Transportation Research Part F 45 (2017) 157–168

Contents lists available at ScienceDirect

Transportation Research Part F

journal homepage: www.elsevier.com/locate/trf

Contributing factors to motorcycle injury crashes in Victoria,

Australia
T. Allen a,⇑, S. Newstead a, M.G. Lenné a, R. McClure b, P. Hillard a, M. Symmons a, L. Day a
a
Monash University Accident Research Centre, Monash University, Australia
b
Harvard Injury Control Research Center, Harvard School of Public Health, United States

a r t i c l e i n f o a b s t r a c t

Article history: Introduction: The increased popularity of powered two wheelers (PTWs) in Australia, com-
Received 17 December 2015 bined with their vulnerability in the event of a crash, necessitates new strategies to prevent
Received in revised form 23 July 2016 serious injury crashes. The purpose of this study was to use case-series data collected from
Accepted 14 November 2016
a recent motorcycle case-control study to analyse contributing factors to crashes using a
Available online 7 January 2017
safe systems approach.
Methods: A total of 235 injured riders were recruited and completed a questionnaire-based
Keywords:
interview, each followed by a detailed inspection of the case motorcycle and crash site by a
Powered two wheeler
Injury
trained crash investigator. Primary and secondary contributors to the crash were judged
Safe system based on all available information sources. Analysis of the most frequent contributing fac-
Accident tors included separation of cases into single and multi-vehicle crashes. A stepwise logistic
Investigation regression was used to test for factors associated with human error for multi-vehicle
LBFTS crashes.
Results: Two thirds of crashes investigated involved another vehicle or road user(s). For
multi-vehicle crashes the most common crash scenario involved another vehicle failing
to give way to the rider, and the primary contributor was a perception failure or traffic scan
error on the part of the other road user. A number of secondary factors were found to be
significantly associated with human error type (other road user or rider error), including
rider age, traffic density, inappropriate speed of the PTW, and a road design issue. For single
vehicle crashes, the most common primary contributor was a misjudgement or control
error on the part of the rider, with inappropriate speed as the most frequent secondary
contributor.
Conclusions: Despite the complexity of factors involved in PTW crashes resulting in injury,
a number of significant associations exist between road users as the primary contributing
factor (rider or other road user) and secondary factors, including rider age, traffic density,
speed and road design issues.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction

Motorcycles and scooters, jointly known as Powered Two-Wheelers (PTWs), have shown a recent surge in popularity in
Australia. In the state of Victoria, the number of registered motorcycles increased by 33% between 2007 and 2012 (ABS,
2012), which was more than 3 times the increase in passenger vehicle registrations over the same period. This increase likely

⇑ Corresponding author at: Accident Research Centre, Monash University, Clayton, Victoria 3800, Australia.
E-mail address: [email protected] (T. Allen).

http://dx.doi.org/10.1016/j.trf.2016.11.003
1369-8478/Ó 2016 Elsevier Ltd. All rights reserved.
158 T. Allen et al. / Transportation Research Part F 45 (2017) 157–168

reflects the mobility benefits offered by PTWs in traffic-congested and parking-restricted urban areas, as well as the popu-
larity of motorcycles for recreational use.
The challenge of increased PTW popularity facing road safety stakeholders is the over-representation of motorcyclists in
road trauma statistics, due at least in part to their vulnerability in the event of a crash. While there has been a decline in road
fatalities and serious injuries in Australia over recent years, motorcyclists have represented an increasing proportion of those
seriously injured or killed. Less than 1% of all vehicle kilometres travelled are by motorcycle or scooter (ABS, 2012), yet 18% of
those seriously injured on Victoria roads in 2013–14 were motorcyclists (including pillions), which has increased from 11%
in 2003–4. Similarly, in 2013–14 about 14% of all fatally injured road users in Victoria were motorcyclists (TAC, 2015).
The most recent comprehensive study investigating serious motorcycle crashes in Australia was conducted almost
20 years ago. Since that time, in addition to strong growth in motorcycle numbers (a doubling of registered motorcycles
between 1996 and 2012), there have been significant changes to road infrastructure, PTW characteristics and types, motor-
cycle licensing rules, and characteristics of the rider population. These changes include the introduction of new barrier types
on some arterial routes (including flexible barriers, and W beam under-run protection), increased popularity of cruiser style
motorcycles, a greater proportion of motorcycles fitted with anti-lock brakes and other assistive technologies, the introduc-
tion of power-to-weight restrictions for novice riders, and an increase in the average age of the active rider population (Allen
et al., 2016). Updated information on factors that contribute to motorcycle injury crashes is needed in order to continue to
mount progressive countermeasures.
Another change to occur in the two decades since the last comprehensive in-depth motorcycle crash study is the adoption
of a safe systems approach to road safety. This viewpoint recognises the need to understand the interaction between the key
elements of the road system (road users, the road environment, and vehicles and travel speeds), and aims to foster a system
that makes allowance for human error to minimise risk of serious injury or death (Bambach & Mitchell, 2015). In the context
of motorcycle safety, this approach can be applied to data obtained from real-world crashes to identify potential
countermeasures.
In-depth crash investigation methods aim to determine those factors that have likely contributed to a crash and injury
outcomes, using data collected from the vehicle(s), the crash site, the road users involved, and police or traffic incident
reports where available. Previous studies have reported that human error (both on the part of the rider and other road users)
is the most common primary contributor to most PTW crashes (ACEM, 2004; Haworth, Smith, Brumen, & Pronk, 1997). In the
safe systems context, this indicates a failure of other elements of the road system to accommodate for human error. Iden-
tifying factors that are commonly associated with human error offers potential to find modifiable elements of the road sys-
tem to prevent future serious injury motorcycle crashes.
One strategy for identifying common factors associated with crashes involving PTWs is to separate multi-vehicle crashes
from single vehicle crashes, since the primary contributing factors has been shown to vary between these two crash types
(ACEM, 2004; Haworth et al., 1997). An earlier study conducted in the same geographic region targeted for the current study
(Haworth et al., 1997) found that excessive speed for the conditions contributed to a greater proportion of single vehicle
crashes (35%) than multi-vehicle crashes (17%), and ineffective braking was a more much common contributor to single
vehicle crashes (Haworth et al., 1997). While a greater proportion of multi-vehicle crashes were judged as involving no rider
contribution, inappropriate rider positioning or a failure to respond were more prevalent in the multi-vehicle crashes com-
pared to single vehicle crashes. A mechanical fault was judged as a more prevalent contributor to single vehicle crashes com-
pared to multi-vehicle crashes. In a European based motorcycle crash investigation study (ACEM, 2004), traffic-scanning
related errors were much more common for other vehicle drivers, while a decision failure was a more common human error
for motorcyclists.
Therefore the purpose of the current study was to identify the factors contributing to motorcycle serious injury crashes,
including separate investigation of multi-vehicle and single vehicle crashes and testing for associations between multiple
factors.

2. Material and methods

2.1. Eligibility & recruitment

The population recruited for this study were riders of powered two wheelers (PTWs) who had recently been injured in a
crash, and admitted to one of 14 study hospitals (9 metropolitan, 5 regional) in Victoria, Australia. Recruitment of partici-
pants occurred between January 2012 and August 2014, as part of a larger case-control study (Day et al., 2013). Eligibility
criteria included that the crash occurred on a public road in Victoria between the hours of 6 am and midnight, and that
the rider was aged 18 years or over. The time-based selection criteria was chosen for practical and safety reasons related
to recruitment of control riders for the larger case-control study (a researcher had to be physically present in the crash loca-
tion around the crash time to observe control rider participants). This criteria resulting in an exclusion 2% of all injury
crashes in Victoria that occurred between midnight and 6 am. All recruitment and experimental procedures were approved
by Monash University Human Research Ethics Committee, and ethics committees representing each of the study hospitals.
Eligible riders were approached by a research nurse while in hospital and invited to participate. Where possible, riders
discharged from hospital before being approached were contacted by mail for a follow-up telephone interview. Riders that
 T. Allen et al. / Transportation Research Part F 45 (2017) 157–168 159

agreed to participate completed an interview-based questionnaire. The questionnaire was structured into 5 sections: Crash-
related, Trip-related (eg. Purpose, time of day, duration, breaks), Motorcycle and clothing (e.g. PTW type, age, condition, pro-
tective clothing), Demographics (including licence and experience items), and Medical-related questions (e.g. Pre-existing
conditions, recent sleep patterns). In the crash-related section, riders were asked to describe the events leading up to the
crash in their own words, as well as what specific factors contributed to the crash from their perspective. Riders were also
asked for their judgement of the relative contributions of factors to the crash, including the road environment, other road
user(s), and themselves. The crash investigation took place following the completion of the questionnaire, which included
inspection of the case motorcycle where possible (73% of cases) and inspection of the crash site by a trained crash investi-
gator who was also an experienced motorcyclist.

2.2. Assigning contributing factors to crash event

The crash investigator conducted systematic investigations of the crash scenes and motorcycles according to the study
protocol using structured assessment forms. He then used these assessments, the rider interviews, and police reports where
available to identify and listed all possible contributing factors. The 1st author collated information from each case and
coded them according to the ACEM (2004) definitions, in consultation with the crash investigator where there was any
uncertainty. If an identified contributing factor did not fall under any of the ACEM categories, a new contributing factor def-
inition was created (see Table 1 for list and the Glossary for full definitions). Additional factors included ‘speed inappropri-
ate’, defined as a travel speed (of the rider or the other road user) immediately before the crash judged as not appropriate for
the conditions (see Glossary for full definition). For each case, a single primary contributing factor was determined by the
first author in consultation with the crash investigator. Where required, one or more secondary factors were also determined
as either definite or probable contributors to the crash.

3. Results

3.1. Case recruitment

During the period January 2012 to August 2014, 543 eligible riders were identified from the study hospitals in metropoli-
tan Melbourne and regional Victoria (Australia). We were able to approach 291 of these injured riders, and 235 (81%) agreed
to participate in the study. The questionnaire interview was completed in person while in hospital by 94% of the participants,
and by the remaining 6% by telephone after discharge. Over 90% of recruited riders sustained at least one injury of moderate
severity or higher as defined by the Abbreviated Injury Scale (AIS).

3.2. Rider characteristics

The age of case riders ranged from 18 to 67 years, with median age of 41 years. This compares to a median age of 42 years
for a population of control riders observed passing the crash sites as part of the larger case-control study (Allen et al., 2016).
The vast majority (93%) of recruited riders were male (see Table 2), which is the same proportion as the recruited control
rider population. There was no significant difference in age of those riders involved in a single vehicle crash (mean
41 yrs) compared with those involved in a multi-vehicle crash (mean 40 yrs).

Table 1
List of contributing factors (primary and secondary) included in analysis. Asterisk (⁄) indicates those
that employed definitions of ACEM (2004).

Rider/other road user factors Environment factors

– Perception failure⁄ – Road design issue⁄
– Traffic scan error⁄ – Road maintenance issue⁄
– Decision/Reaction failure⁄ – View obstruction (mobile/stationary)⁄
– Misjudgement/Control error – Natural light conditions (dusk, dark, glare)
– Speed inappropriate (for conditions) – Temporary traffic hazard⁄
– Speed excessive (cf. speed limit) – Adverse weather⁄
– Position of vehicle unsafe or high-risk – Shoulder surface
– Conflicting behaviours or pre-occupied⁄ – Slippery road due to weather
– Faulty traffic strategy⁄ – Slippery road due to loose material
– Unsafe act or high risk behaviour⁄ – Slippery road due to grease/oil
– Alcohol or drug involvement⁄ – Roadside environment factor, incl. animal⁄
– Physical or physiological failure⁄ – Tram tracks
– Experience/exposure to similar situations⁄ – Traffic control problem⁄
– 3rd or 4th vehicle involvement⁄
Vehicle factors (case PTW/other vehicle)
– vehicle failure – incl. mechanical⁄
– pre-existing mechanical or other issue⁄
160 T. Allen et al. / Transportation Research Part F 45 (2017) 157–168

Table 2
General characteristics of case riders.

Characteristic n %
Age
18–24 43 18
25–34 45 19
35–44 56 24
45–54 52 22
55–64 31 13
65+ 8 3
Gender
Male 218 93
Female 17 7
Licence status (motorcycle)
Full 164 70
Probationary 29 12
Learner 33 14
Unlicenced 7 3
Riding experience (yrs full licence)
More than 20 yrs 54 23
10–20 yrs 36 15
5–9 yrs 34 14
1–4 yrs 49 21
Less than 1 yr 55 23
Not known 7 3

In 70% of cases the rider reported holding a current full motorcycle licence. One quarter of riders reported holding either a
probationary licence (12%) or learner permit (14%). A small proportion of riders (3%) reported holding no current motorcycle
licence (see Table 2). About one quarter of case riders reported holding a full motorcycle licence for less than 1 year, while
another quarter reported having their full licence for more than 20 years.

3.3. Trip characteristics

One third of riders reported they were commuting at the time of the crash, and another third reported riding riding for
recreational purposes. The remaining third reported their trip purpose as either riding to/from a social venue or sporting
event (13%), shopping or similar errand (10%), work-related travel (5%), or doing a test ride or service-related ride (5%).
Over two-thirds of crashes occurred in urban areas (see Table 3), and of these crashes 82% involved another vehicle. For
crashes in rural areas, two-thirds were single-vehicle crashes.
Crashes occurred relatively evenly across all days of the week. The highest number of crashes occurred on Wednesdays
(n = 42, 18%) and Thursdays (n = 37, 16%). Weekend crashes (Saturday & Sunday) comprised 29% of cases. About half of cases
involved a crash occurring between Noon and 6 pm (see Table 3).

Table 3
General trip characteristics for case riders.

Characteristic N %
Trip purpose
Commuting 75 32
Recreation 80 34
(Solo) (47) (20)
(Group) (33) (14)
Social 30 13
Errand 24 10
Work related 12 5
Test-ride/Service related 12 5
Not known 2 1
Crash location
Urban 161 69
Rural 74 31
Day of week
Weekdays 167 71
Weekends 68 29
Time of day
6 am to noon 73 31
Noon to 6 pm 122 52
6 pm to midnight 40 17
 T. Allen et al. / Transportation Research Part F 45 (2017) 157–168 161

3.4. Contributing factors to crash outcome

Human error was judged to be the primary contributor to the crash in 94% of cases, and this was split almost equally
between the case rider (48%) and the other road user (46%). For multi-vehicle crashes, which represented two-thirds of cases,
the most common primary contributor was a traffic scan error on the part of the other road user (n = 86, 54% of multi-vehicle
crashes, see Fig. 1). For single vehicle crashes, the most common primary contributor was a rider misjudgement or control
error (n = 39, 51% of single vehicle crashes, see Fig. 1). It is important to note that in over 90% of cases one or more secondary
factors were judged to either definitely or probably contribute to the crash (mean 3 ± 2 secondary factors). We therefore
wanted to test if any of these secondary factors were significantly associated with road user error. This was possible for
multi-vehicle crashes (see Section 3.5), while for single vehicle crashes the smaller sample size (77 cases) limited our ability
to perform the same analysis.
The road environment was judged as the primary contributor in 5% of crashes, all of which were single vehicle cases. The
most common road environment factors were an animal (n = 5) and a road maintenance issue (n = 4). As a secondary con-
tributor, road environment was noted as a definite or probable factor in 78% of crashes, most of which involved either a road
design issue or road maintenance issue.
A mechanical issue with the motorcycle was judged as the primary contributor in <1% of crashes, and was judged as a
secondary contributor in 4% (n = 9) of crashes. The most common mechanical issue involved the recent fitting of new tyres,
which have lower grip levels during early use.

3.5. Multi-vehicle crashes: secondary factors associated with human error

For multi-vehicle crashes, the primary crash contributor was judged almost exclusively as human error (i.e. either other
road user error or case rider error). Based on sufficient counts of human error amongst both road user types, a stepwise logis-
tic regression was carried out to test if any secondary factors were associated with road user error as the primary contributor
(i.e. case rider error or other road user error). In addition to contributing factors determined from the crash investigation, a

Fig. 1. Crash investigation-based judgements of primary & secondary contributing factors to crashes, separated by number of vehicles involved (single
vehicle or multi-vehicle). Note that for most (91%) cases 2 or more secondary contributing factors were identified (average 3 secondary factors). See
Glossary for list of abbreviations.
162 T. Allen et al. / Transportation Research Part F 45 (2017) 157–168

Table 4
List of additional factors included in the analysis of secondary factors associated with road user error (logistic regression).

Case rider factors

– Age
– Licence status: holds a current full motorcycle licence (self-report)
– Experience: total riding experience of 10,000 km or more (self-report)
– Riding frequency: ride 200 km/week or more over the past 12 months (self-report)
Environment factors
– Traffic density immediately before the crash (light, medium or heavy, self-report)
– Presence of any road surface issues, including wet, bumpy, loose material, greasy, cracked or damaged (crash investigation)
Other factors
– Case rider wearing bright coloured helmet or upper body clothing (self-report)
– Case rider wearing any fluorescent or reflective clothing (self-report)

number of additional factors were included in the logistic regression. These included factors were chosen based on rider,
environment, or other factors previously reported in the literature as being associated with risk of crash or injury amongst
PTW riders (ACEM, 2004; Haworth et al., 1997; Wells et al., 2004). These included factors are listed in Table 4. Of those fac-
tors, four were found to be significantly associated with human error (rider or other road user) – age, traffic density, rider
inappropriate speed and a road design issue. For the logistic regression, the classification table showed 75% of human error
judgements were correctly classified by the model, while the Hosmer-Lemeshow test indicated a significance of 0.179. These
suggest a good level of fit for the applied model.

3.5.1. Rider age

Rider age showed a significant negative association with rider error as the primary contributor for multi-vehicle crashes.
The results of the logistic regression suggested that for multi-vehicle PTW crashes, for every 10 year increase in rider age
there was a 38% decrease in rider error as the judged primary contributing factor (Tables 5 and 6).

3.5.2. Traffic density

A significant association was found between traffic density and road user error as the primary contributor. That is, light
traffic density was associated with other road user error, whereas heavy traffic density was associated with rider error.
(Tables 5 and 6).

3.5.3. Inappropriate speed

The determination of inappropriate rider speed as secondary crash contributor was significantly associated with other
forms of rider error in multi-vehicle crashes. (Tables 5 and 6).

3.5.4. Road design issue

A road design issue as a secondary crash contributor was significantly associated with rider error as the primary crash
contributor in multi-vehicle crashes. (Tables 5 and 6).

4. Discussion

Recent increases in the popularity of motorcycles, combined with significant changes to the road environment and licens-
ing rules in Australia necessitate up-to-date research to understand key crash contributing factors, so that effective counter-
measures can be implemented. The purpose of this paper was to employ a safe systems approach to investigate contributing
factors in PTW injury crashes and their associated factors, using crash investigation and rider self-report data. Cases were
collected from 235 injury crashes that occurred on public roads in Victoria, Australia between 2012 and 2014.

4.1. Contributing factors to multi-vehicle crashes

Multi-vehicle crashes represented two-thirds of cases investigated, and for 54% of these cases a perception failure or traf-
fic scan error on the part of the other road user was judged as the primary contributing factor. The spread of human error (as
the primary factor) amongst both other road users and riders allowed us to examine whether any secondary or other factors
were associated with road user error type (case rider or other road user). A significant association was found between 4 fac-
tors and road user error as the primary contributor: rider age, traffic density, rider inappropriate speed, and a road design
issue.
The negative association between rider age and rider error suggests that older riders were less likely to be the primary
contributor to crashes involving another vehicle. Given our rider experience measures (licence years and weekly riding dis-
tance) showed no significant association, this finding suggests rider age was a more important protective factor than rider
 T. Allen et al. / Transportation Research Part F 45 (2017) 157–168 163

experience. A previous case-control study in Victoria (Haworth et al., 1997) showed reduced risk of serious injury crash risk
for those aged over 35 years (compared to those aged under 25 years), but in contrast to our study they showed reduced risk
for riders with more years of on-road riding experience. However the age of the rider population appears to have increased
dramatically since that study, with a mean rider age of 40 years in the current study compared to 30 years in the earlier
study. A more recent case-control study conducted in Europe showed riders aged 41–55 years had reduced risk of being
in an accident (ACEM, 2004). The same study also showed riders with more experience were less likely to be the primary
contributor to the crash. Therefore our findings with respect to a protective effect of increasing age are similar to those of
comparable studies, but in contrast we did not find protective effects with respect to experience measures. It is possible that
a protective effect of rider experience may have been found with more sensitive experience measures or with greater case
numbers. Alternatively, given most participants in the current study also held a full driver licence, there may be a positive
effect of age on hazard perception skills in traffic through experience driving all forms of motor vehicles (not just motorcy-
cles). It might also be true that the type of riding most common for older riders now is different to that for the older riders in
the earlier studies; for example, older riders now might be more likely to ride recreationally and have a different pattern of
risk. More research is needed to understand what differences might exist between younger and older riders (independent of
riding experience) that might influence their crash risk, including factors such as hazard perception, type of riding, and so on.
The results suggested that as density of traffic increased riders were more likely to be the primary contributor to a crash
with another vehicle, and conversely that for crashes occurring amongst lower traffic density the other road user was more
likely to be the primary contributor. The latter finding is similar to that reported by a European case-control study (ACEM,
2004), which found that over half of crashes occurred when traffic density was light. The authors concluded ‘‘These findings
indicate that PTW accidents occur more frequently in light traffic conditions, suggesting that other road users may not be
expecting the presence of a PTW” (p. 98). In both studies a large number of crashes involved another road user failing to give
way while turning either into the path or across the path of the rider. Our own observations of motorcycles passing crash
sites investigated here showed that motorcyclists were almost three times more likely to have larger leading and trailing
time gaps (4–6 s) compared to other vehicle types travelling in the same direction (Allen et al., 2016). If a driver is perform-
ing a turn across a stream of opposing lanes of traffic at an uncontrolled intersection, it seems reasonable that a traffic scan
error is more likely when traffic density is very low and a motorcyclist (or other vehicle type) is the only approaching vehicle.
Therefore in addition to conspicuity factors contributing to multi-vehicle PTW crashes, motorcyclists may have a greater
exposure (per km travelled) to situations where a traffic scan error from another driver is more likely. It is also possible that
a motorcycle rider not otherwise ‘‘encumbered” by surrounding traffic might be riding at a higher travel speed, reducing the
time available for a driver to see and respond to their approach. This information potentially provides a new level of under-
standing to one of the most common motorcycle crash types, and may lead to new educational or other safety strategies for
riders and other road users. In addition to testing assumptions around increased driver perceptual errors for motorcycles
travelling alone in traffic, more research is needed to determine if particular riding strategies in higher density traffic con-
ditions could reduce rider exposure to common crash scenarios involving another driver failing to give way.
For conditions of high traffic density, the increased likelihood of rider error may be related to unique strategies used by
riders to negotiate heavy traffic due to their smaller size (e.g. lane filtering), or the greater level of difficulty in avoiding an
unexpected event on a PTW due to their lower stability. In higher density traffic riders employ strategies such as lane filter-
ing to move ahead of other vehicles and remove themselves from greater potential of collisions in a busier environment.
These manoeuvres in close proximity to other vehicles requires greater perception and skill on the part of the rider. They
also require other drivers to either see the rider approaching, or for the rider to predict driver behaviours, or that the traffic
will simply continue in the same configuration while the rider negotiates passage. If any of these assumptions are not rea-
lised an altercation occurs. A recent simulator-based study reported that higher traffic flow density can lead to traffic conflict
for PTWs due to a greater probability of sudden braking (Nguyen, Hanoka, & Kawasaki, 2014).
The association between inappropriate speed of the motorcyclist and other forms of rider error for multi-vehicle crashes
could be explained by a higher speed reducing the time available to react and avoid a potential crash, such as a vehicle ahead
braking unexpectedly. Previous research has identified an association between both rider hazard perception and driving vio-
lation behaviours and crash risk (Cheng, Ng, & Lee, 2011). It may simply be that rider error such as a hazard perception failure
manifests itself as an inappropriate choice of travel speed for the prevailing traffic or road conditions. Alternatively, an inap-
propriate travel speed may simply reflect a riders conscious choice of travel speed even if a higher risk is perceived. Further
research is needed to understand the relative contributions of factors that determine rider’s choice of travel speed.
The association between a road design issue and rider error for multi-vehicle crashes may have a number of explanations.
Differences in dynamics between motorcycles and four wheeled vehicles previously mentioned, such as reduced stability
and greater difficulty performing unexpected braking or avoidance manoeuvres are likely contributing factors. It seems plau-
sible that the presence of a design issue or flaw at a given site, such as poor intersection design, fixed obstructions affecting
vision (for the rider or other drivers), poor road signage, narrow lanes, or a transitory issue such as loose gravel or oil spill,
will pose a proportionally greater crash and/or injury risk to motorcyclists compared to that for other motor vehicles.

4.2. Contributing factors to single vehicle crashes

Single vehicle crashes represented one-third of cases, and in half of these cases a rider misjudgement or control error
(including braking errors) was judged as the primary contributor. The most common secondary factor associated with rider
164 T. Allen et al. / Transportation Research Part F 45 (2017) 157–168

Table 5
Multi-vehicle crashes (n = 158) – factors identified as significantly associated with road user error type (other road user or case rider).

Primary contributing factor Rider age (Mean ± SD) Traffic density n (%) Rider speed inappr. n (%) Road design issue n (%)
Light Medium Heavy
Other road user error 42 (13) 58 (47) 32 (26) 25 (20) 21 (17) 10 (8)
Rider error 36 (13) 11 (32) 6 (18) 15 (44) 12 (35) 4 (12)
Total 40 (13) 69 38 40 33 14

Table 6
Secondary contributing or associated factors found to be significantly associated with human error type as the primary crash contributor (other road user or
case rider).

Contributing or associated factor Exp (B) p-value 95% CI for Exp (B) lower 95% CI for Exp (B) upper
Age 0.96 0.01 0.93 0.99
Traffic density 6.25 <0.001 2.3 16.9
PTW speed inappropriate 5.53 0.001 2.1 14.7
Road design 4.51 0.045 1.0 19.6

error for single vehicle crashes was a judgement of inappropriate speed (which we acknowledge could itself be defined as a
rider error). The ACEM (2004) findings similarly showed increased impact speed for single vehicle crashes compared to
multi-vehicle crashes. Both findings may at least partly reflect differences in road environment or trip factors between
the two crash types. For example, single vehicle crashes tend to occur in higher speed zones in rural areas where recreational
riding is more common.
The road environment was judged as a primary contributor to crashes in a higher proportion of single vehicle crashes
(14%) when compared to multi-vehicle crashes (0%). Given two-thirds of single vehicle crashes occurred outside of urban
areas, this suggest a greater contribution of road environment factors to crashes occurring in rural areas, including issues
of road design or road maintenance. This finding suggests that a targeted approach to both road infrastructure improvements
and maintenance for rural roads popular amongst recreational riders may be of benefit to reducing single vehicle motorcycle
injury crashes. The larger case-control study (Day et al., 2013) aims to investigate in more detail which aspects of road design
or characteristics are associated with increased (or decreased) crash risk, which will provide valuable information about
effective road design countermeasures specific to motorcyclists.

4.3. Potential countermeasures

Based on our findings, a number of countermeasures for reducing serious injury PTW crashes are worth considering.
While human error was judged as the primary contributing factor in over 90% of cases, the more important and novel finding
was factors we identified as significantly associated with rider or other road user error. These include road design, traffic
density, inappropriate rider speed, and rider age. Employing a safe systems approach it could be argued that these associated
factors are equally if not more important to finding effective strategies to reduce injury crash risk than the primary con-
tributing factor.
The frequently reported contributing factor of perception failure (particularly for the other driver), when combined with
factors associated with human error, point to a number of potential countermeasures. While the lower conspicuity of PTWs
to other road users has been partly attributed to their low prevalence in traffic (Beanland, Lenne, & Underwood, 2014), mod-
ifiable risk factors include hazard perception of drivers (Horswill, Taylor, Newnam, Wetton, & Hill, 2013) and increasing the
physical conspicuity of riders or PTWs (Rößger, Lenné, & Underwood, 2015). Hazard perception of drivers has been associ-
ated with crash risk (Boufous, Ivers, Senserrick, & Stevenson, 2011; Pelz & Krupat, 1974) and recent laboratory-based
research has shown improvements in hazard perception following a training intervention (Horswill et al., 2013). More
research is needed to determine the long-term effectiveness of hazard perception training for real-world driving and its
long-term effectiveness. With respect to physical conspicuity of PTWs, evidence from a New Zealand study reported that
wearing any fluorescent or reflective clothing was associated with a 33% reduced risk, and wearing a white helmet was asso-
ciated with a 24% lower risk of an injury crash, when compared to wearing a black helmet (Wells et al., 2004). There is also
evidence of improved conspicuity of PTWs using modulating headlights, (Shaheed et al., 2012). Observation of control riders
in the present study (as part of the linked case-control study) found that over half of passing riders were wearing dark cloth-
ing with no fluorescent or reflective surfaces (Allen et al., 2016) suggesting there is scope to improve physical conspicuity of
riders. Since the completion of this study, changes to local regulations now require Learner riders to wear high visibility
vests. This may provide an opportunity to test whether such an intervention can reduce risk of multi-vehicle crashes involv-
ing another vehicle failing to give way. The larger case-control study linked to this case study will test whether bright or
reflecting clothing is associated with reduced crash risk (Day et al., 2013).
 T. Allen et al. / Transportation Research Part F 45 (2017) 157–168 165

The association found between a road design issue and rider error suggests motorcyclists are more at risk to sub-optimal
road design factors. Arguably investments in road design improvements should target vulnerable road users including
motorcyclists (i.e. those over-represented in injury statistics), particularly at sites where their exposure levels or injury rates
are highest, although invariably making a section of road infrastructure safer for one road user type will increase the level of
safety for other road user types, though possibly not to the same extent. The most common road design issues identified here
for multi-vehicle crashes were poor intersection design (including visibility factors), fixed obstructions (such as vegetation),
poorly signed roads (including advisory speeds) and reduced lane widths. Road maintenance factors were also identified as a
common secondary (and to a lesser extent primary) contributor for single vehicle PTW crashes. In the rider questionnaire,
the most common road maintenance issues identified included poor road surface condition, and loose material on the road.
There is evidence that recent ‘black spot’ treatments aimed at reducing motorcycle crashes have been effective in Australia
(Meuleners, Hendrie, & Fraser, 2011; Scully, Newstead, & Corben, 2008). A similar targeting approach could potentially apply
to maintenance of roads popular amongst motorcyclists, but we are not aware that this approach has been trialled or
evaluated.
Inappropriate rider travel speed was frequently judged as secondary contributing factor for both single and multi-vehicle
crashes. Under a safe systems approach speed-related interventions may therefore be effective at reducing injury crash rates.
This may include reviewing whether speed limits on popular motorcycle routes are appropriate for existing road infrastruc-
ture (including existing run off or barrier types), or whether the infrastructure needs to be upgraded to suit the current speed
limit. A proportion of cases judged as inappropriate speed also involved excessive speed (above the posted limit). Radar data
from the control arm of this study showed that a significantly greater proportion of motorcyclists were exceeding the posted
speed limit compared to other vehicle types (Allen et al., 2016). If excessive speed is determined to be a risk factor as part of
the larger case-control study (Day et al., 2013), improving rider compliance to speed limits (through strategies such as speed
enforcement) would be expected to reduce injury crash rates.
Increased use of vehicle crash avoidance technologies for both passenger cars and motorcycles is another potential coun-
termeasure for accommodating for human error by both riders and other road users. For passenger cars, autonomous emer-
gency braking (AEB) has shown to be the most promising recent technology (Fildes et al., 2015). However, its reported
effectiveness is so far restricted to slow speed rear-end crashes, which was not a common crash type seen in the cases inves-
tigated. Therefore investigating effectiveness of AEB in common crash scenarios involving a PTW, such as those where the
driver has failed to give way to an approaching motorcycle, would be valuable. Given the inherent difficulties in avoiding
a crash for PTW riders when compared to cars, the equivalent AEB technology fitted to motorcycles could provide significant
safety benefits in future. While motorcycle autonomous emergency braking (MAEB) is currently in the development stage
(Giovannini, Savino, Pierini, & Baldanzini, 2013; Savino, Giovannini, Baldanzini, Pierini, & Rizzi, 2013; Savino et al., 2014),
motorcycle anti-lock brakes (ABS) is a currently available technology and has been associated with reduced risk of serious
injury crash (Rizzi, Strandroth, Kullgren, Tingvall, & Fildes, 2014, 2015). Therefore strategies to increase the proportion of
ABS-fitted PTWs amongst the active fleet are recommended based on current evidence. ABS has recently been mandated
by the European Commission for all new PTWs (>125 cc) from January 2016.
A number of crash avoidance technologies are currently in development that have the potential to reduce common
motorcycle injury crashes. For passenger cars (and potentially motorcycles) Intersection Movement Assist (IMA) uses
Vehicle-to-Vehicle (V-V) communications to warn a driver when it is not safe to enter an intersection. A limitation of this
and similar V-V based technologies is the requirement for both vehicles to be fitted with the technology for it to be operate.
Therefore if a proven technology of this type was mandated for all new vehicles, we would expect many years before a suf-
ficient proportion of the vehicle fleet have the technology to see significant reductions in multi-vehicle crashes, and in the
meantime risk compensation on the part of drivers assuming the lack of an alert may pose unintended risks for motorcy-
clists. One technology designed specifically for motorcyclists is Advance Rider Assistance Systems or ARAS (SAFERIDER,
2010). The system aims to provide advanced warnings to the rider, including a ‘‘curve warning” if the approach speed is
assessed as inappropriate and an ‘‘intersection support system” that warns the rider if they need to reduce speed to safely
manage an approaching intersection. Both technologies appear well matched to common crash types and contributing fac-
tors identified here. If we assume the technology will be effective at identifying hazards before the rider has perceived them,
an important factor to overcome involves understanding how well technologies like ARAS might be accepted and used by the
motorcycle rider community (Huth & Gelau, 2013), as well as the impact of pricing relative to the cost of the motorcycle.

4.4. Study limitations

A number of limitations of this study need to be considered in light of these findings. With respect to how representative
the case population was of the injured rider population for Victoria, the average age of all eligible riders was 36 years
(n = 543), whereas the mean age of our recruited riders was 40 years (n = 235), showing a small age bias in our recruited
population. We also do not know about those factors not asked about or included as part of the crash investigation or logistic
regression. Therefore it is possible that there may be other contributing factors not identified here.
The availability of evidence for crash investigation was limited for some cases. Inspection of the recruited rider’s motor-
cycle was possible for 73% of cases, due to difficulties locating the motorcycle or gaining permission to inspect. This is a sim-
ilar success rate to that reported by an earlier study in the same region (Haworth et al., 1997). Site inspection often occurred
several days after the crash (median 9 days), which was mainly due to unavoidable delays in approaching injured riders after
166 T. Allen et al. / Transportation Research Part F 45 (2017) 157–168

the crash. However for some cases an earlier inspection may not have necessarily improved the availability of evidence, as
crash-related evidence (e.g. skid marks, debris, scrape marks) can be limited even immediately post-crash. For multi-vehicle
crashes, the lack of information available about other road users is also a limitation. Privacy laws restricting identification
and potential willingness of participation by other parties, particularly when there may be legal issues involved, are barriers
to accessing this information. While this study had indirect access to police traffic incident reports for a small proportion of
crashes, we believe access to incident reports for all crashes would improve the accuracy of judgements on contributing fac-
tors, particularly if it included reports from witnesses or other road users. In a follow-up study we endeavoured to make con-
tact with the other road user for multi-vehicle crashes, without much success – a publication is forthcoming.
A small proportion of cases investigated in this study occurred on unsealed public roads. Data from Victorian Admitted
Episodes Dataset (VAED) on motorcycle riders admitted to hospital suggests that around 30% of all seriously injury PTW
crashes occur ‘‘off-road” (VAED, 2015). The actual number of PTW injury crashes occurring ‘‘off-road” is currently not known,
due to limitations in reporting or accurately defining crashes within hospital or other injury surveillance databases. The con-
tributing factors to motorcycle crashes occurring on unsealed rural roads are likely to be markedly different to the majority
of crashes occurring on sealed roads, which are mostly in built-up urban areas. This suggest that crashes occurring on
unsealed publics roads remain an important consideration if we are to maximise opportunities to reduce injury rates
(Victorian Road Safety Committee, 2012). Future motorcycle safety research will need to investigate modifiable risk factors
for ‘‘off-road” injury crashes before effective countermeasures can be developed for this rider population.

4.5. Conclusions

Crash investigation of 235 cases involving an injured motorcyclist found that human error was the most frequent primary
contributing factor, which was almost equally distributed between the (case) rider and the other road user. However, in a
safe systems context the role of secondary and associated factors in crash outcome must be acknowledged. For multi-
vehicle crashes, while the other road user was the most frequent primary contributor, both rider and road environment fac-
tors featured strongly as secondary contributors. Younger riders, higher traffic density, inappropriate rider speed and road
design issues were all significantly associated with rider error in multi-vehicle crashes. For single vehicle crashes, while a
rider misjudgement or control error was the most frequent primary contributor, road environment factors such as road
maintenance and design issues featured strongly as secondary contributors. Vehicle-related factors were not commonly
identified as primary or secondary contributors to crash outcome. These findings provide a greater understanding of the
interaction between rider, other road user and road environment factors in real-world crashes, with implications for more
effective motorcycle safety countermeasures.

Acknowledgements

The study was funded by the Australian Research Council (LP110100057), VicRoads, the Transport Accident Commission
of Victoria, and the Victorian Government Department of Justice, with in-kind support from Victoria Police and the Victorian
Automobile Chamber of Commerce. Ambulance Victoria provided daily crash notification and location information. The
authors would like to thank the MICIMS project research team: Geoff Rayner, Josie Boyle, Rob Jackel and all hospital-
based research nurses.

References

ABS (2012). Survey of Motor Vehicle Use, 30 June 2012. Australian Bureau of Statistics, Canberra (9208.0).
ACEM (2004). In-depth investigations of accidents involving powered two wheelers. Association of European Motorcycle Manufacturers report.
Allen, T., McClure, R., Newstead, S. T., Lenne, M. G., Hillard, P., Symmons, M., & Day, L. (2016). Exposure factors of Victoria’s active motorcycle fleet related to
serious injury crash risk. Traffic Injury Prevention, 17(8), 870–877.
Bambach, M. R., & Mitchell, R. J. (2015). Safe system approach to reducing serious injury risk in motorcyclist collisions with fixed hazards. Accident; Analysis
and Prevention, 74, 290–296. http://dx.doi.org/10.1016/j.aap.2014.06.002.
Beanland, V., Lenne, M. G., & Underwood, G. (2014). Safety in numbers: Target prevalence affects the detection of vehicles during simulated driving.
Attention, Perception, Psychophysics, 76(3), 805–813. http://dx.doi.org/10.3758/s13414-013-0603-1.
Boufous, S., Ivers, R., Senserrick, T., & Stevenson, M. (2011). Attempts at the practical on-road driving test and the hazard perception test and the risk of
traffic crashes in young drivers. Traffic Injury Prevention, 12(5), 475–482.
Cheng, A. S., Ng, T. C., & Lee, H. C. (2011). A comparison of the hazard perception ability of accident-involved and accident-free motorcycle riders. Accident
Analysis and Prevention, 43(4), 1464–1471.
Day, L., Lenne, M. G., Symmons, M., Hillard, P., Newstead, S., Allen, T., & McClure, R. (2013). Population based case-control study of serious non-fatal
motorcycle crashes. BMC Public Health, 13, 72. http://dx.doi.org/10.1186/1471-2458-13-72.
Fildes, B., Keall, M., Bos, N., Lie, A., Page, Y., Pastor, C., ... Tingvall, C. (2015). Effectiveness of low speed autonomous emergency braking in real-world rear-end
crashes. Accident; Analysis and Prevention, 81, 24–29. http://dx.doi.org/10.1016/j.aap.2015.03.029.
Giovannini, F., Savino, G., Pierini, M., & Baldanzini, N. (2013). Analysis of the minimum swerving distance for the development of a motorcycle autonomous
braking system. Accident; Analysis and Prevention, 59, 170–184. http://dx.doi.org/10.1016/j.aap.2013.05.020.
Haworth, N., Smith, R., Brumen, I., & Pronk, N. (1997). Case-control study of motorcycle crashes (CR 174). Canberra: Federal Office of Road Safety.
Horswill, M. S., Taylor, K., Newnam, S., Wetton, M., & Hill, A. (2013). Even highly experienced drivers benefit from a brief hazard perception training
intervention. Accident; Analysis and Prevention, 52, 100–110. http://dx.doi.org/10.1016/j.aap.2012.12.014.
Huth, V., & Gelau, C. (2013). Predicting the acceptance of advanced rider assistance systems. Accident; Analysis and Prevention, 50, 51–58. http://dx.doi.org/
10.1016/j.aap.2012.03.010.
 T. Allen et al. / Transportation Research Part F 45 (2017) 157–168 167

Meuleners, L., Hendrie, D., & Fraser, M. (2011). An evaluation of the effectiveness of the state black spot programs on motorcycle crashes in Western
Australia. Curtin Monash Accident Research Centre, 2011. http://c-marc.curtin.edu.au/local/docs/16-12-11_Final_DEC_Motorcycle_black_spot_report_
WA.pdf.
Nguyen, L. X., Hanoka, S., & Kawasaki, T. (2014). Traffic conflict assessment for non-lane-based movements of motorcycles under congested conditions.
International Association of Traffic and Safety Sciences, 37, 137–147.
Pelz, D. C., & Krupat, E. (1974). Caution profile and driving record of undergraduate males. Accident Analysis and Prevention, 6, 45–58.
Rizzi, M., Strandroth, J., Kullgren, A., Tingvall, C., & Fildes, B. (2014). Effectiveness of Motorcycle Antilock Braking Systems (ABS) in Reducing Crashes, the
First Cross-National Study. Traffic Injury Prevention. http://dx.doi.org/10.1080/15389588.2014.927575.
Rizzi, M., Strandroth, J., Kullgren, A., Tingvall, C., & Fildes, B. (2015). Effectiveness of Motorcycle Antilock Braking Systems (ABS) in Reducing Crashes, the
First Cross-National Study. Traffic Injury Prevention, 16(2), 177–183. http://dx.doi.org/10.1080/15389588.2014.927575.
Rößger, L., Lenné, M. G., & Underwood, G. (2015). Increasing Road User Conspicuity: Design and Assessment of Interventions to Enhance Rider Safety. Aldershot:
Ashgate.
SAFERIDER (2010). Advanced telematics for enhancing the safety and comfort of motorcycle riders. from <www.saferider-eu.org>.
Savino, G., Giovannini, F., Baldanzini, N., Pierini, M., & Rizzi, M. (2013). Assessing the potential benefits of the motorcycle autonomous emergency braking
using detailed crash reconstructions. Traffic Injury Prevention, 14(Suppl.), S40–49. http://dx.doi.org/10.1080/15389588.2013.803280.
Savino, G., Rizzi, M., Brown, J., Piantini, S., Meredith, L., Albanese, B., ... Fitzharris, M. (2014). Further development of motorcycle autonomous emergency
braking (MAEB), what can in-depth studies tell us? A Multinational Study. Traffic Injury Prevention, 15(Suppl. 1), S165–172. http://dx.doi.org/10.1080/
15389588.2014.926009.
Scully, J., Newstead, S. V., & Corben, B. F. (2008). Evaluation of the Motorcycle Blackspot Program.
Shaheed, M. S., Gkritza, K., & Marshall, D. (2012). Motorcycle conspicuity – what factors have the greatest impact, Center for Transportation Research and
Education. MTC Project 2011–01.
TAC (2015). Motorcycle Crash Data. Transport Accident Commission (Victoria, Australia). 2015.
VAED (2015). Data from Victorian Admitted Episodes Dataset.
Victorian Road Safety Committee, V. R. S. (2012). Inquiry into Motorcycle Safety Parliamentary Paper (Victoria, Australia) (Vol. 197).
Wells, S., Mullin, B., Norton, R., Langley, J., Connor, J., Lay-Yee, R., & Jackson, R. (2004). Motorcycle rider conspicuity and crash related injury: Case-control
study. British Medical Journal, 328(7444), 857. http://dx.doi.org/10.1136/bmj.37984.574757.EE.

Glossary:
Definitions of contributing factors (including abbreviations)

Note: all conclusions/judgements were reached following assembly of all available evidence, including rider self-reported information, inspection of the vehicle and
crash scene, and police reports.

Rider/Other road user factors

Alcohol or drug involvement: Medical or self-reported evidence of excessive alcohol (>0.05 g/dL) or use of illegal drug known to impair driving or riding
near the time of crash.
Conflicting behaviours or pre-occupied (Conflict. behav.): The PTW rider or Other Road User acted in such a way that was unexpected by other road
users, or they were distracted by another activity.
Decision/Reaction failure (Decision/React.): The PTW rider or the Other Road User failed to make the correct decision to avoid the dangerous condition, or
failed to react, based upon his strategy.
Experience/exposure to similar situations: Lack of experience or exposure specific to the vehicle, physical road environment, weather, or traffic condi-
tions present at the time of the incident
Faulty traffic strategy (Faulty Traf. Strat.): PTW rider or Other Road User made a poor decision to perform a manoeuvre or movement in traffic
Misjudgement/Control error (Misjudge/Cont.): The PTW rider or Other Road User misjudged the situation and/or failed to control the vehicle within its
capabilities (including braking and steering actions).
Perception failure (Perception): The PTW rider or the Other Road User failed to detect the dangerous condition based upon the strategy that he was using
to detect dangerous conditions.
Physical or physiological failure: Physical or physiological failure to PTW rider or Other Road User, including fatigue, acute or chronic medical condition
(eg. seizure, cardiac failure).
Position of vehicle unsafe or high-risk (Pos. unsafe/high risk): The road user (rider or driver) positioned their vehicle in a way that was deemed as high
risk of crash. This includes positioning the vehicle where it was more difficult to be seen by other road users, positioning the vehicle too close to other
vehicles, or positioning the vehicle such that negotiating an approaching curve was made more difficult.
Speed inappropriate (Speed Inapp.): The estimated travel speed immediately before the precipitating crash event was judged as not appropriate for the
conditions, including the physical road environment, weather, traffic conditions, or the vehicle.
Speed excessive (Speed Exc.): The estimated travel speed immediately before the precipitating crash event was judged as above the posted speed limit.
Traffic scan error (Traf. Scan Err.): The other vehicle driver or PTW rider did not observe or perceive oncoming traffic or traffic that may have been
entering the roadway from some other direction
Unsafe act or high risk behaviour (Unsafe act): The action of the PTW rider or Other Road User was judged as unsafe or high risk.
3rd or 4th vehicle involvement: The presence or actions of a 3rd or 4th vehicle contributed to the cause of the crash.

Environment factors

Adverse weather (Adv. Weather): A weather event, including strong winds, heavy rain, or fog.
Natural light conditions (Natural Light Cond.): The natural light conditions at the time of the crash reduced visibility for the PTW rider and/or Other Road
User (including dawn, dusk, dark, and sun glare)
Road design issue: A condition which presented a danger to PTW riders based solely upon the design of the road.
Road maintenance issue (Road Maint.): Any road condition that was in poor repair or in need of repair
Roadside environment factor, incl. animal: Unexpected roadside environmental factor in close proximity to PTW rider, including an animal on or near the
road.
Shoulder surface: The road shoulder surface significantly reduced the chance of a rider recovering from an incident, where this part of the road was used
(eg. unsealed shoulder).
Slippery road due to weather: Road surface grip at or near crash site was significantly reduced as a result of recent weather (eg. rain)
168 T. Allen et al. / Transportation Research Part F 45 (2017) 157–168

Slippery road due to loose material: Road surface grip at or near crash site was significantly reduced due to presence of loose material, including gravel
(fine or coarse) and leaf litter.
Slippery road due to grease/oil: Road surface grip at or near crash site was significantly reduced as a result of grease or oil
Temporary traffic hazard: A danger or risk on a carriage-way (excluding a road design or maintenance defect)
Traffic control problem⁄: A fault or issue with a traffic control device affecting its function.
Tram tracks: The presence of tram tracks at or near the crash site.
View obstruction (mobile/stationary): A temporary (mobile) or fixed obstruction that reduced or blocked visibility for the PTW rider and/or the Other
Road User.

Vehicle factors (case PTW/other vehicle)

Pre-existing mechanical or other issue: A pre-existing issue with the PTW or Other Vehicle affecting its safe operation or ability to avoid a crash.
Vehicle failure – incl. mechanical: An acute failure to the PTW or Other Vehicle affecting its safe operation or ability to avoid a crash