Overtaking, rear-end, and door crashes involving bicycles: An empirical investigation
Research highlights
▶ Three main non-junction crash types involving bicycles are overtaking, rear-end, and door crashes. ▶ Buses/coaches as collision partners were associated with overtaking crashes. ▶ Bicycles’ traversing manoeuvres were associated with overtaking and rear-end collisions. ▶ Given a bicycle-taxi/motorcycle accident, it is more likely a door/rear-end crash, respectively.
Introduction
Recent emphasis on bicycling as an alternative to automobile transportation has generated the need for research efforts directed at bicycle safety when sharing roadways with motorised vehicles. The UK National Travel Survey (DfT, 2009a) that surveys about 3500 households per year reports that the level of cycling in the UK is low, accounting for only 1.6% of trips made by adults, and only 0.5% of mileage. Despite the low level of cycling in the UK, official statistics (DfT, 2009b) reveals that bicyclists’ relative risk of being killed or seriously injured (KSI) per kilometre travelled is about 23 times that for car drivers. National statistics on road casualties in accidents reported to the police in Great Britain (see DfT, 2009c) shows that, given a bicycle accident involving another motorised vehicle, bicyclists in non-junction accidents were 1.7 times more likely to be KSI than those in junction cases. These statistics provide an indication of the magnitude of bicyclist-safety problem in the UK, in particular on segments between junctions.
Surprisingly much of the research attention (either in the UK or in the rest of the world) is focused on intersection accidents (i.e., three-/four-legged junction or roundabout) where motorists infringe upon bicycles’ right of way (ROW) by encroaching on bicycles’ path (e.g., Hunter et al., 1995, Summala et al., 1996, Rasanen and Summala, 1998, Rasanen and Summala, 2000, Stone and Broughton, 2003, Herslund and Jorgensen, 2003, Wang and Nihan, 2004, Walker, 2005, Hels and Orozova-Bekkevold, 2007, Daniels et al., 2008, Daniels et al., 2009). One common problem behind bicycle–automobile accidents at intersections seems to be that motorists tend to look more frequently for major dangers (i.e., automobiles) but ignore less frequent and imminent dangers such as bicycles (Rasanen and Summala, 2000). Other researchers (e.g., Summala et al., 1996) revealed that the approaching speed of an automobile (that is about to enter an intersection) would play a part in the visual scanning strategies of the motorist – higher approaching speed results in the motorist being more likely to scan for a more threatening road user (e.g., a conflicting automobile) but being less likely to allocate much attention to a bicycle. Research (e.g., Hills, 1980, Brown, 2002) has identified these crashes as “looked-but-failed-to-see” accidents in which in many cases the motorist has actually been looking in the direction where the other road user was but has not seen the bicyclist.
Aside from the abovementioned contributory factors to bicycle–automobile accidents, inappropriate design features of roundabouts such as inadequate entry deflection and flared entries have also been documented in literature to allow high entry speeds, thereby increasing bicycle accident risks (Maycock and Hall, 1984). Engineering measures such as marking bicycle crossing and raised cycle crossing at junctions are aimed to provide bicycles with greater priority, and there have been numerous studies that have investigated the effectiveness of these engineering measures (e.g., Schoon and van Minnen, 1994, Garder et al., 1998, Hunter et al., 2000, Jensen, 2008).
Compared with studies of intersection accidents, non-intersection accidents involving bicycles have received relatively less attention in literature. Early research work specifically examining non-intersection accidents includes the large-scale analyses of police accident data conducted in the United States by Cross and Fisher (1977) and in New Zealand by Atkinson and Hurst (1983). These researchers revealed that a same-direction crash (i.e., an overtaking and a rear-end crash) was one of the most common crash types for automobile–bicycle crashes. They differentiated between overtaking and rear-end crashes in which bicycle behaviours and motorist behaviours play a part, respectively. It was observed that bicycles’ traversing manoeuvres played a role in the occurrence of an overtaking crash in which a bicyclist (particularly young bicyclists), without being attentive to the traffic behind and without signalling, executed a turning manoeuvre and was struck by an overtaking automobile from behind. The overtaking automobile-drivers observed the bicyclist well in advance, but had less efficient evasive reactions once the bicyclist initiated a turn. A rear-end crash was observed to be associated with motorist behaviours in that a bicyclist travelling straight ahead was rear-ended by a motorist getting too close while passing (Cross and Fisher, 1977). The role of bicycle poor conspicuity especially in nighttime has also been routinely identified for rear-end collisions (e.g., Hoque, 1990, Wood et al., 2009). Evidence in literature showed that a majority of bicycles involved in nighttime accidents was struck by automobiles (travelling from the same directions) from the rear, which pointed to inadequacy of street lighting and bicycle taillights. Recommended countermeasures for reducing the frequency of these two crash types include increased rear conspicuity of bicycles or bicyclists, and the education of juvenile bicyclists (Watts, 1979).
Later studies (e.g., McCarthy and Gilbert, 1996, Stone and Broughton, 2003) further reported that more severe injuries to bicyclists in overtaking collisions (that are generally non-junction accidents) may result from motorised vehicles’ velocities being greater than those in junction accidents where motorised vehicles have to slow to manoeuvre. In addition, overtaking collisions involving large vehicles (e.g., buses/coaches, heavy goods vehicles) other than cars appear to be a safety problem to bicyclists in terms of both accident occurrence and consequence. A behavioural study by Walker (2007) pointed out that drivers of large vehicles were particularly likely to leave narrow safety margin to bicyclists. Adding to the finding of Walker (2007), Parkin and Meyers (2010) further pointed out that HGVs (heavy goods vehicles) and buses were observed to pass bicycle traffic at closer proximities than cars did in the presence of a cycle lane. High involvement of drivers of large vehicles in overtaking crashes may be because bicycles may delay buses or HGVs that have to load and upload passengers/goods punctually. A biomechanical study by Riley and Bates (1980) showed that HGVs commonly caused bicyclist deaths by side impact, and the majority of bicyclists died from multiple injuries through being run over by the wheels of the HGVs. Other biomechanical studies (e.g., Maki et al., 2003, Simms and Wood, 2009) reported that large automobiles with a higher hood caused greater injuries to bicyclists as the grill section hits the middle or upper body rather than the feet. Although not specifically addressing same-direction accidents, past empirical studies employing statistical models of bicyclist injury severity (e.g., Kim et al., 2007, Eluru et al., 2008) confirmed that bicyclists colliding with large automobiles such as heavy trucks were more injurious.
Another common crash type that has been identified by Dennerlein and Meeker (2002) and Transport for London (2005), but received little attention in published literature, was crashes involving bicycles striking an open door of a parked automobile. Collision impacts resulting from first contact with the door (thereby causing cyclists to tumble) and/or second contact with the ground can be devastating, especially to unhelmeted bicyclists. Studies of bicycles striking an open door of a disembarking vehicle occupant seem sparse, although there exists unpublished work that discussed the risk of door collision arising from improper bike lane placement next to on-street parking. It can be noted that most of the typical bicycle lane is located within the so-called “door zone” of parked automobiles because a typical door extends 3 to 3-1/2 feet and bicycle lanes often are just slightly wider than that. The extension of a car door predisposes bicyclists to a risk of door collision despite the bicycle travels in the centre of the bicycle lane. Regardless of the presence of a cycle lane, bicyclists are instructed to ride at least a door's width from parked automobiles, which may also improves sight triangles and increases bicyclists’ conspicuity from a perspective of a parking motorist (Hunter and Stewart, 1999).
Based on the review of previous research, significant gaps remain in understanding the causes leading to these three crash types (i.e., overtaking crashes, rear-end crashes, and door collisions) because junction accidents have been a focus of countless research efforts. The main objective of the present study is to estimate an appropriate statistical model of the three crash types of bicycle accidents on roadways that are 20 m away from junctions (the term “undivided roadway” that represents roadways that are 20 m or more away from junctions is used throughout the remainder of the paper). On a cautionary note, it is worth indicating that these three crash types can occur not only on undivided roadways but also at junctions. The focus on undivided roadway is to avoid the variability in crash occurrence introduced when some other exogenous factors are present at junctions (e.g., junction geometry, junction control measures).
The present paper proceeds with a description of the real-life data (i.e., British Stats19 accident data) used in model estimation. This is followed by a discussion of the proposed methodological approach. Model estimation results are then presented, together with a discussion of the findings. The concluding remarks as well as recommendations for future research are finally provided.
Section snippets
Data source
Data on bicycle accidents were extracted from the British Stats19 national road accident database that is owned by Department for Transport, Great Britain, and maintained and disseminated by United Kingdom Data Achieve. Conditioned on the premise that an accident has involved personal injury and has been reported to local police, appropriately qualified and experienced police accident investigators complete the Stats19 report form. This form records general accident information, on time/date,
Estimation results
Table 2 presents the model estimation results for the mixed logit model of the three bicycle crash types. The results show that the parameters are of plausible sign and that overall goodness of fit is good. The present study also estimates a standard multinomial logit model to predict the likelihood of a non-junction collision being of one particular crash type among the three crash types (Table 3). It is noteworthy that the two models exhibit a good fit to the data as it rejects the null
Discussion
Using mixed logit models to handle the unobserved heterogeneity, the current research has uncovered the determinants of the three bicycle crash types that occurred on undivided roadways in the UK. The model estimation results show that, while some factors appeared to have consistent effects across the observed observations, the effects some other variables have were found to vary across the observations (as indicated by the statistically significant random parameters). When developing
Conclusions and recommendations
By employing mixed logit models, this paper provides insights into contributory factors to the most three common bicycle crash types that occurred on undivided roadways in the UK. Mixed logit models estimated in the current paper were found to be superior to traditional multinomial logit models. In addition, they allow the flexibility to capture individual-specific heterogeneity that may arise, for example, from a set of physical causes of collision types (e.g., travel speed, roadway/geometric
Acknowledgment
The author would like to thank the reviewers for their insightful comments that significantly improved the original version of this paper.
References (48)
- et al.
A note on modelling vehicle accident frequencies with random-parameters count models
Accident Analysis and Prevention
(2009) A Policy on Geometric Design of Highways and Streets
(2004)- et al.
Collisions between cyclists and motorists in New Zealand
Accident Analysis and Prevention
(1983) Quasi-random maximum simulated likelihood estimation of the mixed multinomial logit model
Transportation Research Part B
(2001)Simulation estimation of mixed discrete choice models using randomised and scrambled Halton sequences
Transportation Research Part B
(2003)A review of the ‘look but failed to see’ accident causation factor
Behavioural Research in Road Safety XI
(2002)- et al.
A Study of Bicycle/Motor Vehicle Accidents: Identification of Problem Types and Countermeasure Approaches
(1977) - et al.
Injury crashes with bicyclists at roundabouts: influence of some location characteristics and the design of cycle facilities
Journal of Safety Research
(2009) - et al.
The effects of roundabouts on traffic safety for bicyclists: an observation study
Accident Analysis and Prevention
(2008) - et al.
Occupational injuries among Boston bicycle messengers
American Journal of Industrial Medicine
(2002)
National Travel Survey 2008
Transport Statistics Bulletin—Compendium of Motorcycling Statistics
Road Casualties in Great Britain 2008
A mixed generalized ordered response model for examining pedestrian and bicyclist injury severity level in traffic crashes
Accident Analysis and Prevention
Measuring the safety effect of raised bicycle crossings using a new research methodology
Transportation Research Record
Mixed logit analysis of safety-belt use in single- and multi-occupant vehicles
Accident Analysis and Prevention
The effect of roundabout design features on cyclist accident rate
Accident Analysis and Prevention
Looked-but-failed-to-see-errors in traffic
Accident Analysis and Prevention
Applied Choice Analysis: A Primer
Vision, visibility and perception in driving
Perception
An analysis of fatal bicycle accidents in Victoria (Australia) with a special reference to nighttime accidents
Accident Analysis and Prevention
Evaluation of blue bike-lane treatment in Portland, Oregon
Transportation Research Record
Bicycle–motor vehicle crash types: the early 1990s
Transportation Research Record
Cited by (58)
Wider view over bicycle crashes: Complementing and extending bicycle crash statistics in urban areas using surveys
2023, Journal of Safety ResearchFactors impacting bike crash severity in urban areas
2022, Journal of Safety ResearchCars overtaking cyclists on different urban road types – Expectations about passing safety are not aligned with observed passing distances
2022, Transportation Research Part F: Traffic Psychology and BehaviourCyclists’ interactions with other road users from a safety perspective
2022, Advances in Transport Policy and PlanningSafety of bicyclists in roundabouts with mixed traffic: Video analyses of behavioural and surrogate safety indicators
2021, Transportation Research Part F: Traffic Psychology and BehaviourThe impact of a child bike seat and trailer on the objective overtaking behaviour of motorized vehicles passing cyclists
2020, Transportation Research Part F: Traffic Psychology and Behaviour