3. Errors Made by Bicyclists and Drivers in Accidents
  In this section, human errors are identified and analyzed from the 293 accidents investigated by ITARDA. Interview studies on some bicyclists and drivers however had to be omitted for their loss of life and other compelling reasons. Similarly, drivers were not interviewed if they had encountered an accident while their cars were parked. In total, 220 bicyclists and 290 automobile drivers were the subjects of the human error analysis.

  (1) Overall Profiling of Errors
- Few visual errors but many misjudgments by bicyclists -
  The number of errors made per bicyclist and automobile driver are shown in Figure 6. "Recognition" errors mean failure to perceive the approaching vehicle, traffic lights or other things that need to be recognized for safe riding or driving. "Judgment/prediction" errors mean the incorrect judgment/prediction about what is recognized. "Operation/action" errors mean failure to operate or act as intended.The analysis found 2.7 errors by the driver and 2.3 errors by the bicyclist per accident. Said inversely, both drivers and bicyclists had two or three chances to avoid the accident. Breakdown by error types indicates that bicyclists made only few recognition errors, thus confirming that bicyclists usually have a good open view of their surroundings. But their judgment/prediction errors outnumbered those of drivers, thereby canceling out their open view advantage. Recognition errors and judgment/prediction errors are explained in greater detail in the next section.

Fig.6 Number of Errors per Accident
  (2) Objects of Recognition Errors
-1- What Are Easily Overlooked
  - Bicyclists and drivers often overlook counterparts approaching sideways -
  Given in Figure 7 are the number of recognition errors made and the counterparts not recognized per accident. Having a better view of the surroundings, bicyclists made recognition errors numbering only about 70% of those by drivers. For both bicyclists and drivers, "crossing vehicles" rank as the most unrecognized counterpart and accounted for roughly 70% of the total unrecognized counterparts.

Fig.7 Recognition Errors by Person/Counterpart
  As shown in Figure 8, "crossing vehicles" mean bicycles and automobiles passing or about to pass through a crosswalk. The second most frequent recognition errors were recorded with "oncoming vehicles in straight progress", but they accounted for no more than 6-8% of the total. A "preceding vehicle" in analysis denotes a bicycle which, while running forward to an automobile, suddenly enters into the course of progress of the automobile.

Fig.8 Definition of Crossing Vehicles
  -2- Whys Recognition Errors Occur
- Bicyclists and drivers both basically inattentive -
- Bicycles not easily detectable from moving automobiles -
  Major reasons for making recognition errors are listed in Figure 9. "Not alert to other vehicles," "drop in concentration" and "attention to other things" can be put together into the bracket of "basic inattentiveness" and together accounted for 60%. "Looked but could not see" accounted for about 40%. Reasons for basic inattentiveness were first analyzed below, as some bicyclists and drivers remained basically inattentive even at a blind intersection.
  Reported in Figure 10 are the reasons why bicyclists and drivers were inattentive to each other. The most common reasons were: 1) "A" took it for granted that "B" would stop, because the green light had turned on for "A" or because a stop rule was on "B". 2) "A" took it for granted that no other vehicle was around, because the traffic volume had always been small at the intersection. These reasons suggest the possible sedimentation of daily riding/driving practices into habits.

Fig.9 Recognition Errors on Crossing Vehicles by Person/Reason (covering all cases)

Fig.10 Errors: Reasons for 'not alert to other vehicles'
  Similarly, reasons for "drop in concentration," "attention to other things" and "looked but could not see" are reported in Figures 11, 12 and 13, respectively. Figure 11 differ from other Figures 12 and 13 in that a higher error rate was registered by bicyclists. The probable reason is that bicyclists, because bicycles are handy, each have little awareness of being a vehicle driver.
  As shown in Figure 13, many automobile drivers failed to recognize bicyclists although they were watchful. A major reason was that bicycles were inconspicuous within the general traffic, and this was compounded in some cases by bicyclists wearing clothes of dark colors and keeping their bicycle lamps unlit.

Fig. 11 Errors: Reasons for 'drop in concentration'

Fig.12 Errors: Reasons for 'attention to other things'

Fig.13 Errors: Reasons for 'looked but could not see'
  (3) Frequent Judgment/Prediction Errors
- One's good view not ensuring the same for the other -
  The number of and items involved in judgment/prediction errors per person are shown in Figure 14. For both bicyclists and drivers, about 40% of errors related to accident counterparts. About 10% of errors related to traffic lights. As reported in Figure 15, these errors led bicyclists and drivers to think they could proceed safely before their counterparts. Although detailed data are omitted here, about 40% of these errors were based on the misunderstanding on the part of bicyclists that drivers did recognize them. This result, combined with their low rate of recognition error, indicates that a good open view enjoyed by bicycles actually misled them to think they were seen by drivers.
  Automobile drivers' responses indicate difficulty in predicting the course of progress of a bicycle. This is probably because bicyclists make a turn or change their courses without giving a signal. In fact, of the 293 accidents under the present study, 56 cases involved a situation require the bicyclist to give a signal, but only in one case a signal was actually given. The Road Traffic Law of Japan does require bicyclists to signal before changing courses or stopping. Giving a signal has an effect not only of promoting communication with other vehicles but also of creating an intermission in vehicle operation so that hasty action will be refrained.

Fig.14 Judgment/Prediction Errors and Items

Fig.15 Judgment/Prediction Errors about Counterparts
  Types of judgment/prediction errors made are reported in response to traffic signals are Figure 16. Among both bicyclists and drivers, there were many accident cases where the vehicle was started as soon as the green light turned on. Since some vehicles rush into an intersection at a yellow light, a safety action is to take a full breath when the green light turns on, and then make a start. Moreover some bicyclists thought they were not required to observe traffic lights, while on the contrary they must do so even though there is no mandatory bicyclist license scheme.

Fig.16 Judgment/Prediction Errors at Traffic Signals
  Of the total investigated accidents that occurred at places with a stop requirement, the percentage of cases where the traffic light or stop sign was ignored is shown in Figure 17. Bicyclists recorded a non-observance rate over two times higher than that by drivers, both for traffic lights and stop signs.

Fig.17 Non-Observance of Traffic Signs and Lights
  The percentage of bicyclists failing to observe traffic signals and rules is shown in Figure 18, according to the age of bicyclists. The lowest non-observance of traffic lights was registered by the group 15 or younger bicyclists. But the non-observance of stop rule tended to increase in the youngest and oldest groups. (The number of samples for a middle 25-54 old group was too small to yield a reliable result.) Knowledge of stop rule may be limited among young people, and elementary education on traffic rules and manners is necessary to reduce accidents stemming from limited knowledge.

Fig.18 Non-Observance Rate of Traffic Signs and Lights by Bicyclists' Age
UP
Institute for Traffic Accident Research and Data Analysis (ITARDA)