trips.⁽¹²⁵⁾

The Greater Cleveland Regional Transit Authority (GCRTA) through its operations had a 3-year dataset of over 160,000 trips with bicycle on bus boardings (BoBBs) between 2008 and 2011. The research study sought to answer the following two questions:

• On a typical day in the operations of a large public transit system, what determines the number of bus riders who decide to travel with their bicycles?
  
  
• What factors influence the highly variable number of BoBBs observed on GCRTA’s motor bus network in recent years?

Similar to the case of bicycle-rail transit travelers in Los Angeles County, the bicycle on bus travelers in the Cleveland, OH, region represent a subset of overall bicycle trips within a region. At the same time, since the public transportation operator has a means and desire to collect data in the course of operations, this example is unusual in that it includes a 100 percent sample of all bicycle trips within the subset of trips in question. For those buses with bicycle racks installed, it was possible for GCRTA to collect and maintain data on the BoBB trips. A summary of the key data obtained is shown in table 14.

Table 14. Summary of Cleveland, OH, BoBB data.⁽¹²⁴⁾

The study found that the number of BoBB travelers showed seasonal variation, with the lowest levels in the winter and highest levels in the summer. Use also varied by bus route.

While data on individual traveler attributes were not part of the dataset, the researchers were able to develop a model predicting what external factors had a significant influence on bicycle on bus trips. A summary of the major determining factors is shown in table 15.

The researchers concluded that weather was the most important variable in predicting the number of daily BoBBs. For every increase of one degree Fahrenheit in the mean daily temperature, there was an average of 2.21 more BoBBs. In addition, the occurrence of significant levels of precipitation was associated with an average of 22.06 fewer BoBBs.

Table 15. Example of bicycle use data linked to other traveler attributes: bicycle trip
purpose data from Los Angeles County.

SUMMARY AND CONCLUSIONS

This case study focused on identifying information and data to help understand the factors underlying traveler choices to use bicycling as an active transportation mode as well as the development of models of bicycle mode shift and usage patterns that may be incorporated in regional and operational travel demand forecasting frameworks. The examination included review of information and data collected by local areas in regional case studies consisting of the following urban metropolitan regions: Washington, DC, region, Southern California metropolitan region (SCAG region), San Francisco Bay Area, and the Cleveland, OH, region.

Data collected from four metropolitan regions confirm that bicycle travel is increasing both as an active transportation mode and as a means of travel demand management. However, bicycle travel supply and demand variables collected by local agencies vary considerably in quality and robustness. While leading edge travel demand modeling agencies are beginning to integrate bicycle use data into travel forecasting, simple trend extrapolation remains the primary approach. Significant data gaps limit the ability to fully incorporate bicycling choice and use in activity-based models of travel demand.

Examples from the four metropolitan study areas were presented, focusing on overall bicycle use and limited evidence for potential modal shift in connection with bike on transit service options and bikesharing plans. The importance of factors such as weather in bicycle use decisions is strongly evident through the available data. Recommended data needed to advance the state of the art and the practice were identified and presented.