A Comparative Analysis of Bicycle Lanes Versus Wide Curb … · 2000-03-01 · Subcontractor: Bicycle Federation of America 16. Abstract This report is a comparative analysis of bicycle

A Comparative Analysis of BicycleLanes Versus Wide Curb Lanes:Final ReportPUBLICATION NO. FHWA-RD-99-034 DECEMBER 1999

Research, Development, and TechnologyTurner-Fairbank Highway Research Center6300 Georgetown PikeMcLean, VA 22101-2296

Technical Report Documentation Page 1. Report No.

FHWA-RD-99–0342. Government Accession No. 3. Recipient's Catalog No.

4. Title and Subtitle

A COMPARATIVE ANALYSIS OF BICYCLE LANES VERSUS WIDE CURB LANES: FINAL REPORT

5. Report Date

6. Performing Organization Code

7. Author(s) William W. Hunter, J. Richard Stewart, Jane C. Stutts, Herman H. Huang, and Wayne E. Pein

8. Performing Organization Report No.

9. Performing Organization Name and Address

University of North CarolinaHighway Safety Research Center730 Airport Road, CB #3430Chapel Hill, NC 27599

10. Work Unit No. (TRAIS)

11. Contract or Grant No.

DTFH61-92-C-00138

12. Sponsoring Agency Name and Address

Office of Safety and Traffic Operations Research & DevelopmentFederal Highway Administration6300 Georgetown PikeMcLean, VA 22101-2296

13. Type of Report and Period Covered

Final Report March 1995 - May 199814. Sponsoring Agency Code

15. Supplementary Notes

Contracting Officer’s Technical Representative (COTR): Carol Tan Esse, HSR-20Subcontractor: Bicycle Federation of America

16. Abstract

This report is a comparative analysis of bicycle lanes (BLs) versus wide curb lanes (WCLs). The primaryanalysis was based on videotapes of almost 4,600 bicyclists (2,700 riding in BLs and 1,900 in WCLs) in the cities ofSanta Barbara, CA, Gainesville, FL, and Austin, TX, as the bicyclists approached and rode through eight BL andeight WCL intersections with varying speed and traffic conditions. The intent was to videotape bicyclists whoregularly ride in traffic. The videotapes were coded to learn about operational characteristics (e.g., intersectionapproach position and subsequent maneuvers) and conflicts with motor vehicles, other bicycles, or pedestrians. Aconflict was defined as an interaction between a bicycle and motor vehicle, pedestrian, or other bicycle such that atleast one of the parties had to change speed or direction to avoid the other. Both bicyclist and motorist maneuvers inconflict situations were coded and analyzed. This covered maneuvers such as a bicyclist moving incorrectly from thebicycle lane into the traffic lane prior to making a left turn, or conversely, a motor vehicle passing a bicyclist and thenabruptly turning right across its path. Bicyclist experience data were also collected separately from the videotaping ateach of the 16 data collection sites in each city through use of a short oral survey. Slightly more than 2,900 surveyswere completed. These data were analyzed to learn about the age, riding habits, and experience levels of thebicyclists riding through these intersections. Bicycle-motor vehicle crash data were also analyzed to determine ifthere were parallels to the videotape data.

In addition to this final report, there is a separate report (FHWA-RD-99-035) containing a synopsis of the keyfindings of the final report and recommended countermeasures, as well as a guidebook (FHWA-RD-99-036) aboutinnovative bicycle accommodations. 17. Key Words:

Bicycle lane, wide curb lane, bicycle operations, bicyclemaneuvers, conflicts

18. Distribution Statement

No restrictions. This document is available to the public through the National Technical InformationService, Springfield, Virginia 22161

19. Security Classif. (of this report)

Unclassified20. Security Classif. (of this page)

Unclassified21. No. of Pages

vii, 10422. Price

Form DOT F 1700.7 (8-72) Reproduction of form and completed page is authorized

Table of Contents

Chapter 1 - Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1Objective and Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2Brief Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Bicycle lanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Wide curb lanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4Other facilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5Intersection treatments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

Organization of the Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Chapter 2 - Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11City Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Santa Barbara, California . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12Gainesville, Florida . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12Austin, Texas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Site Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12Videotaping of Bicyclists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

Bicyclist Experience Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15Coding of Videotape Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16Creation of Project Database . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16Coding and Analysis of Crash Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

Chapter 3 - Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19Bicyclist Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Videotape data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19Bicyclist experience survey results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

Midblock Actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23Movements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23Midblock spacing between bicycles and motor vehicles . . . . . . . . . . . 25Behaviors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

Intersection Actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29Movements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

iv Table of Contents

Behaviors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34Conflicts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

Midblock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36Intersections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39Midblock and intersection combined conflict rates . . . . . . . . . . . . . . . 43

Statistical Modeling of Conflicts Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43Analysis of midblock bike/motor vehicle conflicts . . . . . . . . . . . . . . . . 43Intersection car/bike /motor vehicle conflicts . . . . . . . . . . . . . . . . . . . . 46Reanalysis of conflicts based on data from more “typical” sites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

A Clinical Analysis of High Conflict BL and WCL Sites . . . . . . . . . . . . . . 50Midblock conflicts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50Intersection conflicts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

Examination of Serious Conflicts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Serious midblock conflicts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Serious intersection conflicts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

Comparisons with Crash Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

Chapter 4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69Summary of Main Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

Bicyclist characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69Midblock movements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69Statistical modeling of spacing between bicycles and motor vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69Intersection movements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70Midblock conflicts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71Intersection conflicts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71Statistical modeling of conflict data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72Clinical examination of high conflict rate sites . . . . . . . . . . . . . . . . . . . 72Clinical examination of serious conflicts . . . . . . . . . . . . . . . . . . . . . . . . 73Comparisons with crash data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

Further Comment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73Level of experience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73Wrong-way riding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74Turning and other maneuvers at intersections . . . . . . . . . . . . . . . . . . . 74Conflicts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

Recommended Countermeasures for Certain High Conflict Rate Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

Parked motor vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

Table of Contents v

Driveways and intersecting streets . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76Additional lanes at intersections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

Appendix A - Experience Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

Appendix B - Coding Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

vi Table of Contents

List of Figures

Figure 1. Typical bike lane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Figure 2. Typical wide curb lane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Figure 3. Oregon bike lane standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Figure 4. Oregon wide curb lane standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

Figure 5. Profiled marking at a bus stop to separate bicyclists from bus passengers . . . . . . . 5

Figure 6. Oregon rumble strip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Figure 7. BL dashed to intersection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Figure 8. Colored bicycle crossing in Montreal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Figure 9. A European raised and painted bike path (crossing) . . . . . . . . . . . . . . . . . . . . . . . . . 7

Figure 10. T-intersection marking in Denmark . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Figure 11. Modern roundabout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

Figure 12. Bike box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

Figure 13. Recessed stop line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

Figure 14. Map of project cities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

Figure 15. Typical data collection setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

Figure 16. BL and WCL intersection types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

Figure 17. Standard no parking signs for bike lanes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

Figure 18. Double striped BL with parking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

Figure 19. Combination BL with parking T’s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

Figure 20. Typical conflict situations at a driveway crossing a sidewalk . . . . . . . . . . . . . . . . 76

Figure 21. Example of traffic splitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

Figure 22 Dashed BL stripe at right-turn situation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

Figure 23. European bike box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

Table of Contents vii

List of Tables

Table 1. Videotaped bicyclist characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

Table 2. Experience survey bicyclist characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

Table 3. Experience survey riding characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

Table 4. Midblock actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

Table 5. Model for distance from curb for bikes not being passed . . . . . . . . . . . . . . . . . . . . . 26

Table 6. Model for distance from curb for bikes being passed . . . . . . . . . . . . . . . . . . . . . . . 27

Table 7. Model for distance between bikes and passing motor vehicles . . . . . . . . . . . . . . . . 27

Table 8. Midblock behaviors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

Table 9. Intersection actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

Table 10. Intersection movements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

Table 11. Intersection turning information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

Table 12. Intersection behaviors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

Table 13. Midblock conflict information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

Table 14. Midblock conflict bicycle and motor vehicle actions . . . . . . . . . . . . . . . . . . . . . . . 38

Table 15. Intersection conflict information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

Table 16. Intersection conflict bicycle and motor vehicle actions . . . . . . . . . . . . . . . . . . . . . 42

Table 17. Example of model results for midblock conflicts . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

Table 18. Midblock conflicts model including bike lane width . . . . . . . . . . . . . . . . . . . . . . . . 44

Table 19. Bike lane width by midblock conflicts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Table 20. Intersection conflicts by facility type and bicyclist movement . . . . . . . . . . . . . . . . 46

Table 21. Intersection conflicts by intersection type for straight through and right turning bicyclists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

Table 22. Most frequent occurring crash types rank ordered in each of the three study sites, based on 1995 police-reported crash data . . . . . . . . . . . . . . . . . . . . . . 67

Chapter

1 Introduction

Figure 1. Typical bike lane.

BackgroundA number of recent events renders a

study of bicycle facilities as appropriate andtimely. The passage of the 1991 IntermodalSurface Transportation Efficiency Act(ISTEA) legislation meant a variety of fundscould be more readily used by local and stateofficials to plan and build such facilities. Indications are that many governments andagencies have taken advantage of the oppor-tunity. Publication of the National Bicyclingand Walking Study in 1994 with the U.S.Department of Transportation (USDOT)goals of doubling the percentage of tripsmade by bicycling and walking andsimultaneously reducing by 10 percent thenumber of bicyclists and pedestrians injuredor killed in traffic crashes adds emphasis tothe need to accommodate non-motoristswith well-designed facilities. User surveyrespondents have clearly stated that morefacilities are desired and will increase theamount of travel by bicycle.

In addition to the recent activitiesmentioned above, during the past 20 yearsbicycle facilities have been planned andimplemented in communities nowconsidered as pro-bicycling, including Seattle,WA; Davis and Palo Alto, CA; Madison, WI;Eugene and Corvallis, OR; Boulder, CO;Gainesville, FL; Tucson, AZ; and others. These communities tend to have a localbicycle coordinator and bicycling advisorycommittee in place. Not all of the

implemented facilities have been ideallyconstructed. However, what has tended tooccur in all of these communities is thatmotorists have adapted to bicyclists wherebicycle facilities have been implemented,and most facilities appear tofunction effectively, although not withoutsome problems. What has not beendone and reported to the bicycling andtraffic engineering community is a thoroughevaluation of the various kinds of facilities incommunities like these.

Given the information presented above,considerable effort was devoted to decidingwhat kinds of bicycle facilities should beevaluated in this project. A long-standingissue in the bicycling community centers onwhether bicycle lanes or wide curb lanes are

preferable. A bicycle lane (BL) is a portionof a roadway that has been designated bystriping and pavement markings for thepreferential or exclusive use of bicyclists(figure 1). BL width is normally in the range

of 1.2 to 1.8 m. A wide curb lane (WCL) isthe lane nearest the curb that is wider than astandard lane and provides extra space sothat the lane may be shared by motorvehicles and bicycles (figure 2). Thus, WCLs

2 Chapter 1

Figure 2. Typical wide curb lane.

may be present on normal two-laneroadways or on multilane roadways. A desirable width for WCLs is 4.3 m. Laneswider than 4.6 m sometimes result in theoperation of two motor vehicles side byside. Many bicyclists report feeling saferwhen riding on BLs, while BL opponentsventure that these facilities make it difficultfor bicyclists to handle turning maneuvers atintersections, especially left turns. WCLadvocates believe that these wider lanesencourage cyclists to operate more likemotor vehicles and thus lead to more correctmaneuvering at intersections. Bothperspectives have merit and should beaddressed in any evaluation of these facilities.Because a WCL is a wider-than-normaltraffic lane that is shared with motor vehicles,some do not refer to this layout as a bicyclefacility. However, for purposes of this study,both BLs and WCLs will be referred to asbicycle facilities.

The debate over whether BLs or WCLsare preferable has been heated for manyyears and is not unlike the seat belts versus airbags dichotomy that prevented a concertedapproach to the promotion of occupantrestraints in the United States in the 1970sand 1980s. While both BLs and WCLs areacceptable facilities in many locations, thedebate has sometimes forced decisionmakers to choose which facility type theyprefer, to the exclusion of the other. Morebicycle facilities might be in place in thiscountry except for this long-standing divisionof opinion. Because of the interest in BLsand WCLs, it was decided to make thesefacilities the focus of this project, with anemphasis on operations and interactionsbetween bicyclists and motorists atintersections.

Objective and ScopeThe primary objective of the current

study was a comparative analysis of BLs

versus WCLs. Bicyclists riding in either a BLor WCL were videotaped as theyproceeded through BL and WCLintersections with varying speed and trafficconditions in three U.S. cities. Thevideotapes were coded to learn aboutoperational and safety characteristics.Operational characteristics pertained to howbicyclists maneuvered through the sites,while safety characteristics pertained toconflicts with motor vehicles, other bicycles,or pedestrians. A conflict was defined as aninteraction between a bicycle and motorvehicle, pedestrian, or other bicycle such thatat least one of the parties had to changespeed or direction to avoid the other.Exposure/experience data were alsocollected separately from the videotaping ateach of the data collection sites in each citythrough use of a short oral survey.Information was obtained about the age,gender, race, helmet use, levels ofexperience, etc., of the bicyclists ridingthrough these intersections.

A secondary study objective was todevelop a guidebook of current innovativebicycling activities, with a primary focus onintersection treatments that pertained to BLsand WCLs. The innovative treatment“shopping list” included advance stop bars(often called bike boxes) where bicycles areallowed to proceed ahead of motor vehicletraffic at an intersection; painting a modifiedversion of the bicycle logo near the curb ina WCL to alert drivers that bicycles wouldbe operating in this space; colored

Chapter 1 3

Figure 3. Oregon bike lane standards.Source: Oregon Bicycle and Pedestrian Plan,1995

pavement designating the appropriate pathfor the bicycle through an intersection; trafficcalming measures like diagonal diverters andspeed humps with "slots" in the pavementfor bikes and buses; bicycle traffic signals;combination bus/bike lanes; differenttechniques for separating bike lanes; andothers. The Bicycle Federation of America(BFA) was responsible for locating therelevant examples and developingappropriate descriptions. This guidebook isone of the final products of this contract.

Brief Literature Review The National Bicycling and Walking

Study (1994) established USDOT goals ofdoubling the percentage of trips made bybicycling and walking, while simultaneouslyreducing the number of bicyclists andpedestrians injured or killed in traffic crashesby 10 percent. To realize these goals, ourtransportation system needs better ways toaccommodate bicycling and walking. The1991 ISTEA allowed cities and States tospend Federal transportation funds onfacilities for bicycling and walking. Local

bicycle planners can choose amongconventional roadway treatments such as

BLs and WCLs, and more innovativetreatments such as modern roundabouts andadvanced stop bars (popularly referred to asbike boxes in the United States).

Bicycle lanesA bicycle lane is a section of the roadway

that is delineated from the adjacent motorvehicle travel lane by a stripe. BLs areusually along the right edge of the roadway,but may be designated to the left of parkingor right-turn lanes. Recommended widthsfor bicycle lanes (figure 3) are generally 1.2to 1.8 m

(see, for example, North Carolina DOT,1994; New Jersey DOT, 1995; OregonDOT, 1995). A Dutch design manual(C.R.O.W., 1994) suggests 2.0 m so thatbicyclists can ride side-by-side, and anotherDutch study (Botma and Mulder, 1993) callsfor a width of 2.5 m when the 1-hour peakvolume exceeds 150 bicycles to allowbicyclists to pass one another. In anationwide survey of U.S. cyclists takenmany years ago, 85 percent considered BLs

4 Chapter 1

wider than 1.8 m to be adequate; only 41percent considered BLs narrower than 1.5 mto be adequate (Kroll and Sommer, 1976).

Ninety-three percent of U.S. cyclistsusing BLs thought the street was safer withthe lanes than without them, although therewas no conclusive evidence that they actuallyimproved cyclist safety (Kroll and Sommer,1976). Two other studies credited BLs withreducing bicycle-motor vehicle crashes bymore than half in Corvallis, Oregon, and bytwo-thirds in Eugene, Oregon (Ronkin, nodate; City of Eugene, 1980). The installationof BLs along a one-way arterial pair inMadison, Wisconsin, was associated with asignificant increase in the number of crashesassociated with turning movements;however, crashes decreased sharply after thefirst year of operation (Smith and Walsh,1988).

A manual prepared for the FederalHighway Administration (FHWA) usesvarious factors to make recommendationsfor roadway design treatments foraccommodating bicyclists. The factorsinclude definitions of design bicyclists, typeof roadway, traffic volume, average motorvehicle operating speeds, traffic mix, on-street parking, sight distance, and number ofintersections and entrances. BLs are oftenrecommended when most bicyclists on theroute are less experienced (Wilkinson, Clarke,Epperson, and Knoblauch, 1994).

In Denmark, roadway stretches withBLs or bicycle paths tended to have a lowerfrequency of crashes involving cyclistcasualties than stretches without lanes orpaths (Herrstedt et. al., 1994). Anotherevaluation of BLs in Denmark found nochange in the number of overall crashes orbicycle-motor vehicle crashes at signalizedjunctions, but did find an increase in thenumber of bicycle-motor vehicle crashes atpriority junctions (unsignalized junctions,usually signed, where one roadway has

priority over the other). There was also areduction in all crashes on stretches (thesections of roadway between intersections)(Jensen, 1997).

The presence of a stripe separatingbicyclists and motorists (as with a BL orpaved shoulder) has been shown to result infewer erratic driver maneuvers, morepredictable bicyclist riding behavior, andenhanced comfort levels for both groups ofusers (Harkey and Stewart, 1997; Kroll andRamey, 1977; McHenry and Wallace, 1985).The principal findings from the 1997 studyof bicyclists riding in midblock situations byHarkey and Stewart for the Florida DOTwere the following:

! The separation distance betweenbicyclists and motorists was about 1.8 m andvaried only a small amount by facility type(BLs, WCLs, and paved shoulders).

! The distance between the bicyclistand the edge of the roadway wasconsiderably greater on BL and pavedshoulder facilities (0.8 m) than on WCLs (0.4m).

! Motor vehicles moved about 0.4 mfurther to the left when passing a bicyclist onWCLs compared with BL and pavedshoulder facilities.

! Motor vehicle encroachment into theadjacent lane to the left when passing abicycle was much greater on WCLs (22.3percent) than on BL (8.9 percent) and pavedshoulder facilities (3.4 percent).

! For a BL facility, the change in lateralposition of the motor vehicle when passing abicycle was approximately 0.3 m regardlessof BL width.

Wide curb lanesWide curb lanes can accommodate both

bicyclists and motorists and allow sufficientroom for passing. These are sometimesdesignated when right-of-way constraints

Chapter 1 5

Figure 4. Oregon wide curb lanestandards.Source: Oregon Bicycle and Pedestrian Plan,1995

preclude the installation of “full width” BLs. WCLs should be 4.0 to 4.6 m wide (figure 4) to provide enough width for lane sharing but not so much width that motorists formtwo lanes at intersections (McHenry andWallace, 1985). Wilkinson et al. (1994)recommend WCLs in many kinds ofroadway situations where most bicyclists areexperienced riders. The Harkey and Stewartstudy (1997) performed for the FloridaDOT showed that motorists encroach intothe adjacent lane of traffic significantly moreoften when WCLs are used as comparedwith BLs.

At present there appears to be a trendtoward more use of BLs at the State andlocal levels, perhaps due to preferences citedby bicyclists. (See, for example, a statement inthe Florida Bicycle Facilities Planning and DesignManual (Florida DOT, 1995) that WCLsshould be used as a last resort because “onlyfive percent of bicyclists feel comfortableusing these facilities.”) On the other hand, theNorth Carolina DOT (1994) refers to a1970s FHWA publication to list principalproblems with BL applications in its bicyclefacilities planning and design guidelines,including: (1) provision of inadequate lanewidth or use of unrideable street surface asthe BL area, (2) abrupt termination of lanesat hazard or constraint situations, creating afacility that leads bicyclists to a trap, as well astransitions that force awkward bicyclistmovements at other termination points, (3)use of non-standard and poorly visible lanedemarcation signs and markings that createuncertainties in motorist and bicyclistunderstanding of lane presence and purpose,(4) lane configuration and lane use ordinancesthat prevent the bicyclist from establishingproper position with respect to motorvehicle traffic at intersections, as well as formid-block turns into driveways, and (5) laneuse ordinances that conflict with reasonablebicyclist desires to leave the lane in order to

avoid road hazards or to overtake otherbicyclists, motor vehicles, or pedes- triansoccupying the bike lane.

Other facilitiesA combined bus and bike lane in Toronto

was found to increase bicycle traffic andlower accident rates. More than 75 percent

of riders felt safer riding along the new busand bike lane (Egan, 1992). Combinedbus/bike lanes should be 3.1 to 3.7 m wide(Harrison, Hall, and Harland, 1989). Withbus/bike lanes, the potential exists forconflicts between buses and bikes at thecrossing points. One design places a bicyclelane to the right of the through traffic lanesand to the left of the bus and right-turn lane. This allows bicyclists to ride without leap-frogging past stopped buses (Berchem andSomerfeld, 1985).

Other Danish designs are aimed atreducing conflicts between bicyclists and buspassengers due to the high incidence ofcrashes in bus stop areas. These designsinclude: (1) a pedestrian crossing combined withprofiled markings (figure 5); (2) a profiledmarking on the offside of the bicycle path; and (3) apainted pattern with a visual brake. The intentwas to use pavement markings to highlight the conflict area at bus stops and move

6 Chapter 1

Figure 6. Oregon rumble strip.Source: Oregon Bicycle and Pedestrian Plan,1995

Figure 5. Profiled marking at abus stop to separate bicyclistsfrom bus passengers.Source: Safety of Cyclists in UrbanAreas, 1994

bicyclists away from the passengers alightingfrom buses. The proportion of cyclists whowait for bus passengers to cross the bicyclepath did not change with any of the designs.

All three designs reduced the speed ofcyclists when there was a bus at the bus stop. The distance from where cyclists first reactedto a bus to the nearest conflict pointincreased. The number of serious conflictsdecreased with the painted pattern(Herrstedt, 1994).

The expected number of bicycle-motorvehicle crashes is much lower when bicyclistsride along paved highway shoulders than whenbicyclists and motorists share the travel lanes(Khan and Bacchus, 1995). Operationally ithas been shown that paved shouldersessentially function like BLs with respect tobicycle and motor vehicle interactions (i.e.,the stripe separating bicyclists from motoristsresults in a lower risk environment for bothmodes of travel (Harkey and Stewart, 1997).

One potential hazard is that an inattentiveor sleepy driver may drift off the roadwayonto the shoulder and strike a bicyclist ridingon the shoulder. Although there isconsiderable debate regarding the mosteffective design, a shoulder rumble strip (figure6) is an efficient device to waken drivers who are drifting off the roadway(Garder, 1995). On highways with posted

speeds of less than 100 km/h, a minimumwidth of 1.5 m of paved shoulder issufficient space to accommodate both arumble strip and bicyclist travel (Khan andBacchus, 1995).

In the Netherlands, separate bicycle pathsare recommended when motor vehiclespeeds exceed 50 km/h or when trafficvolumes exceed 1,200 vehicles per hour. One-way cycle paths should be at least 1.8 mwide, and two-way cycle paths should be 2.8m wide (Diepens and Okkema TrafficConsultants, 1995). In an earlier survey ofU.S. cyclists, bike paths were rated as beingsafer than bike lanes, and most thought thatpaths wider than 2.8 m were “good” (Krolland Sommer, 1976).

Intersection treatmentsIntersections and intersection-related locations

account for 50 to 70 percent of bicycle-motor vehicle crashes (Hunter, Stutts, Pein,and Cox, 1996). Countermeasures such asgrade separation can be adopted to reduceintersection conflicts between bicycles andmotor vehicles. More than 50 interchanges

Chapter 1 7

Figure 7. BL dashed to intersection.

Figure 8. Coloredbicycle crossing inMontrealSource: Pronovost andLusignan, 1996

Figure 9. A European raised andpainted bike path (crossing).Source: Oregon Bicycle and Pedestrian Plan,1995

in Beijing, China, provide for gradeseparation between bicyclists and motorvehicles (Liu, Shen, and Ren, 1993; Burden,Wallwork, and Guttenplan, 1994).

Grade separation is expensive, though,and thus lower cost, at-grade treatments aremore widely used. For example, bicycle pathcrossings of roadways can be offset awayfrom the intersection to enhance bicyclists’view of motorists (NCHRP, 1976). On onestreet in Cupertino, California, a BL stripewas dashed to guide cyclists riding in the BL(next to the curb) to the left of right-turningvehicles (Grigg, no date). The Florida DOT

(1995) is one of a number of State DOTsrecommending that BLs be discontinued ordashed in advance of an intersection, so thatbicyclists and motorists can merge (figure 7). Right-angle bicycle crossings with good sightlines are recommended at intersections.

At five intersections in Montreal, coloredbicycle crossings were installed (figure 8), with the pavement colored blue at bicycle pathcrossing points. After the markings werepainted, bicyclists were more likely to obeystop signs and to stay on designated cyclepath crossings. Improved bicyclist behaviorled to a decline in the level of conflictbetween cyclists and motorists (Pronovostand Lusginan, 1996). In Denmark, themarking of bicycle travel paths (raised overpasses) at

signalized junctions resulted in 36 percentfewer accidents with motor vehicles and 57percent fewer cyclists who were killed orseverely injured (Jensen, 1997). Some ofthese crossings also used blue color on thepavement.

A raised and painted bicycle path (crossing)(figure 9) introduced at 44 intersections in

Gothenburg, Sweden, reduced motorvehicle speeds (by 35 to 40 percent for right-turning motor vehicles) and increased cyclistspeeds (by 10 to 15 percent). The safetyimprovement was estimated by using aquantitative model and by surveying

8 Chapter 1

Figure 10. T-intersection marking inDenmark.Source: Safety of Cyclists in Urban Areas, 1994

Figure 11. Modern roundabout.Source: Innovative Bicycle Accommodations,in press

bicyclists and experts. The model estimatedthe combined effect of lower motoristspeeds and higher bicyclist speeds to be a 10percent reduction in the number of bicycle-motor vehicle crashes. Bicyclists perceived a20 percent improvement in safety after thebicycle path was raised and painted. Expertsestimated a 30 percent improvement insafety. However, the authors suggested thatthe total numbers of crashes should beexpected to increase due to a 50 percentincrease in bicyclists using the improvedcrossings (Leden, 1997). A follow-on paperusing a Bayesian approach for combining theresults of the model and surveys estimated arisk reduction of approximately 30 percentattributable to the raised and painted crossing(Gårder, Leden, and Pulkkinen, 1998).

A different report based on a review ofthe literature, interviews with bicyclists, andexpert opinion concluded that the crash riskwould increase by about 40 percent when abicycle path is added at a signalizedintersection (Leden, Gårder, and Thedéen,

1993).Profiled pavement marking aimed at

reducing the lateral distance betweenmotorists and cyclists and increasingattentiveness between bicyclists and motoristschanged motorist and cyclist behavior at T-intersections and four-way intersections inDenmark (figure 10).

Profiled markings were placed to guideapproaching cyclists closer to the travel lanes. At the intersection, the cyclists were guidedaway from the travel lanes. More motoristsadapted their speeds to the cyclists’ speedsand stayed behind the stop line. Motoristswere less likely to turn right in front ofcyclists. At T-intersections, cyclists becamealert earlier (Herrstedt et al., 1994).

Many bicycle-motor vehicle crashes atroundabouts occur when motorists cut in frontof bicyclists or fail to yield the right-of-way. Small roundabouts with flared entry roadsare the most dangerous design, whereaslarge roundabouts are the most feared bybicyclists. Crash rates for bicyclists atroundabouts in the United Kingdom aretwo to three times higher than thoseexperienced by bicyclists at traffic signal-controlled intersections. Mini-roundaboutshave a much better crash record, similar tothat of four-way traffic, signal-controlledintersections. Lane markings, warning signs,sharper entry angles, and visibilityimprovements have helped reduce bicyclistcrashes in roundabouts (figure 11). Smaller roundabouts, where motorists cannotovertake bicyclists, are recommended (Allottand Lomax, 1993; Balsiger, 1995). In acomparison of Swedish, Danish, and Dutchroundabouts, a separate cycle path with an

Chapter 1 9

Figure 13. Bike box.

Figure 13. Recessed stop line.Source: Safety of Cyclists in Urban Areas, 1994

ordinary cycle crossing was found to be thesafest design when motor vehicle trafficflows were large, compared with a cycle lanewithin the roundabout or no specific bicyclefacility at all (Brüde and Larsson, 1996). Results were based on observed versusexpected crashes, with expected crashesobtained from a predictive model. Theauthors noted that data were limited. Aroundabout on the University of Californiaat Davis campus allows five times as manycyclists to pass through, compared withwhen the intersection was controlled by stopsigns, and bicycle crashes that result in injury

are rare (Burden, Wallwork, and Guttenplan,1994).

On roads with marked BLs, an advancedstop line (ASL) or bike box may be placed inthe BL at a signalized intersection. The bikebox is placed in front of the motor vehiclestop line to give bicyclists a space to wait infront of motorists and to allow them to passthrough first when the green phase starts. Being in the box makes bicyclists morevisible to motorists and can reduce conflictswith turning motor vehicles (figure 12).Under a single-signal design, one trafficsignal is placed at the box. With a two-signaldesign, used in the United Kingdom,motorists are held by a red signal, while aspecial green signal directs bicyclists ahead tothe box (U.K. Department of Transport,1993; Zegeer et al., 1994).

Bike boxes have worked successfully onroads in the United Kingdom with up to1,000 vehicles per hour passing through theintersection. Wheeler (1995) and Wheeler etal. (1993) monitored schemes at nineintersections. Two-thirds or more of thebicyclists used the cycle lane and the reservedwaiting area. Signal violations by bicyclistswere less than 20 percent. As many as 16percent of motorists encroached into theBLs. At one intersection, more than half ofall lead motorists encroached into thecyclists’ reserved waiting area. The single-signal design is likely to be as effective as thetwo-signal design if a mandatory cycle laneand a distinctly-colored road surface in thecyclist areas are used. In Denmark, recessedstop lines (figure 13) for motor vehicles

significantly reduced the number of crashesbetween right-turning motorists and cyclistsgoing straight through the intersection(Herrstedt et al., 1994).

Organization of the ReportThe results of this research are provided

in three documents. This final reportcontains the comprehensive results pertainingto operations and conflicts. A researchsummary provides planners, engineers, andpedestrian/bicycle coordinators a tool with

10 Chapter 1

information about the operational and safetyproblems associated with BLs and WCLs,along with some suggested countermeasuresfor problem situations. A third document isa guidebook of current innovative bicyclingpractices. The guidebook is wide-rangingand covers topics such as on-street designsapplicable to BLs or WCLs, retrofittingstreets for bikes, use of colored pavement,bicycle traffic signals, and others.

In this final report, chapter 2 contains adescription of the project researchmethodology and data collection techniques.Chapter 3 focuses on the comparativeoperational and safety differences betweenBLs and WCLs. Chapter 4 summarizes themain results and offers discussion about keyissues.

Chapter

2 Methods

Austin

Santa Barbara

Gainesville

Figure 14. Map of project cities.

OverviewBicyclists riding in either a BL or WCL

were videotaped as they approached andproceeded through eight BL and eight WCLintersections with varying speed and trafficconditions in three cities. Approximately4,600 bicyclists were videotaped in the threecities (2,700 riding in BLsand 1,900 in WCLs). Thevideotapes were coded tolearn about operationalcharacteristics (e.g.,intersection approachposition and subsequentmaneuvers) and conflictswith motor vehicles, otherbicycles, or pedestrians. Aconflict was defined as aninteraction between a bicycleand motor vehicle,pedestrian, or other bicyclesuch that at least one of theparties had to change speedor direction to avoid theother. Both bicyclist andmotorist maneuvers inconflict situations werecoded and analyzed. Thiswould cover maneuverssuch as a bicyclist movingincorrectly from the bicyclelane into the traffic lane prior to making aleft turn, or conversely, a motor vehiclepassing a bicyclist and then abruptly turningright across its path. Bicyclist experience datawere also collected separately from thevideotaping at each of the 16 data collection

sites in each city through use of a short oralsurvey. Slightly more than 2,900 surveys werecompleted. These data were analyzed tolearn about the age, riding habits, andexperience levels of the bicyclists ridingthrough these intersections.

City SelectionConsiderable effort in the early part of

the project was spent in identifying possiblecities for study. Candidates were narrowedand visits made to Santa Barbara, CA; thePalo Alto area of CA; Madison, WI;Gainesville, FL; and Austin, TX. Based onkey factors such as amount and type offacilities, number of riders, willingness andeagerness of local contacts to participate, andwindows of opportunity (i.e., climate) forvideotaping, Santa Barbara, CA, Gainesville,

FL, and Austin, TX, were selected as primaryproject cities (figure 14). These were spreadgeographically across the United States andprovided for a good comparative analysis. More detail about each follows.

12 Chapter 2

Santa Barbara, California This city has a population of about

90,000 with another 80,000 located inneighboring communities. This is an oldercity, and many of the streets have low motorvehicle speeds which, in turn, is good forbicycling. About 3 to 5 percent ofcommuting trips are estimated to be madeby bicycling. The University of California atSanta Barbara is located about 11 km westof downtown (in Isla Vista) and may bereached by bicycle through an off-roadfacility. About 24 km of bicycle lanes arepresent and probably at least this amount ofwide curb lanes, although no official measureexists. A large majority of the wide curblanes were not planned as a specific bicyclefacility; the curb lanes are simply wider inthese locations due to repav-ing over thegutter pan, remarking of lanes after removalof parking, etc. The number of bicycle lanesstarted increasing after the adoption of the1974 Bicycle Master Plan.

Gainesville, Florida This city has a population of about

200,000 and is recognized as Florida's mostbicycle-friendly community. The city ishome to the University of Florida, andbicycle traffic is greatest in and around thecampus. At present, Gainesville has about130 km of roadways with bicycle lanes orpaved shoulders and an additional 30 km ofroadways with wide curb lanes. Anestablished Bicycle/Pedestrian Program hasbeen in place since 1983, and a full-timeprogram coordinator has been employed fornearly all of that period. The bicycle programoperations are centered in the trafficengineering department, which has done avery good job of tracking bicycle/motorvehicle crashes in the city.

Austin, Texas

Austin is the capital of Texas and has apopulation of almost 500,000. A bicycleplan

was developed in the late 1960s and manybike lanes were installed in the late 1970s andearly 1980s. Approximately 85 km ofbicycle lanes are now in place. No estimateis available for the number of wide curblanes. A local policy states that streetrestriping will provide for wide curb lanes ifpossible. The basic city bicycle map is about15 years old and is being replaced by aGeographic Information System version. The University of Texas is located in thecentral core and accounts for many riders,but there are also dedicated commuters. Between 1 and 2 percent of work trips areby bicycle. Recreational riders often usebicycle facilities to exit the city area for longerrides. The city bicycle program is locatedwithin the Public Works and TransportationDepartment, which provides access to otherplanners and transportation engineers.

Site CharacteristicsThe objective was to achieve a group of

sites within each city that varied by width ofBL or WCL (two levels), speed limit (twolevels), and traffic volume (two levels). Sucha matrix yields a total of eight sites. Thus,eight BL and eight WCL sites were selectedfor videotaping in each city. Selectedbreakpoint values were:

BL width - 1.5 m or less, >1.5 mWCL width - 4.3 m or less, >4.3 mSpeed limit - 50 km/h or less, >50 km/hTraffic volume - Low volume up to

7,500 vpd for 2 lanes; 15,000 vpd for 4 lanes, +etc. High volume greater than 7,500 vpd for 2 lanes; 15,000 vpd for 4 lanes, etc.

Chapter 2 13

We also tried to satisfy an objective of having20 to 30 bicyclists per hour riding throughthe selected intersections. The following BL

and WCL matrices show the overall mix forall three cities combined:

Bike Lane Sites

Width of BL 1.5 m or less >1.5 m

Traffic Volume Low High Low High

50 km/h or less FL FL FLCA CATX TX TX TX TX

FL FLCA

FL FL FLCATX TX

CA CA

>50 km/hCATX

CA

Wide Curb Lane Sites

Width of WCL 4.3 m or less >4.3 m

Traffic Volume Low High Low High

50 km/h or less FL

TX TX

FL FL

TX TX TX

FL FL FLCA CA CATX TX

>50 km/hCATX

CAFL FLCA CA CA

As potential sites were selected in each city,we attempted to develop a mix based on thevariable parameters shown above, as well asattempting to have variety in the sites that isrepresentative of real-world conditions (e.g.,BL and WCL sites with and without parking,BL sites with a weaving area and a bikepocket, BL sites with and without the stripecarried all the way to the intersection, BL andWCL sites where turning lanes were added atthe intersection). In all three cities thepreliminary site list of top candidates had to

be altered, usually due to a small number ofriders available for videotaping. BL siteswere generally popular and tended to havea reasonably high number of bicyclistsavailable. Sometimes the preliminary list ofBL sites was altered because it wasdiscovered that the viewing angle forvideotaping was not good. It was difficultto find eight suitable WCL sites in any ofthe selected cities due to small numbers ofbicyclists riding on WCL facilities.

14 Chapter 2

In Gainesville and Austin, the selectedsites were quite close to the universitycampuses, because this is where the majorityof the bicyclists were located, and data couldbe collected in an efficient manner. In SantaBarbara, the university campus was remote,and student bicyclists were a much smallerpart of the mix. In the project cites, BL sitestended to concentrate at low traffic speedand low traffic volume locations, while WCLsites tended to concentrate at high trafficvolume locations. Overall, the matrices offinal sites indicate a reasonable mix ofvariation.

Besides the items mentioned above, avariety of other descriptive data items werecollected at each site. These included type ofarea, pavement marking (striping)information for the BLs and WCLs, trafficcontrol device present, number of lanes,estimated driving speed, presence of parking,average annual daily traffic (AADT), andothers.

Videotaping of BicyclistsThe initial plan was to videotape

bicyclists both at midblock and intersectionlocations. However, it became apparent thatsample sizes would be relatively small if thevideotaping task was divided in this fashion.Thus, the decision was made to forego themidblock videotaping and instead videotapeeach intersection twice for the followingreasons:

! Intersections account for about halfof all bicycle-motor vehicle crashes.

! Because of the need to make turningmovements, intersections were expected tolead to more conflicts between vehicles,pedestrians, and other bicycles.

! It was of interest to learn about themaneuvers bicyclists make to travel throughintersections, such as the ways left turns aremade.

! The camera position would allowviewing of the approaching bicyclists from aconsiderable distance back from theintersection (not unlike a midblock situation).

Intersections and the approaches tointersections (referred to as midblock hereafter)were thus the focus of the data collectioneffort. Bicyclists were videotapedin the oncoming direction as they approachedthe selected intersections. The two-person datacollection team usually mounted the camera ona 3-m stepladder set up some 30 to 40 m onthe far side of the intersection. The locationwas such that the oncoming bicyclists generallywere not aware of the camera until close to theintersection. The stepladder was quite beneficialin providing a viewing angle above traffic. In afew of the Gainesville intersections, a platformtruck belonging to the city was used to enable abetter viewing position than could be affordedby a stepladder.

Normally the camera position allowed fora view of more than 150 m back from theintersection. Five 46-cm traffic cones were setup at 30-m intervals from the intersection stopbar location (at 30, 60, 90, 120, and 150 m).Approaching bicyclists were usually capturedbefore reaching the 150-m cone and followedthrough the intersection (figure 15). The datacollector would zoom in on the bicyclist toenable a better view of any kinds of bicycle-motor vehicle interactions. If the bicyclist hadto stop for a traffic signal, the data collectorwould ascertain if it were possible to videotapeanother approaching bicyclist. If so, thisbicyclist would be followed up to theintersection, and then both bicyclists would betaped as they rode through. Each intersectionwas videotaped twice for 2 hours at eachsession. The basic plan was to videotapeselected intersections during both weekday andweekend times if riders were present duringthese periods. However, if

Chapter 2 15

Figure 15. Typical data collection setup.

riders tended to travel through anintersection mainly on weekends (say, as partof a heavily traveled recreation route), thenvideotaping would be done twice onweekends. On the other hand, if commuterswere the norm and little riding was done onweekends, then videotaping would be donetwice during weekdays. Generally, all 16 siteswere videotaped once before the second round of taping began. Asstated earlier, approximately 4,600 bicyclistswere videotaped in the three cities (2,700 atBL sites and 1,900 at WCL sites).

Besides the bicyclist videotapingdescribed above, 15-minute samples oftraffic were also videotaped thatcorresponded to the time of the bicyclevideotaping. The camera was positioned at alocation where all the legs of the intersectioncould be observed. This videotape enabledcounts of motor vehicles traveling throughthe intersection and thus some measure ofexposure to traffic.

Bicyclist Experience Data

A one-minute oral survey was used tocollect information about the bicyclists ridingthrough the intersections. Data collectorspositioned themselves such that bicyclists couldbe safely stopped before reaching theintersection proper. The first data collectorwould stand about 50 m in front of the seconddata collector and ask approaching bicyclists tostop for a 1-minute bicycle survey ahead. Fourquestions were asked:

! What is your age? ! On average, how many days a week do

you ride your bike?! On average, about how many miles do

you ride each week?

! How would you classify yourself withrespect to the experience you have riding oncity streets? (1 or 2, shown below)

1. I feel comfortable riding under mosttraffic conditions, including major streets withbusy traffic and higher speeds.

OR

16 Chapter 2

2. I only feel comfortable riding onstreets with less traffic and lower speeds, onstreets with bicycle lanes, or on sidewalks.

In addition, information was coded pertain-ing to where the bicyclists were riding (road,sidewalk, or other location) as theyapproached the survey station, and the race,gender, and helmet use of the bicyclist. (Theexperience data collection form is shown inappendix A.) The data enabled us to gaininformation about the experience level ofbicyclists riding through the particularintersection. Such knowledge could bedirectly relevant to the types of maneuversand conflicts seen at the site. These data wereentered directly to a spreadsheet for analysis.

Each experience data collection sessionlasted 2 hours and was matched to thevideotaping (i.e., same basic time of day andday of week), and almost all were done afew days after the videotaping. Thus, ifvideotaping was done on both a weekdayand a weekend, then experience data wouldbe collected on a weekday and weekend. Iffilming was done only on weekdays or onlyon weekends, then experience data would belimited to this time period as well. Using thismethod, slightly more than 2,900 surveyswere completed. Generally, about two-thirdsof the bicyclists passing any given siteconsented to an interview. The most likelyreason for not stopping was being in a hurryto get to class, a meeting, etc. Sometimes thebicyclists unable to stop came back later andcompleted a survey. Characteristics of theriders not completing a survey were notobtained, but there was no evidence that thisgroup was different from the groupcompleting the survey.

Coding of Videotape DataA form for coding a variety of items

associated with a bicyclist riding through anintersection was developed, tested, and

revised several times before the form wassatisfactory. The objective was to code actionsassociated both with a “midblock” (theintersection approach) and an intersectionarea. Midblock was thus defined as the areabetween the third and fifth traffic cones set upon the approach leg ( 90 to 150 m from theintersection stop bar location). The intersectionwas defined as the area covered by the firstthree traffic cones (0 to 90 m back from thestop bar location).

The following are examples of the types ofvariables that were coded:

! Bicyclist riding wrong way.! Bicyclist demographics and helmet use.! Midblock positions and movements.! Bicyclist spacing from the curb or

gutter pan seam and from a passingmotor vehicle.

! Bicyclist midblock behaviors (e.g.,turning across a lane of traffic).

! Midblock conflict information.! Intersection positions and movements.! Bicyclist straight, left turn, and right turn

methods.! Bicyclist straight, left turn, and right turn

conflict information.The complete working form may be found inappendix B.

During initial coding practice, questionswere discussed and resolved. Data coders wereconstantly exchanging information at this stageso as to develop consistency. Once the codingprocess was finalized, a computerized entryscreen was created, pilot tested, and“debugged.” From this point onward, allvideotape data were coded via the computerscreen and automatically stored in a database.

Creation of Project DatabaseOnce videotape coding had been

completed, a database of various files wasassembled. This included:

! The coded videotape items.! The experience file.

Chapter 2 17

! The motor vehicle traffic count filebased on the 15-minute samples.

! The file describing the intersectionfeatures.

Coding and Analysisof Crash Data

Two years of recent (1994 and 1995)bicycle-motor vehicle crash data wereobtained from each of the three cities.Crashes from one complete year (1995)from each city were “typed” following themethodology originally developed by theNational Highway Traffic SafetyAdministration (NHTSA) in the late 1970s1

and being modified in partnership with theFHWA for computer application. Thecomputer software will be known asPBCAT (Pedestrian and Bicycle CrashAnalysis Tool), a user-friendly softwarepackage developed for FHWA by theUniversity of North Carolina Highway SafetyResearch Center.

Crashes from each city were relativelysparse, and very few matched theintersections selected for videotaping.However, city trends were examined todetermine if overall crash patterns weresimilar to the types of behaviors and conflictscoded from the videotape data.

The chapter that follows discusses cyclistcharacteristics and operational and safetyfindings associated with BLs and WCLs.

1For more specific background oncrash typing, see Hunter, Stutts, Pein, andCox (1996).

Chapter

3 Results

Using the methods described in theprevious chapter, this chapter presents resultsof the analysis of the data. The sections thatfollow are descriptive and focus on cyclistcharacteristics and midblock (or intersectionapproach) and intersection operational andsafety findings associated with BLs and WCLs.Tables are sometimes grouped for ease inpresentation. Findings from statistical modelsare then presented. The chapter concludeswith a clinical analysis of high-rate conflict sitesand the serious conflicts, and an examinationof the bicycle-motor vehicle crash data fromthe project cities.

Bicyclist Characteristics

Videotape dataSeveral variables describing the 4,589

videotaped bicyclists are summarized in table1. This table is typical of others that follow.The variables are cross-tabulated by whetherat BL or WCL sites. Frequencies and columnpercentages are routinely presented. Totalsdiffering from 4,589 are due to missing values.The text occasionally includes someinformation that was not placed in a table.

Statistical testing of relationships wasdone using chi-square tests to determine ifdifferences between BL and WCL distributionswere significant or due to chance alone. Whenthe distributions are significantly different,asterisks (**) are placed beside the name ofthe variable (e.g., “Gender” in table 1), and thelevel of significance, or p-value, is shown withthe appropriate number of asterisks at thebottom of the table. Using “Gender” fromtable 1 as an example, a p-value of < .05(single *) means that the difference in thedistributions could be due to chance less than5 times out of 100.

Generally, the tables show all levels of avariable to convey more information to thereader; however, categories were groupedwhen necessary to permit appropriatestatistical testing. In the text that follows, asingle triangle is used to indicate a majorindividual cell chi-square contribution to asignificant chi-square value for the overalldistribution. Chi-square testing was notperformed in cases where the distributionsproduced zero cells due to all effects of avariable being directly related to either a BLor WCL (e.g., turning left from a BL couldnot be done from a WCL).

Table 1 shows that slightly more thanthree-fourths of the bicyclists observed onthe videotapes were male. The proportion ofmales was slightly larger on BLs (77 percent)than WCLs (74 percent), while there wereslightly more females riding on WCLs (23percent) than BLs (20 percent).

Age of bicyclists was estimated fromobserving the videotapes and categorized intothe following groups: less than 16, 16-24, 25-64, and greater than 64 years. Overall, almost55 percent of the bicyclists were age 16-24,and another 44 percent age 25-64. This resultwas not surprising, given that we were tryingto capture cyclists riding in traffic, as well asthe fact that cyclists going to and fromcollege campuses were prevalent, particularlyin Gainesville and Austin. Some 58 percentof the bicyclists riding on WCLs were judgedto be age 16-24, compared with 53 percentfor those riding on BLsþ. More than 46percent of the cyclists riding on BLs were age25-64þ, as opposed to 40 percent for thoseon WCLs.Observed helmet use was 32 percent andvaried only slightly by facility type. Overall,5.6 percent of the bicyclists were riding thewrong way (i.e., facing traffic), 1.3 percent ofthese in the road and 4.3 percent on the

20 Chapter 3

Gender* BLs WCLs Total Helmet Use BLs WCLs Total

Male 2055(77.3)

1434(74.2)

3489(76.0)

Yes 822(31.0)

622(32.5)

1444(31.6)

Female 538(20.3)

442(22.9)

980(21.4)

No 1827(69.0)

1295(67.6)

3122(68.4)

Unsure 64(2.4)

56(2.9)

120(2.6)

Total 2649(58.0)

1917(42.0)

4566(100.0)

Total 2657(57.9)

1932(42.1)

4589(100.0)

* p < .05

Age*** BLs WCLs TotalWrong

Way Riding***

BLs WCLs Total

< 16 23(0.9)

27(1.5)

50(1.1)

Yes, in road 26(1.0)

32(1.7)

58(1.3)

16-24 1350(52.6)

1077(58.1)

2427(54.9)

Yes, sidewalk 61(2.3)

136(7.0)

197(4.3)

25-64 1183(46.1)

746(40.2)

1929(43.6)

No 2566(96.7)

1763(91.3)

4329(94.4)

65+ 10(0.4)

5(0.3)

15(0.3)

Total 2653(57.9)

1931(42.1)

4584(100.0)

Total 2566(58.0)

1855(42.0)

4421(100.0)

*** p < .001

*** p < .001 Level of Significance * p < .05 ** p < .01 *** p < .001

Table 1. Videotaped bicyclist characteristics.

sidewalk. Wrong-way riding was prevalent on percent for BL sites) versus correctly riding1

sidewalks (7 percent at WCL sites versus 2.3 with traffic. Wrong-way riding was found topercent at BL sites). These results were re- be significantly associated with WCL sitesþ.examined by eliminating the sidewalk ridingand comparing the wrong-way riding in the road (1.7 percent for WCL sites and 1.0

While wrong-way riding on a sidewalk is1

not necessarily illegal or improper behavior, it canlead to operational and safety problems withmotor vehicle traffic. Thus, it has been defined andused in this report as a behavioral characteristic ofbicyclists.

Chapter 3 21

Gender BLs WCLs Total Helmet Use BLs WCLs Total

Male 1225(74.4)

922(73.6)

2147(74.1)

Yes 604(36.7)

484(38.6)

1088(37.5)

Female 421(25.6)

331(26.4)

752(25.9)

No 1044(63.4)

769(61.4)

1813(62.5)

Total 1646(56.8)

1253(43.2)

2899(100.0)

Total 1648(56.8)

1253(43.2)

2901(100.0)

Age BLs WCLs

Total Race*** BLs WCLs Total

< 16 22(1.3)

15(1.2)

37(1.3)

White 1316(79.9)

1061(84.7)

2377(82.0)

16-24 641(38.9)

516(41.2)

1157(40.0)

Black 31(1.9)

39(3.1)

70(2.4)

25-64 982(59.6)

717(57.2)

1699(58.5)

Hispanic 174(10.6)

88(7.0)

262(9.0)

65+ 4(0.2)

5(0.4)

9(0.3)

Asian 103(6.3)

52(4.2)

155(5.3)

Total 1649(56.8)

1253(43.2)

2902(100.0)

Other 23(1.4)

12(1.0)

35(1.2)

Total 1647(56.8)

1252(43.2)

2899(100.0)

Level of Significance * p < .05 ** p < .01*** p < .001

*** p < .001

Table 2. Experience survey bicyclist characteristics.

Bicyclist experience survey resultsIn addition to the videotape data, bicyclist

experience data were gathered from a separatesurvey administered to bicyclists passingthrough each of the data collection sites.Information was gathered on the gender, age,race, and helmet use of the bicyclist; averagenumber of days a week ridden; averagenumber of miles a week ridden; location whereriding at the time of the survey (roadway,sidewalk or other); and the rider’s opinion ofhis or her level of riding experience (either

experienced or casual). (See appendix A) Information was gathered from 2,907bicyclists, 1,653 (57 percent) at BL locationsand 1,254 (43 percent) at WCL locations.Data were collected during both weekday andweekend time periods, although theoverwhelming majority (97 percent) ofsurveys were completed by bicyclists ridingon a weekday.

Table 2 provides the demographics ofthe riders completing the surveys.

22 Chapter 3

Nearly three-fourths (74 percent) of the survey station: nearly one out of fivebicyclists were male, slightly lower than the 76- bicyclists surveyed at WCL sites was observed78 percent male riders (adjusting for riding on the sidewalk, compared with only 6unknowns) captured on videotape. Bicyclists percent of those surveyed at BL sites. participating in the survey were also less likely Since the bicyclists in the exposureto be in the 16-24 year age category and more sample were shown to be older than those inlikely to be aged 25-64: 40 percent of the the videotape sample, rider age (<25, surveyed bicyclists were age 16-24, compared was cross-tabulated by the various ridingwith 55 percent of the videotaped bicyclists, experience variables. Generally, the youngerand 59 percent were age 25-64, versus 44 riders (<25) were more likely to ride fivepercent for the videotape. These differences days a week (p<.001), more likely to ride 25may reflect actual differences in the videotape or fewer miles per week (p<.001), and moreand survey samples or a greater willingness on likely to be observed riding on a sidewalkthe part of the older cyclists to participate in (p<.001). They were also much less likely tothe survey. However, they most likely reflect a be observed wearing a helmet (p<.001).tendency on the part of the data coders to There were no significant differences in theunderestimate the ages of the bicyclists viewed number of younger and older bicyclists whoon videotape. viewed themselves as experienced riders.

Helmet use was also higher for the These findings suggest that the videotapedexposure data sample than for the videotape riders, while no more “experienced” than theriders — 38 percent versus 32 percent. exposure riders, may ride fewer miles perInformation on bicyclist race was collected for week and may be more likely to peak atthe exposure sample only, based on the riding five days a week. They may also have ajudgment of the data collector, and showed lower overall helmet wearing rate. However,higher percentages of whites and blacks riding without more specific information on thein WCL locations,þ and more Hispanics and nature and extent of any misclassification ofAsians riding in BL locations.þ This was the rider age, no adjustments were made to theonly rider characteristic variable to show exposure data, and the data weresignificant differences by facility type. incorporated into the statistical modeling

Information on self-reported riding (described later in this chapter) as site-experience is provided in table 3. Eighty dependent variables.percent of the bicyclists surveyed reportedriding five or more days per week. Thosesurveyed at WCL locations were especiallylikely to ride five days a week, and less likelyto ride only one to three days a week. Justover 40 percent reported riding 10-25 milesper week, and another 25 percent 26-50 milesa week. Less than 7 percent rode more than100 miles a week, and for the totaldistributions there were no differences withrespect to facility type. Finally, 34 percent ofthe bicyclists considered themselves to beexperienced, versus merely casual, riders. Thispercentage also did not vary by WCL versusBL facility. Bicyclists surveyed at WCL sites,however, were more frequently observedriding on sidewalks as they approached the

Chapter 3 23

Days Ride Per

Week***BLs WCLs Total

MilesRide Per

WeekBLs WCLs Total

1 37(2.3)

8(0.6)

45(1.6)

< 10 205(12.5)

126(10.1)

331(11.5)

2 68(4.1)

23(1.8)

91(3.1)

10-25 662(40.3)

506(40.7)

1168(40.5)

3 142(8.6)

81(6.5)

223(7.7)

26-50 408(24.9)

325(26.2)

733(25.4)

4 112 (6.8)

88(7.1)

200(6.9)

51-75 156(9.5)

118(9.5)

275(9.5)

5 612(37.3)

560(44.9)

1172(40.6)

76-100 98(6.0)

87(7.0)

185(6.4)

6 141(8.6)

96(7.7)

237(8.2)

101-150 57(3.5)

42(3.4)

99(3.4)

7 530(32.3)

392(31.4)

922(31.9)

151+ 55(3.4)

38(3.1)

93(3.2)

Total 1642(56.8)

1248(43.2)

2890(100.0)

Total 1641(56.9)

1242(43.1)

2883(100.0)

*** p < .001

WhereRiding When Surveyed***

BLs WCLs Total Type Rider

BLs WCLs Total

Road 1556(94.4)

1001(80.0)

2557(88.2)

Experienced 560(34.5)

425(34.0)

985(34.3)

Sidewalk 93(5.6)

235(18.8)

328(11.3)

Casual 1064(65.5)

824(66.0)

1888(65.7)

Other 0(0.0)

15(1.2)

15(0.5) Level of Significance

* p < .05 ** p < .01 *** p < .001

Total 1649(56.9)

1251(43.1)

2900(100.0)

*** p < .001

Table 3. Experience survey riding characteristics.

Midblock Actions

MovementsMidblock (or the intersection approach)

was defined as between 90 and 150 m fromthe intersection (see figure 15). Within thismidblock zone, about 50 percent of thebicyclists approaching the intersections were inBLs, another 27 percent in WCLs, 7

percent on the sidewalk, and 9 percent in atraffic lane (generally the lane adjacent to a

BL or WCL). Whatever their initial midblockposition, 83 percent of the bicyclists did notchange this position throughout the midblockzone. Other variables describing midblockmovements are presented in table 4.

24 Chapter 3

BicyclistMidblock

Movement***BLs WCLs Total

Motor Vehicle Encroached into AdjacentTraffic Lane***

BLs WCLs Total

Straight tointersection

2409(95.4)

1597(93.1)

4006(94.5)

Yes 22(6.8)

28(16.7)

50(10.1)

Crossed to leftbefore intersection

82(3.3)

48(2.8)

130(3.1)

No 289(88.9)

131(78.0)

420(85.2)

Turned rightbefore intersection

31(1.2)

63(3.7)

94(2.2)

Unsure 14(4.3)

9(5.4)

23(4.7)

Other movement 4(0.2)

7(0.4)

11(0.3)

Total 325(65.9)

168(34.1)

493(100.0)

Total 2526(59.6)

1715(40.4)

4241(100.0)

*** p < .001

*** p < .001

Bicyclist RidingNext to Parked

Vehicle***BLs WCLs Total

Encroachment Led to Conflict BLs WCLs Total

Yes 615(23.2)

236(12.3)

851(18.6)

Yes 1(4.2)

1(3.3)

2(3.7)

No 2039(76.8)

1690(87.8)

3729(81.4)

No 23(95.8)

29(96.7)

52(96.3)

Total 2654(58.0)

1926(42.1)

4580(100.0)

Total 24(44.4)

30(55.6)

54(100.0)

*** p < .001

Level of Significance * p < .05

** p < .01 *** p < .001

Table 4. Midblock actions.

Ninety-five percent rode straight to theintersection. About 3 percent crossed thestreet to the left in advance of the intersection(proportionally more from BLs than WCLs),

and another 2 percent turned right prior to the

intersection (proportionally more from WCLsþthan BLs).2

It is important to note that these making a left turn at the intersection proper, the2

movements prior to the intersection may be more advance crossover would likely be made whether aa function of the destination preference of bicyclists BL or WCL was present.

than a statistically significant finding associatedwith either a BL or WCL. For example, if bicyclistsdesire to make an advance crossover to the leftprior to an intersection because of the difficulty of

Chapter 3 25

Proportionally twice as many bicyclists the form,were riding next to parked vehicles in BLsþ(23 percent) than in WCLs (12 percent), butthere were a greater number of BL sites thatcontained parking as part of the BL. Withoutproper enforcement, it would appear thatmotor vehicles are not hesitant to park (legallyor illegally) in either kind of facility.

Overall 10 percent of motor vehiclespassing bicycles on the left encroached into theadjacent motor vehicle traffic lane; 17 percentof the motor vehicles traveling in a WCLencroached while 7 percent of the vehiclestraveling on facilities with a BL encroached. This tendency agrees with results from arecent study for the Florida DOT (Harkeyand Stewart, 1997). Of the definiteencroachments into the adjacent traffic lane,only two (one each from a BL and WCL)resulted in a conflict with another motorvehicle.

Midblock spacing between bicyclesand motor vehicles

Whenever possible, measurements weremade of distances between the bicycle and thecurb face or edge of road (or gutter pan seam,if present), and between the bicycle and apassing motor vehicle. Two separate files weredeveloped, one that contained only the firsttype of measurement, bike to curb or gutterpan seam (“curb space”) made fromobservations where the bike was not beingpassed. The second file contained both typesof measurements, “curb space” and “carspace,” made from observations where thebike was being passed by a motor vehicle.

Statistical models were developed toinvestigate how these spacing variablesdiffered, on average, as a function of thevarious roadway and traffic characteristics.Least squares regression analysis was used indeveloping these models. Thus, it was assumedthat the observed distances followed a normaldistribution with mean value, µ , of

µ = þ + þ X + ... + þ X o 1 1 k k

, where the variables X , X , ..., X represent1 2 k

characteristics of the roadway, bicyclist, andtraffic conditions, and the þ’s are coefficientsestimated by fitting the model to data.Consider first a model for average curb spacefor bicycles that were not being passed andwhere the bicycle was not being ridden besidea parked vehicle. As usual, this model wasdeveloped by trying various combinations ofexplanatory variables potentially associatedwith the response variable and retaining thosefor which the association was statisticallysignificant. Results from carrying out thisprocess are presented in table 5. A total of1,393 observations were used in this model.

The model shows that both a variableindicating the presence of a BL and the widthof the BL were statistically significant, whileWCL width was not significant when includedin the model. Thus, average curb space for a

WCL is given simply by the constant plus thetraffic volume effect. For BLs with widths < 1.6 m, the average bicycle distance from thecurb was less than that for WCLs having thesame traffic volume. For BLs greater than1.6 m wide, however, the average bicycledistance from the curb was greater than forWCLs with similar traffic volumes. Forexample, using traffic volume of 400 vehiclesper hour, the model gives the value of 0.9 mfor the average distance from bike to curb atWCL sites. The corresponding averagedistances on BLs of 1.2 m, 1.8 m, and 2.1 mare 0.7 m, 1.0 m, and 1.2 m, respectively.

26 Chapter 3

Table 5. Model for distance from curb forbikes not being passed.

Variable estimate P-value(þ)

Constant 3.21 .0001

Traffic volume (vph) -.001 .0001

BL indicator -2.71 .0001

BL width (m) 1.74 .0001

It should be kept in mind that the estimatedmodels are very much dependent on thenature of the data from which they aredeveloped. In particular, the BLs tended to bequite wide, ranging from 1.2 to 2.9 m, with anaverage of 1.8 m. Total width (BL + adjacenttraffic lane) averaged 5.6 m for BL sitesversus 4.7 m for WCL sites. This variability inwidths across facility types makes it difficult tocompare the model results for Bls and WCLs.

On-street parking was part of the bicyclefacility at some sites (e.g., a shared parking andBL). At these sites, some bikes were observedriding next to parked vehicles. In these cases,spacing distance was measured from the biketo the parked vehicle rather than to the curb.There were 270 such observations of bicyclesnot being passed by a motor vehicle. Theseobservations were added to the data set and amodel was fitted to both these bike-to-curband bike-to-parked vehicle distances, whichcon- tained, in addition to the variables oftable 5, a variable indicating that the bike wasriding beside a parked vehicle. This variablewas statistically significant, with an estimatedcoefficient of -.347. This would seem toindicate that bicyclists ride closer to parkedvehicles than they do to the curb when parkedvehicles are not present. However, average BLwidths were greater by the amount 0.2 mwhen parking was part of the facility, and theestimated BL width effect in this model was1.76. So the increased width contri- buted anamount (1.76) (0.2) = .353 to the estimated

distance between bike and or parked vehicle.Thus, the estimated parked vehicle effect wassimply off- setting the increased BL width atthese sites and should not be interpreted ashaving any general applicability. In otherwords, on average, bicyclists rode about thesame distance from parked vehicles at BLsites with parking as they did to the curb (orgutter pan seam) when parked vehicles werenot present.

A model for average curb space basedon 319 measurements made on bikes beingpassed by motor vehicles is presented in table 6.For this model and the one presented earlierin table 5, both the variables included in themodels and the estimates are quite similar.The major difference is in the constant term,which shows the bikes to be positioned about0.3 m closer, on average, to the curb whenbeing passed than when not. This model againpredicts distances from bike to curb to besmaller for narrow Bls (width < 1.5 m) andlarger for wider BLs (width > 1.5 m) than forWCLs with similar traffic volumes.

Table 6. Model for distance from curbfor bikes being passed.


Constant 2.25 .0001


BL indicator -2.28 .0001

BL width (m) 1.57 .0001

Chapter 3 27

Table 7. Model for distance betweenbikes and passing motor vehicles.


Constant .37 .6164


Total width (m) 1.08 .0001

Speed (km/h) .04 .000

Results from a model for distancebetween bike and passing vehicle are presented intable 7. This model was based on 318observations. Note that this model does notcontain either the BL indicator variable or BLwidth as a separate variable. These variableswere tried and found to be non-significant,indicating that type of facility (BL versusWCL) does not affect the separation distance.Instead it contains a total width variable, whichconsists of BL width plus the width of theadjacent traffic lane or the width of the WCLwhen no BL is present. Thus, this width effectapplies to the total space available to thebicycle and motor vehicle, whether the space isa WCL or a BL plus an adjacent traffic lane. Estimated driving speed is also a factor in thismodel, with spacing between vehicles andbikes tending to increase as speed increases.

BehaviorsVarious behaviors of bicyclists in the

midblock area were coded from the videotape.A complete list of possible behaviors may befound as part of the coding form in appendixB. The behaviors were divided into threesections: (1) bicycle-motor vehicle commonactions, (2) motor vehicle initiated events, and(3) bicycle initiated events. Common actions3

were behaviors that either a bicycle or motorvehicle could perform, such as failing to stopfor a stop sign or traffic signal, improperturns, improper lane use, etc. Examples ofmotor vehicle initiated events included long-term travel in the BL or right edge of theWCL, turning right soon after overtaking abicyclist, and entering/ exiting on-streetparking. Examples of bicycle initiated eventsincluded turning or swerving across a lane oftraffic, riding in between slowed/stoppedtraffic, and stunt riding or other erratic riding.Also included in this last category wereencounters with pedestrians or other bicyclists(e.g., a bicyclist avoiding another bicyclisttraveling in the opposing direction in BL, or abicyclist avoiding a jogger in a BL).

The motor vehicle portions of the three3

items discussed above were only coded for conflictsbetween bicycles and motor vehicles. Conflicts arediscussed later in the chapter.

28 Chapter 3

Behavior*** BLs WCLs TotalVehicle in Vicinity for Midblock

Behaviors***

BLs WCLs Total

None 2085(84.8)

1458(89.9)

3543(86.8)

Yes 1790(72.1)

1371(77.8)

3161(74.5)

Bike did not stop for stopsign

13(0.5)

1(0.1)

14(0.3)

No 679(27.4)

389(22.1)

1068(25.2)

Bike slowed/stopped/swerved for traffic not atintersection

55(2.2)

17(1.1)

72(1.8)

Unsure 14(0.6)

3(0.2)

17(0.4)

Bike entered roadwayfrom sidewalk or overcurb

9(0.4)

10(0.6)

19(0.5)

Total 2483(58.5)

1763(41.5)

4246(100.0)

Bike turned or swervedacross a lane of traffic

111(4.5)

49(3.0)

160(3.9)

*** p < .001

Bike stunt riding 12(0.5)

7(0.4)

19(0.5)

Bikes riding two abreast 23(0.9)

17(1.1)

40(1.0)

Bike encounter withpedestrian

21(0.9)

15(0.9)

36(0.9)

Bike encounter withanother bike

16(0.7)

2(0.1)

18(0.4)

Other behaviors 113(4.6)

46(2.8)

159(3.9)


** p < .01 *** p < .001

Total 2458(60.3)

1622(39.8)

4080(100.0)

*** p < .001

Table 8. Midblock behaviors.

Behavior*** BLs WCLs TotalVehicle in Vicinity for Midblock

Behaviors***

BLs WCLs Total

None 2085(84.8)

1458(89.9)

3543(86.8)

Yes 1790(72.1)

1371(77.8)

3161(74.5)

Bike did not stop for stopsign

13(0.5)

1(0.1)

14(0.3)

No 679(27.4)

389(22.1)

1068(25.2)


55(2.2)

17(1.1)

72(1.8)

Unsure 14(0.6)

3(0.2)

17(0.4)


9(0.4)

10(0.6)

19(0.5)

Total 2483(58.5)

1763(41.5)

4246(100.0)


111(4.5)

49(3.0)

160(3.9)

*** p < .001


7(0.4)

19(0.5)


17(1.1)

40(1.0)


21(0.9)

15(0.9)

36(0.9)


16(0.7)

2(0.1)

18(0.4)


46(2.8)

159(3.9)


** p < .01 *** p < .001

Total 2458(60.3)

1622(39.8)

4080(100.0)

*** p < .001

Table 8. Midblock behaviors.

Table 8 shows a list of behaviors a motor vehicle pulls into the BL from adistributed by type of site (BL or WCL). Some driveway)þ.grouping of behaviors has been performed to - The bicyclist turning or swerving acrossallow valid statistical comparisons. Behaviors a lane of trafficþ.seen proportionally more often in BLsincluded:

- The bicyclist not stopping at a stop sign(there was an intersecting street in themidblock area at a few sites).

- The bicyclist having to slow/stop/swerve due to traffic not influenced by theintersection ahead (e.g., a bicyclist slows when

Chapter 3 29

Midblock Behaviorswith a Motor Vehicle

in the Vicinity***

BLs WCLs Total

None 1412(84.1)

1088(90.1)

2500(86.6)

Bike did not stop forstop sign

5(0.3)

1(0.1)

6(0.2)


48(2.9)

17(1.4)

65(2.3)


8(0.5)

7(0.6)

15(0.5)


76(4.5)

34(2.8)

110(3.8)


5(0.4)

14(0.5)


11(0.9)

31(1.1)


12(0.7)

10(0.8)

22(0.8)


8(0.5)

1(0.1)

9(0.3)


34(2.8)

116(4.0)

Total 1680(58.2)

1208(41.8)

2888(100.0)

*** p < .001 Level of Significance

* p < .05 ** p < .01 *** p < .001

Table 8. Midblock behaviors (con’t).

- Bicyclist encounters with otherbicyclistsþ.

- “Other” kinds of behaviorsþ.Examples of the “other” kinds of behaviors(low frequencies of each) included improperturns and lane changes, motor vehicles parkedin BLs, weaving while riding in the BL,weaving in between motor vehicles, andcarrying a bag in hand.

Behaviors seen proportionally more oftenin WCLs included:

- The bicyclist entering the roadway from

the sidewalk or over the curb (only slightlyhigher than for BL).

- Bicyclists riding two abreast.When any of the midblock behaviors

were observed, a code was used to denotewhether there was a motor vehicle in thevicinity, obviously implying that the behaviorswere potentially more dangerous if this wasthe case. “In the vicinity” was defined asapproximately 66 m. Overall, vehicles werein the vicinity nearly three-fourths of thetime a bicyclist was observed performing thebehavior (72 percent for BL sites versus 78percent for WCL sites).

The next part of table 8, “MidblockBehaviors with a Motor Vehicle in theVicinity,” amplifies the information above byallowing specific comparisons for thebehaviors. For example, of the 160 bicyclistswho turned or swerved across a lane oftraffic, 110 did so with a motor vehicle in thevicinity.

Intersection Actions

MovementsFor bicyclists proceeding toward the

camera, the beginning of the intersection wasdefined as starting 90 m upstream of thestop bar, and included the intersection proper(see figure 15). Table 9 shows that 49percent of the bicyclists were riding in a BLand 31 percent in a WCL as they approachedthe intersection. (No statistical test of theoverall distribution was performed due to theseparation of bicyclists into the initialapproach of either a BL or WCL.) Almost 8percent were on the sidewalk and 6 percentin a motor vehicle traffic lane (i.e., in neither aBL or WCL).

30 Chapter 3

Bicyclist InitialIntersectionApproachPosition

BLs WCLs Total Bicyclist ObeyedTraffic Signal

BLs WCLs Total

Bike lane 2028(81.1)

0(0.0)

2028(48.5)

Obeyed signal 1401(91.2)

1136(92.0)

2537(91.6)

Wide lane 0(0.0)

1278(76.0)

1278(30.6)

Disobeyed signal 135(8.8)

99(8.0)

234(8.4)

Sidewalk 82(3.3)

245(14.6)

327(7.8)

Total 1536(55.4)

1235(44.6)

2771(100.0)

Traffic lane 216(8.6)

53(3.2)

269(6.4)

Left turn lane 5(0.2)

2(0.1)

7(0.2)

Bicyclist ObeyedStop Sign*** BLs WCLs Total

Right turn lane 29(1.2)

5(0.3)

34(0.8)

Obeyed sign 554(80.6)

116(55.2)

670(74.7)

Other 61(2.4)

36(2.1)

97(2.3)

Disobeyed sign 133(19.4)

94(44.8)

227(25.3)

Not throughintersection

79(3.2)

63(3.8)

142(3.4)

Total 687(76.6)

210(23.4)

897(100.0)

Total 2500(59.8)

1682(40.2)

4182(100.0)

*** p < .001

Bicyclist Changed

IntersectionApproach

Position To

BLs WCLs Total Result ofDisobeying

Signal

BLs WCLs Total

Bike lane 87(3.6)

0(0.0)

87(2.2)

Maneuver safe 109(84.5)

75(78.1)

184(81.8)

Wide lane 0(0.0)

34(2.1)

34(0.8)

Maneuversomewhat unsafe

17(13.2)

19(19.8)

36(16.0)

Sidewalk, path 66(2.7)

41(2.5)

107(2.7)

Maneuver definitely unsafe

3(2.3)

2(2.1)

5(2.2)

Traffic lane 404(16.8)

258(16.0)

662(16.5)

Total 129(57.3)

96(42.7)

225(100.0)

Left turn lane 109(4.5)

27(1.7)

136(3.4)

Right turn lane 173(7.2)

103(6.4)

276(6.9)

Result of Dis-obeying Sign*** BLs WCLs Total

Other 30(1.3)

30(1.9)

60(1.5)

Maneuver safe 100(78.1)

87(94.6)

187(85.0)

No change 1540(63.9)

1122(69.5)

2662(66.2)

Maneuversomewhat unsafe

24(18.8)

5(5.4)

29(13.2)

Total 2409(59.9)

1615(40.1)

4024(100.0)

Maneuver definitely unsafe

4(3.1)

0(0.0)

4(1.8)

Total 128(58.2)

92(41.8)

220(100.0)

Level of Significance * p < .05 ** p < .01 *** p < .001

*** p < .001

Table 9. Intersection actions.

Chapter 3 31

A large number of bicyclists in the “other” type of facility. Of the 17 cyclists who walkedcategory came into the intersection area froma driveway or other intersecting street.Proportionally more bicyclists approached theintersection on a sidewalk when the facility wasa WCL (15 percent) compared with a BL (3percent). On the other hand, proportionallymore bicyclists approached the intersection ina motor vehicle traffic lane when a BL waspresent (9 percent) compared with a WCL (3percent).

Overall, two-thirds of the bicyclists did notchange their position as they traveled throughthe 90-m intersection area (similar to above,no statistical test performed). There wereproportionally more bicyclists who changed toleft turn lanes from BLs. Most of the “other”movements involved bicyclists turning off toeither the left or right before the intersection.

Overall, 92 percent of bicyclists obeyedthe traffic signals that were in place. There waslittle variation by type of facility. The resultwas different for stop signs, however. Overall,75 percent of bicyclists obeyed the stop signsthat were present. Proportionally morebicyclists in WCLs (45 percent) disobeyed signsthan bicyclists in BLs (19 percent). Of 225instances where a signal was disobeyed, 36 (16percent) were considered somewhat unsafeand 5 (2 percent) definitely unsafe. There wasno variation by type of facility. The stop signsituation again produced a different outcome.Of 220 instances where a stop sign wasdisobeyed, 29 (13 percent) were consideredsomewhat unsafe and 4 (2 percent) definitelyunsafe. However, the proportion of bicyclistswith somewhat unsafe (19 versus 5 percent)and definitely unsafe (3 versus 0 percent)movements was higher in the BL facilities. Astatistically significant difference was detectedwhen the somewhat unsafe and definitelyunsafe categories were combined.þ

Almost 72 percent of the bicyclists wentstraight through the intersection, with another15 percent turning left and 13 percent turningright (table 10). There was little variation by

their bike through the intersection, 12 were inWCLs.

Of the 2,700+ bicyclists who wentstraight through the intersections, 27 percentwent from BL to BL, 17 percent from WCLto WCL, 4 percent from WCL to anothertraffic lane, 25 percent maneuvered like amotor vehicle in a traffic lane, and 4 percentused the marked crosswalk. Another 2percent switched to the sidewalk and thencrossed on the marked crosswalk. Some 3percent went straight through from a rightturn lane, and another 10 percent used someother method, such as BL to traffic lane orfrom left side of right turn lane. The 10percent using another method of goingstraight through included 14 percent in WCLsand 8 percent in BLs. (No statistical test ofthe distributions was performed due toexclusive BL and WCL movements.) Ninepercent of the cyclists tended to shy to theright (i.e., move to the right and away fromtraffic) as they went through the intersection (11 percent in BLs and 7 percent in WCLs).Interestingly, more than twice as many WCLusers (17 percent) exited to a sidewalk orpath beside the roadway than BL users (6percent) (no table shown).

Depending on a number of factors, suchas multiple lanes and the speed and volumeof motor vehicle traffic, left turns atintersections can pose problems for bicyclists.The types of left turns that were videotapedbear this out. There were problems in makingleft turns from both BLs and WCLs.

32 Chapter 3

BicyclistDirection

BLs WCLs TotalBicyclistWalkedBike?**

BLs WCLs Total

Straight 1646(72.2)

1078(70.9)

2724(71.7)

Yes 5(0.2)

12(0.6)

17(0.4)

Left 308(13.5)

243(16.0)

551(14.5)

No 2620(99.8)

1905(99.4)

4525(99.6)

Right 312(13.7)

187(12.3)

499(13.1)

Total 2625(57.8)

1917(42.2)

4542(100.0)

Other 13(0.6)

13(0.9)

26(0.7)

** p < .01

Total 2279(60.0)

1521(40.0)

3800(100.0)

Bicyclist Straight Through Method BLs WCLs Total

Bicyclist Tracking*** BLs WCLs Total

Straight from BL toBL

732(44.4)

0(0.0)

732(26.8)

Straight 1445(88.9)

941(93.0)

2386(90.5)

BL to traffic lane 210(12.7)

0(0.0)

210(7.7)

Shy to right 173(10.6)

70(6.9)

243(9.2)

WCL to WCL 0(0.0)

478(44.3)

478(17.5)

Shy to left 8(0.5)

1(0.1)

9(0.3)

WCL to traffic lane 0(0.0)

116(10.8)

116(4.3)

Total 1626(61.6)

1012(38.4)

2638(100.0)

Like motor vehiclein traffic lane

455(27.6)

215(19.9)

670(24.6)

*** p < .001

In street but oncrosswalk

67(4.1)

43(4.0)

110(4.0)

Moved to sidewalk & then to crosswalk

19(1.2)

37(3.4)

56(2.1)


** p < .01*** p < .001

Straight from rightturn lane

41(2.5)

33(3.1)

74(2.7)

Other 126(7.6)

156(14.5)

282(10.3)

Table 10. Intersection movements.

Of the 550 bicyclists making left turns at theintersections (table 11), 44 percent did so likea motor vehicle with proper lane destinationpositioning (41 percent from BLs and 48percent from WCLs). Another 8 percent mademotor vehicle style left turns but withimproper lane destination positioning (3

percent from BLs and 14 percent fromWCLs). Almost 17 percent of the cyclistsmade pedestrian style left turns, where thecyclist would ride all the way to theintersection and then use the crosswalk to getacross the street like a pedestrian.

Chapter 3 33

Bicyclist Left Turn Method BLs WCLs TotalMotor vehicle style - proper destinationpositioning

126(41.0)

116(47.9)

242(44.1)

Motor vehicle style - improperdestination positioning

10(3.3)

34(14.1)

44(8.0)

Pedestrian style - near side ofintersection

26(8.5)

40(16.5)

66(12.0)

Pedestrian style - far side of intersection 10(3.3)

17(7.0)

27(4.9)

Right hook 19(6.2)

2(0.8)

21(3.8)

Left from BL 88(28.7)

0(0.0)

88(16.0)

Advance crossover 10(3.3)

14(5.8)

24(4.4)

Motor vehicle-pedestrian combination 15(4.9)

14(5.8)

29(5.3)

Other 3(1.0)

5(2.1)

8(1.5)

Total 307(55.9)

242(44.1)

549(100.0)

Bicyclist Right Turn Method** BLs WCLs TotalStandard 277

(89.9)154

(81.1)431

(86.6)Non-standard 31

(10.1)36

(19.0)67

(13.5)Total 308

(61.9)190

(38.2)498

(100.0) ** p < .01 Level of Significance

* p < .05 ** p < .01 *** p < .001

Table 11. Intersection turning information.

Twelve percent crossed on the near side, and site, a t-intersection, where most of the motor5 percent on the far side, of the intersection. vehicle and bicycle traffic made leftProportionally more cyclists made thesepedestrian style left turns from WCLs. Of the4 percent who turned right to allow a motorvehicle to pass before executing a u-turn (a“right hook” left turn), the vast majority werein BLs. Overall, 16 percent of the bicyclistsmade left turns from the BL (29 percent ofthe bicyclists in BLs made left turns from theBL, but the vast majority occurred at a single

turns).No statistical test of the distributionswas made because of this exclusive BLmaneuver. Finally, 4 percent made anadvance crossover before reaching theintersection proper, and 5 percent made acombination motor vehicle-pedestrian leftturn (such as approaching the intersection ina left turn lane and then using the crosswalkto complete the turn). Upon completing theleft turn, 21 percent of the cyclists on WCLs

34 Chapter 3

exited to a sidewalk or path beside the road, swerve due to traffic not influenced by thecompared with 15 percent of the cyclists on intersection ahead (e.g., a motor vehicleBLs (not shown in table). pulling into the BL from a driveway).

For bicyclists making motor vehicle style - Bicyclist encounters with other bicyclistsleft turns, the distance back from the stop bar (low frequencies).that the cyclist started to merge to the left wascoded. Where BLs were present, more than Behaviors seen proportionally more often inhalf of the cyclists merged left in the last 60 WCLs included:m. Where WCLs were present, about half of - The bicyclist not stopping at a trafficthe cyclists merged left in the last 30 m. The signal (only slightly higher than for BL).median value was 63 m for BLs and 35 m for - Improper lane use by the bicyclist.WCLs. Seventeen percent of the cyclists - The bicyclist having to slow/merging left and making these motor vehicle stop/swerve due to intersection traffic.style left turns were considered to have - The bicyclist turning or swerving acrossincorrect lane position through the maneuver a lane of traffic.(11 percent from BLs and 23 percent from - The bicyclist passing slow or stoppedWCLs — not shown in table). vehicles on the rightþ (Note: this could be

Almost 500 bicyclists made right turns at coded in a BL situation when the BL stripethe intersections, and 87 percent were judged was terminated or dashed to the intersection).to have done so correctly (or in a standard - Bicyclists riding two abreast (lowfashion), and 13 percent in a non-standard frequencies).fashion (e.g., from BL or WCL/traffic lane to - Bicyclist encounters with pedestrians.wrong way position on cross street, or from - “Other” types of behaviors.BL or WCL/traffic lane to sidewalk). Bicyclists Examples of the “other” kinds of bicyclistin WCLs were more likely to make right turns behaviors (low frequencies of each) includedin a non-standard manner (19 percent in turning left from the right side of the trafficWCLs and 10 percent in BLs).þ Such turns lane, right on red maneuvers, and walking theincluded the use of the sidewalk and going bicycle.through parking lots or yards. Upon When any of the intersection behaviorscompleting the right turn, approximately 15 were observed, a code was used to denotepercent of the cyclists at both BL and WCL whether there was a motor vehicle in thesites exited to a sidewalk or path beside the vicinity, obviously implying that the behaviorsroad (not shown in table). were potentially more dangerous if this were

BehaviorsSimilar to the midblock area, various

behaviors in the intersection were coded.Table 12 shows a list of behaviors distributedby whether a BL or WCL site. Some groupingof behaviors has been performed to allowvalid statistical comparisons. Specific behaviorsseen proportionally more often in BLs thanWCLs included:

- The bicyclist not stopping at a stop sign(only slightly higher than for WCL).

- The bicyclist having to slow/stop/

the case. “In the vicinity” was defined aswithin approximately 66 m. Overall, vehicleswere in the vicinity 83 percent of the time(85 percent at BL sites versus 80 percent atWCL sites).þ

The next part of table 12, “MidblockBehaviors with a Motor Vehicle in theVicinity,” amplifies the information above byallowing specific comparisons for thebehaviors. For example, of the 81 bicyclistswho turned or swerved across a lane oftraffic, 66 did so with a motor vehicle in thevicinity.

Chapter 3 35

Behavior*** BLs WCLs TotalVehicle inVicinity for IntersectionBehaviors***

BLs WCLs Total

None 1929(81.1)

1140(70.9)

3069(77.0)

Yes 2238(84.9)

1539(80.3)

3777(83.0)

Bike did not stop for stop sign 140(5.9)

91(5.7)

231(5.8)

No 398(15.1)

377(19.7)

775(17.0)

Bike did not stop for trafficsignal

76(3.2)

60(3.7)

136(3.4)

Total 2636(57.9)

1916(42.1)

4552(100.0)

Bike improper lane use 12(0.5)

30(1.9)

42(1.1)

*** p < .001

Bike slowed/stopped/swervedfor intersection traffic

25(1.1)

26(1.6)

51(1.3)

Bike slowed/stopped/swerved for traffic not at intersection

21(0.9)

7(0.4)

28(0.7)

Bike turned or swerved acrossa lane of traffic

39(1.6)

42(2.6)

81(2.0)

Bike passed slow or stoppedvehicle on right

31(1.3)

86(5.4)

117(2.9)


11(0.7)

25(0.6)


10(0.6)

14(0.4)

Bike encounter with pedestrian 7(0.3)

18(1.1)

25(0.6)

Bike encounter with anotherbike

9(0.4)

2(0.1)

11(0.3)



85(5.3)

156(3.9)

Total 2378(59.7)

1608(40.3)

3986(100.0)

*** p < .001

Table 12. Intersection behaviors.

36 Chapter 3

Intersection Behaviors witha Motor Vehicle in the

Vicinity***

BLs WCLs Total

None 1690(81.6)

1042(74.1)

2732(78.6)

Bike did not stop for stop sign 94(4.5)

32(2.3)

126(3.6)

Bike did not stop for trafficsignal

71(3.4)

52(3.7)

123(3.5)

Bike improper lane use 10(0.5)

29(2.1)

39(1.1)

Bike slowed/stopped/swerved for intersectiontraffic

25(1.2)

25(1.8)

50(1.4)


20(1.0)

7(0.5)

27(0.8)


32(1.5)

34(2.4)

66(1.9)

Bike passed slow or stoppedvehicle on right

31(1.5)

85(6.1)

116(3.3)


10(0.7)

23(0.7)


8(0.6)

12(0.4)


5(0.2)

16(1.1)

21(0.6)

Bike encounter with anotherbike

9(0.4)

2(0.1)

11(0.3)


64(4.6)

132(3.8)

Total 2072(59.6)

1406(40.4)

3478(100.0)

*** p < .001 Level of Significance * p < .05

** p < .01 *** p < .001

Table 12. Intersection behaviors (con’t).

Conflicts

MidblockA conflict was defined as an interaction

between a bicycle and motor vehicle,pedestrian, or other bicycle such that at leastone of the parties had to

change speed or direction to avoid the other.There were 188 conflicts in the midblockarea, and totals less than this in the tables thatfollow are due to missing values. Theconflicts in the midblock area were of thefollowing type (table 13):

- 71 percent bicycle/motor vehicle.- 10 percent bicycle/bicycle.- 19 percent bicycle/pedestrian.

Almost all of the bike/bike conflicts occurredin BLs. Compared with BLs, bicyclists inWCLs experienced more bike/pedestrianconflicts (30 versus 16 percent)þ and lessbike/bike conflicts (2.5 versus 12 percent).

A scale was used to code level ofavoidance response for both bicycles andmotor vehicles (but not for pedestrians). Thescale for bicycles was used in a study ofpainted bike crossings at intersections inCanada (Pronovost and Lusginan, 1996) andhad the following categories:

- No change in riding.- Stops pedaling.- Slight change of direction.- Applies brakes.- Major change of direction.- Full stop.- Collision or near crash.

Examining the full distribution for themidblock conflicts shows that BL users hadproportionally more minor responses, such asstopping pedaling and slight change ofdirection. Conversely, WCL users hadproportionally more serious responses, suchas braking, major change of direction, andfull stop. In the midblock area, there were noinstances of collisions/near crashes. Aftercombining several categories to enable a validchi-square test, no significant differences inthe BL and WCL distributions were found.

Chapter 3 37

Conflict Type* BLs WCLs TotalBicycle

AvoidanceResponse

BLs WCLs Total

Bike/motorvehicle

99(71.7)

27(67.5)

126(70.8)

No change inriding

4(2.8)

5(12.2)

9(4.9)

Bike/bike 17(12.3)

1(2.5)

18(10.1)

Stops pedaling 12(8.4)

2(4.9)

14(7.6)

Bike/ped 22(15.9)

12(30.0)

34(19.1)

Slight change ofdirection

98(68.5)

19(46.3)

117(63.6)

Total 138(77.5)

40(22.5)

178(100.0)

Applies brakes 10(7.0)

4(9.8)

14(7.6)

* p < .05 Major change ofdirection

17(11.9)

8(19.5)

25(13.6)

Full stop 2(1.4)

3(7.3)

5(2.7)

Total 143(77.7)

41(22.3)

184(100.0)

Motor VehicleDriver

AvoidanceResponse

BLs WCLs Total Seriousness ofConflict

BLs WCLs Total

No change indriving

98(78.4)

21(75.0)

119(77.8)

Minor 140(98.6)

38(97.4)

178(98.3)

Slows 10(8.0)

2(7.1)

12(7.8)

Serious 2(1.4)

1(2.6)

3(1.7)


5(4.0)

2(7.1)

7(4.6)

Total 142(78.5)

39(21.6)

181(100.0)


0(0.0)

2(1.3)

Major change ofdirection

1(0.8)

1(3.6)

2(1.3)


Full stop 9(7.2)

2(7.1)

11(7.2)

Total 125(81.7)

28(18.3)

153(100.0)

Table 13. Midblock conflict information.

The motor vehicle scale was created for the distributions. Categories were again combinedcurrent project and used similar categories: to enable a valid chi-square test, and no

- No change in driving. significant difference in the BL and WCL- Slows. distributions was found.- Slight change of direction. The distributions were examined further- Applies brakes. by utilizing the level of the scale (1 for no- Major change of direction. change in riding, 2 for stops pedaling, etc.) to- Full stop. develop a mean value for the responses. For- Collision or near crash. the bicycle avoidance response, the BL mean

There was little variation in the BL and WCL was 3.21 and the WCL mean was 3.41. Using

38 Chapter 3

Bike Actions in MidblockConflicts***

BLs WCLs Total

None 6(4.4)

10(25.0)

16(9.0)

Bike slowed/stopped/swerved for trafficnot at intersection

85(62.0)

15(37.5)

100(56.5)

Bike turned or swerved across a lane oftraffic

14(10.2)

1(2.5)

15(8.5)

Bike encounter with pedestrian 12(8.8)

11(27.5)

23(13.0)

Bike encounter with bicyclist 5(3.7)

0(0.0)

5(2.8)

Other bike actions 15(11.0)

3(7.5)

18(10.2)

Total 137(77.4)

40(22.6)

177(100.0)

*** p < .001

Motor Vehicle Actions in Midblock Conflicts**

BLs WCLs Total

None 31(25.4)

3(9.1)

34(21.9)

Motor vehicle turned right in front ofbike after overtaking

6(4.9)

3(9.1)

9(5.8)

Motor vehicle illegally parked inBL/WCL

29(23.8)

2(6.1)

31(20.0)

Motor vehicle entering/exiting on-streetparking or driver/passenger entering orexiting a parked or stopped vehicle

19(15.6)

4(12.1)

23(14.8)

Other motor vehicle actions 37(30.3)

21(63.6)

58(37.4)

Total 122(78.7)

33(21.3)

155(100.0)

** p < .01 Level of Significance * p < .05 ** p < .01 *** p < .001

Table 14. Midblock conflict bicycle and motor vehicle actions.

a t-test to examine the hypothesis of no Bicycle actions more associated with BLsdifference between the means, the difference in these midblock conflicts (table 14)was not statistically significant. included: The motor vehicle avoidance

The seriousness of the conflicts was coded responses were similarly compared. The BLas minor, serious, or unsure. In the midblock mean was 1.60 and the WCL mean 1.71.area, 98 percent of the conflicts were coded as Once again the difference was insignificant.minor in nature, and there were no differencesin the BL and WCL distributions.

Chapter 3 39

- The bicycle slowing/stopping/ swerving - 79 percent bicycle/motor vehicle.for traffic not influenced by the intersection. - 10 percent bicycle/bicycle.

- The bicycle turning or swerving across a - 10 percent bicycle/pedestrian.lane of traffic.

- Encounters with other bicyclists. The differences in the BL/WCL distributions- “Other” actions. were statistically significant. There were

Examples of the “other” kinds of bicyclist proportionally more bike/bike conflicts inactions (low frequencies of each) included BLs (15 percent) and less in WCLs (4improper left turns and passing slow moving percent). Conversely, there wereor stopped motor vehicles on the right proportionally more bike/pedestrian conflicts(sometimes occurred at a BL site after the BL in WCLs (17 percent) and less in BLs (6stripe was terminated or dashed to the percent).intersection, or where a bicyclist was turning For the intersection conflicts, the positionright in a right turn lane). of the motor vehicle with respect to theBicycle actions more associated with WCLs bicyclist was coded. Overall, 66 percent ofincluded: the motor vehicles were traveling in the same

- No action takenþ. direction, 6 percent in the opposing direction,- Encounters with pedestriansþ. 5 percent approaching from the left, 15Motor vehicle actions more associated percent approaching from the right, and 7

with BLs included: percent from some other position.- No action taken. Proportionally more BL conflicts involved- Illegally parked in the BL. motor vehicles from the same direction and- Entering/exiting on-street parking or cross street traffic from the right.

driver or passenger entering/exiting a parked Proportionally more WCL conflicts involvedor stopped vehicleþ. motor vehicles from the opposing direction,(Note: The statistical contributions to the cross street traffic from the left, and trafficoverall chi-square value noted for BLs above from some other position. The differences byare actually due to less of the actions in WCLs type of facility were not statistically significantthan expected.) (cross street traffic from right and left

Motor vehicle actions more associated withWCLs included: There was variability in the position of

- Turning right in front of a bicyclist after the motor vehicle depending on theovertaking. maneuver of the bicyclist (no table shown).

- “Other” actionsþ. For example, when the bicyclist went straightExamples of the “other” kinds of motor through the intersection (82 cases), the motorvehicle actions (low frequencies of each) vehicle position was:included failure to yield, improper right turns, - Same direction - 77 percent ofand crowding bikes. conflicts.

IntersectionsThere were 198 conflicts in the

intersection area, and totals less than this in thetables that follow are due to missing values.The conflicts in the intersection area were ofthe following type (table 15):

combined for statistical test).

40 Chapter 3

Conflict Type* BLs WCLs TotalBiycle Avoidance

Response BLs WCLs TotalBike/motor vehicle 78

(78.8)72

(78.3)150

(78.5)No change in riding 11

(10.6)11

(12.5)22

(11.5)Bike/bike 15

(15.2)4

(4.4)19

(10.0)Stops pedaling 12

(11.5)17

(19.3)29

(15.1)Bike/pedestrian 6

(6.1)16

(17.4)22

(10.0)Slight change ofdirection

53(51.0)

33(37.5)

86(44.8)

Total 99(51.8)

92(48.2)

191(100.0)


7(8.0)

18(9.4)

* p < .05 Major change ofdirection

9(8.7)

9(10.2)

18(9.4)

Motor VehiclePosition BLs WCLs Total

Full stop 8(7.7)

9(10.2)

17(8.9)

Same direction asbike

68(73.1)

40(57.1)

108(66.3)

Collision or nearcrash

0(0.0)

2(2.3)

2(1.0)

Opposing directionto bike

3(3.2)

7(10.0)

10(6.1)

Total 104(54.2)

88(45.8)

192(100.0)

Cross street fromleft

3(3.2)

5(7.1)

8(4.9)

Cross street fromright

15(16.1)

10(14.3)

25(15.3)

Other 4(4.3)

8(11.4)

12(7.4)

Total 93(57.1)

70(42.9)

163(100.0)

Motor Vehicle

Driver AvoidanceResponse

BLs WCLs TotalSeriousness ofIntersection

ConflictBLs WCLs Total

No change in driving 61(64.9)

43(60.6)

104(63.0)

Minor 92(93.9)

84(91.3)

176(92.6)

Slows 19(20.2)

8(11.3)

27(16.4)

Serious 6(6.1)

8(8.7)

14(7.4)


5(5.3)

9(12.7)

14(8.5)

Total 98(51.6)

92(48.4)

190(100.0)


5(7.0)

8(4.9)

Major change ofdirection

0(0.0)

1(1.4)

1(0.6)

Full stop 6(6.4)

5(7.0)

11(6.7)


Total 94(57.0)

71(43.0)

165(100.0)

Table 15. Intersection conflict information.

Chapter 3 41

- Opposing direction - 4 percent of distributions were found.conflicts. The distributions were examined further

- Cross street from left - 2 percent of by utilizing the level of the scale (1 for noconflicts. change in riding, 2 for stops pedaling, etc.) to

- Cross street from right - 11 percent of develop a mean value for the responses. Forconflicts. the bicycle avoidance response, the BL mean

- Other position - 6 percent of conflicts. was 3.18 and the WCL mean was 3.24. Using

When the bicyclist made a left turn (24 cases), difference between the means, the differencethe motor vehicle position was: was not statistically significant.

- Same direction - 42 percent of conflicts. The motor vehicle avoidance responses- Opposing direction - 17 percent of were similarly compared. The BL mean was

conflicts. 1.72 and the WCL mean 1.99, and the- Cross street from left - 13 percent of difference was insignificant.

conflicts. The seriousness of the conflicts was- Cross street from right - 21 percent of coded as before. In the intersection area, 93

conflicts. percent of the conflicts were coded as minor- Other position - 8 percent of conflicts. in nature, and there were no differences in

When the bicycle made a right turn at the Bicycle actions more associated with BLsintersection (8 cases), the motor vehicle in these intersection conflicts (table 16)position was: included:

- Same direction - 63 percent of conflicts - The bicycle slowing/stopping/ swerving(5 cases). for intersection traffic.

- Opposing direction - 0 percent of - The bicycle slowing/stopping/ swervingconflicts. for traffic not influenced by the intersectionþ.

- Cross street from left - 13 percent of - The bicycle turning or swerving acrossconflicts (1 case). a lane of traffic.

- Cross street from right - 0 percent ofconflicts. Bicycle actions more associated with WCLs

- Other position - 25 percent of conflicts included:(2 cases). - No action taken.

Similar to midblock conflicts, the - Passing slow moving or stopped motoravoidance response scale for both bicycles and vehicles on the right.motor vehicles was used at intersections. - Encounters with pedestrians.Examining the full distribution for the - “Other” actions.intersection conflicts shows that bicyclists inWCLs had proportionally more of the seriousresponses, such as major direction changes,full stops, and near collisions; however, therewere no significant differences in thedistributions.

In regard to the motor vehicle avoidanceresponse scale, there was little variation in theBL and WCL distributions. Categories werecombined to enable a valid chi-square test, andno significant differences in the BL and WCL

a t-test to examine the hypothesis of no

the BL and WCL distributions.

42 Chapter 3

Bike Actions in Intersection Conflicts***

BLs WCLs Total

None 8(7.8)

10(11.1)

18(9.4)

Bike slowed/stopped/swerved forintersection traffic

30(29.4)

22(24.4)

52(27.1)

Bike slowed/stopped/swerved fortraffic not at intersection

37(36.3)

8(8.9)

45(23.4)

Bike turned or swerved across a lane oftraffic

7(6.9)

4(4.4)

11(5.7)

Bike passed slow or stopped vehicle onright

0(0.0)

10(11.1)

10(5.2)

Encounter with pedestrian 2(2.0)

16(17.8)

18(9.4)

Other bike actions 18(17.7)

20(22.2)

38(19.8)

Total 102(53.1)

90(46.9)

192(100.0)

*** p < .001

Motor Vehicle Actions in Intersection Conflicts

BLs WCLs Total

None 36(37.5)

28(43.1)

64(39.8)

Motor vehicle slowed/stopped/ swervedfor intersection traffic

11(11.5)

12(18.5)

23(14.3)

Motor vehicle turned right in front ofbike after overtaking

4(4.2)

7(10.8)

11(6.8)

Motor vehicle illegally parked inBL/WCL

7(7.3)

1(1.5)

8(5.0)

Motor vehicle entering/exiting on-streetparking

10(10.4)

6(9.2)

16(9.9)

Other motor vehicle actions 28(29.2)

11(16.9)

39(24.2)

Total 96(59.6)

65(40.4)

161(100.0)


Table 16. Intersection conflict bicycle and motor vehicle actions.

Examples of the “other” kinds of bicyclist Examples of the “other” kinds of motoractions (low frequencies of each) included vehicle actions (low frequencies of each)improper left turns, merging onto the road included a driver or passenger entering orfrom a sidewalk, and stunt riding. exiting a parked or stopped vehicle, and

Motor vehicle actions more associated crowding of the BL or edge of travel lane.with BLs in these intersection conflicts Motor vehicle actions more associated withincluded: WCLs included:

- Illegally parked in the BL. - No action taken.- Entering/exiting on-street parking. - Slowing/stopping/swerving for- “Other” motor vehicle actions. intersection traffic.

Chapter 3 43

- Turning right in front of a bicyclist after had 2 conflicts. For bicyclists in WCLs andovertaking. BLs, the percentages were 98.5, 1.5, and 0

Midblock and intersection combinedconflict rates

Combining the number of conflicts inboth the midblock and intersection areasresulted in a total of 237 conflicts at BL sitesand 132 at WCL sites. Conflict rates perentering bicyclist by type of facility were thefollowing:

All Conflicts (Bike/motor vehicle, bike/bike, and bike/pedestrian)

BL sites: 237 conflicts/2,657 bicyclists = 8.9 conflicts per 100 entering bicyclists WCL sites: 32 conflicts/1,932 bicyclists = 6.8 conflicts per 100 entering bicyclists

Bike/Motor Vehicle ConflictsBL sites: 177 conflicts/2,657 bicyclists = 6.7 conflicts per 100 entering bicyclistsWCL sites: 99 conflicts/1,932 bicyclists = 5.1 conflicts per 100 entering bicyclists

Thus, BL sites had slightly higher rates ofconflicts than WCL sites. Examining thebike/motor vehicle rates shows that more than6 percent of the bicyclists riding through theproject sites had a conflict with a motorvehicle. These raw frequency results are morethoroughly explored in the statistical modelingsection that follows.

Statistical Modeling of Conflicts Data

Analysis of midblock bike/motorvehicle conflicts

Each bicyclist observed in the midblockarea was coded as being involved in nobike/motor vehicle conflicts or as beinginvolved in one or more such conflicts.Overall, out of 4,342 observed bicyclists,4,222 (97.2 percent) had no conflicts, 114 (2.6percent) had 1 conflict, and 6 (.001 percent)

versus 96.4, 3.4, and 0.2 percent, respect-ively. Thus, the rate of midblock bike/ motorvehicle conflicts associated with BLs wasmore than double that associated with WCLsfor these small percentages. However, asexplained below, this difference cannot bedirectly or fully attributed to the type offacility. The analyses that follow represent anattempt to identify other factors that are associated with these conflicts and that mayaccount for some of the observeddifferences.

As noted earlier, a number of bicycliststurned left or right prior to reaching theintersection. In particular, 130 bicyclistscrossed to the left before the intersection and12 were involved in bike/motor vehicleconflicts. All 12 were associated with twoparticular BL sites that are examined in depthclinically in a subsequent section of thischapter.

Using data from those bicyclists notturning left prior to the intersection, ageneralized linear model was developed toexplore relationships between conflicts andfactors associated with site geometrics, motorvehicle traffic, bicyclist characteristics, etc. Inthis type of model, the number of midblockbike/motor vehicle conflicts (for a givenbicyclist) was taken as the response ordependent variable, y, which was assumed tofollow a Poisson distribution with mean µ. Itwas further assumed that the function N log (µ) =þ + þ þ X ,o j j

j=1where X X ... , X , are the explanatory1, 2, N

variables of interest and theþ’s are unknowncoefficients estimated by fitting the model todata. Significance tests of theþ’s show whichvariables are important and how they relateto conflicts. An example of the results offitting such a model is given in table 17.

44 Chapter 3

Table 17. Example of model results formidblock conflicts.


Constant -4.892 .0001

Traffic volume (vph) .001 .0058

Driveways .291 .0263

Bike lane 1.053 .0001

Parked vehicles -.294 .2685

The model of table 17 contains fourexplanatory variables: an hourly count ofmotor vehicle traffic traveling in the samedirection as the bicycle, number of intersectingdriveways in the midblock area, a variableindicating the presence of a BL, and a variableindicating that the bike is being ridden besideparked vehicles. The p-values show that thefirst three variables are statistically significant BL width (m) -1.22 .0001while the fourth is not. The magnitude andalgebraic sign of the estimated the relationships between mean conflicts andthe explanatory variables. Thus, the positiveestimate for the BL indicator variable showsthat even with the other variables taken intoaccount, the mean number of midblockconflicts is still greater in BLs than in WCLs.

Model development was continued bydeleting non-significant variables, such aspresence of parked vehicles, and trying otherpotential explanatory variables in the model. Anumber of other variables such as number ofmidblock intersecting streets; estimated vehiclespeed; number of lanes; bicycle compatibilityindex (BCI); bicyclist age and gender; and4

bicyclist experience level (a site specificvariable based on the results of the oralsurvey) were examined in the model and werealso found not to be significantly related tomidblock bike/motor vehicle conflicts. Anadditional variable that did appear to besignificantly associated with conflicts was BLwidth. Including this variable in the modelproduced the results shown in table 18.

Table 18. Midblock conflicts modelincluding bike lane width.


Constant -5.009 .0001

Traffic volume (vph) .001 .0121

Driveways .508 .0005

Bike lane 3.094 .0001

The BL indicator variable is coded aszero when no BL is present, and one when aBL is present. BL width was coded as zerowhen no BL was present, and the actualwidth (which varies from 1.2 to 2.9 m) whena BL was present. Thus, the last two variablesin the model contribute the amount w = 3.094 - 1.22 BL widthto the linear expression for log (µ) when a BLis present. This indicates that, for BLs withwidths less than 2.5 m, w is positive and theB

likelihood of a bike/motor vehicle conflict isgreater for the BL than for the

The BCI is a tool that can be used to4

assess the “bicycle friendliness” of a roadway.Variables used to develop the BCI for aroadway include a number of geometric and RD- 98-072, Development of the Bicycleoperational characteristics (e.g., curb lane Compatibility Index: A Level of Service Conceptwidth, traffic volume, and vehicle speed). For (Harkey, Reinfurt, Knuiman, Stewart, andmore detail, see FHWA Report No. FHWA- Sorton, 1998).

Chapter 3 45

Bike LaneWidth

Midblock Conflicts

Total0 1 2

0.0 1857(98.6)

27(1.4)

0(0.0)

1884

1.22 372(93.7)

24(6.1)

1(0.3)

397

1.46 113(89.7)

12(9.5)

1(0.8)

126

1.53 890(97.6)

21(2.3)

1(0.1)

912

1.59 39(90.7)

4(9.3)

0(0.0)

43

1.83 109(99.1)

1(0.9)

0(0.0)

110

1.98 197(100.0)

0(0.0)

0(0.0)

197

2.14 186(97.9)

4(2.1)

0(0.0)

190

2.35 161(96.4)

6(3.6)

0(0.0)

167

2.44 281(98.9)

3(1.1)

0(0.0)

284

2.90 146(98.0)

3(2.0)

0(0.0)

149

Total 4351 105 3 4459

Table 19. Bike lane width by midblock conflicts.

WCL. On the other hand, when the BL width attributed to the narrow BLs or to veryis greater than 2.5 m, w is negative and the specific characteristics of a few sites. TheseB

likelihood of a conflict is less for the BL than specific characteristics are examined in morea comparable WCL. To examine these results detail in the clinical examination later in thein more detail, table 19 gives a tabulation of chapter.midblock conflicts corresponding to each As another illustration of the difficultiesspecific BL width in the study. Note the 0.0 in separating general relationships fromwidth category corresponds to all WCLs. From specific site-related characteristics, two WCLthe table it is clear that the higher conflict rates sites were considered as being perhaps

correspond to some of the lowest BL widths. However, the widths of 1.4s and 1.59 m,which correspond to the highest conflict rates,represent one specific site each. Moreover,while four sites have widths of 1.22 m, thehigh conflict rate is again due to a single site,which contributes 257 observations and 21conflicts. The basic question then is whetheror not the higher conflict rates can really be

atypical in that some on-street parkingoccurred in the WCL during the videotaping.Although it can be argued that motor vehiclesparking in a bicycle facility is a frequent “real-world” condition, it was thought that perhapsdata from these sites should be excludedfrom the analyses. Sixty-two observationswere available from the first site. This site

had a high traffic volume (942 vehicles/h), 2

midblock driveways, and

46 Chapter 3

IntersectionMovement

Facility Type

Intersection Conflicts

0 1 2 Total

StraightThrough

WCL 985(96.4)

36(3.5)

1(0.1)

1022

BL 1544(97.0)

47(2.9)

1(0.1)

1592

LeftTurn

WCL 206(95.0)

7(3.2)

4(1.8)

217

BL 284(95.3)

14(4.7)

0(0.0)

298

RightTurn

WCL 160(98.8)

2(1.2)

0(0.0)

162

BL 297(98.0)

6(2.0)

0(0.0)

303

Table 20. Intersection conflicts by facility type and bicyclist movement.

a total of 9 midblock bike/motor vehicle type (BL or WCL) and the bicycle movementconflicts. The second site had a relatively low at the intersection. The table showstraffic volume (193 vehicles/h), 1 mid-block intersection conflict rates to be very similardriveway, and 1 conflict in 50 observations. for BLs and WCLs, and, in fact, when theWhen these 112 observations were deleted two factors (facility type and intersectionfrom the data set and the model of table 18 movement) were included in a model, facilityrefitted to the remaining data, both the type was not estimated traffic volume and driveway effects statistically significant (p = 0.339) whilebecame nonsignificant, (p = 0.126 and p = movement was (p = 0.010). More 0.176, respectively). Factors such as traffic specifically, bicycle left turn movements werevolume and presence of driveways would associated with higher conflict rates thanseem to be logically related to the right-turn or straight-through movements. occurrence of bike/motor vehicle conflicts In addition to simply noting whether thesince the opportunities for such conflicts facility was a BL or WCL, five differentshould increase with increasing values of these facility types were defined based on thefactors. With a limited number of sitesrepresenting a wide variety of configurations,however, it is very difficult to determinestatistically the exact nature of theserelationships. Thus, some clinical analysis wasnecessary to examine high conflict rate sites indetail and gain a better understanding of anydifferences between BLs and WCLs.

Intersection bike/motor vehicle conflicts

For those bicyclists who continued on tothe intersection, table 20 shows a tabulation ofbike/motor vehicle conflictsin the intersection area as a function of facility

nature of the BL or WCL at the intersection(figure 16) as shown below:

Type 1: BL striping continued tointersection (one site also had a bike pocket),

Type 2: BL terminated prior tointersection,

Type 3: BL dashed to intersection (again one site had a

bike pocket).Type 4: WCL continued to

intersection,Type 5: WCL narrowed (due to turn

lanes, etc.) at intersection.

Chapter 3 47

No separate types were considered for experienced bicyclists increased. No otherbike pockets since these only occurred at two bicyclist variables (e.g., age and gender) weresites having different types of striping. When significant.this new five-level type variable was includedin a model for intersection conflicts along withintersection movement, both factors weresignificant (p = 0.0004 for type and p =0.0017 for movement). Separate models were then estimated for left-turning bikes andfor those going straight or right. For left-turning bikes, the five-level type factor was notsignificant (p = 0.3580). However, the conflictrate for type 1 intersections (BL stripecontinued to intersections) was lower than thatfor the other types combined, at a levelapproaching significance, p = 0.0646. Ninety-seven percent of the bikes in left-turnmovements from type 1 intersections had noconflicts versus 94.2 percent from the othertypes combined.

For bicycles going straight or turning right,the five-level type factor was significant (p =0.0013). In particular, types 1 and 4, where theBL or WCL was unchanged to the intersection,tended to be associated with significantly lowerconflict rates than the other three types.Conflict frequencies and rates for bikestraveling straight through the intersection ormaking right turns are shown in table 21 forthe five intersection types.

Attempts were made to include a numberof other variables that might seem to belogically related to intersection conflicts suchas motor vehicle traffic volumes and speeds,number of driveways in the intersection area(i.e., within 90 m of the intersection stop bar),bicyclist experience level, etc., into the models.None of the geometric or traffic variableswere found to have significant or consistentmeaningful relationships with intersectionconflicts. However, bicyclist experience level,or more specifically the percentage ofexperienced bicyclists at a site, was statisticallysignificant with a p-value of 0.0037. Theestimated effect of this variable was -0.0275,indicating that the likelihood of intersectionconflicts decreased as the percentage of

48 Chapter 3

Type 3: BL dashed tointersection.

Type 1: BL stripingcontinued to intersection.

Type 2: BL terminated prior to intersection.

Type 4: WCLcontinued tointersection.

Type 5: WCLnarrowed atintersection.

Figure 16. BL and WCL intersection types.

Table 21. Intersection conflicts by intersection type for straight through and rightturning bicyclists.

Intersection TypeIntersection Conflicts

0.00 1 2 Total

BL continued to intersection 897 15 0 912(98.4) (1.6) (0.0)

BL terminated prior to 767 31 0 798intersection (96.1) (3.9) (0.0)

BL dashed to intersection 461 21 1 483(95.5) (4.4) (0.2)

WCL unchanged at intersection 869 18 5 892(97.4) (2.0) (0.6)

WCL narrowed at intersection 482 27 0 509(94.7) (5.3) (0.0)

Total 3476 112 6 3594

Chapter 3 49

Reanalysis of conflicts based on datafrom more “typical” sites

A reanalysis of the conflicts modelswas carried out using data from only 28 ofthe 48 sites. These 28 sites might bethought of as more “typical” of BL andWCL sites in the sense that sites with on-street parking as part of the bicycle facilitywere eliminated, as well as sites where theWCL was narrowed at the intersection dueto the provision of turn lanes. The datafrom these 28 sites included 2,566 of theoriginal 4,459 observations used in themidblock conflicts models. Sites 403 and615, where motor vehicles parked in partof the WCL during the videotaping, wereamong the excluded sites, and earlierresults had shown that eliminating thesesites would eliminate the traffic volumeand midblock driveway variables from themodel. This was also the case with thereanalysis. Results from a midblockconflicts model fitted to the “typical” sitesagain showed that BL presence and BLwidth were significant variables.

Models for intersection bike/motorvehicle conflicts were also reanalyzed. With the non-typical sites excluded, theconflict rate for left-turning bicycles nolonger differed significantly from thecombined rate for bikes traveling straightthrough the intersection or making rightturns. A model fitted to all bicycleintersection maneuvers (straight, left, andright) showed the conflict rate for type 3intersections (BL dashed to intersection) tobe significantly higher than for types 1 (BLstriping carried to intersection), 2 (BLterminated prior to intersection), and 4(WCL carried to intersection). Type 5 wasnon-typical and already eliminated. However, all the type 3 conflicts camefrom one busy intersection. There wereno significant differences between conflictrates for types 1 and 4, but the conflictrate for type 2 was significantly higher than

the rates for types 1 and 4. In addition,intersections with higher proportions ofexperienced bicyclists were significantlyassociated with lower intersection conflictrates.

The section of text that follows is aclinical examination of high rate conflictsites and offers some insight to the site-specific differences present in the data.

50 Chapter 3

A Clinical Analysis of HighConflict BL and WCL Sites

The lack of consistent results of themodeling efforts described in the precedingtext forced a closer examination of the siteswith the highest conflict rates (based onnumbers of entering bicycles). What followsis a clinical or in-depth look at each of thesesites to see if there were apparent factors orsituations that led to the conflicts.

Midblock conflicts

1. Intersection of State and Las Positas -Santa Barbara, CA

! WCL site with turn lanes added at theintersection.

! 11 conflicts, 17.7 conflicts per 100entering bicyclists.State and Las Positas is an intersection in a

busy bicycle corridor in Santa Barbara. Theintersection has four legs and is controlled bya traffic signal. The 5.2-m WCL is on StateStreet, which is four lanes, with AADT of33,000. The estimated driving speed is 56km/h. At the intersection, the two approach-ing lanes become five lanes. Beginning at thecenterline, there is a double left turn lane,then two straight through lanes, and then aright turn lane added by widening the

intersection. There are two intersectingdriveways in the midblock area. Ninety-minute parking is allowed on the street, andvehicles sometimes park in the WCL, butthe number of parked vehicles is generallysmall.

Of these 11 midblock conflicts, 9 werebike/motor vehicle. One of the nine wascoded as serious in nature and one unsure.In the serious conflict the bicyclist turned orswerved across a lane of traffic as themotor vehicle was entering or exiting an on-street parking location. Five of the bikeactions in the conflicts were having to slow,stop, or swerve for traffic not influenced bythe intersection (e.g., a motor vehicleentering the roadway from a driveway), andin one instance a second bike actioninvolved turning or swerving across a laneof traffic. In five other conflicts the bikeaction was “none” (i.e., no action detected).The most frequent motor vehicle actionswere entering/exiting on-street parking andturning right after overtaking a bike.

There was one bike/pedestrian conflict,and the seriousness was minor.

The most prevalent midblock conflict problemfor this location pertained to motor vehiclesentering/exiting from on-street parking.

2. Intersection of State and Cabrillo -Santa Barbara, CA

! BL site with dashed stripe leading tothe intersection.

! 45 conflicts, 16.2 conflicts per 100entering bicyclists.

Chapter 3 51

State and Cabrillo is an intersection in thetourist area of Santa Barbara. Theintersection has four legs and is signalcontrolled, but the leg opposite theapproaching BL is an entrance to a popularwharf that contains shops and restaurants. The 1.2-m BL is on State Street, which isfour lanes, with AADT of 9,000. Theestimated driving speed is 53 km/h. At theintersection, the two approaching lanesbecome three lanes. Beginning at thecenterline, there is a left turn lane, then a left-and-through lane, and then a right turn lane.There is one intersecting street in themidblock area.

Of these midblock conflicts, 26 werebike/motor vehicle, and 24 of these wereminor. In one of these serious conflicts, themotor vehicle turned right in front of thebicyclist after overtaking, and in the second abicyclist turned or swerved across a lane oftraffic. Twenty-three of the bike actions inthe conflicts were slowing/stopping/swerving for traffic not influenced by theintersection. In two cases the bike turned orswerved across a traffic lane, and in one casethe bike made an improper left turn. Themost frequent motor vehicle actions wereturning right after overtaking a bike, illegallyparking in the BL, and having a driver orpassenger enter or exit a parked or stoppedvehicle.

There were 10 bike/bike and 8 bike/pedestrian conflicts, and all were minor. Thebike/bike conflicts typically ocurred as thebicyclists maneuvered toward their desiredposition at the intersection.

The most prevalent midblock conflict problem forthis location pertained to the motor vehicle trafficturning right across the bike lane to enter the sidestreet, or parking in the BL to drop off or pick uppassengers. Conflicts also arose from the generallyhigh volumes of bike and pedestrian traffic.

3. Intersection of Coast Village and HotSprings - Santa Barbara, CA

! BL site with a short dashed stripeterminated prior to the intersection.


Coast Village and Hot Springs is a three-legged, stop-sign-controlled intersectionseveral miles away from the downtownarea of Santa Barbara. The approaching 1.5-m BL on Coast Village has a short dashedstripe at the end of the marking but isterminated well back from the intersectionbecause most bicyclists desire to shift to theleft to go straight through the intersection.Coast Village has two lanes and an AADTof 8,300. The estimated driving speed is 64km/h. At the intersection, the approachinglane becomes two lanes. Beginning at thecenterline, there is a straight through laneand then a right turn lane. There is an

52 Chapter 3

intersecting driveway from a parking area inthe midblock area.

Of these 10 midblock conflicts, 9 werebike/motor vehicle, and all were minor. Allof the bike actions in the conflicts wereslowing/stopping/swerving for traffic notinfluenced by the intersection. Eight of themotor vehicle actions were failing to yield at

a driveway entrance/exit. There was one bike/bike conflict, and it

was minor.The most prevalent midblock conflict problem for

this location pertained to the motor vehicle trafficentering or exiting the parking area driveway.

4. Intersection of SW 13th and SW 16th -Gainesville, FL

! BL site with stripe terminated justprior to the intersection.

! 19 conflicts, 15.6 conflicts per 100entering bicyclists.SW 13th and SW 16th is a four-legged,signal-controlled intersection south of themain campus of the University of Florida inGainesville. The 2.1-m BL is on SW 13thStreet, which has four lanes, with AADT of23,500. The estimated driving speed is 64km/h. At the intersection, the twoapproaching lanes become four lanes.Beginning at the centerline, there are two leftturn lanes, then a through lane, and then a

through-and-right lane. There is oneintersecting street and two driveways in themidblock area.

Of these 19 midblock conflicts, 15were bike/motor vehicle, and all wereminor. In 15 conflicts the bicycle wasmaking an advance crossover to the left.

Eleven of the bike actions in the 15bike/motor vehicle conflicts were turningor swerving across a lane of traffic. Themost frequent motor vehicle action was“none” (i.e., no action detected).

There were three bike/bike and onebike/pedestrian conflicts, and all wereminor.

The most prevalent midblock conflict problemfor this location pertained to bicyclists trying tomake an advance crossover to the left prior to thebusy intersection with its two left turn lanes formotor vehicle traffic.

5. Intersection of Speedway and 38th -Austin, TX

! BL site with stripe terminated priorto the intersection.

! 15 conflicts, 11.9 conflicts per 100entering bicyclists.Speedway and 38th is a four-legged, signal-controlled intersection north of the maincampus of the University of Texas inAustin. The 1.5-m BL is on Speedway,which has two lanes, with AADT of 5,600.

Chapter 3 53

The estimated driving speed is 53 km/h. Atthe intersection, the approaching lane splitsinto two through lanes. There is oneintersecting street and one driveway in themidblock area. This street has many busesshuttling students to campus, and this was amajor contributor to the conflicts.

Of these 15 midblock conflicts, 14 werebike/motor vehicle. Twelve were minor andthe two others were coded as unsure as towhether serious or minor. In one of these“unsure” conflicts, the motor vehicle failedto yield at a driveway, and in the second a

driver or passenger was entering or exiting aparked or stopped vehicle. Eleven of thebike actions in the conflicts were slowing/stopping/ swerving for traffic notinfluenced by the intersection. The mostfrequent motor vehicle actions were relatedto interacting with parked or stoppedvehicles (six cases where a driver orpassenger was entering or exiting a vehicle,three cases where a vehicle was parked orstopped in the BL, and two cases where avehicle was entering or exiting on-streetparking).

The most prevalent midblock conflict problem forthis location pertained to the interactions betweenbicycles and parked or stopped vehicles.

6. Intersection of Anacapa and Figueroa- Santa Barbara, CA

! WCL site with turn lanes added atthe intersection.

! 5 conflicts, 9.4 conflicts per 100entering bicyclists.Anacapa and Figueroa is a four-legged,signal-controlled intersection in the businessdistrict of Santa Barbara. The 5.2-m WCL ison the right hand side of Anacapa, which isa one-way street with two lanes and AADTof 12,300. The estimated driving speed is 53km/h. At the intersection, the twoapproaching lanes become three lanes.Beginning at the far left, there are twothrough lanes, and then a right turn lane.There is one intersecting street in themidblock area. Parking is allowed on theleft side of Anacapa (90 minutes, from 9a.m. to 6 p.m.).

Of these five midblock conflicts, allwere bike/motor vehicle and minor innature. All of the bike actions in the conflictswere slowing/stopping/swerving for trafficnot influenced by the intersection. Two ofthe motor vehicle actions involved illegalparking in the WCL and another motorvehicle turned right in front of a cyclist afterovertaking.

Although conflicts were few, there appeared tobe a possible midblock problem with parked orstopped vehicles.

7. Intersection of E. 30th and Speedway- Austin, TX

! BL site with stripe carried to theintersection.


54 Chapter 3

E. 30th and Speedway is an intersection closeto the University of Texas campus. Theintersection has five legs and is stop signcontrolled. The 1.6-m BL is on E. 30th,which is two lanes, with AADT of 6,300.The estimated driving speed is 53 km/h.There are no intersecting streets or drivewaysin the midblock area.

Of these four midblock conflicts, allwere bike/motor vehicle and minor. All ofthe bike actions in the conflicts wereslowing/stopping/swerving for traffic notinfluenced by the intersection. The mostfrequent motor vehicle actions were illegalparking in the BL.

The most prevalent midblock conflict problem forthis location pertained to motor vehicles illegallyparking in the BL.

8. Intersection of 22nd and Nueces -Austin, TX

! BL site with striping carried to the

intersection.! 7 conflicts, 7.9 conflicts per 100

entering bicyclists.22nd and Nueces is an intersection near thecampus of the University of Texas. Theintersection is four-legged and is stop signcontrolled. The 1.5-m BL is on 22nd,which is two lanes, with AADT of 4,100.The estimated driving speed is 32 km/h.There is one intersecting driveway in themidblock area.

Of these seven midblock conflicts, allwere bike/motor vehicle and minor in

nature. All of the bike actions in the conflictswere slowing/stopping/swerving for trafficnot influenced by the intersection. The mostfrequent motor vehicle actions were illegalparking in the BL (five cases) and a driveror passenger entering/exiting a parked orstopped vehicle (one case).

The most prevalent midblock conflict problemfor this location pertained to motor vehicles parkingor stopping in the BL.

9. Intersection of Barton Springs andDawson - Austin, TX

! WCL site with no turn lanes addedat the intersection.

! 4 conflicts, 7.7 conflicts per 100entering bicyclists.Barton Springs and Dawson is a four-legged, signal-controlled intersection southof the central business district in Austin. The

Chapter 3 55

4.1-m WCL is on Barton Springs, which isfour lanes, with AADT of 22,000. A two-way left turn lane is present. The estimateddriving speed is 56 km/h. There are nointersecting streets or driveways in themidblock area.

Of these four midblock conflicts, onewas a minor bike/motor vehicle conflict,and the other three were minor bike/pedestrian conflicts. Because of the trafficvolume, a good many bicyclists ride on theadjacent sidewalk along this street, whichwould lead to conflicts with pedestrians.

The most prevalent midblock conflict problem forthis location pertained to bicyclists interacting withpedestrians on the sidewalk.

10. Intersection of Nueces and 22nd -Austin, TX

! BL site with striping carried to theintersection.


Nueces and 22nd is an intersection near thecampus of the University of Texas. Theintersection is four-legged and is stop signcontrolled. [Note: The other side of theintersection was also videotaped (see No. 8)

because of the frequency of bicycle riders.]The 1.2-m BL is on Nueces, which is a one-way street with two approach lanes andAADT of 3,500. The estimated driving

speed is 40 km/h. There is one intersectingdriveway in the midblock area.

Of these three midblock conflicts, twowere bike/motor vehicle, and both wereminor. One of the bike actions in theconflicts was for crowding vehicles in thetravel lane and the other was “none.” Themotor vehicle actions were crowding theBL and being parked in the BL. There wasone bike/pedestrian conflict, and theseriousness was minor.

Although the conflict rate per entering bicyclistwas high, there were not enough conflicts to identify aprevalent midblock conflict problem for this location.

Intersection conflicts

1. Intersection of Barton Springs andDawson - Austin, TX

! WCL site with no turn lanes addedat the intersection.

! 13 conflicts, 25.0 conflicts per 100entering bicyclists.Barton Springs and Dawson is a four-legged, signal-controlled intersection southof the central business district in Austin. The4.1-m WCL is on Barton Springs, which hasfour lanes, with AADT of 22,000. A two-way left turn lane is present. The estimateddriving speed is 56 km/h. There are nointersecting streets or driveways in the

intersection area.

56 Chapter 3

Of these 13 intersection conflicts, 12were bike/motor vehicle, and one of thesewas serious. In this conflict the bicyclistturned or swerved across a lane of traffic.For the bike/motor vehicle conflicts, sevenof the bicycles were going straight throughthe intersection, three making left turns, andtwo making right turns. The bike actions inthe conflicts were bicyclist failing to stop atthe traffic signal (four conflicts), turning orswerving across a lane of traffic (twoconflicts), passing slow or stopped vehicleson the right (two conflicts), and in oneconflict the bicycle entered the roadwayfrom the sidewalk. In four other conflicts thebike action was “none” (i.e., no actiondetected). The most frequent motor vehicleactions were slowing, stopping, or swervingdue to intersection traffic (eight conflicts). Infour conflicts the motor vehicle action was“none.” There was one bike/pedestrianconflict, and the seriousness was minor.

This was quite a busy intersection at late-daypeak hour, and instead of a “most prevalent conflictproblem,” it would appear that conflicts take placefrom a variety of bicyclist actions in maneuveringthrough this busy intersection.

2. Intersection of 43rd and Duval -Austin, TX

! WCL site with no turn lanes added atthe intersection.

! 12 conflicts, 23.1 conflicts per 100entering bicyclists.43rd and Duval is an offset four-legged,stopsign-controlled intersection in aneighbor- hood area north of the Universityof Texas campus in Austin. The 5.7-m WCLis on 43rd, which has two lanes, with AADTof 1,200. The estimated driving speed is 53km/h. There is one intersecting driveway inthe intersection area. The street is usually freeof parked motor vehicles, but not always.

Of these 12 intersection conflicts, 10were bike/motor vehicle, and all were

minor. For the bike/motor vehicle conflicts,seven of the bicycles were making a left turnat the intersection, one a right turn, and twosome other movement (e.g., pulling into astore area before getting through theintersection). Although most of the bikeactions in the conflicts were slowing orstopping or swerving for intersectiontraffic, in one conflict a bicyclist made animproper left turn and in another conflict abicyclist failed to yield when changing lanes.The most frequent motor vehicle actionswere entering or exiting on-street parking(five conflicts).

There were two bike/pedestrian con-flicts, and the seriousness was minor inboth.

The most prevalent intersection conflict problemfor this location pertained to bicycle interactions withparked vehicles.

3. Intersection of Speedway and 43rd -Austin, TX

! BL site with striping terminatedprior to the intersection.


Chapter 3 57

Speedway and 43rd is a four-legged, stop-sign-controlled intersection in a busy bicyclecorridor that is used by bicyclists to accessthe University of Texas campus in Austin.The 1.5-m BL is on Speedway, which is twolanes, with AADT of 5,600. The estimateddriving speed is 56 km/h. There are twointersecting driveways in the intersection areathat lead to a post office.

Of these 11 intersection conflicts, allwere bike/motor vehicle, and none wereserious. Ten of the bicycles were goingstraight through the intersection and one wasmaking a left turn. Nine of the bike actions inthe conflicts were bicyclist having to slow,stop, or swerve for traffic not influenced bythe intersection. In two other conflicts thebike action was “none” (i.e., no actiondetected). The most frequent motor vehicleactions were related to parking: illegallyparking in the BL (six conflicts), entering orexiting on-street parking (three conflicts), anda driver or passenger entering or exitingfrom a parked or stopped vehicle (oneconflict). One other motor vehicle actioninvolved turning right in front of the bicyclistafter overtaking.

The most prevalent intersection conflict problemfor this location pertained to bicycle interactions withparked or stopped vehicles near a busy post office.

4. Intersection of 34th and Guadalupe -Austin, TX

! WCL site with turn lanes added at theintersection.

! 7 conflicts, 17.1 conflicts per 100entering bicyclists.34th and Guadalupe is a four-legged, signal-controlled intersection west of theUniversity of Texas campus in Austin. The5.6-m WCL is on 34th, which has twolanes, with AADT of 8,950. At theintersection, the approaching lane becomestwo lanes. Beginning at the centerline, thereis a left turn lane, and then a through andright lane. The estimated driving speed is 48km/h. There is one intersecting driveway inthe intersection area. This is an intersectionat which many bicyclists approach in theWCL but then turn right to head towardcampus or the downtown area.

Of these seven intersection conflicts, sixwere bike/motor vehicle, and all wereminor. For the bike/motor vehicle conflicts,four of the bicycles were going straightthrough the intersection, one making a rightturn, and one was not coded. Five of thebike actions in the conflicts were having toslow, stop, or swerve for intersectiontraffic, and in one conflict a bicyclist waspassing slow or stopped vehicles on theright. The most frequent motor vehicleactions were slowing, stopping, or swervingdue to intersection traffic (three conflicts). Infour conflicts the motor vehicle action was“none.” In one conflict the motor vehiclewas crowding the bicyclist in the WCL.

58 Chapter 3

There was one bike/pedestrian conflict, andthe seriousness was minor.

Instead of a “most prevalent conflict problem,” itwould appear that conflicts take place as a result ofnormal intersection maneuvering.

5. Intersection of SW 13th and SW 16th -Gainesville, FL

! BL site with stripe terminated justprior to the intersection.

! 20 conflicts, 16.4 conflicts per 100entering bicyclists.SW 13th and SW 16th is a four-legged,signal-controlled intersection south of themain campus of the University of Florida inGainesville. The 2.1-m BL is on SW 13thStreet, which has four lanes, with AADT of23,500. The estimated driving speed is 64km/h. At the intersection, the twoapproaching lanes become four lanes.Beginning at the centerline, there are two leftturn lanes, then a through lane, and then athrough-and-right lane. There are threeintersecting driveways in the intersection area.

Of these 20 intersection conflicts, 18were bike/motor vehicle, and one wasserious. In this serious conflict the bicyclistturned or swerved across a lane of trafficand the motor vehicle used an improperlane. For the bike/motor vehicle conflicts, sixof the bicycles were going straight throughthe intersection, five making left turns, andone a right turn. An “other” type ofmovement was taking place, such as cuttingacross the street to the left, in seven conflicts.The most frequent bike actions in theconflicts involved a bicycle turning orswerving across a lane of traffic (sixconflicts), crowding vehicles in the travel lane(four conflicts), and having to slow, stop, orswerve for intersection traffic (four conflicts).In another conflict a bicyclist was riding inbetween slow or stopped vehicles. The mostfrequent motor vehicle actions were slowing,stopping, or swerving due to intersection

traffic (10 conflicts). In seven conflicts the

motor vehicle action was none. In oneconflict the motor vehicle was using animproper lane (e.g., through from a rightlane).

There was one bike/bike conflict, andthe seriousness was minor.

The most prevalent intersection conflict problemfor this location pertained to bicyclists trying to crossthe street to the left prior to the busy intersectionwith its two left turn lanes for motor vehicle traffic.The speed and volume of traffic was such that it wasrare that a bicyclist made a motor vehicle style leftturn. In many cases an advance crossover to the leftwas made before reaching the intersection area.

6. Intersection of State and Cabrillo -Santa Barbara, CA

! BL site with dashed stripe leading tothe intersection.


Chapter 3 59

State and Cabrillo is an intersection in thetourist area of Santa Barbara. Theintersection is four-legged and signalcontrolled, but the leg opposite theapproaching BL is an entrance to a popularwharf that contains shops and restaurants. The 1.2-m BL is on State Street, which hasfour lanes, with AADT of 9,000. Theestimated driving speed is 53 km/h. At theintersection, the two approaching lanesbecome three lanes. Beginning at thecenterline, there is a left turn lane, then a left-and-through lane, and then a right turn lane.There are two intersecting driveways in theintersection area.

Of these 35 intersection conflicts, 21were bike/motor vehicle, and five of thesewere serious. In two of these serious

conflicts the motor vehicle failed to yield at adriveway, and in the other three the motorvehicle made an “other,” or non-standard,movement. For the bike/motor vehicleconflicts, 14 of the bicycles were goingstraight through the intersection, 3 makingleft turns, and 2 making a right turn. Twowere making some other kind of movement.Twelve of the bike actions in the conflictswere for slowing, stopping, or swerving forintersection traffic. In eight conflicts the bikeactions were for slowing, stopping, orswerving for other traffic not influenced bythe intersection. In one conflict a bicyclistused an improper lane, such as straight

through from a right turn lane. The mostfrequent motor vehicle actions were “none”(seven conflicts), and making an “other”movement (seven conflicts). Other actionsincluded failing to yield at a driveway (twoconflicts), entering the BL to make a rightturn but without yielding to the bicyclist(two conflicts), crowding (one conflict), andhaving a driver or passenger enter or exit aparked or stopped vehicle (one conflict).

There were 12 bike/bike and 2 bike/pedestrian conflicts, and all were minorexcept for one bike/bike conflict where abicyclist was riding erratically.

The most prevalent intersection conflict problemfor this location pertained to the weaving of bikesamong themselves and with motor vehicles after thesolid BL stripe ended. There was also a variety ofmaneuvering taking place at the intersection proper. Conflicts also arose from the generally high volumesof bike and pedestrian traffic.

7. Intersection of Speedway and 26th -Austin, TX

! WCL site with parking and with turnlanes added at the intersection.

! 21 conflicts, 10.8 conflicts per 100entering bicyclists.Speedway and 26th is a four-legged, signal-controlled intersection near the heart of theUniversity of Texas campus in Austin. The7.3-m WCL is on Speedway, which has twolanes, with AADT of 6,400. The estimateddriving speed is 40 km/h. The approach

60 Chapter 3

7.3-m WCL narrows to 4.1 m when angledparking and the addition of a left turn lane

begin about 85 m from the intersection.There are no intersecting streets or drivewaysin the intersection area. The angled parkingarea is usually filled with parked motorvehicles.

Of these 21 intersection conflicts, 17were bike/motor vehicle, and 3 of thesewere serious. Two of these occurred when abicyclist stopped at the controlledintersection but failed to yield. The thirdinvolved a motor vehicle turning right infront of a bicyclist after overtaking. For thebike/motor vehicle conflicts, 12 of thebicycles were going straight through, and 5were making a left turn at the intersection.Although most of the bike actions in theconflicts were slowing, stopping, or swervingfor intersection traffic (12 conflicts), in twoconflicts the bicyclist stopped at theintersection but failed to yield, and in oneconflict the bicyclist failed to stop at thetraffic signal. The most frequent motorvehicle action was “none.”

There was one bike/bike conflict andtwo bike/pedestrian conflicts, and theseriousness was minor in all three.

The most prevalent intersection conflict problemfor this location pertained to the variety ofmaneuvering taking place at the intersection proper.Interactions between bicycles and parked motorvehicles were not a problem.

8. Intersection of Rio Grande and 24th -Austin, TX

! WCL site with turn lanes added atthe intersection.

! 8 conflicts, 10.4 conflicts per 100entering bicyclists.Rio Grande and 24th is a four-legged,signal-controlled intersection near the heartof the University of Texas campus inAustin. The 4.5-m WCL is on Rio Grande,which is a one-way street with two lanes

and AADT of 6,600. The estimated drivingspeed is 40 km/h. The approach WCLnarrows at the intersection with the additionof a left turn lane. There is one intersectingstreet and one driveway in the intersectionarea.

Of these eight intersection conflicts,seven were bike/motor vehicle, and nonewere serious. For the bike/motor vehicleconflicts, five of the bicycles were goingstraight through, and two were makinganother kind of maneuver near theintersection. The most frequent bike actionin the conflicts was passing a slow movingor stopped vehicle on the right (fiveconflicts). Bicyclists would often make thismaneuver to get up to the front of theintersection. The traffic signal has a long redphase, and often a bicyclist would stop andthen proceed straight through if the crossstreet traffic allowed. The most frequentmotor vehicle action was “none.”

Chapter 3 61

There was one bike/pedestrian conflict,and the seriousness was minor.

The most prevalent intersection conflict problemfor this location pertained to bicyclists passing slowmoving or stopped vehicles on the right.

9. Intersection of SW 13th and University- Gainesville, FL

! WCL site with full width carriedthrough the intersection.

! 11 conflicts, 6.7 conflicts per 100

entering bicyclists.SW 13th and University is a four-legged,signal-controlled intersection near the heartof the main campus of the University ofFlorida in Gainesville. The 4.4-m WCL is onSW 13th Street, which has four lanes, withAADT of 32,000. The estimated drivingspeed is 56 km/h. At the intersection, thetwo approaching lanes become three lanes asa left turn lane is added. However, the fullwidth of the WCL is carried through theintersection. There is one intersecting streetand two driveways in the intersection area.

Of these 11 intersection conflicts, 6 werebike/motor vehicle, and none were serious.For the bike/motor vehicle conflicts, four ofthe bicycles were going straight through theintersection, one making a right turn, and one“other” type of maneuver. The mostfrequent bike actions in the conflicts involveda bicycle passing slow moving or stoppedvehicles on the right (two conflicts), and

having to slow, stop, or swerve for trafficnot influenced by the intersection (twoconflicts). The most frequent motor vehicleaction was “none.” In one conflict themotor vehicle turned right in front of thebicyclist after overtaking.

There was one bike/bike and threebike/pedestrian conflicts, and theseriousness was minor in all four.

The most prevalent intersection conflict problemfor this location pertained to bicyclists passing motorvehicles on the right and interacting with pedestriansat the intersection. This is a busy intersection formotor vehicles, bikes, and pedestrians. Manybicyclists entering from the north will ride to thefront of the intersection, move to the sidewalk, andthen cross in the crosswalk area before turning rightto head into campus.

10. Intersection of Rio Grande and 21st- Austin, TX

! BL site with striping carrried to the

intersection.! 3 conflicts, 6.4 conflicts per 100

entering bicyclists.

Rio Grande and 21st is a four-legged, stop-sign-controlled intersection near the heart ofthe University of Texas campus in Austin.The 1.2-m BL is on Rio Grande, which is aone-way street with two lanes and AADTof 4,100. The estimated driving speed is 40

62 Chapter 3

km/h. There is one intersecting street in theintersection area.

Of these three intersection conflicts, allwere bike/motor vehicle. Two were codedas minor in nature and one as unsure. For thebike/motor vehicle conflicts, one of thebicycles was going straight through, oneright, and one left. In two of the conflicts thebikes were having to slow, stop, or swervefor traffic not influenced by the intersection,and in the third conflict the bicyclist made animproper left turn. In two of the conflictsthe motor vehicle action was a driver orpassenger entering or exiting a parked orstopped car.

With only a few conflicts, no prevalentintersection conflict problem for this location could beidentified with certainty, although parked motorvehicles again appear to be a problem.

62 Chapter 3

Examination of Serious Conflicts

There were 17 serious conflicts coded fromthe videotape, 3 in the midblock (i.e.,intersection approach) area and 14 at theintersection. These are described in the textthat follows.

Serious midblock conflicts

Intersection of State and Cabrillo - BL SiteSanta Barbara, CABicyclists were proceeding toward theintersection in the BL. A motorist overtookthe bicyclists and then turned right into anintersecting street. The bicyclists had toswerve to avoid the turning vehicle.

Intersection of State and Cabrillo - BL Site

Santa Barbara, CA

The bicyclist was approaching theintersection in the BL. He switched to theadjacent traffic lane and then turned across alane of traffic to cross the street to the left.The motorist applied brakes to avoid thebicyclist.

Intersection of State and Las Positas -WCL SiteSanta Barbara, CAThe bicyclist was approaching in a WCL. Astraffic stopped for a motorist who wasbacking into a parking space, the bicyclistswerved quickly into the adjacent traffic laneand then back into the WCL.

Serious intersection conflicts

Intersection of SW13th and SW 16 TH -BL Site

Chapter 3 63

Gainesville, FL

The bicyclist approached in the BL and thenswitched to the adjacent traffic lanes as hecrossed the street to the left prior to theintersection. The motorist was trying tomove to the left turn lane by going throughthe channelized area. The bicyclist made arapid direction change.

Intersection of SW13th and SW8th - WCL SiteGainesville, FLThe bicyclist was approaching the intersec-tion in the WCL and was overtaken by amotorist who then started to turn right into adriveway. Both parties came to a full stop.

Intersection of State and Cabrillo - BL SiteSanta Barbara, CA(Note: Two serious conflicts)

The bicyclist was approaching the intersec-tion in the BL. The motorist was pulling outfrom a driveway and entered the BL. Thebicyclist swerved to the left and came to afull stop. A second bicyclist (behind themotorcyclist) also had a serious conflict withthis motorist.

Intersection of State and Cabrillo - BL SiteSanta Barbara, CA(Note: Two serious conflicts)Several bicyclists were approaching theintersection in the BL. The motorist passedthe bicyclists and then turned right into adriveway. Two bicyclists had to changedirection quickly to avoid being struck bythe motorist.

Intersection of State and Cabrillo - BL Site

64 Chapter 3

Santa Barbara, CAThe bicyclist was approaching theintersection in the BL. A passenger in amotor vehicle opened the right rear door,and the bicyclist had to swerve to the right toavoid striking the door.

Intersection of State and Las Positas -WCL SiteSanta Barbara, CAA young male bicyclist was riding on thesidewalk approaching the intersection when amotorist pulled from a driveway into thesidewalk area. The bicyclist did a quick u-turnto avoid the vehicle.

Intersection of State and Ontare - WCL SiteSanta Barbara, CAThe motorist and bicyclist were approachingthe intersection in the WCL. The motorist

overtook the bicyclist and turned right intoa driveway. The bicyclist had to quicklyswing around the vehicle.

Intersection of Barton Springs andDawson - WCL SiteAustin, TXThe bicyclist approached in the WCL andthen crossed over to the two-way left turnlane. He then attempted to continue to crossto the left but had to brake and changedirection quickly when he decided there wasnot time to cross in front of an oncomingmotorist. The motorist had to brake.

Intersection of Guadalupe and 30th -WCL SiteAustin, TXThe bicyclist was going through theintersection in the WCL. He decided to shiftto the left to avoid a section of rough

Chapter 3 65

pavement. A following motorist had to shiftto the left quickly to avoid the bicyclist.

Intersection of Speedway and 26th -WCL SiteAustin, TXThe bicyclist and motorist were near theintersection. From the inside lane, themotorist turned right across the outside lane.The bicyclist had to adjust quickly to avoidthe vehicle.

Intersection of Speedway and 26th -WCL SiteAustin, TX(Note: Two serious conflicts)Two bicyclists had approached in the WCL.Both decided to disobey the traffic signal but

had to quickly stop when a motoristapproached from their left.

For the 17 serious conflicts describedabove, the primary causes were thefollowing:

! Motorist turned right soon afterovertaking bicyclist - six conflicts

! Motorist pulling from driveway tostreet - three conflicts

! Motorist parking maneuver/passenger exiting stopped vehicle - two conflicts

! Bicyclist turned or swerved across alane of traffic - three conflicts

! Bicyclist disobeyed traffic signal -two conflicts

! Bicyclist shifted to path of motoristto avoid rough pavement - one conflict

66 Chapter 3

Comparisons with Crash Data

In addition to videotaping thebehavior of bicyclists at WCL and BLsites,two years (1994 and 1995) of recentpolice-reported bicycle-motor vehiclecrash data were obtained for each of thethree participating cities (Gainesville,Austin, and Santa Barbara). Actual hardcopies of the police reports werereviewed by project staff, and the 1995crashes were “typed” following themethodology originally developed byNHTSA1 and being modified inpartnership with FHWA for computerapplication. The computer software willbe known as PBCAT (Pedestrian andBicycle Crash Analysis Tool). Althoughthe crash typing was done primarily toidentify and examine interesting parallelsto the videotape data, the typing alsoprovided a way to generate some usefulfeedback to the participating cities. Thissection highlights the results of these crashand videotape comparisons.

At the outset, it should beemphasized that, whereas the crash datareflect events occurring throughout a city,the videotape data are limited to 16intersections within that city with veryspecific, predefined characteristics (e.g., aBL or WCL on a low-speed roadway). Thus, the videotape data cannot beregarded as “exposure data” for thecrash events. On the other hand, the 16selected sites in each of the cities dorepresent locations with high levels ofbicycle traffic, so that any patterns ofriding behavior at these sites mightreasonably be expected to extend to the

city as a whole, and be reflected in theircrash data.

Table 22 summarizes the primarycrash types identified for each city.During 1995, there were 158 reportedbicycle-motor vehicle crashes in Gaines-ville, 173 in Austin, and 77 in SantaBarbara. Overall, the most frequentlyoccurring crash types were on average: Average

PercentMotorist Drive Out at Stop Sign 16.4 Motorist Drive Out at Midblock 10.9 Bicyclist Ride Out at Intersection 10.2

However, there are distinctive crash typesthat stand out in each city. In SantaBarbara, for example, one of the twomost frequently occurring crash types wasthe Bicyclist Striking a Parked Vehicle: 16percent of all reported crashes. A typicalsituation was the cyclist riding alongside arow of parked vehicles and running intoan opening car door. Only 3 percent ofAustin’s crashes were parked vehicle-related, and none of the crashes inGainesville. From the videotape data, 41percent of the Santa Barbara bicyclistswere recorded as riding next to parkedvehicles (e.g., a combined BL and parkingfacility), 21 percent of the Austinbicyclists, and none of the Gainesvillebicyclists. Thus, the availability of bicyclefacilities with on-street parking appears tobe directly related to the higherproportion of crash types involvingparked vehicles.

In Gainesville, the most frequentlyoccurring crash type was Motorist DriveOut at Stop Sign — nearly a fourth of itsreported crashes. In three out of four ofthese crashes, the bicyclist was travelingon the sidewalk, facing traffic (i.e.,approaching from the motorist’s right). The videotape data also exhibited a high

1For more complete backgroundon bicycle-motor vehicle crash typing, seeHunter, Stutts, Pein, and Cox (1996).

Chapter 3 67

Rank Gainesville(N=158)

Austin(N=173)

Santa Barbara(N=77)

Overall *

1 Drive Out at Stop Sign (24.7%)

Drive Out at Stop Sign (15.6%)

Bicyclist Strikes ParkedVehicle (15.6%)

Drive Out at StopSign (16.4%)

2 Right Turn On Red (14.6%)

Ride Out at Intersection (13.9)

Motorist Right Turn(15.6%)

Drive Out atMidblock (10.9%)

3 Drive Out at Midblock (12.7%)

Ride Out at Midblock (11.6%)

Drive Out at Midblock (13.0%)

Ride Out atIntersection (10.2)

4 Ride Out at Intersection (10.8%)

Motorist Left Turn- FacingBicyclist (8.7%)

Drive Out at Stop Sign (9.1%)

Motorist Right Turn(8.3)

5 Motorist Left Turn-Facing Bicyclist (7.0%)

Drive Out at Midblock (6.9%)

Motorist Overtaking (7.8%)

Right Turn on Red(7.6)

6 Motorist Right Turn(6.3%)

Right Turn On Red (6.9%)

Ride Out at Intersection(6.5%)

On-Street ParkingRelated (7.1)

7 Ride Out at Midblock (5.1%)

Bicyclist Left Turn In FrontOf Motorist (5.8%)

Motorist Left Turn- FacingBicyclist (5.2%)

Motorist Left Turn -Facing Bicyclist (6.9)

8 Drive Through(4.4%)

Assault (4.0%)

Bicyclist Lost Control (5.2%)

Ride Out at Midblock(6.0)

9 Motorist Left Turn-Bicyclist Same Direction(4.0%)

Drive Out From On-StreetParking (3.9%)

Motorist Overtaking(3.9)

10 Motorist Overtaking (4.0%)

85.4% of Total 81.5% of Total 77.9% of Total 77.3% of Total

* Percentage based on an average of the percents for each city (i.e., not weighted by sample size).

Table 22. Most frequent occurring crash types rank ordered in each of the three study sites, based on 1995 police-reported crash data.

level of sidewalk and wrong way ridingin Gainesville. Seventeen percent of theGainesville bicyclists captured on

videotape were observed approachingthe

targeted intersections on the sidewalk; thiscompares with less than 3 percent of theAustin bicyclists and less than 2 percent ofthe Santa Barbara bicyclists. In addition,9 percent of Gainesville bicyclists wereobserved traveling the “wrong” direction(i.e., facing oncoming motor vehicle

traffic) on a sidewalk, compared withonly 1 percent of both the Austin andSanta Barbara bicyclists. Wrong-waysidewalk riding was also afactor in 87 percent of Gainesville’s RightTurn on Red crashes, and 75 percent oftheir Drive Out at Midblock crashes.

68 Chapter 3

Sidewalk riding, and in particular wrong-way sidewalk riding, was clearly acontributing factor in Gainesville’scrashes.

Ride-out at Intersection crashes occurwhen the bicyclist runs a stop sign ortraffic signal, or fails to yield at anuncontrolled intersection, and strikes avehicle approaching on a crossing path. This crash type occurred most frequentlyin Austin (14 percent), followed byGainesville (11 percent) and then SantaBarbara (6 percent). The percentages ofbicyclists observed on the videotapesdisobeying signals or stop signs (mostlythe latter) were:

% Disobeying

Signal

%DisobeyingStop Sign

GainesvilleAustinSanta Barbara

1285

382017

The overall lower rates for Santa Barbaraparallel the crash results. Gainesville hada considerably higher rate of “ride out”behavior than Austin, however, which isnot reflected in the crash data. Part ofthis may be due to the locations selectedfor the videotaping and the nature of thetraffic flow at these locations. In particu-lar, there were several sites in Gainesvillewhere many bicyclists ran stop signs, butwhere motor vehicles had adapted to thisbehavior and crash risk was minimal.

Other riding behaviors observedamong the videotaped Gainesvillebicyclists were displayed in the crash data. For example, there was a much higherpercentage of incorrect left turnsobserved among the Gainesville riders —11 percent of the riders made impropermotor vehicle style turns, and nearly 25percent made a left turn from the bikelane (most of these from a single site, a t-intersection where almost all of the bikeand motor vehicle traffic made a left

turn). This compares with 5 percent orfewer of the Austin and Santa Barbarabicyclists noted for such movements. Only one of Gainesville’s reportedcrashes, however, involved a left-turningbicyclist. Gainesville cyclists were alsomore likely to employ a non-standardright turn method (17 percent of rightturn movements, compared with 11 per-cent for both Austin and Santa Barbara).But although Gainesville had more right-turn crashes than the other sites, overallnumbers were low (three right-turningcrashes for Gainesville, one in Austin, andnone in Santa Barbara).

The one left-turning behaviorobserved in the videotape data that mighthave a parallel in the crash data is a higherpercentage of “advance crossover” intersection maneuvers for the Austinbicyclists (11 percent of all left turns,compared with 3 percent in bothGainesville and Santa Barbara). Nearly 6 percent of the crashes reported forAustin were Bicyclist Left Turn in Frontof Motorist. None of these type crasheswere reported for Gainesville or SantaBarbara.

Clearly, one cannot expect behaviorsobserved on one or two occasions at arelatively small subset of intersections tomimic precisely overall riding behaviorwithin a city. Nevertheless, certainconditions — on-street parking, sidewalkriding, wrong-way riding, even heavytraffic that makes turning maneuversdifficult — may persist and may indeedhave counterparts in the crash data. Thecomparisons drawn in this section wouldlikely have been stronger if youngchildren had been excluded from thecrash data. Children under age 15 wereinvolved in 27 percent of the Austincrashes, 15 percent of the Gainesvillecrashes, and 10 percent of the SantaBarbara crashes. In contrast, less than

Chapter 3 69

1 percent of the videotaped bicyclistswere categorized as children. Anotherdifference in the two samples is that all ofthe videotaping was done during thedaytime, whereas the percentage ofcrashes reported under daylightconditions in the three cities varied from81 to 89 percent. In each city, however,certain riding conditions prevail, and theseappear to have been reflected in both thevideotape and the crash data.

Chapter

4 Discussion

This comparative analysis was based onvideotapes of almost 4,600 bicyclists in threeU.S. cities approaching and then ridingthrough intersections for which theassociated bicycle facility was either a BL orWCL. In two of the three cities, the vastmajority of bicyclists were traveling to orfrom college campuses, and the intersectionsselected were generally in bicycle commutingcorridors. The intent was to videotapebicyclists who regularly ride in traffic. Theresult was a group of sites with varying “real-world” characteristics such as different BLstriping techniques (e.g., using a solid ordashed BL stripe all the way to theintersection), presence of parking (e.g., acombination BL and parking lane), andprovision of turn lanes at intersections thatsometimes narrow the nominal width of theBL or WCL at the intersection proper. Whatfollows is a brief summary of the mainoperational and safety (conflict) results andsome further elaboration of key issues.

Summary of Main Results

Bicyclist characteristics! The overwhelming majority of

videotaped bicyclists were between the agesof 16 and 64. Slightly more than three-fourths were male.

! Overall, 5.6 percent of the bicyclistswere riding the wrong way (i.e., facingtraffic). This included 1.3 percent in the roadand 4.3 percent on sidewalks. However,wrong-way riding was much more prevalent

on the sidewalk at WCL sites (7.0 percent)compared with BL sites (2.3 percent).Eliminating sidewalk riding from thecomparison, however, still resulted insignificantly more wrong-way ridingassociated with WCL sites (1.7 percent) thanBL sites (1.0 percent).

! A bicyclist experience oral surveywas administered to bicyclists proceedingthrough the project sites on days whenvideotaping was not being done. Therewere no statistically significant differences inthe age, gender, and helmet use of bicyclistsby type of facility. Higher proportions ofWhites and Blacks rode in WCL situationsand higher proportions of Asians andHispanics in BL situations, and thedifferences were significant.

! Bicyclists surveyed at WCL sitestended to ride more days per week, but themiles per week for bicyclists at BL versusWCL sites were equivalent. Overall, aboutone-third of the riders at both BL andWCL sites considered themselves to beexperienced bicyclists.

! When bicyclists were surveyed, theirriding location (i.e., in the street or on thesidewalk) when approaching the surveystation was recorded. Surveyed bicyclistsshowed the same tendency as thevideotaped bicyclists in that sidewalk ridingwas more associated with WCL sites.

Midblock movements! In the midblock or intersection

approach area (between 90 and 150 mfrom the intersection), significantly moremotor vehicles passing bicycles on the leftencroached into the adjacent traffic lanefrom WCL situations (17 percent)compared with BL situations (7 percent).This is in agreement with results from arecent Florida DOT study (Harkey andStewart, 1997). However, encroachmentsinto the adjacent traffic lane very rarelyresulted in a conflict with another motorvehicle.

70 Chapter 4

Statistical modeling of spacing betweenbicycles and motor vehicles

Least squares regression analysis wasused to develop models to investigate howspacing between bicycles and motor vehiclesdiffered, on average, as a function of variousroadway and traffic characteristics. Mainresults were:

! When bicycles were not being passedby motor vehicles, for BL widths equal to orless than 1.6 m, the average bicycle distancefrom the curb (or gutter pan seam, whenpresent) was less than for WCLs having thesame traffic volume. For BLs greater than1.6 m wide, however, the average bicycledistance from the curb was greater than forWCLs having the same traffic volume.

! When bicycles were being passed bymotor vehicles, results were similar to theabove. Bicycles tended to be positionedabout 0.3 m closer to the curb when beingpassed than when not. The relevant modelagain predicted distances from bike to curbto be smaller for BLs less than 1.5 m andlarger for BLs equal to or wider than 1.5 mthan for WCLs with similar traffic volume.Distance from the bicycle to the passingmotor vehicle was developed from anothermodel and was primarily a function of thetotal width available, along with slight drivingspeed and traffic volume effects. Total widthwas defined as the BL width plus the widthof the adjacent traffic lane, or simply thewidth of the WCL when no BL was present.Thus, for comparable speed and trafficconditions, the distance from the bicycle tothe passing motor vehicle was a directfunction of total width, whether the primarybicycle facility was a BL or WCL.

! On average, bicyclists rode about thesame distance from parked vehicles at BL siteswith parking as they did to the curb (orgutter pan seam) when parked vehicles werenot present.

Intersection movements! The intersection was defined as

starting 90 m upstream from the stop barand included the intersection proper.Proportionally more bicyclists approachedthe intersection on a sidewalk when thefacility was a WCL (15 percent) than a BL(3 percent).

! Overall, 92 percent of bicyclistsobeyed the traffic signals that were present,and there were no differences by facilitytype. When a signal was disobeyed, 16percent of the actions were consideredsomewhat unsafe and 2 percent definitelyunsafe. There were no differences by facilitytype.

! Overall, 75 percent of bicyclistsobeyed existing stop signs. Proportionallymore bicyclists obeyed stop signs at BL sites(81 percent) than at WCL sites (55 percent). When a stop sign was disobeyed, 13 percentwere considered somewhat unsafe and 2percent definitely unsafe. The proportion ofbicyclists with both somewhat unsafe (19versus 5 percent) and definitely unsafe (3versus 0 percent) movements was higher atBL sites. The differences between BL andWCL sites were significant when thesomewhat unsafe and definitely unsafecategories were combined.

! Seventy-two percent of the bicyclistswent straight through the intersection, withanother 15 percent turning left and 13percent turning right. There were nodifferences by facility type. Nine percent ofthe bicyclists tended to shy to the right (i.e.,move to the right and away from traffic) asthey went straight through the intersection(11 percent in BLs and 7 percent in WCLs),and this difference was significant.

! Left turns presented a problem forbicyclists and were made in a variety ofways. Overall, 44 percent made left turnslike a motor vehicle with proper lanedestination positioning (41 percent from BL

Chapter 4 71

sites and 48 percent from WCL sites). Onthe other hand, 14 percent of bicyclists atWCL sites made motor vehicle style leftturns with improper lane destinationpositioning compared with 3 percent fromBL sites. There were proportionally morepedestrian style left turns from WCL sites (24percent versus 12 percent from BL sites).Both findings may reflect the generally highertraffic volumes and speeds and greaternumber of lanes at WCL sites.

! Right turns for bicyclists were aneasier maneuver, with only 13 percent madein a non-standard fashion (e.g., from a BL orWCL to a wrong-way position on the crossstreet). Nineteen percent of the right turnsmade at WCL sites were non-standardversus 10 percent of right turns at BL sites,and the differences were significant.

Midblock conflicts! Of the 188 midblock conflicts, 71

percent were bicycle/motor vehicle, 10percent bicycle/bicycle, and 19 percentbicycle/pedestrian. Almost all of thebike/bike conflicts occurred in BLs.Compared with BLs, bicyclists in WCLsexperienced more bike/pedestrian conflicts(30 percent versus 16 percent, and reflectiveof the increased sidewalk riding in WCLsituations) and less bike/bike conflicts. Thedifferences by facility type were statisticallysignificant.

! There were no differences in thebicycle or motor vehicle avoidance responsescales by facility type. The scales ranged fromno change in riding or driving up to collisionor near crash.

! Overall, 98 percent of the midblockconflicts were coded as minor, and therewere no differences by facility type.

! Bicycle actions more associated withBLs in these midblock conflicts included thebicycle having to slow, stop, or swerve fortraffic not influenced by the intersection; the

bicycle turning or swerving across a lane oftraffic; encounters with other bikes; and“other” bike actions (such as an improperleft turn). The bicycle action moreassociated with WCLs in these midblockconflicts was encounters with pedestrians.

! Motor vehicle actions moreassociated with BLs in these midblockconflicts included illegal parking in the BLand entering/exiting on-street parking or adriver or passenger entering/exiting aparked or stopped vehicle. Motor vehicleactions more associated with WCL conflictsincluded turning right in front of a bicyclistafter overtaking and “other” actions such asfailing to yield, improper right turns, andcrowding bikes.

Intersection conflicts! Of the 198 intersection conflicts, 79

percent were bike/motor vehicle, 10percent bike/bike, and 10 percentbike/pedestrian. The differences in theBL/WCL distributions were statisticallysignificant. There were proportionally morebike/bike conflicts in BLs (15 percent) andless in WCLs (4 percent). Conversely, therewere proportionally more bike/pedestrianconflicts in WCLs (17 percent, and againreflective of sidewalk riding) and less in BLs(6 percent).

! The position of the motor vehiclewith respect to the bicycle in the intersectionconflicts was 66 percent in the samedirection, 6 percent in the opposingdirection, 5 percent approaching from theleft, 15 percent approaching from the right,and 7 percent approaching from someother position. There were no differencesby facility type.

! There were no differences in thebicycle or motor vehicle avoidance responsescales by facility type.

72 Chapter 4

! Overall, 93 percent of the intersectionconflicts were coded as minor, and therewere no differences by facility type.

! Bicycle actions more associated withBLs in these intersection conflicts includedthe bicycle having to slow/stop/swerve forintersection traffic, the bicycle having toslow/stop/swerve for traffic not influencedby the intersection, and the bicycle turning orswerving across a lane of traffic. Bicycleactions more associated with WCLs includedpassing slow moving or stopped vehicles onthe right, encounters with pedestrians, and“other” actions such as improper left turnsand merging onto the road from a sidewalk.

! Motor vehicle actions moreassociated with BLs included illegal parkingin the BL and “other” actions such as adriver or passenger entering/exiting a parkedor stopped vehicle and crowding the BL.Motor vehicle actions more associated withWCLs included having to slow/stop/swervefor intersection traffic and turning right infront of a bicyclist after overtaking.

Statistical modeling of conflict data! Raw frequency conflict rates per

entering bicyclist were slightly higher at BLsites than WCL sites when midblock andintersection conflict data were combined (6.7versus 5.1 bike motor vehicle conflicts per100 entering bicyclists).

! The rate of midblock bike/motorvehicle conflicts associated with BLs wasconsiderably higher than the rate for WCLs,although the rates were small. Generalizedlinear models fitted to the data showed thatboth the presence of a BL and the BL width,along with traffic volume and the presenceof driveways, were significant variables in themidblock conflict rate models. The practicaleffect of such models was that the midblockbike/motor vehicle conflict rate was higherat sites with BLs less than 2.5 m wide than atWCL sites. However, a closer examination

of the data revealed that the highermidblock BL conflict rates were attributableto only a few sites. The midblock conflictsat the 10 highest rate sites were thusexamined clinically.

! An initial model fitted to theintersection conflicts showed no differences inthe conflict rate by type of bicycle facility,but higher conflict rates for left turnmovements. A subsequent model wasdeveloped that included differentintersection types based on the type of BLstriping (e.g., solid stripe to the intersection,dashed stripe to the intersection) andwhether the typical WCL cross section wasmaintained through the intersection (ornarrowed due to the provision of turnlanes). This model showed lower conflictrates for straight through and right turningbicycles where the BL stripe continued allthe way to the intersection and the WCLwas not narrowed at the intersection.

Clinical examination of high conflictrate sites

! The 10 highest conflict rate sites forboth the midblock and intersection areaswere examined clinically to determine if anytypical conflict patterns existed. In themidblock area, there were seven BL and threeWCL sites. The predominant motor vehicleactions in the midblock conflicts pertainedto motor vehicles entering or exiting on-street parking (there were several sites whereparking was part of the facility), parking orstopping in the bicycle facilities to let apassenger enter or exit the vehicle, andpulling across the BL or WCL into anintersecting street or driveway. Thepredominant bicycle actions were turning orswerving across a lane of traffic (usually toavoid making a left turn at the intersectionahead) and interacting with pedestrianswhen riding on the sidewalk. If “fault” inthe conflicts had been assigned, the large

Chapter 4 73

majority of the fault would have been due tomotor vehicle actions.

! In the intersection area, there were fourBL and six WCL sites. The predominantmotor vehicle actions again pertained toentering or exiting on-street parking andparking or stopping in the bicycle facility tolet a passenger enter or exit the vehicle. Thepredominant bicycle actions were turning orswerving across a lane of traffic, passingslow or stopped motor vehicles on the right,and interacting with pedestrians. Some of theconflicts resulted simply from the typicalmaneuvering that might occur when bicyclesand motor vehicles position themselves tomake turns at intersections. If “fault” in theconflicts had been assigned, the majoritywould have been due to bicycle actions.

! Identifiable situations leading toconflicts from this clinical analysis werepresence of parked motor vehicles (eitherentering/exiting legal parking or illegalparking/stopping) in the BL or WCL,presence of driveways or intersecting streets,and provision of turn lanes at intersectionsthat typically (but not always) resulted in anarrowing of the BL or WCL at theintersection proper (normally in the last 30 to50 m before the stop bar). Except forcombined BL and parking facilities, thesesituations did not appear to be related towhether a BL or WCL was present. In otherwords, the conflicts that resulted were site-specific and likely would have occurredwhether a BL or WCL was present.

Clinical examination of seriousconflicts

! Seventeen conflicts were coded asserious, 10 at WCL and 7 at BL sites. If“fault” had been assigned, 11 would havebeen the fault of the motorist and 6 the faultof the bicyclist. The motorist turned rightsoon after overtaking the bicyclist in six ofthe conflicts, pulled from a driveway to the

street in three conflicts, and was involved ina parking situation in the other two cases.The bicyclist turned or swerved across alane of traffic in three conflicts, disobeyed atraffic signal in two cases, and shifted infront of a motor vehicle in the process ofavoiding rough pavement in the other.Examining these situations clinically, thereappeared to be no differences between BLand WCL serious conflicts.

Comparisons with crash data! One year (1995) of police-reported

crash data was “typed” using the NHTSAmethodology for all three of the projectcommunities. There were parallels to thevideotape data.

! In Santa Barbara, one of the twomost frequently occurring crash types wasthe bicyclist striking a parked vehicle. SantaBarbara had a number of individualintersections where parking was part of thebike facility, and overall 41 percent of thebicyclists were recorded as riding next toparked vehicles, as compared with 21percent of the Austin bicyclists and none ofthe Gainesville bicyclists.

! In Gainesville, the most frequentlyoccurring crash type was Motorist DriveOut at Stop Sign. In three out of four ofthese crashes, the bicyclist was riding thewrong way (facing traffic) on the sidewalk.Seventeen percent of the Gainesvillebicyclists were observed approaching thetargeted intersections on the sidewalk, ascompared with less than 3 percent of theAustin bicyclists and less than 2 percent ofthe Santa Barbara bicyclists. In addition, 9percent of the Gainesville bicyclists wereobserved riding the wrong direction on asidewalk, compared with 1 percent of boththe Austin and Santa Barbara bicyclists.Wrong-way sidewalk riding was also afactor in 87 percent of Gainesville’s Right

74 Chapter 4

Turn on Red crashes, and 75 percent of itsDrive Out at Midblock crashes.

! In Austin, 11 percent of the bicyclistsmade “advance crossovers” to the left priorto the intersection, as compared with 3percent in both Gainesville and SantaBarbara. Nearly 6 percent of the crashesreported for Austin were Bicyclist Left Turnin Front of Motorist. None of these types ofcrashes were reported for Gainesville orSanta Barbara.

Further Comment

Level of experience Many in the bicycling community have

assumed that more experienced bicycliststend to use WCLs and that lesserexperienced bicyclists use BLs. This issue wasexplored in this project by use of an oralquestionnaire, where each surveyed bicyclistwas asked to read or listen to a statementbeing read to them about their experience orcomfort level on certain types of facilities(see chapter 2 for details). Overall resultsshowed that 34 percent of the bicyclistsconsidered themselves to be experienced,and there were no differences by type offacility.

Wrong-way ridingWrong-way riding, or riding facing

traffic, was present for approximately 6percent of the videotaped bicyclists. Thereseems to be a prevailing feeling that thispractice is more widespread in BLs, but inthis study a higher proportion of the wrong-way riding tended to occur at WCL sites,whether in the roadway or on the sidewalk.Proportionally more of the WCL wrong-way riding took place on the sidewalk;however, eliminating sidewalk riding fromthe tabulation still showed significantly morewrong-way riding in the street associated

with WCL sites. This may be related to thefact that WCLs are often associated withhigher volume roadways and thatmaneuvering through intersections on theseroadways can be a complex task. Thus, thebicyclist may choose what seems to be asafer route by riding the wrong way on anadjacent sidewalk or in the street. It may notbe safer in actuality, as wrong-way ridingeither in the street or on a sidewalk is afrequent factor in bicycle-motor vehiclecrashes (See Hunter, Stutts, Pein, and Cox,1996).

Turning and other maneuvers atintersections

Besides the sidewalk riding mentionedabove, complexity of traffic at the WCLintersections in this study may also berelated to the operational findings that moreincorrect left-turn destination positioningand pedestrian-style left turns wereassociated with WCL intersections. Inaddition, WCL sites had proportionallymore non-standard right turns than BL sites.Left turns presented problems at BL sites aswell. An intersection conflict model showedhigher conflict rates for straight and rightturning bicycles where the bike lane wasterminated prior to the intersection, dashedto the intersection, or the nominal width ofthe BL or WCL was narrowed due to theprovision of turn lanes. A prevalent conflictin these situations, whether at a BL or WCLsite, was for a motor vehicle to pass abicyclist and then turn right soon after theovertaking maneuver was made.Experienced bicyclists can prevent some ofthese conflicts by taking control of the lanewith their positioning, particularly within theintersection, so that the motor vehiclecannot pass. More bicyclists need trainingrelated both to turning maneuvers atintersections and to safely negotiating theseareas if merely going straight through.

Chapter 4 75

Intersections continue to account for abouthalf of all bicycle-motor vehicle crashes(Hunter, Stutts, Pein, and Cox, 1996).

ConflictsThere were nearly 400 midblock and

intersection conflicts noted, but the vastmajority were minor in nature. There was nodifference in the severity level of the conflictsfor BL versus WCL sites as measured bybicycle or motor vehicle response scales toconflicts. Bike/bike conflicts were moreassociated with BLs, while bike/pedestrianconflicts were more associated with WCLs.Unadjusted conflict rates showed BL sites tohave slightly higher rates per entering bicyclistthan WCL sites.

Many midblock and intersection conflictmodels were attempted to identify significantvariables related to the occurrence ofconflicts. A midblock conflict model showedthat presence and width of a BL weresignificantly related to conflicts, along withtraffic volume and presence of driveways.Conflicts increased with traffic volume,number of driveways, presence of a BL, andnarrower BLs. The interpretation questionwas whether the higher conflict rates werereally attributable to these variables,particularly narrower BLs, or to site-specificcharacteristics for a few locations. Furtheranalysis showed that a few sites withnarrower BLs and high conflict rates tendedto greatly affect the results.

As another example of the interpretationproblem, the same data described abovewere modeled by eliminating two atypicalWCL sites for which vehicles parking on thestreet during the videotaping reduced theeffective space available to that of a regulartraffic lane, instead of a WCL. One of thesesites had a high traffic volume, twodriveways, and a high conflict rate. The otherhad low traffic volume, one driveway, and alower conflict rate. Eliminating these sites

and refitting the model to the data led tothe elimination of both the traffic volume anddriveway factors. However, both factorsseem intuitively related to higher conflictrates.

The difficulty of statistically interpretingoutcomes that seemed so dependent onsite-specific characteristics led to clinicalanalysis of higher conflict rate sites, both atmidblock and intersection locations. Resultsof this clinical analysis showed severalfactors to be consistently related to theoccurrence of the conflicts: (1) presence ofparked motor vehicles (eitherentering/exiting legal parking or illegalparking/stopping) in the BL or WCL, (2)presence of driveways or intersecting streets,and (3) provision of additional (usually turn)lanes at intersections that typically (but notalways) resulted in a narrowing of the BL orWCL. Fortunately, these are factors forwhich some countermeasures are available.

RecommendedCountermeasures for CertainHigh Conflict Rate Problems

Parked motor vehiclesMotor vehicle parking conditions vary

widely, and there can be large differencesbetween all day parking with low turnoverand high turnover parking that typicallyserves retail stores. High turnover from on-street parking was one of the situations that led to conflicts with bicycles in this study.The other problem situation was illegalparking or stopping in the bicycle facility.Many communities in the United Statesallow motor vehicles to park in bicyclefacilities, particularly BLs, during someportions of the day, generally when bicycle

76 Chapter 4

Figure 17. Standard no parking signs forbike lanes.Source: Oregon Bicycle and Pedestrian Plan,1995

Figure 18. Double striped BL withparking.

Figure 19. Combination BL withparking T’s.

traffic is low. In other words, there is nobicycle facility when motor vehicles areallowed to park. This practice can onlyfunction effectively if police enforcementkeeps the motor vehicles out of the facilityduring the time parking is prohibited.However, this kind of enforcement is

difficult to maintain, and violations of theseparking provisions are apparent even inbicycle-friendly communities. Eliminatingparking altogether in the bicycle facility is amuch stronger statement. If bicycling is to bea truly integrated and useful form oftransportation, then bicyclists should havefacilities available throughout the day.

In like fashion, motor vehicles do nothesitate to pull into BLs to allow passengersto enter or exit. In areas of busy bicycletraffic, this can lead to many conflicts. At theleast, standard “no parking in bike lane” signs

(figure 17) should be used liberally. Moreoften than not, however, this is anenforcement issue.

Besides enforcement, good designpolicy can help to eliminate some of theconflicts. If motor vehicle parking is anintended part of a BL, then a double-striped 1.5-m BL that positions the rightmost BL stripe at least 0.9 m from parkedvehicles is recommended to provide thebest channelization of bicyclists (figure 18).At least 2.4 m should be allowed forparking. When available right-of-way doesnot allow the double striped BL describedabove, then a combination lane, intendedfor both motor vehicle parking and bicycleuse, is an alternative. Such a lane should beat least 3.7 m wide, with 4.3 m beingpreferable, and contain parking T’s(sometimes referred to as tick marks) todenote the parking spaces (figure 19).

Bicyclist education about correctposition when riding on streets with on-street parking is also highly recommended.Bicyclists should be at least 0.9 m fromparked vehicles, and riding should be in astraight line. Such recommendations can beeasily highlighted on a community bicyclemap.

Driveways and intersecting streets

Chapter 4 77

Figure 20. Typical conflict situationsat a driveway crossing a sidewalk..

Driveways and intersecting streets ineither the midblock or intersection area canlead to bike/motor vehicle conflicts.Driveways or alleys in commercial areas arenormally the culprits because more motorvehicle traffic is present. Sometimes theproblem is the motorist driving out of adriveway or alley and failing to stop beforecrossing a sidewalk or an implied sidewalkarea that has bicycle travel (figure 20). Clearsight lines should be provided for themotorist if possible. If the sidewalk ends atthe driveway cut, a crosswalk could bepainted (with optional advance stop bar), orthe sidewalk could be extended across thedriveway cut. A “WATCH FORBICYCLISTS” sign could also be installed.

Treatments can also be developed forthe bicyclist riding on the sidewalk. First andforemost, education should be providedabout the dangers of sidewalk riding, and especially wrong-way riding that places thebicyclist out of the normal viewing patternfor a motorist exiting from a commercialdriveway or alley. Bicyclists should also becautioned to ride slowly in these areas thatare primarily designed for walking speeds.Painting “USE CAUTION” on the sidewalkat hazardous driveways is alsorecommended.

Most of the problems noted at the highconflict rate sites in this project involvedbicyclists riding in the street, however, andnot on the sidewalk. From anecdotalobservation, it would seem as thoughmotorists are not hesitant to use a BL as abuffer when they exit from a commercialdriveway or alley into the street. A remedy isto provide a stop bar for the motorist priorto the BL. “WATCH FOR BICYCLISTS”or “YIELD TO APPROACHINGBICYCLISTS” signs might be helpful in thissituation. Dashing the BL stripe at busydriveways is also recommended, not only toalert a motorist that a bicyclist may be

approaching because of the presence of theBL but also to alert a bicyclist that amotorist may be emerging from thedriveway adjacent to the dashed stripe.

Equally important is the problem ofmotorist overtaking where a right turn ismade into the driveway soon after theovertaking is completed. Bicyclist educationabout the danger of driveways is warranted,with the message focusing not only onmotorist-drive-out but also on motorist-overtaking situations. Motorist educationrelating to the overtaking situation above isalso needed.

Additional lanes at intersectionsThere are several problems with

additional lanes at intersections. One has todo with the loss of space to the BL orWCL when additional turn lanes areprovided with the same width of crosssection. It is common practice now to usenarrower lanes for turning movements orto calm traffic. Using narrower widths mayretain the full width of the bike facility at the

intersection.Another practice involves terminating

either the BL or WCL and splitting theapproaching traffic into two through lanesjust prior to the intersection stop bar area.When this occurs, the two lanes often

78 Chapter 4

Figure 22. Dashed BL stripe at right-turn situation.

Figure 23. European bike box.

Figure 21. Example of traffic splitting.

become one again on the far side of theintersection (figure 21).The idea is to use the extra lane to get trafficthrough the intersection faster, but with thenotion comes problems for the bicyclist.First is the loss of space. Second is theweaving among the motorists as they jockeyfor position through the intersection andbeyond as they must merge again.

Right turn lanes present another problemfor bicyclists. There may be weavingbetween bicycles and motor vehicles in theapproach to the right turn in a designated BL

if there is a high volume of right turningmotor vehicle traffic. Use of a dashed stripegives notice that weaving will take place(figure 22). Bicyclists may also have a

tendency to overtake or stop on the right ofmotor vehicles turning right. Education onthe hazards associated with this maneuver isrecommended.

Whether right turn lanes are present ornot, right turning motor vehicles atintersections pose a problem for bicyclists.Similar to the driveway conflict mentionedabove, in the intersection area a motorvehicle may also turn right to another street

soon after overtaking a bicyclist. Onetreatment made popular in Europe thathelps to counter this problem is the use ofan advanced stop bar or bike box (the termnow frequently used to refer to thistreatment in the United States). In Europe, abike box is typically placed at the end of aBL (figure 23) so that bicyclists mayproceed easily to the head of the trafficqueue and thus get through the intersectionahead of right turning motor vehicle traffic.The bike box is gaining some popularity inthe United States, and different versions ofthe technique are being tried or considered,such as placing a bike box at a WCLintersection, or using the bike box at a BLlocation to get left turning bicyclists to thehead of the queue. The bike box appears tobe well accepted in Europe. However,evaluations of such applications in theUnited States need to be made to determine

Chapter 4 79

if the applications are understood andaccepted.

ConclusionsThe debate over whether BLs or WCLs

are preferable has been heated for manyyears. While both BLs and WCLs areacceptable facilities in many locations, thedebate has sometimes forced decisionmakers to choose which facility type theyprefer, to the exclusion of the other. Morebicycle facilities might be in place in thiscountry except for this long-standing divisionof opinion.

This comparative analysis of BL versusWCL sites utilized an extensive data base toexamine many factors related to theoperations and safety of these facilities.Forty-eight sites were videotaped in thestudy, and these produced 369 total conflicts,276 of which were bike/motor vehicleconflicts. In reality this is relatively fewconflicts, which is an encouraging outcome.On the other hand, approximately 6 percentof the bicyclists had a conflict with a motorvehicle, which is not a trivial amount. Manymore sites would have been necessary toproduce a wholesale increase in the numberof conflicts available for analysis.

Across the board these facilities workwell, with the vast majority of identifiedconflicts in this study being minor in nature.Both behavioral actions and geometriccharacteristics were identified as problems inthe study of these bicycle facilities, and thereare remedies for these, but in most cases thenoted problems at the higher conflict ratesites could not be labeled as particular BL orWCL deficiencies. The destination patternsof bicyclists traveling through the project sitesled to maneuvers and conflicts that in manycases would have occurred whether thebicycle facility present was either a BL orWCL.

This is an important point that plannersand engineers should heed. With theirrelative freedom of movement (i.e., notbeing as confined to a traffic lane as amotor vehicle), bicyclists will use a variety ofways to get through an intersection and ontoward their destination. The chosenmethods usually reflect perceived timesavings/efficiency or improved safety. Asan example, difficulties in making left turnsbecause of heavy motor vehicle flows willlikely lead to advance crossovers or otheralternate maneuvers. Even though standarddesign templates for bicycle facilities shouldbe applied wherever possible to promoteconsistency in understanding and propermovements through intersections, it isapparent that such templates cannot beapplied across the board to achievestandard or desired bicyclist movements.Some tailoring will be necessary to take intoaccount desired or frequent movements bybicyclists, just as is done for locations withhigh motor vehicle movements and/orcrash rates.

The overall conclusion of this researchis that both BL and WCL facilities can andshould be used to improve ridingconditions for bicyclists, and this should beviewed as a positive finding for thebicycling community. The identifieddifferences in operations and conflicts wererelated to the specific destination patterns ofbicyclists riding through the intersectionareas studied. Given the stated preferencesof bicyclists for BLs in prior surveys (e.g.,Rodale Press, 1992), along with increasedcomfort level on BLs found in developingthe Bicycle Compatibility Index (Harkey etal., 1998), use of this facility isrecommended where there is adequatewidth, in that BLs are more likely toincrease the amount of bicycling thanWCLs. Increased bicycling is importantbecause in the United States there are but a

80 Chapter 4

few communities that have a significant shareof trips made by this mode. Overall, wehave not yet reached the critical massnecessary to make motorists and pedestriansaware of the regular presence of the bicycle.When this critical level of bicycling is reached,gains in a “share the road” mentality willcome much more quickly than at present.Certainly not all the problems will disappear,but the ability to develop and implementsolutions will be greatly enhanced.

Appendix

A Experience Form

Bicycle Facilities Evaluation Study - Exposure Data

Site _______________ and _______________ Data Collector Initials ________________

Site # ________ Day of Week __________ Start Time (24 hour) _________________

Date ____ / ____ / ____ Time 1 or 2 _____ End Time (24 hour ) _________________

Obs. # Age Daysper

Week

Milesper

Week

RidingExperience

12

WhereRidingR=RoadS=SdwlkO=Other

RaceW=WhiteB=BlackH=Hispan.A=AsianO=Other

SexM=MaleF= Fem

HelmetUseY= YesN= NoO=Other

Comments

Survey Questions: Enter ? if unknown or missing. Ask questions 1-4. Observe and record questions 5-8.1. What is your age? (record actual number years)2. On average, how many days a week do you ride your bike? (record actual number of days, 0-7)3. On average, about how many miles do you ride each week? (record actual number of miles reported)

4. How would you classify yourself with respect to the experience you have riding on city streets? (1 or 2)5. Location where cyclist was riding (Roadway, Sidewalk, Other)6. Race of cyclist (White, Black, Hispanic, Asian, Other or mixed) 7. Sex of cyclist (Male, Female)8. Cyclist helmet use (Yes, No, Other)

Appendix

B Coding Form

Coding Items for the Comparative Analysis of Bike Lanes versus Wide Curb Lanes (TaskOrder 7)

GeneralDate

Time of day data collected - inclusive Use military time (e.g., 1200-1400 hours)

Tape time

Was bicyclist riding wrong way? (Has to be for some duration - not like a left turn) 1. Yes, in road2. Yes, on sidewalk3. No4. Other 0. Unk.

If yes to riding wrong way, then code the 3 items below and then describe in a briefnarrative any conflicts that resulted because of wrong way riding. Then skip to next screen for anew bicyclist.

BicyclistBicyclist gender Bicyclist Age_____1. Male 1. <162. Female 2. 16-240. Unsure 3. 25-64

4. 65+Bicyclist carrying passenger 0. Unsure1. Yes2. No0. Unsure

Bicyclist helmet use 1. Yes2. No0. Unsure

84 Appendix B

Describe any wrong way riding conflicts Midblock (between 300-500 foot cones)

Bicyclist initial midblock position (i.e., at the 500 foot cone) 1. Bike Lane2. Wide lane3. Sidewalk4. Path beside roadway5. Traffic lane (e.g., lane next to bike lane or wide curb lane)6. Other (e.g., weaving) 0. Unk.

Changed position midblock to 1. Bike Lane2. Wide lane3. Sidewalk4. Path beside roadway5. Traffic lane (e.g., lane next to bike lane or wide curb lane)6. Did not change positions0. Unk.

Midblock movement 1. Straight to intersection2. Left crossover in advance of intersection3. Right in advance of intersection4. Other 0. Unk.

Spacing and associated passing conditions (see list below) from curb or gutter pan seam to right side biketire at approximate 500-foot cone / (With no motor vehicle present, if possible) To nearesttenth of a foot (e.g., 1.2 ft) - use transparency)

Associated passing conditions: 1. Bicycle alone and free flowing - no vehicle in vicinity(200-300 ft)2. Bicycle alone and free flowing - vehicle in vicinity(200-300 ft)3. Bicycle being passed by motor vehicle

4. Bike passing motor vehicles5. Other (e.g., bike avoiding debris or other obstacle)

0. Unk.

For any of the above conditions, was bike riding next to parked vehicles?1. Yes2. No

Appendix B 85

Spacing and associated passing conditions from first passing motor vehicle that can be viewedapproximately between 300-500 foot cones / (Nearest tenth of a foot - left side bike tire to right side m.v. tire) / (Nearest tenth of a foot - from curb or joint to right side bike tire)

Did motor vehicle encroach into next lane when passing the bicyclist? 1. Yes2. No0. Unsure

If yes, did this result in a motor vehicle to motor vehicle conflict? 1. Yes2. No0. Unsure

Bicyclist midblock (between 300-500 foot cones) behavior / / (Note: Use attached BehaviorList. Code no more than 3 items. Also use this list for the coded behavior in midblock conflict codingbelow)

For the above bicyclist behavior, was there a vehicle in the vicinity — approximately 200-300 feet — (i.e.,such that there was some chance for a conflict)? 1. Yes2. No0. Unsure

Midblock Conflict No./Type / Type 1= bike/motor vehicleType 2= bike/bikeType 3= bike/pedestrian

For midblock conflict: Bike action (Use attached behavior list. Try to code a single item. Use more codes if necessary. )Motor vehicle action (Use attached behavior list. Try to code a single item. Use more codes if necessary.)

Bicyclist level of avoidance response 1. No change in riding2. Stops pedaling3. Slight change of direction4. Applies brakes5. Major change of direction6. Full stop7. Collision or near crash0. Unsure of response

86 Appendix B

Motor vehicle level of avoidance response 1. No change in driving2. Slows3. Slight change of direction4. Applies brakes5. Major change of direction6. Full stop7. Collision or near crash0. Unsure of response

Seriousness of conflict 1. Minor (some braking or maneuvering to avoid each other, but there is time to do so)2. Serious (braking or maneuvering that has to be done quickly to avoid contact)0. Unsure

Describe any midblock conflicts

Within Intersection (from 300-foot cone to intersection stop line/crosswalk)Bicyclist initial intersection approach position 1. Bike Lane2. Wide lane3. Sidewalk4. Path beside roadway5. Traffic lane (e.g., lane next to bike lane or wide curb lane)6. Left turn lane7. Right turn lane8. Other 9. Never made it through intersection (If this, then skip following)0. Unk.

Changed Intersection approach position to 1. Bike Lane2. Wide lane3. Sidewalk4. Path beside roadway5. Traffic lane (e.g., lane next to bike lane or wide curb lane)6. Left turn lane7. Right turn lane8. Other 9. Did not change positions0. Unk.

Appendix B 87

Intersection movement 1. Straight2. Left3. Right4. U turn5. Other 0. Unk.

Walked bike? 1. Yes2. No0. Unk.

For straight, left, or right movement, bicyclist intersection traffic control 1. Obeyed signal2. Obeyed sign3. Disobeyed signal 4. Disobeyed sign5. Other 6. Unsure

If disobeyed signal/sign 1. Maneuver safe2. Maneuver somewhat unsafe3. Maneuver definitely unsafe4. Other 5. Unk.

Note: From here on, coder chooses between straight, left and right “sections.” If straight, thenignore left and right, etc. Need good way to get screen cursor to shift if section n/a.

Bicyclist straight through method 1. Straight through from bike lane to bike lane on other side of intersection2. Bike lane to traffic lane3. Wide lane to wide lane4. Wide lane to other traffic lane5. Like motor vehicle in traffic lane (i.e., using neither bike lane nor wide lane beginning the

approach)6. Stayed in street but crossed on marked crosswalk or crosswalk area7. Moved to sidewalk and then to crosswalk area8. Straight through from right turn lane9. Other 0. n.a./unk.

88 Appendix B

Straight method exit facility 1. Bike Lane2. Wide lane3. Sidewalk4. Path beside roadway5. Traffic lane (e.g., lane next to bike lane or wide curb lane)6. Other 0. n.a./unk.

Tracking 1. Straight2. Shy to right3. Shy to left (e.g., on one-way street)0. Unk.

Bicyclist straight through intersection behavior / / (Use attached list. Try to code a singleitem, but no more than 3.)

For the above bicyclist behavior, was there a vehicle in the vicinity — 200-300 feet — (i.e., such thatthere was some chance for a conflict)? 1. Yes2. No0. Unsure Straight through intersection Conflict No./Type /

Type 1= bike/motor vehicleType 2= bike/bikeType 3= bike/pedestrian

For straight conflict: Bike action (Use attached list. Try to code a single item. Use more codes ifnecessary. )Motor vehicle action (Use attached list. Try to code a single item. Usemore codes if necessary.)

For straight conflict: Motor vehicle position 1. Same direction2. Opposing direction3. Cross street from left4. Cross street from right5. Other (e.g., driveway) 0. Unk.

Appendix B 89




Describe any straight through conflicts

Bicyclist left turn method 1. Motor vehicle style - proper destination positioning2. Motor vehicle style - improper destination positioning3. Pedestrian style - near side4. Pedestrian style - far side5. Right hook6. Left turn from bike lane (new item)7. Advance crossover8. MV - Ped hybrid9. Other/weird 0. n.a./unk.

90 Appendix B

Specify exit facility on left turn 1. Bike Lane2. Wide lane3. Sidewalk4. Path beside roadway5. Traffic lane (e.g., lane next to bike lane or wide curb lane)6. Unknown0. n.a.

Merge distance for bicyclist making motor vehicle style left turnEstimated distance from intersection stop line when bicyclist started merge to left (Use traffic cones toapproximate) (Nearest 10 feet)

If left turn like motor vehicle, was lane position correct/appropriate? 1. Yes2. No0. Unsure

Bicyclist left turn intersection behavior / / / (Use attached list. Try to code a single item, butno more than 3.)

For the above bicyclist behavior, was there a vehicle in the vicinity (i.e., such that there was somechance for a conflict)?1. Yes2. No0. Unsure

Left turn Conflict No./Type / Type 1= bike/motor vehicleType 2= bike/bikeType 3= bike/pedestrian

For left turn conflict: Bike action (Use attached list. Try to code a single item. Use more codes ifnecessary. )Motor vehicle action (Use attached list. Try to code a single item. Usemore codes if necessary.)

For left turn conflict: Motor vehicle position 1. Same direction2. Opposing direction3. Cross street from left4. Cross street from right5. Other (e.g., driveway) 0. Unk.

Appendix B 91




Describe any left turn conflicts Bicyclist right turn method 1. Done correctly (e.g., from bike lane or traffic lane to another bike lane or traffic lane)2. Done incorrectly (e.g., from bike lane or traffic lane to wrong way position on cross street)3. From bike lane or traffic lane to sidewalk4. Weird 5. Unsure6. n.a./unk.

Right turn exit facility 1. Bike Lane2. Wide lane3. Sidewalk4. Path beside roadway5. Traffic lane (e.g., lane next to bike lane or wide curb lane)6. Other 0. n.a./unk.

92 Appendix B

Bicyclist right turn intersection behavior (Use attached list. Try to code a single item. Use morecodes if necessary.)

For the above bicyclist behavior, was there a vehicle in the vicinity (i.e., such that there was some chancefor a conflict)? 1. Yes2. No0. Unsure

Right turn Conflict No./Type / Type 1= bike/motor vehicleType 2= bike/bikeType 3= bike/pedestrian

For right turn conflict: Bike action (Use attached list. Try to code a single item. Use more codes ifnecessary. )Motor vehicle action (Use attached list. Try to code a single item. Usemore codes if necessary.)

For right turn conflict: Motor vehicle position 1. Same direction2. Opposing direction3. Cross street from left4. Cross street from right5. Other (e.g., driveway) 0. Unk.



Appendix B 93


Describe any right turn conflicts

For any Additional Observations to the movements through the intersection that should be mentioned,describe in a narrative

94 Appendix B

List of Behaviors

Bicycle/Motor Vehicle Common Actions

10. B/MV NONE11. B/MV failed to stop at stop sign (Note: 11,12,13 include main & other intersections)12. B/MV failed to stop at traffic signal; stopped but proceeded when gap13. B/MV stopped at a controlled intersection but failed to yield14. B/MV failed to yield at driveway or other uncontrolled midblock entrance/exit

15. B/MV improper left turn (turned left in front of approaching traffic) 16. B/MV improper right turn (swing wide) - other than overtaking (see 53)17. B/MV failed to yield changing lanes/merging (specific intent of changing lanes/merging)18. B/MV improper lane use (through from right, right from through, etc.)

19. B/MV slowed/stopped/swerved due to intersection traffic20. B/MV slowed/stopped/swerved due to other traffic not influenced by the intersection (e.g.,

vehicle pulled off the road)

Motor Vehicle Initiated Events

50. MV traveling in bike lane/edge of wide curb lane (long term)51. MV crowding (includes encroaching ) bike lane, edge of travel lane52. MV entered bike lane for right turn/failed to yield to cyclist53. MV turned right in front of bicyclist after overtaking54. MV (illegally) parked in bike lane/wide curb lane55. MV entering/exiting on-street parking56. MV driver or passenger entering/exiting a parked/stopped vehicle57. MV other (describe)

Bicycle Initiated Events

80. B entering roadway from sidewalk/over curb81. B turned or swerved across a lane of traffic82. B passing slow or stopped vehicle on right in wide curb lane83. B riding in-between slowed/stopped traffic84. B crowding vehicles in travel lane85. B stunt riding or other erratic riding (Note: Only code 84, 85 if mv nearby)86. B riding 2+ abreast (only if causes conflict)87. B lost control due to curb face or gutter seam88. B lost control due to other surface irregularity or debris in road89. B lost control - other or unknown cause90. B encounter with pedestrian91. B encounter with another bicyclist92. B other (describe)

References

Allott and Lomax. Cyclists and Roundabouts: A Review of Literature, 2nd Edition, Cyclists’Touring Club, London, England, 1993.

Balsiger, O. “Cycling at Roundabouts: Safety Aspects,” Proceedings of the 8th Velo-CityConference, Basel, Switzerland, 1995, pp. 215-217.

Berchem, S. and Somerfeld, W.O. UniqueRoadway Design Reduces Bus-Bike Conflicts, TRNews, No. 117, 1985, pp. 2-3.

Botma, H. and Mulder, W. “RequiredWidths of Paths, Lanes, Roads and Streetsfor Bicycle Traffic,” 17 Summaries of MajorDutch Research Studies About Bicycle Traffic,Grontmij Consulting Engineers, TheNetherlands, 1993.

Brüde, U. and Larsson, J. The Safety of Cyclistsat Roundabouts - A Comparison between Swedish,Danish, and Dutch Results, Swedish NationalRoad and Transport Research Institute,Linköping, Sweden, 1996.

Burden, D., Wallwork, M., and Guttenplan,M. Bicycle and Motor Vehicle Roundabouts: ANew Safety Tool to Aid Bicyclists, Pedestrians,Motorists and Transit in China and OtherLocations, 1994.

City of Eugene, 18th Avenue Bike Lanes - OneYear Report, Memorandum to City Council,City of Eugene, Oregon, 1980.

C.R.O.W. Sign Up for the Bike: Design Manualfor a Cycle-Friendly Infrastructure, C.R.O.W. Record 10, Centre for Research and Contact

Standardization in Civil and TrafficEngineering, The Netherlands, 1994.

Diepens and Okkema Traffic Consultants. International Handbook for Cycle NetworkDesign, Delft University of Technology, TheNetherlands, 1995.

Egan, D. “A Bicycle and Bus SuccessStory,” The Bicycle: Global Perspectives, VeloQuebec, Montreal, Canada, 1992.

Florida DOT, Florida Bicycle Facilities Planningand Design Manual, Final Draft, FloridaDepartment of Transportation, Tallahassee,Florida, 1995.

Garder, P. “Rumble Strips or Not AlongWide Shoulders Designated for BicycleTraffic?” Transportation Research Record 1502,1995, pp. 1-7.

Gårder, P., Leden, L., and Pulkkinen, U. “Measuring the Safety Effect of RaisedBicycle Crossings Using a New ResearchMethodology,” Transportation Research Record1636, Washington, DC, 1998.

Grigg, G. “Cupertino’s Bicycle Facilities,”Bicycle Forum, Number 6, pp. 12-15, no date.

Harkey, D.L., Reinfurt, D.W., Knuiman, M.,Stewart, J.R., and Sorton, A. Development ofthe Bicycle Compatibility Index: A Level of ServiceApproach, Report No. FHWA-RD-98-072,Federal Highway Administration, McLean,Virginia, 1998.

Harkey, D.L. and Stewart, J.R. “Evaluationof Shared-Use Facilities for Bicycles andMotor Vehicles,” Transportation ResearchRecord 1578, 1997, pp. 111-118.

Harrison, J.H., Hall, R.D., and Harland,D.G. Literature Review of Accident AnalysisMethodologies and Cycle Facilities, Transport and

96 References

Road Research Laboratory, ContractorReport 163, Berkshire, England, 1989.

Herrstedt, L., Nielsen, M.A., Agústson, L,Krogsgaard, K.M.L., Jørgensen, E., andJørgensen, N.O.. Safety of Cyclists in UrbanAreas: Danish Experiences, Danish RoadDirectorate, Copenhagen, Denmark, 1994.

Hunter, W.W., Stutts, J.C., Pein, W.E., andCox, C.L. Pedestrian and Bicycle Crash Types ofthe Early 1990's, Publication No. FHWA-RD-95-163, Federal HighwayAdministration, McLean, Virginia, 1996.

Jensen, S.U. “Junctions and Cyclists.” Insertto Proceedings of Velo City ‘97 - 10th InternationalBicycle Planning Conference, Barcelona, Spain,1997.

Jensen, S.U. and Nielsen, M.A. “The Markingof Bicycle Crossings at SignalizedIntersections” (in Danish). Summarized inNordic Road & Transport Research, No. 1,1997, p. 27.

Khan, A.M. and Bacchus, A. “Bicycle Use ofHighway Shoulders,” Transportation ResearchRecord 1502, 1995, pp. 8-21.

Kroll, B. and Ramey, M. “Effects of BikeLanes on Driver and Bicyclists Behavior,”Transportation Engineering Journal, Volume 103,March 1977.

Kroll, B. and Sommer, R. “BicyclistsResponse to Urban Bikeways,” Journal of theAmerican Institute of Planners, January 1976.

Leden, L. “Has the City of GothenburgFound the Concept to Encourage Bicyclingby Improving Safety for Bicyclists?” Proceedings of Velo City ‘97 - 10th InternationalBicycle Planning Conference , Barcelona, Spain,1997, pp. 271-274.

Leden, L., Gårder, P., and Thedéen, T. “Safety Implications of Bike Paths inSignalized Intersections -- Reviewing Usinga Bayesian Approach.” Proceedings of theConference Strategic Highway Research Program(SHRP) and Traffic Safety on Two Continents,Hague, Netherlands, 1993, pp. 24-35.

Liu, X., Shen, L.D., and Ren, F. “Operational Analysis of BicycleInterchanges in Beijing, China,”Transportation Research Record 1396, 1993, pp.18-21.

McHenry, S.R. and Wallace, M.J. Evaluationof Wide Curb Lanes as Shared Lane BicycleFacilities, Maryland State HighwayAdministration, Baltimore, Maryland, 1985.

National Bicycling and Walking Study,Publication No. FHWA-PD-94-023,Federal Highway Administration,Washington, DC, 1994.

NCHRP. “Design and Control of FreewayOff-Ramp Terminals,” National CooperativeHighway Research Program Synthesis of HighwayPractice No. 35, Transportation ResearchBoard, Washington, DC, 1976.

New Jersey DOT, Bicycle Compatible Roadwaysand Bikeways: Planning and Design Guidelines,Draft, New Jersey Department ofTransportation, Trenton, New Jersey, 1995.

North Carolina Department ofTransportation, North Carolina Bicycle FacilitiesPlanning and Design Guidelines, NCDOTOffice of Bicycle and PedestrianTransportation, Raleigh, North Carolina,1994.

Oregon DOT, Oregon Bicycle and PedestrianPlan: An Element of the Oregon Transportation

References 97

Plan, Oregon DOT, Bicycle and PedestrianProgram, Salem, Oregon, 1995.

Pronovost, J. and Lusginan, M. “Behavior ofRoad Users Following the Application of aSpecial Bikeway Crossing Marking,” Pro Bike/ Pro Walk 96 Resource Book. BicycleFederation of America and PedestrianFederation of America, Portland, Maine,1996.

Rodale Press Inc., Pathways for People,Emmaus, Pennsylvania, 1992.

Ronkin, M.P. “Bike Lane or SharedRoadway?” Pro Bike News, Vol. 13, No. 3,pp. 4-5, no date.

Smith, R.L. and Walsh, T. “Safety Impactsof Bicycle Lanes,” Transportation ResearchRecord 1168, 1988, pp. 49-59.

U.K. Department of Transport. AdvancedStop Lines for Cyclists, Traffic Advisory Leaflet,U.K. Department of Transport, London,England, 1993.

Wheeler, A. “Advanced Stop-Lines forCyclists - A Simplified Layout,” TrafficEngineering and Control. Vol. 36, No. 5, pp.283-289, May 1995.

Wheeler, A.H., Leicester, M.A.A., andUnderwood, G. “Advanced Stop-Lines forCyclists,” Traffic Engineering and Control, Vol.34, No. 2, pp. 54-60, February 1993.

Wilkinson, W.C., Clarke, A. Epperson, B.and Knoblauch, R.L. Selecting Roadway DesignTreatments to Accommodate Bicycles (FHWA-RD-92-073). Washington, D.C.: FederalHighway Administration, 1994a.

Zegeer, C.V., Cynecki, M., Fegan, J., Gilleran,B., Lagerwey, P., Tan, C., and Works, B.

FHWA Study Tour for Pedestrian and BicyclistSafety in England, Germany, and the Netherlands,Report No. FHWA-PL-95-006, FederalHighway Administration, Washington, DC,1994.

Documents

A Comparative Analysis of Bicycle Lanes Versus Wide Curb … · 2000-03-01 · Subcontractor: Bicycle Federation of America 16. Abstract This report is a comparative analysis of bicycle