U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000


Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

Report
This report is an archived publication and may contain dated technical, contact, and link information
Publication Number: FHWA-HRT-10-041
Date: December 2010

Evaluation of Shared Lane Markings

CHAPTER 3. METHODS

The experimental design was to collect data from bicycles and motor vehicles operating in the traffic stream before and after the installation of the sharrows. While it would have been advantageous to have used an experimental design with comparison data, no adequate comparison sites were available. This is often the case in bicycle safety studies because slight differences in traffic flow, grade, pavement surface, or some other variable can greatly influence outcomes related to the bicyclist. One way to possibly obtain a comparison site is to install a treatment on part of a route and to use the remainder as a comparison. However, when a community is installing a treatment, almost invariably, the desire is to install the treatment along the entire route where the cross section is continuous.

Videotape data were collected by local staff hired and trained by HSRC. A camera was set up on a tripod in line with the outside edge of a parked motor vehicle or the edge of the roadway to provide a clear view of oncoming bicycles and motor vehicles. Zooming was used to follow the bicycles for several hundred feet. Videotaping was performed on weekdays when it was not raining during various times of the day in Cambridge, MA, and Chapel Hill, NC, and at early morning commuting time in Seattle, WA.

Supplemental data were collected in Cambridge, MA, and Chapel Hill, NC. Radar was used in both cities to collect speeds of free-flowing motor vehicles before and after the placement of the sharrows. In addition, data collectors in Cambridge, MA, measured the distance from the curb for both the front and rear tires of parked motor vehicles. In Chapel Hill, NC, data collectors used a form to note instances of sidewalk riding and wrong-way riding. No supplemental data were gathered in Seattle, WA.

From the before and after videotape data, a number of measures of effectiveness and other attributes were coded. The bicycle was the basic unit of analysis. Coding of the videotapes was performed to obtain information about the bicyclists and to examine the operations of bicycles and motor vehicles when a motor vehicle was following or passing a bicycle. In Cambridge, MA, and Seattle, WA, events related to the presence of parked motor vehicles were also examined, such as existing open doors or near dooring events and motorists pulling in or out of parking spaces.

For each bicyclist included in the selected video clips, gender and helmet use were coded along with the direction of travel.

For motor vehicles, coding was performed for the following:

  • Whether the motor vehicle was following, passing, other, etc.
  • Whether the motor vehicle moved to the adjacent lane part way, all the way, or not at all.
  • Whether the motor vehicle overtaking the bicyclist (following and passing, if applicable) did so safely (i.e., no dangerous slowing or abrupt movements).

For bicycles, coding was performed for the following:

  • Whether the bicyclist rode over the sharrows (in the after period).
  • Whether the bicyclist was near a parked car, near center of the lane, or in another position.
  • Whether the bicyclist took control of the lane to prevent a motor vehicle from passing.
  • Whether there was an interaction between a bicycle and a parked motor vehicle such as a dooring event (bicycle struck an opened door), near dooring event (door opened as bicycle was in close proximity), bicycle struck existing open door, a motorist pulling in or out of the parked position, etc.

The occurrence of avoidance maneuvers and conflicts between bicycles and motor vehicles was coded. An avoidance maneuver was defined as a change in speed or direction by either party to avoid the other, while a conflict was defined as a sudden change in speed or direction by either party to avoid the other. If a yielding event took place, the party that yielded was coded, such as a bicyclist slowing down and giving way to a motorist pulling out of a parking space or a motorist slowing and giving way to a bicyclist to move to the center of the lane.

Avoidance maneuvers by bicyclists were coded as follows:

  • The bicyclist kept moving safely.
  • The bicyclist kept moving unsafely.
  • The bicyclist made no change.
  • The bicyclist slowed and stopped pedaling.
  • The bicyclist made a slight direction change (typically a slight lateral movement).
  • The bicyclist used the brakes.
  • The bicyclist made a major direction change (typically a rapid shift to avoid a motor vehicle, object, pavement discontinuity, etc.).
  • The bicyclist made a full stop.
  • Unsure whether the bicyclist made an avoidance maneuver.

Avoidance maneuvers by motorists were coded as follows:

  • The motorist made no change.
  • The motorist slowed down.
  • The motorist made a slight direction change (typically a slight lateral movement).
  • The motorist changed lanes.
  • The motorist used the brakes.
  • The motorist made a major direction change (typically a rapid shift to avoid a bicycle, another motor vehicle, object, pavement discontinuity, etc.).
  • The motorist made a full stop.
  • Unsure whether the motorist made an avoidance maneuver.

The following spacing data were also obtained from images extracted from the videotapes:

  • Distance between bicycles and parked motor vehicles (tire to tire).
  • Distance between bicycles and the curb at the edge of the road (tire to curb) where there was no parking.
  • Distance between bicycles and passing motor vehicles (tire to tire).
  • Distance between motor vehicles in the travel lane and parked motor vehicles (tire to tire) or to the curb (tire to curb) when no bicycles were present.

SigmaScan® software was used to examine images from the videotapes so that the spacing measures of interest could be obtained.(13) SigmaScan® uses a calibrator of known length or height to determine the spacing distance. In Cambridge, MA, and Seattle, WA, where parking was present on both sides of the street, four sets of 3-ft grid lines were painted on the street at 75 ft intervals. In Chapel Hill, NC, 36-inch traffic cones were placed on the sidewalk beside the roadway. Observation indicated that neither the grid lines nor the traffic cones had an effect on the position of the bicyclist in the street. Once the sharrows were installed, these markings also served as a calibrator.

Chi-square tests were performed to examine the distributions of variables before and after placement of the sharrows. Analysis of variance (ANOVA) models were used to study the effect of sharrows on spacing and other performance measures. The independent variables were site characteristics, type of treatment, and a dummy variable to indicate whether it was a before or after condition. The sign and significance of the coefficient of the dummy variable was used to assess the effectiveness of the markings. None of the data were combined across sites because of the differences in the uses of the sharrows in each city.

Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000
Turner-Fairbank Highway Research Center | 6300 Georgetown Pike | McLean, VA | 22101