Bike-GPS Project

Read the Final Report!

The "Bike-GPS: Understanding and Measuring Bicycling Behavior" project was completed in December 2008. The study, funded by the Robert Wood Johnson Foundation, began with a survey, then underwent a second phase of GPS data collection. On May 16, 2008, project lead Jennifer Dill presented many of the findings at the CTS Friday Transportation Seminar.

The key research questions of the study were: How does the built environment influence bicycling behavior; and what routes did they take?

The data Dr. Dill presented documents seven days worth of trips for 149 participants, covering 1,689 total trips. 45% of the participants were women. The data covers trip purposes, length, destinations, average speed, reasons for biking rather driving, and more. She presents in detail the different sets of data, how they are being processed, and in what ways the data can be used in the future.

In addition to addressing the two key research questions, the study also looks at what barriers are preventing people from bicycling more, and how to attract different groups, such as women, to cycling.

Click here to watch the stream of Dr. Dill's presentation. (Note: there is a seven minute delay before the video feed begins.)

Bike GPS study in the news:

• Portland Tribune - "Bike lanes work, PSU professor says"

• The Oregonian – "Secret habits of timid cyclists studied"

• Portland Tribune – "Why we bike – or don't"

• KGW – "Researchers try to get into the heads of bike commuters"

• The Oregonian – "Professor wants to follow you on your bicycle"

• BikePortland.org – "PSU study to track cyclists with GPS"

Oct. 2008 Update:

BikeGPS: Understanding and measuring bicycling behavior

Principal Investigator:      

Jennifer Dill, Ph.D.
Director, Center for Transportation Studies
Associate Professor, Nohad A. Toulan School of Urban Studies and Planning

Funded by: Robert Wood Johnson Foundation Active Living Research program, OTREC

Background

With rates of obesity, heart disease, and related health problems increasing in the U.S., many policy makers are looking for ways to increase physical activity in everyday life. Using a bicycle instead of a motor vehicle for a portion of everyday travel could help address these problems. Current rates of bicycling for transportation in major urban areas in the U.S. are very low, indicating that our current infrastructure and urban form is not attractive to potential bicyclists. However, over 60% of all personal trips are five miles or less in length – a reasonable distance to ride a bike – and nearly 40% are two miles or less. Moreover, bicycling is a popular form of recreation throughout the country. Given the potential for bicycling for utilitarian travel, why aren't more people cycling? There is very little research in the U.S. on bicycling. One area where data are lacking is on the effect of different types of infrastructure, such as bicycle lanes or paths, on bicycling. This research project aims to fill that data gap. The project used global positioning system (GPS) technology to record adults in the Portland, OR region rode their bicycles. This project uses that data to address the following questions:

• How often, why, when, and where do cyclists ride? How does this vary based upon rider characteristics?


• How do cyclists’ routes differ from the shortest network distance?


• do cyclists choose their routes? How do network characteristics (e.g. bike lanes, heavy traffic, traffic lights, etc.) influence those decisions? How do other factors, such as cycling with a child or the age or gender of the cyclist, influence these decisions?


• What is the difference in travel time between bicycling and driving?

Methodology

This study took place in the Portland, Oregon metropolitan region. The region’s population is about 1.6 million, with over one-third living within the City of Portland. The City of Portland has received attention for its commitment to providing bicycle infrastructure and other supportive policies, including being named by Bicycling magazine as the best city for bicycling in the country and by the League of American Bicyclists as one of only two platinum-level bicycle-friendly cities in the U.S. Bicycle infrastructure in the region includes about 550 miles of bike lanes on streets, 130 miles of separate bike paths, and 30 miles of “bicycle boulevards.” Bicycle boulevards are low-traffic residential streets, usually running parallel to a major road, that use traffic calming features to give priority to bicycles over motor vehicles. For example, barriers at some intersections force cars to turn while bikes can continue on a through path. Traffic signals allow bikes traveling on the boulevard to cross busy streets safely. The routes are signed and usually connect with other bicycle infrastructure, including lanes and bridge crossings.

The study collected GPS data from 166 bicyclists from March to November 2007. The participants were then selected using a stratified random sampling method. The objective was to get a range of types of cyclist (frequent vs. infrequent), home location (City of Portland versus the remainder of the region), age, and gender. The sample was not intended to exactly represent the general cycling public. If the sample had done so, there would likely be too few of certain types of cyclists to examine their behavior separately. Each participant used a specially programmed Garmin iQue to collect data. Once on and within view of three or more satellites, the unit recorded its position (“point”) and speed every three seconds. Each participant was asked to carry the unit for at least seven days. At the end of the test period, a project team member downloaded the GPS data and produced a series of maps of each trip. Participants then logged into a secure website to view their trips. For each trip and map, participant was asked about the accuracy of the data collected, along with questions about their route choice priorities. Data was corrected based upon survey responses.

Key Findings: Trip Frequency, Purpose, and Distance

The participants in this study were primarily regular bicyclists. While participating in the study, they made an average of 1.6 bicycle trips per day. Most participants (77%) made an average of two or few bicycle trips per day while they had the GPS device. Aside from riding back home, riding to work was the most frequent trip purpose (Figure 1). The vast majority of the bicycle travel recorded by the participants was for utilitarian purposes. Only five percent of the trips were purely for exercise. About 18% of the trips were for shopping, dining out, or other personal business, and 12% were for social/recreation purposes (such as going to the movies, the gym, or visiting friends).

Participants rode an average of 6.2 miles per day. The median bicycle trip distance was 2.8 miles (Figure 2). The average overall speed for the bicycle-only trips was 10.8 miles per hour (standard deviation 3.2), including times where the bicycle was not moving. Removing times when the GPS recorded zero velocity, the average speed was 11.1 miles per hour (standard deviation 3.2). The median speeds were also 10.8 and 11.1 mph, respectively.

Key Findings: Route Choices

When asked about their route choices and preferences for utilitarian trips, participants placed highest importance on minimizing distance and avoiding streets with lots of vehicle traffic (Table 1). Riding on a street with a bicycle lane was usually ranked third in importance, followed by reducing waiting time at stop lights and signs. These top four preferences reflect two sometimes conflicting sets of objectives. Most utilitarian bicyclists want to minimize their travel time. That is a fundamental assumption in travel demand modeling and planning for all travelers, no matter the mode (car, transit, etc). However, depending upon the network available, the quickest route for bicyclists may not satisfy their second major set of objectives, which is related to avoiding motor vehicle traffic.

When the bicyclists were riding for utilitarian purposes, they rode mainly on facilities with bicycle infrastructure. For over half (52%) of the miles bicycled on bicycle-only utilitarian trips were made on facilities with bicycle infrastructure, including lanes, separate paths, or bicycle boulevards. Over one-quarter of the mileage (28%) occurred on streets (arterials or minor streets) with bike lanes. An equal share (28%) of the mileage occurred on minor streets without bike lanes. These are typically low traffic volume, residential streets. Therefore, only 19% of the travel was on streets that would be expected to have high volumes of motor vehicle traffic and no separate facility for a bicycle. Bicycling for exercise purposes followed a different pattern. Nearly two-thirds (64%) of this travel was on roads without bike lanes. This reflects, in part, a significant amount of exercise travel in more rural areas. Overall, smaller shares of the exercise riding occurred on more urban facilities – streets with bike lanes and bicycle boulevards. The exception is the use of regional multi-use paths; 15% of the exercise travel occurred on these facilities.

The route of each non-exercise bicycle trip was compared to the route of the shortest path between the start and end of the trip. We developed shortest paths for 1,599 trips that met the following criteria: (1) made 100% on bicycle; (2) started and ended at least 200 feet apart; and (3) were not for exercise or an organized ride. Nearly all of the shortest paths were shorter in distance than the observed bicycle trips. The average difference in distance between the actual bicycle trip and the shortest path between the same origin and destination was 0.95 miles, though the median was 0.27 miles. The difference between the shortest path and the observed route increases with trip distance. Looking only at the trips 10 miles or shorter in distance, the median difference between the observed route and the shortest path was just under a quarter mile (0.24 miles). This represents about an extra 1.5 minutes of travel, given the average speed on the trips.

Comparing the facilities used for the observed trips to the shortest paths reveals some preferences in facility type. Bicyclists spent a higher share of their miles on facilities with bicycle infrastructure and on low traffic streets than the shortest paths predicted. In particular, they rode 14% of their miles on paths, compared to 6% of the miles for the shortest paths, a difference of eight percentage points. Many bicyclists are avoiding arterials and highways that do not have bike lanes. Those facilities represented 19% of the bicyclists’ miles, compared to 36% of the shortest path miles. This also indicates that the major streets without bike lanes are often part of the shortest path between two points.

The stated and revealed preference data comparing men and women found that women are more likely to prefer to bicycle on low traffic streets and bicycle boulevards, and less likely to prefer riding on busier streets with bike lanes. Similarly, less experienced bicyclists placed higher importance on factors that make the trip easier – routes with less traffic and requiring less physical effort. They were more likely to go out of their way to use multi-use paths and less likely to divert from the shortest path to use a street with a bike lane.

Conclusions

The study has several policy implications. The findings and analysis to date indicate that bicyclists do use and value the infrastructure provided (lanes, paths, and boulevards). Well-connected low-traffic streets, bicycle boulevards, and separate paths may be more effective than bicycle lanes on busy streets at getting more women and new adults bicycling. A well-connected street network also appears to be important, both for minimizing travel distances and allowing for an efficient network of low-traffic streets and bicycle boulevards.

While the data indicate that bicycle boulevards and paths may be more effective than bike lanes on arterials at encouraging more bicycling among groups of people who currently do not bicycle much, the importance of bike lanes should not be ignored. Over one-quarter (28%) of all of bicycle travel occurred on streets with bike lanes. The data indicate that adding bike lanes to more arterials might reduce travel times and distances, particularly for experience bicyclists. This could increase bicycle travel. Finally, for many short trips (3 miles or less), the bicycle was time- competitive with the automobile. Shorter trips are most likely to occur in areas with a greater mix of land uses and higher network connectivity, making potential origins and destinations closer. Therefore, policies that promote these features are likely to support more bicycling for transportation.

While this study collected more detailed information on bicycling behavior than any of the other studies found in the literature, there are still many limitations. One limitation is that the study was only conducted in one region. Caution must always be used when conclusions based upon data from one area are used to make recommendations for another area. In addition, the bicyclists participating in this study do not represent all bicyclists. The sample included primarily regular bicyclists who bicycle mainly for utilitarian purposes. This makes it more difficult, though not impossible, to draw conclusions about the behavior of infrequent cyclists. Because the study was intended to examine revealed preferences, opinions and preference of non-cyclists are not addressed. Finally, the GPS units presented a few limitations, including potential errors when linking the GPS point data to the network. In addition, some bicycle travel was not recorded. Based upon participant-provided information, the total number of bicycle trips may be underreported by about eight percent. Another potential issue is that participation in the study and carrying the GPS device could influence their behavior. Survey responses indicated that this may have been a minor problem.

Next Steps

We are currently analyzing the data in more detail. We have also partnered with Metro to use the data to help improve their ability to model and predict bicycle behavior.