2.1 Motorcycle accidents

Motorcyclists have an especially poor safety record when compared to other road user groups. Their killed and serious injury (KSI) rate in the UK, per million vehicle kilometres, is approximately twice that of pedal cyclists and over 16 times that of car drivers and passengers. Motorcyclists make up less than 1% of vehicle traffic but their riders suffer 14% of total deaths and serious injuries on Britain's roads (DETR, 2000).

26,192 motorcyclists (this figure includes moped and scooter riders) and pillion passengers were injured in reported accidents in Great Britain in 1999. 6,361 of these injuries were considered serious, and 547 motorcyclists and passengers were killed. In the same year 205,735 car drivers and passengers were injured in accidents. 18,681 of these injuries were considered serious, and 1,687 drivers and passengers died.

In 1999 a motorcyclist was killed or seriously injured for every 665,894 kilometres ridden. Car drivers, however, covered an average of 18,661,626 kilometres before a serious injury or death occurred. According to these figures, in 1999 motorcyclists were approximately 28 times more likely to be killed or seriously injured on the roads in Great Britain than car drivers.

Chesham et al. (1993) compared distance travelled with injuries sustained and found that in 1990 a motorcyclist was 35 times more likely to be killed or seriously injured than a car driver. Although the figures have improved over the last 10 years, the risk factor for motorcyclists when compared to car drivers is still very high. It is important to remember that, compared in an accident, a motorcyclist is much more vulnerable to personal injury than a car driver. Safety and accident avoidance for motorcyclists is therefore of paramount importance.

A similar picture is given by research in other countries. Motorcycle riders in New Zealand accounted for approximately 20% of fatalities and 25% of hospitalisations for road traffic accidents as a whole, but motorcycles represented only 5% of licensed vehicles and accounted for only 1.4% of estimated total vehicle mileage in that country (Reeder et al., 1999). Young male riders, in particular, were identified as a problem; riders aged 15-24 years accounted for 67% of all motorcycle accident fatalities. This led to the introduction of a graduated licensing scheme in New Zealand, which has reportedly reduced casualties in the target group of 15-19 year olds by 22%, though this mainly occurred by reducing that groups' overall exposure to motorcycle riding.

Research in Norway by Kopjar (1999) investigated young riders' moped accidents, the use of mopeds in that country apparently being relatively widespread in the 16- 17 year old age group. Kopjar discovered that moped-related accidents accounted for 50% of hospitalisations for traffic accidents as a whole, and that 43% of moped accidents were single vehicle incidents. He concluded that moped injuries were a serious problem in late adolescence, and that road safety professionals often overlooked the moped problem.

Though a fall in the amount of motorcycling and changes in motorcycle training in the UK both contributed to a fall in the number of motorcyclist casualties in the early 1980s and mid-1990s, this trend has reversed in more recent years. The possible reasons for this include the increasing sales of mopeds and scooters: the number of motorcycles, scooters and mopeds that are newly registered has more than doubled between 1991 and 2001, with over 177,000 being registered in 2001. This is despite an earlier fall in registrations to below 59,000 in 1993. Between 2001 and 2002, two-wheeled motor vehicle traffic rose more than other categories, with motorcycle traffic seeing a rise of 5.5% (DfT, 2003). There are also increasing numbers of older motorcyclists returning to the road on fairly powerful machines after a long break in riding motocycles; KSI casualties in age groups between 30 and 59 have increased in the last 10 years and, similarly, Ormston et al. (2003) found that, in Scotland, motorcyclists aged over 30 have accounted for an increasing proportion of casualties since 1997. According to the National Statistics/DfT (2003) National Travel Survey (NTS), average trip lengths have increased from 6.1 miles to

9.7 miles, while the average motorcycle rider now spends 3.2 hours a week riding a motorcycle compared with 3.0 hours between 1989 and 1991.

The reasons for riding tended not to vary between males and females, although the NTS found that men were nearly seven times more likely to make a motorcycle trip than women, and young male motorcyclists were the heaviest users in terms of both the average number of trips they made (10.0 per driver per week) and distance travelled (77 miles). Distance travelled increased with engine size. In the period 1992-2001, 43% of motorcycles in the NTS sample had engines over 500cc in capacity (16% below 50cc). Although smaller motorcycles were used mainly for commuting, business or education and less for leisure purposes than larger motorcycles (over 500cc), these too were also used mainly for commuting, business or education purposes. Unsurprisingly, in the light of this, four out of five motorcycle trips were made during the week and Sundays showed the lowest level of motorcycle use.

Very little research has been done in this country on the types of crashes experienced by motorcyclists. Motorcycle accidents have somewhat different characteristics when compared with other vehicle groups. KSI casualties in the UK peak through the 20-39 rider age bands, and motorcycles are over-represented in right of way violation (ROWV) accidents, accidents involving running off the road on bends, and accidents that are related specifically to the sort of manoeuvres that motorcycles can perform, e.g. overtaking other traffic without crossing the centre line or ‘filtering' between lines of traffic. Preusser et al. (1995) found that a subset of fatal motorcycle accidents with characteristics similar to these accounted for around 85% of the total, in a sample of over 2,000 such accidents. One of the most widely quoted in-depth US study, by Hurt et al. (1981), also highlighted the high frequency of ROWVs and single vehicle accidents on bends in a sample of over 3,000 motorcycle accidents. Moss (2000) focussed attention on rural motorcycle accidents in Cheshire, and found that sports bikes and riders in the 26-40 age group accounted for the majority of rural bend accidents.

Chesham et al. (1993) reviewed the changing focus of research activity in the area of motorcycle safety. They found two main periods of time with a different emphasis. The first period covering the 1970s was based on actual accident analysis in which control factors that contribute to the severity of motorcycle accidents were examined (such as drink driving). The second period in the 1980s considered the actual process of motorcycle riding, focussing on, for example, rider skills. They go on to say that the 1990s brought a third area of consideration in which riders' beliefs and attitudes about road safety are being considered.

Mannering and Grodsky (1995) point out several reasons why the characteristics of motorcycle accidents differ from those of other vehicles. Firstly, they claim car drivers ‘tend to be inattentive with regard to motorcyclists and have conditioned themselves to look only for other [cars] as possible collision dangers'. Motorcyclists themselves often repeat anecdotal stories of the car driver's ‘sorry I didn't see you' explanation for collisions. Secondly, Mannering and Grodsky (1995) also claim that motorcycle operation is typically a more complex task than car driving, requiring excellent motor skills, physical co-ordination and balance. Motorcycle riding can also involve counterintuitive skills, such as ‘counter-steering, simultaneous application of [mechanically separate] front and rear brakes, and opening the throttle while negotiating turns'. Any impairment (for example, from medication or alcohol) would therefore more greatly affect a motorcyclist's risk of an accident when compared with a similar level of impairment while car driving. Sun et al. (1999) have argued that, for this reason, legal blood alcohol levels should be lower for motorcyclists than the level set for other drivers in the USA.

Mannering and Grodsky (1995) also state that, because motorcycle riding is well known to be a dangerous activity, it ‘may tend to attract risk-seeking individuals, in all age and socio-economic categories', which would have a corresponding effect on the total motorcycle accident figures. Some evidence of another aspect of risk-seeking among motorcyclists was found by Sunderstrom et al. (1999) in their study of casualties at a US hospital's trauma centre. They discovered evidence that the use of illegal pharmaceuticals had declined markedly in car driver casualties over a 10year period, but not among motorcycle rider casualties. In addition, the use of alcohol, cocaine and PCP (‘angel dust') was found to be higher among injured motorcyclists than among injured car drivers.

Lastly, Mannering and Grodsky (1995) point out that most motorcycles offer a substantially better performance when compared with cars, due mainly to their far higher power to weight ratio. The higher engine capacity sports machines offered for sale in the USA and UK frequently boast acceleration rates of 0-60 mph in less than three seconds. Mannering and Grodsky report that ‘some motorcyclists cite as the primary reason for selling their sport bike the fact that they are unable to resist the temptation to ride at dangerous speeds'.

Rutter and Quine (1996) looked at age and experience in motorcycling safety, and from a national prospective survey of over 4,000 riders in the UK found that age played a much greater role than inexperience in explaining why young age groups are over-represented in casualty statistics. Rutter and Quine explain that more emphasis should be given to the consequences of dangerous riding and why training is so important: ‘Only when a proper set of underpinning beliefs and perceptions is provided for behaviour will skills be turned into safety'.

Concerning ‘ROWV' accidents, Hurt et al.'s (1981) study found that, in multiple vehicle accidents, the driver of the other vehicle violated the motorcyclist's right of way and caused the accident in two-thirds of all such accidents. Keskinen et al. (1998) found that older drivers have problems at intersections but seem to be able to cope with other traffic situations comparatively better than younger drivers. Young, middle-aged and older male drivers' habits at T-junctions were examined, the focus of interest being on the attention and interaction between drivers of different age groups. Time differences were noted (i.e. the time passing from the moment the turning driver had completed his/her turn until the opponent driver on the main road reached the centre of the intersection). Two notable conclusions were reached. Firstly, older drivers have a habit of driving and accelerating slowly, and accelerating slowly may shorten the time difference when entering the main road. Young drivers/riders, on the other hand, tend to travel faster. ‘An older driver turning and a young driver approaching [could therefore] create a potentially dangerous combination with a low safety margin' (Kestinen et al. 1998). Crucially, time differences were also found to be particularly short when the opponent driver was riding a motorcycle.

A review by Brown (2002) of work relating to drivers' ‘looked but failed to see' accidents examined some of the psychological processes that might occur in drivers reporting this type of accident. Of particular note was Duncan's (1996) ‘integrated competition hypothesis', which suggests that attention to some kinds of object in road traffic scenes may be inhibited as drivers concentrate on features of the traffic scene which their experience has shown to be of critical importance. In addition, Treisman's (1996) ‘feature integration theory'suggested that drivers might rapidly scan the traffic scene for a single feature of a potential hazard, such as proximity, and decide to proceed without noticing the approach of a more distant but rapidly approaching vehicle. Perception experiments by Mack and Rock (1998) have also shown that subjects may be less likely to perceive an object if they are looking at it directly than if it falls outside the centre of the visual field, a phenomenon which they call ‘inattentional blindness'. Brown (2002) notes that theorising in this area has been considered somewhat controversial, but further investigations of such perceptual factors could clearly have great importance in the understanding of motorcycle accident causation.

Much of the literature suggests that rider attitudes and the perception of the risks involved in motorcycling are the most important consideration when deciding to what extent motorcyclists are at risk from injury compared with other road users. Mannering and Grodsky (1995) surveyed motorcyclists' perceived likelihood of being involved in an accident and concluded that motorcyclists do have a reasonable grasp of factors that can increase the likelihood of accident involvement, highlighting in particular miles ridden, speeding and dangerous overtaking manoeuvres. Their findings suggest that, for the most part, motorcycle accidents are not the result of misjudgement regarding the overall risk of motorcycling.

Everett et al. (2001), however, examined national trends in transportation-related injury risk and safety behaviours among US high school students, and they found that many young people place themselves at unnecessary risk from motor vehicle and bicycle related crash injuries and fatalities because of drink driving and the improper use of safety equipment (including motorcycle helmet use).

Cross-national differences in risk-perceptions in Japan and the US were examined by Hayakawa et al. (2000) who found that objective differences in risk environments combine with cultural influences, which, in turn, leads to differences in risk perception. Raising a motorcyclist's perception of risk would, therefore, seem to be a logical way of improving a motorcyclist's riding behaviour.

Reeder et al. (1996) examined the opinions and behaviours (and therefore the perceived risk) of young motorcyclists in New Zealand. An examination of the literature found three broad areas of concern; the extent and use of protective gear, motorcyclists' level of conspicuity, and risky and illegal behaviour. Very importantly they found that there was a difference between the riders' opinions and actual practice. ‘Protective opinions were much more common than protective behaviours despite a widespread belief in their efficacy' (Reeder et al. 1996). Protective clothing, for example, was often not worn because it was, in many cases, unacceptable to the users. Fashion may be a contributory factor to this, as may be the cost of the clothing itself. The use of headlights in daytime was, however, found to be much more common.

2.2 Methodological issues

The causality of real road accidents can be a difficult phenomenon to study. One possible solution to this is the use of methodology that investigates road accidents after they have occurred, rather than the more familiar psychological research that relies for its method on examination of driver behaviour in controlled environments.

One such well-known approach involves the use of multi-disciplinary accident investigation (MDAI) teams that travel to the site of accidents soon after they occur to collect data. Research such as that of Sabey and Taylor (1980) is based on the work of MDAI teams. Findings were concerned with the proportional contributions to road accidents of the user, environment and vehicle. It is from this work that the much quoted figure of 95% was identified as the proportion of road accidents involving human error. Sabey and Taylor (1980) quoted research carried out in the US that produced much the same figure. They went on to assess driver errors behind this figure by examining the contribution of perceptual errors, lack of skill, manner of execution and various forms of impairment, such as alcohol. One of the most detailed motorcycle accident studies ever made in the US, the Hurt Report (Hurt et al., 1981), also used MDAI teams to produce detailed findings over a high number of case examples.

However, in a review of the work of multi-disciplinary team research worldwide, Grayson and Hakkert (1987) pointed out several disadvantages to such a method. Operational costs are very high, and typically only a small number of accidents can be studied, unless research can be carried out over several years. Although Sabey and Taylor (1980) did study over 2,000 accidents, such a figure is the exception rather than the rule. There is a bias towards injury accidents due to the notification procedure. The accidents sampled are bound to be of a heterogeneous nature, which works against any approach that aims to study a specific problem.

A further criticism concerns the conclusions reached. Despite the vast amount of information collected in such work, ‘definitive conclusions are very limited' (Grayson and Hakkert, 1987) and have been applied mainly to vehicle design and engineering efforts rather than to human behaviour and road design. According to Grayson and Hakkert (1987), these limitations tend to disappear ‘if an in-depth but not immediate response on-the-spot approach is taken'. They comment that it is also important that any in-depth technique is only really of use if applied to specific areas rather than a large heterogeneous sample of information.

Many studies have used in-depth techniques applied to secondary data sources, such as police reports, interviews and questionnaires. Fell (1976) was among the first to claim that an ‘accident causal schema' could be constructed from such sources. Fell was of the opinion that in-depth work using police reports, while still having some limitations, could be used to improve the ‘state of the art' in understanding accident causation.

More recently, Malaterre (1990) used police reports to break down and analyse accidents. Malaterre constructed four stages in his analysis - driving, accident, emergency and collision. The factors identified in his analysis stage were next used in synthesis; the building of prototype cases. Such an approach, Malaterre claimed, focussed effectively on functions not correctly carried out by the driver, which are sometimes difficult to locate. Malaterre's sample was, however, quite small (115 cases) and was also heterogeneous. He ended by concluding that more precise analysis needed to be carried out by referring to complete police accident reports, with all their varieties of information.

It is often overlooked that local council initiatives into examining accident causation at specific locations (‘black spots') make much use of police reports to present a full picture of what happened. England (1981) describes the approach as very cost effective when targeting engineering countermeasures, and points out that it has the additional benefit of checking the accuracy of summary statistical information that is held on accidents.

The in-depth technique itself has been used in areas outside accident causation for some time. The examination of in-depth case study techniques by Yin (1984) shows how they are primarily of use in producing analytic generalisations rather than more traditional statistical generalisations. They concentrate on an iterative type of explanation building that often features chronologies, sequences and contingent event analysis.

Case study methods were used by Clarke et al. (1998a) in the analysis of police road accident files in right-turn accidents, a key feature of this work being that it treated accidents as a ‘clinical' problem, rather than just an ‘epidemiological' one as in many traditional approaches. For the first time sequence analysis was used in conjunction with rule-finding computer software. This approach concentrated on the relatively homogeneous class of right-turn accidents to produce new findings. It was, however, felt that much of the information from the original police reports was being lost. The rich nature of an accident report that made it understandable to a human observer had to be left out when the data were being prepared for computer analysis. Subsequent work investigating overtaking accidents, by Clarke et al. (1998b), placed more emphasis on the interpretation of causal patterns by the human coders, but retained the powers of a computer database for the later stages of storing, sifting and aggregating explanatory models of individual cases. This later approach was continued in the present study. The method has been further developed over the years in this research group and is reported in detail in Clarke et al. (2002).

3.1 The accident database

The data were entered into a FileMaker Pro database customised to handle the information and search parameters required for this project. Figure 1 shows the standard data entry set-up.

In-depth Study of Motorcycle Accidents

Data are entered describing the relatively objective facts of each case: time of day, speed limit, class of road, etc. The database includes some fields configured as check boxes or ‘radio buttons'; these provide quick access to selected cases during further analysis. Summary fields are also used to calculate things such as the mean age of the involved riders. Any combination of fields in the database can be used to search for cases matching a variety of criteria. A variety of layouts are also used to present and analyse the data, in addition to the data entry layout above.

A ‘prose account' is also entered for each case giving a step-by-step description of the accident. The causal story is always written from the viewpoint of the motorcyclist, who is labelled as ‘rider 1', though much consideration is also given to other road users' actions and intentions. The prose accounts give a detailed summary of the available facts, including information from witnesses that appears to be sufficiently reliable. Discrepancies can occur between the interviews of riders and the statements of independent witnesses, but these can usually be resolved by considering all the statements together with various other reported facts. These can include the measurement of skid marks by police, vehicle damage reports, etc. Figure 1, it should be noted, only shows part of a typical prose account because the text is held in an ‘expandable field' in the database.

Next, a sketch plan of each accident is made from sources in the file. The orientations of the sketch plan and the icons contained in it are standardised for the speed of entry and to allow direct comparisons between example or prototype cases.

A minimum set of possible explanations for each accident is recorded from a standard checklist adapted and developed from a previous study (Clarke et al. 1998b). The list has subsections for the road environment, vehicle and rider characteristics, and specific rider actions. The emphasis throughout is on giving the finest grain description possible of each accident, not for use as a formal coding scheme but rather to provide search and selection aids to identify homogeneous groups of cases for further qualitative analysis. In addition, we entered data for a version of a national ‘contributory factors in accidents' form developed at the Transport Research Laboratory (TRL) which involves the identification of one major precipitating factor (PF) from a possible list of 15, and a further coding of up to four contributory factors (CFs) together with a confidence rating in the CFs identified. Finally, entries are made in additional fields for comments and quotes from the involved riders.

The ultimate aim of entering facts and figures, prose accounts, standardised graphics and explanatory factors in the database was to build a library of analysed cases stored as a series of case studies. In this sense, the database is used to find groups and recurring patterns, rather than being considered as ‘raw'data awaiting analysis. In this way it was possible to find patterns, sequences and processes within each group of accident. Statistical examinations were not the primary focus of the study, even though simple statistics were used to characterise the sample.

3.2 The questionnaire

An examination of the attitudes of motorcyclists was the next and final phase of the research process. A quantitative approach was used in the form of a questionnaire survey as this seemed to be the most appropriate method given the aims and constraints of the project.

The questionnaire was designed to be completed without any help from the researcher, therefore minimising the possibility for influencing respondent's answers. Secondly, the respondent's anonymity could be assured as respondents were given the option of not supplying any information that would specifically identify them. This was particularly important for this study as some of the questions required the respondents to admit to potentially incriminating information, such as breaking the speed limit or driving while under the influence of alcohol. It was for this reason that the respondents were given the option to remain anonymous.

There were four main sections to the questionnaire focussing on rider's experience, training undertaken, safety issues and personal details (for background purposes). A final version of the questionnaire can be found in the appendix of this report.

The Motorcycle Action Group (MAG) is a political pressure group with approximately 45,000 members in the UK, who exist to promote rider's rights and welfare. All branch representatives were contacted via email to enquire if they would be willing to ask their members to complete a questionnaire. In addition, contact was made with the Public Liaison Manager at the group's central offices who agreed to put a copy of the questionnaire on the group's website for riders to either print out and return by post or to fill in electronically. A notification of the questionnaire was also included in a monthly email sent to all MAG members in summer 2003 alerting them to the questionnaire.

4.1 Right of way violations

Of the total cases, 681 (38%) involve ROWVs. However, less than 20% of these involve a motorcyclist who rated as either fully or partly to blame for the accident. The majority of motorcycle ROWV accidents have been found to be primarily the fault of other motorists. This is an even higher level of ‘non-blameworthiness' in ROWV accidents than that observed in other in-depth studies, e.g. Hurt et al. (1981)

The majority of ROWVs occur at T-junctions, which are three times as common as roundabouts or crossroads. This finding is in accordance with the work of Hole, Tyrell and Langham (1996), who found that the majority of such accidents occurred at ‘uncontrolled' (i.e. no stop light or sign with only give-way markings and/or signs present) T-junctions in urban environments.

When these cases are examined, it can be seen that the most common failure of other drivers in motorcycle accidents is a failure in the continuity of their observation of the road scene. Over 65% of ROWV accidents where the motorcyclist is not regarded as to blame involve a driver who somehow fails to see a motorcyclist who should be in clear view, and, indeed, frequently is in view to witnesses or other road users in the area. Failures of observation that involve drivers failing to take account of restricted views of one kind or another, and failing to judge the approach speed and/or distance of a motorcyclist, are not included in this category. Sometimes, accident-involved drivers in motorcycle accidents fail to see riders even when they are verifiably using visibility aids, such as daytime running lights and high-visibility protective clothing. This occurs in over 12% of such cases (but the level of use of these aids to visibility is felt to be under-reported by police). An example of such a case is given in Figure 6.

Story: It was early in the afternoon on a fine spring day. The rider (M,44) of a Honda CBR1000 motorcycle was travelling along an unclassified urban road at around the 30 mph limit. According to witnesses, he was not going above the speed limit and was displaying daytime lights. As he approached a junction ahead on the offside, he could see a Vauxhall Astra (2), driven by (F,63) waiting to turn right at the give way line to travel in the same direction as him. As he got to within 20 metres of the junction mouth, the car driver began to emerge, making her right turn. The motorcyclist braked heavily and steered nearside in an effort to get his bike between the nearside kerb and the turning car before he hit it. However, he was unsuccessful in this, and he hit the nearside of the car as it turned, causing a severe injury to his right hand that required two operations and several months off work. The Astra driver claimed that she had looked left, but had simply not seen the motorcyclist, despite the fact that visibility was good and the rider was displaying lights. She was charged with driving without due care and attention.

4.2 Losing control on bends

Over 15% of total cases involve loss of control on a bend, corner or curve. This type of accident is almost always regarded as primarily the fault of the motorcyclist rather than other road users, and it has already been shown that such accidents are more associated with riding for pleasure than accidents of other types. Hurt et al. (1981) found that rider error in such cases consisted of ‘slideout and fall due to overbraking, running wide of a curve due to excess [inappropriate] speed, or under-cornering', which seem also to be the most frequent rider errors in bend accidents in this sample.

Riders having this type of accident are nearly three times as likely (compared with the whole sample) to be rated as ‘inexperienced' riders by researchers; this usually occurs when it has been noted by an investigating police officer that some form of inexperience was a factor, e.g. a rider who has only very recently passed the motorcycle test. Interestingly, such accidents appear no more likely than accidents of other types in the sample to occur on damp, wet or icy roads, though there is some evidence of a problem with riders hitting oil, gravel and mud in some rural bend accidents. Though inappropriate speed for the bend is implicated in a large proportion of cases, there are also a number of cases where there seems no evidence of any failure except ones relating to lack of experience, as in the fatal example shown in Figure 7.

Story: It was late in the evening on a fine night in autumn. The rider (M,28) of a Kawasaki ZZR 600 motorcycle (1) was travelling along a rural B road with a 60 mph limit and streetlights. He was out for a pleasure ride with another friend of his who was ahead of him. Both riders had taken direct access intensive motorcycle courses to gain their full licence and rider 1 had passed his test 11 days before. He had picked up the Kawasaki that day. The two riders travelled round a right-hand bend. Although the bend was not so severe that it could not be taken at the legal maximum of 60 mph, rider 1, who was following his slightly more experienced friend, failed to take the bend correctly even though he was not exceeding the limit. His bike hit the nearside kerb and he was thrown into the road. The bike was a write off and the rider received multiple injuries from which he died. Police put the cause of the accident down to an inexperienced rider riding an unfamiliar machine.

Diagram:

The riders in this ‘going out of control' category seem to fall into two groups. There are those riders who have a full motorcycle licence, but perhaps either have not held it for a long time (as in Figure 7) or have returned to motorcycling after passing a test some years ago (so called ‘born again' bikers). The mean age of these riders considered together is 29.7 (standard deviation 10.3), and of those cases with licence records available, they have a lower mean number of years experience (5.3 years of full licensure) than the equivalent group derived from the sample by excluding ‘going out of control' accidents (7.2 years of full licensure). Of accidents where the rider is rated fully or partly to blame for the accident, nearly one-third of the accidents in the 31-35 year old age range involve going out of control on a bend, which perhaps contributes to the peak in this age range shown in Figure 2.

Then there is a group of riders who either have only a provisional licence or, sometimes, no licence at all, who can be regarded as perhaps the least experienced riders in the whole sample. The mean age of this group is 22.6 (standard deviation 8.8). The latter group has approximately twice the proportion of fatalities found in the former group (though caution must be exercised in the interpretation of these findings, as frequencies are quite low). Again, of accidents where the rider is rated fully or partly to blame for the accident, 20% of the accidents in the 16-20 year old age range involve going out of control on a bend, which again perhaps contributes to the peak in this age range shown in Figure 2.

4.3 Motorcycle manoeuvrability accidents

We have identified a subgroup of the sample cases that comprise accidents specifically related to the way motorcyclists are able to manoeuvre their vehicles in ways that are frequently not appreciated by other motorists. Even small capacity motorcycles have a high power-to-weight ratio, and nearly all motorcycles are small enough to make use of unoccupied road space between other traffic that larger vehicles are unable to use.

If all accident cases where the rider is judged either fully or partly to blame are examined (n ¼ 919), 16.5% involve a motorcyclist overtaking other vehicles and causing an accident. These riders have a tendency to be slightly younger than the rest of the sample, and the indications are that they have a tendency to be riding machines of a higher engine capacity than other accident-involved drivers (mean cubic centimetres of capacity ¼ 507, versus 431 for riders to blame/partly to blame in other types of accident). Perhaps surprisingly, there is no evidence that the two peak age ranges identified in Figure 2 (16-20 years and 31-35 years) are overinvolved in overtaking accidents when compared with the rest of the sample.

However, motorcycle accidents also occur when riders take the opportunity to pass slow moving or stationary traffic, which is often referred to as ‘filtering'. Though only slightly more than 5% of the whole sample identifiably involve a rider filtering, other drivers are more than twice as likely to be considered at fault in such accidents as the motorcyclists involved, though there is also evidence for an increased proportion of ‘combined fault' accidents in this category. It seems that motorcyclists are, as it were, ‘subverting' other drivers' expectations of how traffic behaves, in some cases. An example of this is shown in Figure 8.

Story: It was dark on a fine December evening. It was not raining but the road was damp. The rider (M,47) of a Honda CBR600 motorcycle (1) was passing a queue of stationary traffic on a single carriageway A class road with a 60 mph limit when the driver (M,56) of a Vauxhall Astra Van (2) pulled out of the line of traffic with the intention of making a U-turn and collided with the motorcyclist who was knocked from his bike and sustained an black eye, bruised nose and sprained shoulder. Driver 2 admitted to not seeing the motorcycle and not looking properly after assuming that the vehicles behind were stationary, he had clearly not considered the possibility of a motorcycle passing the traffic. Several witnesses clearly put the blame with driver 2 who was charged with driving without due care and attention. One witness however thought the motorcyclist was travelling a little fast but there is no evidence to support this.

At his police interview, the driver in the above case made the following statement: ‘There could be nothing coming from behind me because the car and lorry to my rear were stationary'.

4.4 Other motorcycle accidents

Though the majority (70%+) of motorcycle accidents are covered by the scenarios outlined above, the remainder contain some special areas of interest. One such is rear-end shunt accidents. Rear-end shunts have been found to be among the most common types of accidents for all drivers. West and French (1993), in their analysis of different types of shunt, found that ‘active involvement in shunts was a function of being young and male'. Shunts account for over 11% of all motorcycle accidents in the sample, and riders are typically found to be more likely to be at fault than in accidents of other types. The evidence is that ‘at fault' riders in shunt accidents tend to be younger, more inexperienced riders, on smaller capacity machines. Nearly 40% of them are riding scooters and mopeds, motorcycle types which account for only 17% of machines in other types of ‘at fault' accident. It could be that these relatively inexperienced riders are experiencing difficulties in bringing their machines to a controlled stop, especially in wet and slippery road conditions. Lightweight bikes with separate front and rear brakes are relatively easy to break into a skid on. Inexperience in this area can have fatal consequences that would be unlikely on four wheels, as shown in Figure 9.

Story: It was early in the morning on a cold day in winter. Dawn had just come up, and it had got light, but most vehicles still had at least sidelights on. The weather was frosty, but the roads had been salted and gritted and, as a consequence, they were quite wet. The rider (M,18) of a Suzuki scooter-style moped (1) was travelling along an unclassified urban road close to his home. He was on his way to work.

The rider was following, at some distance, a Fiat Tempra (2), driven by (M,30). The Fiat approached a zebra crossing ahead, and the driver noticed that there was a paperboy on a bike waiting to cross the road, so he slowed down and stopped in a normal fashion. He remained stopped with the clutch depressed and his foot on the footbrake. The car had its lights on, and the brake lights were therefore illuminated in addition.

The paperboy on the bike began to cross the crossing, and had almost reached the other side, when rider 1 approached the rear of the waiting Fiat. He failed to notice it until too late, and when he did, he applied his brakes quite hard in an attempt to stop behind the car. The back wheel of the scooter locked under braking and skidded to the nearside on the wet road. The rider and machine both hit the road sideways and skidded into the rear of the Fiat, getting trapped under the rear of the car. During the short time the rider was skidding down the road, his full-face helmet came off as the strap was not fastened properly. The rider hit his head on the rear of the car bumper when he collided with it, and received a serious head injury, from which he died a short while later in hospital.

Witnesses and police attached no blame whatever to the car driver. No one could understand why the rider's helmet had not been strapped on properly, as it apparently had been when he left home that morning some minutes previously. Had his helmet not come off, it was held to be quite likely that he would have survived the accident. One witness described how he saw the scooter rider most mornings, and that the rider always seemed unsteady and very inexperienced on his machine. The coroner recorded a verdict of accidental death.

Diagram:

4.5 Types of motorcycle and engine capacity

According to DfT figures for 2001, the percentages of motorcycles of various engine capacities registered in the UK were as shown in Table 2 below. The equivalent percentages from our sample are shown alongside.

Figures from our sample of accident-involved motorcycles reveals that the 50- 125cc band above, which contains machines most often used by young, inexperienced and learner riders, accounts for over 25% of the sample; it seems that these machines are over-represented in accidents relative to their registered numbers. Larger engine capacity motorcycles above 500cc are under-represented in the sample with respect to their numbers on the road, even though they are likely to travel higher mileages than smaller machines. A similar finding occurs when percentage sales figures from the Motorcycle Industry Association (MIA) (for 2002) are compared with percentages of various bikes in the sample.

A slightly more detailed table of engine capacities was produced from the study, shown below in Table 3. This included breakdowns across three types of accident: ROWVs, bend accidents and those involving overtaking/filtering.

Riders of machines 100-250cc in capacity are significantly younger than other riders in the sample, but this is perhaps not surprising as most ‘learner legal' machines fall into this cc band. Riders of machines 500cc and above, and 900cc and above in particular, seem to be significantly older than other riders in the sample. Riders of machines 600cc and above seem to have an above average risk of becoming involved in accidents on bends, and there is also some evidence of an increased risk of accidents involving overtaking or filtering for riders of machines 900cc and above.

The MIA also records annual figures for motorcycles sold of various types, e.g. custom, super-sport, traditional, etc. These categories were also used, where possible, in the study. The figures for 2003 are shown in Table 4.

Some of the motorcycle categories were also selected for analysis using standard normal residuals (as used with rider age groups earlier), and Figure 10 shows how machines of various categories are over-or under-represented in the sample with respect to the type of accident they are involved in.

Figure 10: Standard normal residuals for five types of motorcycle in four types of
accident, for cases where motorcyclists have been judged fully or
partially to blame for the accident

It can be seen that super-sport motorcycles are over-represented in bend accidents, but scooters and mopeds are under-represented in this type of accident while being more likely to come to grief in rear-end shunt accidents. Super-sport bikes have a significantly lesser propensity than other types of motorcycle for being involved in both rear-end shunts and ROWV accidents. Somewhat surprisingly, sports-tourer bikes appear significantly over-represented in overtaking/filtering accidents. No type of motorcycle stands out as being over-represented in ROWV accidents, though super-sport bikes appear to be under-represented.

It is clear that the majority of respondents (60%) have not had the level of training offered today, having undertaken the old-style motorcycle test before the introduction of compulsory basic training. The training undertaken was not detailed on the motorcycle accident database and so a comparison could not be made.

4.7.6 Riding habits

Riders were encouraged to comment on their riding habits in question number 13 of the questionnaire (a copy of which is in the appendix of this report). The first two sections of question 13 concerned visibility. Daytime headlight use was fairly common, with over 60% (n ¼ 143) of the respondents always or frequently opting to use lights. The use of bright and/or reflective clothing was, however, less common. Just 14% (n ¼ 143) of the respondents always or frequently chose to wear such items of clothing. Unfortunately, these details were often not recorded in police files used for the accident database so a direct comparison to actual accidents could not be made.

The use of protective clothing was considered next. Protective jackets were always worn by 81% (n ¼ 145) of the respondents and a further 13% frequently wore one. Protective trousers were worn less but even so 69% (n ¼ 145) of the respondents still said they always or frequently wore them.

Only 7.6% (n ¼ 144) of the respondents said they regularly drove their motorcycle while they considered themselves to be tired and a quarter of the respondents claimed they ‘never'drove while feeling tired. Unsurprisingly then, tiredness was a known factor in only four of the accidents on the motorcycle accident database. Similarly, 84.7% (n ¼ 144) of the respondents claimed they never rode ‘under the influence of alcohol or drugs', with the remaining respondents only admitting to have occasionally done so. Excess alcohol was found to be a contributory factor in just 3.4% of all accidents on the motorcycle database.

Speeding was found to be common among the respondents, with 58% (n ¼ 143) admitting to always, or frequently, breaking the speed limit. The remaining respondents admitted to ‘occasionally' breaking the speed limit but only when they thought it was safe to do so. Travelling in excess of the speed limit was considered to be a contributory factor in just 3.5% accidents on the motorcycle accident database. Of these, 62% were a result of the motorcyclist speeding. Misjudging the appropriate speed for conditions, however, accounted for a further 5.6% of accidents where the speed limit itself was not broken. Speeding, whether in excess of the speed limit or travelling too fast for the conditions, was therefore a factor in a total of 9.2% of accidents on the motorcycle accident database.

The respondents also admitted to frequently speeding while overtaking other road users, with 38.6% (n ¼ 145) of the respondents claiming to regularly pass vehicles travelling ‘at or above the speed limit'. Only 9.7% of the respondents claimed to never pass a vehicle travelling at or above the speed limit. Just under half of the respondents regularly passed two or more vehicles at the same time while overtaking, with only 3.4% (n ¼ 145) of the respondents claiming never to have done this. Overtaking accidents only account for a total of 6.6% of motorcycle accidents held on the motorcycle accident database and only a quarter of these were the fault of the rider.

The ‘lifesaver glance' is the last glance over his or her shoulder that a motorcyclist makes before carrying out a manoeuvre, especially a turn in the road as a last check that their path is clear. Only 40.3% (n ¼ 144) of the respondents claimed to always use this but a further 43.1% (n ¼ 144) claim to frequently use it.

The motorcycle accident database has shown that the most common cause of single vehicle accidents is a result of riders misjudging the appropriate speed to negotiate a bend in the road. A question was therefore included on the questionnaire asking the respondents how often, if ever, they have misjudged the speed required to negotiate a bend. Even though 69% of respondents appeared to be very experienced, they admitted to occasionally miscalculating bends. There were over 200 accidents on the motorcycle accident database that were a direct result of a rider losing control on a bend and, although the reasons for losing control were often unknown or not recorded, it is known that inappropriate speed was the main cause of 27.5% of the accidents.

4.7.7 Views and opinions regarding causes of motorcycle accidents

A total of 141 respondents to the questionnaire listed what they considered to be the three main causes of motorcycle accidents. Altogether 361 responses led to a list of 35 causes. These causes were easily broken down into five broad categories that focussed on poor observation and inattention, environmental concerns, inexperience, risk taking and poor training. A direct comparison of perceived and actual causes of motorcycle accidents (as revealed by the accident database) can therefore be shown in Figure 13. Figures for the ‘actual' number of causes use a total number of 2,155 causes entered into the database at the time of analysis.

Poor observation and/or inattention were the most common cause of motorcycle accidents given by the respondents to the questionnaire as well as being the most common actual cause as shown on the motorcycle accident database. The questionnaire respondents clearly stated that it was not poor observation on the part of the motorcyclist but specifically on the part of other road users, with a massive 43% (n ¼ 141) of the respondents saying that one of the main causes of motorcycle accidents was ‘other road users failing to see riders'.

An examination of who is actually responsible for motorcycle accidents involving poor observation or inattention reveals that the rider is only responsible for 4.27% (n ¼ 2,155) of the total number of causes given on the database, whereas another road user is responsible for almost a quarter (23.4%, n ¼ 2,155).

The specific behaviours of other drivers given by the respondents included motorists' inattention, motorists changing lanes without looking properly, and motorists being distracted while in their vehicles by passengers, mobile phones, etc. ‘No continuity of observation' was the biggest cause of accidents in the database; that reason alone being the cause of over a third of all motorcycle accidents. Of these, the riders were only at fault in 10.9% of the accidents, whereas the other road users accounted for 77.1%.

Environmental problems were a major concern for the questionnaire respondents and poor road surfaces were thought to be the biggest threat to the riders, 32.6% (n ¼ 141) of the respondents claiming that poorly maintained roads were one of the main causes of motorcycle accidents. Slippery road surfaces were also mentioned by 17% (n ¼ 141) of the respondents. Slippery road surfaces could be a result of not only wet or icy weather but also spillages on the road, such as oil and diesel. Of the causes listed on the motorcycle accident database, 14.7% (n ¼ 2,155) involved such environmental concerns.

The questionnaire respondents gave inexperience as the third major cause of motorcycling accidents and, with the exception of only one response, it was the rider who was seen to be at fault.

Of the causes listed on the motorcycle accident database, 22.7% (n ¼ 2,155) of the causes listed could be attributed to inexperience. There was, however, more of a split in terms of responsibility, with 7.51% (n ¼ 2,155) of all accident causes listed being the fault of the rider's inexperience and a further 5.1% (n ¼ 2,155) being the fault of the other road user's inexperience.

The fourth major cause of accidents was, perhaps surprisingly, seen by the questionnaire respondents as being the fault of the motorcyclists as opposed to other road users, and concerned accidents that were a result of deliberate risk taking by the riders. 14.7% of the responses (n ¼ 361) referred to causes that were a result of deliberate risk taking, with just under a tenth of the respondents (9.9%, n ¼ 141) simply stating that it was riders taking unnecessary risks that caused motorcycle accidents. A quarter of all respondents (25.5%, n ¼ 141) said that riding bikes ‘too fast for conditions' was a major cause of accidents. The emphasis was placed on riding too fast for conditions and not necessarily breaking the speed limit.

Risk taking was found to account for 20.8% of the causes listed on the motorcycle accident database, and of these causes 12.9% were a result of the motorcyclist taking risks and 4.69% were a result of other road users taking risks. Sub-categories included within this category included all actions that could be seen as putting the rider or other road users at risk, including travelling at inappropriate speed for conditions, driving recklessly, driving while tired or under the influence of alcohol, disobeying road signals, and specific manoeuvres such as overtaking in inappropriate situations or close following.

4.7.8

The remaining causes of accidents mentioned by just 10 of the respondents referred to training issues, four of whom claimed that poor motorcycle training for the riders themselves was to blame, and six of whom said it was poor training for other road users with regard to motorcycling issues.

The remaining causes listed on the motorcycle accident database accounted for only a small number of accidents and referred to mechanical problems (with the bike and traffic signals), which accounted for 2.3% of accidents, and the well-being of the rider, with 0.6% of accidents being a result of the rider being taken ill while riding.

Who is responsible for motorcycle accidents?

Figure 14 shows which category of road user the questionnaire respondents thought were the main cause of motorcycle accidents.

Figure 14: Pie chart showing the category of road users the questionnaire respondents thought were most likely to cause a motorcycle accident

Other 5%

Any road user

Motorcyclist
14%

78%

All but one of the respondents answered this question and, as can be seen in Figure 14, just under 80% of the respondents thought that car drivers were primarily to blame for accidents involving motorcycles. Cyclists, large commercial vehicles, pedestrians and animals in the road were each only mentioned once.

The motorcycle accident database recorded who was to blame for the 1,790 accidents, splitting the accidents into five main categories. Rider 1 was the motorcyclist involved in the accident. Driver 2 was the second vehicle involved in the accident. The third category was where a pedestrian was to blame. The remaining two categories were joint/combined blame and other/unknown. The split can be seen in Table 6.

Table 6 shows an even split of responsibility for accidents with driver 2 and rider 1 to blame in around 40% of the accidents, respectively. However, just under a third of accidents suffered by motorcyclists did not involve a second vehicle and if these are removed from the figures a slightly different picture emerges, as can be seen in Table 7.

Driver 2 is now clearly shown as responsible for over 50% of accidents involving two or more vehicles.