I. INTRODUCTION

In 1982, in response to the Environmental Protection Agency's request for the development of "reasonably available control measures" (RACMs) to reduce air pollution in the Boston metropolitan area, the Metropolitan Area Planning Council developed two projects to increase the use of bicycles for commuting in its area (1). One of these projects was a study of bicycle-motor vehicle accidents in the Boston area, patterned after the Cross-Fisher study completed in 1977 and the Missoula, Montana study of 1981 (2). The purpose of the study was to identify the most common types of accidents occurring in the MAPC region, and to develop a set of countermeasures to reduce the frequency of these accidents.

Several studies and articles had previously suggested the importance of fear for safety as a major deterrent against bicycle commuting (3). It was expected that the study would result in the implementation of recommendations for education and increased enforcement, and directly reduce the number of accidents in the region. In addition, publicity about the study's findings could be used to increase motorists' and bicyclists' awareness about the most frequent accident classes and thereby motivate them to take actions to prevent their occurrence. Ultimately, it was hoped that these measures would result in the increased use of bicycles for commuting, and a concomitant decrease in auto-generated pollution.

In choosing to carry out this study, MAPC was aware of the limitations of the methods used -- review of police and operator accident reports. As has been pointed out in other studies of this type, only a fraction of bicycle-motor vehicle accidents which occur are formally reported. Cross estimated that, between 1972 and 1977, about 1,000 fatal and 40,000 non-fatal bicycle-motor vehicle accidents across the country were reported to police, while another 40,000 injury-producing accidents went unreported. (4).

Still, without an extraordinary effort being put forth, these reports provide the best consistent source of information about bicycle-motor vehicle accidents. Another suggested source of data is hospital records. The forms used would not be standardized and would include only the most serious accidents. They would also lose the advantage of involving the police in the study. It is beneficial for police to be involved, as any recommendations for improved enforcement will largely rely on the police for implementation. Another possible benefit is that use of these forms for research purposes will encourage police, motorists and bicyclists to complete them with greater attention to the quality of description. Currently, the quality of data is mediocre.

Methodology

The study was carried out between November, 1982 and June, 1983. Data from police and operator reports of bicycle-motor vehicle accidents occurring in 1979 and 1980 were obtained by:

1. a paid intern reviewing microfilm of accident reports at the state's Registry of Motor Vehicles; and
2. volunteers reviewing actual reports of accidents at six local police departments.

The area within Route 128, a major beltway in the region encompassing 35 cities and towns including Boston and Cambridge, was chosen for the study (see Figure 1). As almost 2,000 accidents had been reported for 1979 and 1980, it was decided to study a sample of the reported accidents.

Figure 1. Study Area

Larger scan of this image, city and town names legible (153 kB))

The selection of accidents was done using a computer printout provided by the Massachusetts Department of Public Works of all bicycle-motor vehicle accidents occurring in the study area during 1979 and 1980. One in four accidents was selected for review. When accident reports were found to be missing from the Registry of Motor Vehicles or the local police department, additional reports were selected from this printout. This procedure resulted in a sample of 516 reports (5). Of these, 87 provided insufficient information for accident classification purposes, and were included in the results only for purposes of examining other variables such as month of year, time of day and weather conditions. In total, 429 accidents were classified using a modification of the Cross scheme (6).

The MAT Classification Scheme and MAPC Revisions

The Manual Accident Typing (MAT) scheme prepared by the National Highway Traffic Safety Administration (NHTSA) in 1982 was used to classify the accidents (7). This scheme is based upon the classification system created by Kenneth Cross in his 1977 study. This system classified accidents according to four variables

• Pre-collision direction of travel of each operator
• Relative pre-crash motion of the two vehicles
• Operator errors
• Characteristics of accident location In his study, Cross created 36 types (Types 1-36) which he grouped into seven classes (Classes A-G).

The MAT added eight types to the Cross classification system, and fitted these into Classes A-G.

MAPC revised the MAT scheme slightly. It removed accident Type 27 "Cyclist Overtaking" from Class G and Type 35, "Drive-Out: On-Street Parking" and Type 41, "Cyclist Strikes Parked Vehicle" from the MAT's two miscellaneous classes to create a new class labeled "Class G (Revised): Slowed or Parked Car." It was believed that the accident types in this class represent a distinct set that may be addressed by specific distinct countermeasures. "Other" or "Weird" accident types were combined into a "Class H (Revised): Other". The MAT system separated these latter two groups. In all other respects, the MAPC classification scheme is similar to the MAT system (8). Below is a list of the nine classes used in the MAPC system:

• Class A: Bicycle Ride-Out at Driveway, Alley or Mid-Block
  Involve a bicycle emerging from a driveway, alley or other mid-block location (such as over a shoulder or curb) and colliding with a motor vehicle.
  
• Class B: Bicycle Ride-Out at intersection
  Involve a bicycle emerging at an intersection and proceeding straight across the intersection. (Accidents involving bicycles making right or left turns are included in Class E.)
  
• Class C: Motorist Driveout
  Involve a motor vehicle emerging from a mid-block location (driveway, alley) or an intersection, thus paralleling Classes A and B. Only motor vehicles proceeding straight across the intersection or turning right on red are included in this class. (Accidents involving motorists making right or left-turns are included in Class F).
  
• Class D: Motorist Overtaking/Overtaking Threat
  Involve a motor vehicle approaching from behind and colliding or almost colliding with a bicycle.
  
• Class E: Bicyclist Unexpected Turn/Swerve
  Involve a bicycle making a left- or right turn at an intersection or swerving mid-block into the path of an over taking or approaching motor vehicle. Excluded are accidents where the bicyclist swings too sharply or too widely and collides with a motor vehicle on the perpendicular leg of the intersection, which are included in Class H: Other.
  
• Class F: Motorist Turn
  Involve a motorist who is turning right or left at an inter section and collides with a motor vehicle approaching from behind or from the opposite leg of the intersection. Excluded are accidents where the motorist turns right on red (included in Class C: Motorist Driveout) or where the motorist cuts left-hand turn (included in Class H: Other).
  
• Class G: (Revised) Slowed or Parked Cars
  Involve a bicyclist overtaking and colliding with a motor vehicle that is slowed in traffic, parked, or entering or exiting parking. As mentioned above, this class was created by MAPC, and was not included separately in the Cross or Missoula studies or NHTSA's MAT system.
  
• Class H: (Revised) Other
  Involve unrelated accidents that do not fall under any of the above classes. This class, therefore, cannot be analyzed as a class in terms of specific countermeasures; each of the types must be assessed individually. This class differs from the Cross, Missoula and NHTSA's MAT system.