Analysis of High Risk and High Severity Groups among Motorcyclists
Monash University Accident Research Centre - Report #77 - 1995
Authors: D. Carr, D. Dyte & M. Cameron
Full report in .pdf format [2.5MB]
Previous research into motorcycle safety has shown that novice riders, intoxicated riders, and riders on road curves have the highest crash risks if exposure is kept constant, and the highest risks of more-severe injuries when a crash occurs. Police-reported casualty crashes (1990-1993) were analysed using univariate and multivariate techniques to estimate the involvement of various factors in the crash risks of these groups of riders compared with the crashes in the comparison groups of motorcyclists.
While the magnitude of the crash risk within these three groups could not be estimated without appropriate exposure data, the results of this study do identify factors that might lead to future road-safety measures. In all three high risk groups, riders at night or during alcohol-times were over-involved in crashes and more likely to be seriously injured compared to control groups of riders. Each high risk group was also over-represented in crashes involving loss of control of the motorcycle. These two factors are suggested as possible areas for further investigation and, possibly, countermeasure development.
Following from Cameron (1992), this report presents analyses of data obtained from Police-reported crashes involving motorcycles in Victoria for the period 1990-1993. The purpose of the analyses was to attempt to identify factors which were over-represented in casualty crashes in three groups of motorcyclists with high risks of crash involvement or high risks of injury. A range of factors which may be related to each group's high risk level was examined.
In his analysis of earlier crash data, Cameron (1992) demonstrated that novice riders, intoxicated riders, and motorcyclists on curves had high frequencies of crash involvement compared to suitable comparison groups (experienced riders, sober riders, and motorcyclists on straight road lengths respectively). Previous research cited in Cameron (1992) has shown that these groups have high risks of casualty-crash involvement given equal road-use exposure. The same high-risk groups were used in the present analyses, and seriously and fatally injured riders were compared with less seriously injured riders in the analysis of factors related to high risk of severe injury.
Data were analysed using both univariate and multivariate statistical techniques. The factors involved in road crashes interact in complex ways, and simple statistical techniques do not allow accurate conclusions to be drawn about the involvement of various factors. Univariate techniques can be used as a guide for potentially useful targets, but multivariate techniques are able to evaluate the contribution of factors disentangled from the contributions of other factors.
The univariate analysis involved the use of contingency tables of target groups and potentially-related environmental, rider, and crash factors. As an aid to interpreting the large number of crosstabulations, data were also presented as odds ratios. An odds ratio represents the rate of involvement as a relative quotient, allowing conclusions about the likelihood of a particular type of rider (compared to the comparison group) being involved in a crash under specific conditions.
The multivariate analysis involved the modelling of all factors simultaneously using a logistic regression procedure. The number of interactions between factors in this approach is potentially substantial, so limitations were placed on the analysis so factors were included only if there was previous research supporting their inclusion, or if they were seen to be historically important, or if they had been shown to have some importance in the univariate analyses.
The analysis was restricted to police-reported casualty crashes and it was not possible to obtain exposure data to estimate the relative importance of these factors to crash risk.
Novice motorcyclists are defined here as riders with a motorcycle learner permit or a probationary licence. As a result of changes to licencing procedures, novice drivers as defined here are not all subject to the zero blood alcohol concentration limit or the maximum engine-capacity limit imposed in general on inexperienced riders.
The results of the univariate analysis suggested that novice riders were over-represented in crashes in rural cities and towns, in high alcohol hours, and on dark streets with street lights. They were also over-represented in crashes if they were female, although females only made up 6.2% of crash involved novice riders.
The multivariate analysis for novice riders confirmed the over-representation of novices in rural city/town crashes, their over-representation in night-time crashes, and confirmed the over-involvement of female novice riders.
This analysis also indicated that novice riders were over-represented in crashes involving a collision with a fixed object and in high speed zone crashes involving a fall without a collision.
Novice riders were under-represented in crashes in high speed zones on curves, and in crashes in high speed zones at high alcohol times or in wet-road conditions. They were also under-represented in the intoxicated group, possibly a result of the zero-BAC requirement which applies to the majority of this group.
These analyses were based on riders with known blood alcohol concentrations (BAC), and an intoxicated rider was defined as one having a BAC in excess of .05 g/100ml.. As only 30% of riders had known BACs, it was not possible to draw firm conclusions about this factor.
The univariate analysis indicated that intoxicated riders were over-represented in crashes on curves, at weekends, at high alcohol times, and in dark conditions. They were over-represented in crashes involving colliding with a fixed object or hitting an object off the carriageway.
The results of the multivariate analysis confirmed the above results, but also indicated that intoxicated riders were over-represented in low speed zones in general and also in high speed zones in crashes on curves. Intoxicated riders were under-represented in crashes if they were also novice riders, although the opposite was suggested for novice riders involved in crashes between 10pm and 6am.
Motorcyclists on Curves
The decision to code a crash as occurring on a curve is made by the police when completing the accident report form. The police attend about two thirds of casualty-crash scenes in Victoria. The rest are reported at a Police station by a participant.
The univariate analysis indicated that motorcyclists on curves were over-represented in rural (not towns) crashes, high speed zones, crashes in summer, weekend crashes, crashes in high alcohol times, night-time crashes, and crashes in dark conditions without street lights. They were also over-represented in crashes involving collision with a fixed object and crashes involving falling from the motorcycle. Riders on curves were over-represented when intoxicated.
The multivariate analysis for motorcyclists on curves produced a very complicated model, but largely confirmed the univariate results. It also indicated that motorcyclists on curves were over-represented for crashes in the dark with no street lights but were under-represented in crashes in the dark with street lights and crashes at dusk. Similarly, motorcyclists on curves were over-represented in crashes at night. Male motorcyclists on curves and motorcyclists on curves high speed zones were both over-represented in crashes in rural (not towns) conditions.
High Severity Group
The univariate analysis showed that high-severity crashes were over-represented in rural (not towns) crashes, crashes in high speed zones, crashes on curves, weekend crashes, crashes during high alcohol hours and at night, and crashes in the dark. They were over-represented in crashes involving colliding with a fixed object or hitting an object off the carriageway, and younger riders were over-represented in high-severity crashes.
The multivariate analysis largely confirmed the univariate analysis, although the over-representation of high-severity crashes in crashes in rural (not towns) locations was seen only in crashes in high speed zones. Intoxicated drivers were over-represented in high-severity crashes on weekends.
Conclusions and Recommendations
The analyses of the accident involvement of novice riders, intoxicated riders, riders on curves, and seriously injured riders show a pattern of factors that may point towards new areas for the development of road safety measures for this group of road users.
All these groups were over-represented in night-time crashes or crashes at alcohol times. Alcohol times are a surrogate measure for alcohol involvement in crashes and are mostly times at night. Crashes in the dark, similarly, are part of this pattern. Alcohol times may reflect either crashes that involve alcohol or, alternatively, crashes that reflect the life-style related exposure of motorcyclists. However it is viewed, it is clear from the data presented in this report that crashes at alcohol times or at night are a serious problem for riders in these groups. Measures that reduce their exposure or the influence of alcohol, if that is the problem, would be expected to be beneficial to riders.
Each identified high risk group was also over-represented in crashes apparently relating to motorcycle loss of control. Intoxicated riders were over-represented in crashes involving curves, hitting fixed objects, and running off the road; motorcyclists on curves were over-involved in crashes in high speed zones and crashes involving hitting objects; and high severity crashes were over-represented in high speed zones, rural crashes, and crashes on curves. These factors seem to represent problems surrounding the loss of control of the motorcycle, and future countermeasures might be appropriately targetted towards this area.