Analysis of factors associated with injury severity in crashes involving young New Zealand drivers
Introduction
The continued over-representation of young drivers in road crashes and fatal crash injuries is a global public health concern (see, e.g., Williams, 2003, Twisk and Stacey, 2007, Scott-Parker et al., 2013, Zhu et al., 2013). Across the OECD countries, motor vehicle crashes are the leading cause of death, with approximately 25,000 young people between the age of 15 and 24 killed every year. These 25,000 crash fatalities account for 35% of all deaths in this age group. Across the same countries, young drivers represent 28–30% of all driver fatalities, while constituting only 9–13% of the population (OECD, 2006). Young drivers are a disproportionate danger to other road users as well. U.S. data show that about 30% of the fatalities in crashes involving young drivers are their passengers, 24% are other vehicle occupants and 7% are non-motorists (AAA, 2009). OECD estimates indicate that for each young driver killed, 1.3 other road users are likely to die in the same crash (OECD, 2006). The societal costs of crashes involving young drivers have been estimated at 20 billion Euros across OECD countries in 2004 (OECD, 2006).
An ample body of research has been dedicated to studying the problem of high risk of crash involvement among young drivers, the factors that are associated with risk taking behavior of young drivers and the effectiveness of graduated driver licensing (GDL) programs (e.g., Hedlund and Compton, 2005, Williams and Shults, 2010, Williams et al., 2012, Prato and Kaplan, 2013). Understanding the factors associated with crash injury severity of young drivers is also important because of its potential to help designing well-targeted and restrictive measures related to youth-oriented road safety programs. Notably, only a handful of studies has been dedicated to understanding the underlying factors associated with injury severity in young driver crashes, and these studies have often concentrated on specific aspects rather than on the spectrum of factors possibly related with injury severity. Yannis et al. (2005) and Zambon and Hasselberg (2006) focused on young motorcyclist crashes due to the popularity of two-wheelers among young adults. Specifically, Yannis et al. (2005) analyzed the effect of engine size and young rider's age on crash severity and the at-fault young rider's risk in Greece, while Zambon and Hasselberg (2006) looked at the socioeconomic differences in road traffic injuries of young riders. Gray et al. (2008) investigated the factors underlying injury severity of motor vehicle crashes involving male young drivers from 1991 to 2003 in London and the whole of Great Britain. Factors related to the driver, infrastructure and environmental characteristics were analyzed, although the data did not capture behavioral factors such as the number of passengers, driver distraction, speeding or reckless driving, license type and driving experience (Gray et al., 2008). Paleti et al. (2010) concentrated on the correlation of injury severity with aggressive driving style and low driving experience among young drivers. Orsi et al. (2013) focused on the association of higher injury severity with the presence of young passengers.
The current study extends this line of research by developing a conditional model of crash injury severity (i.e., conditional on the fact that a crash has occurred) and thus exploring the entire spectrum of underlying factors associated with the injury severity of crashes involving young drivers in New Zealand. The analysis was conducted on a comprehensive dataset of 43,008 crashes that involved at least one 15–24 year-old driver in New Zealand between 2002 and 2011 and resulted in any minor, serious or fatal injuries to the young drivers, their accompanying passengers, other vehicle occupants and vulnerable road users. The dataset was compiled from police reports collected by the Land Transport Safety Authority (LTSA) that contained detailed information about young drivers’ characteristics and behavior, other road users’ characteristics and behavior, vehicles’ characteristics, crash location, roadway characteristics, and environmental and temporal conditions. The analysis was conducted by means of discrete choice models. A binary logit model, a generalized ordered logit model and a mixed logit model were estimated and their performance comparison illustrated the latter being the best model. The mixed logit model accounted for heterogeneity and heteroscedasticity in the propensity that determines the probability of injury severity outcomes, as well as correlation between serious and fatal injuries.
The paper is organized as follows. The next section provides a general background for the case study context. Then, the data source and estimation sample are described, and the employed methodology is detailed. Last, model estimation results are presented and discussed, and concluding remarks are offered.
Section snippets
Case study context
New Zealand considered youth-oriented driver policies in the mid-80s to counter unfavorable crash rate trends among young New Zealander drivers 15–24 years old. In 1985, their road crash fatality rate was nearly 48 per 100,000 person years, compared to 40 in Australia, 35 in the U.S., 34 in Canada, and 18 in the U.K. and Sweden (Begg and Stephenson, 2003). In 1986, their road crash fatality proportion reached 39%, while their population proportion was only 18%. In light of this, the New Zealand
Data
The dataset for model estimation was derived from police reports of crashes between 2002 and 2011 where at least one 15–24 year-old driver was involved. LTSA collected the police reports that contained one record for each road user who was involved in a crash and suffered a minor, serious, or fatal injury. An injury was considered minor if no medical treatment was required, serious if medical treatment or a period in hospital was necessary, and fatal if death occurred within 30 days of the
Methodology
The current study developed conditional models of injury severity of crashes involving young drivers (i.e., conditional on the fact that a crash involving a young driver occurred) by applying discrete choice models. Given that the LTSA database coded crash injury severity according to a discrete three-point scale (i.e., minor, serious, fatal) that could be considered either ordinal or nominal, three model forms were considered: a binary logit, a generalized ordered logit, and a mixed logit.
From
Model comparison
Discrete choice models of injury severity conditional on the occurrence of a crash were estimated. For each discrete choice model, the estimation process considered separate models for single-vehicle and two-vehicle crashes after testing whether they performed better than the combined model. Given the log-likelihood from the estimation of two separate models and the log-likelihood from the estimation of one model with all crashes, a likelihood ratio test verified the equality of parameter
Conclusions
The current study proposed discrete choice models to investigate injury severity conditional on the occurrence of crashes involving young drivers in New Zealand. After specification testing, a mixed logit model proved to be the best model for its ability to fit the data better than a generalized ordered logit and a binary logit.
Results show that major concerns are the behavior of young drivers, the presence of passengers and the involvement of vulnerable road users. Additional concerns are the
Acknowledgements
The authors would like to acknowledge the significant contribution of two anonymous reviewers who provided insightful and constructive comments to a previous version of this paper.
References (58)
- et al.
Under-reporting of road crash casualties in France
Accident Analysis and Prevention
(2006) - et al.
Identifying factors that predict persistent driving after drinking, unsafe driving after drinking, and driving after using cannabis among young adults
Accident Analysis and Prevention
(2003) - et al.
Graduated driver licensing: the New Zealand experience
Journal of Safety Research
(2003) - et al.
Risky driving habits and motor vehicle driver injury
Accident Analysis and Prevention
(2005) - et al.
Adolescent, and their parents, attitudes towards graduated driver licensing and subsequent risky driving and crashes in young adulthood
Journal of Safety Research
(2011) - et al.
The situational risks of young drivers: the influence of passengers, time of day and day of week on accident rates
Accident Analysis and Prevention
(1998) - et al.
The effect of the learner license Graduated Driver Licensing components on teen drivers’ crashes
Accident Analysis and Prevention
(2013) Other high-risk factors for young drivers—how graduated licensing does, doesn’t, or could address them
Journal of Safety Research
(2003)Improving graduated licensing systems: a conceptual approach and its implications
Journal of Safety Research
(2007)- et al.
Injury severity analysis of accidents involving young male drivers in Great Britain
Journal of Safety Research
(2008)
Cell phone conversing while driving in New Zealand. Prevalence, risk perception and legislation
Accident Analysis and Prevention
Graduated driver licensing research in 2004 and 2005
Journal of Safety Research
On the use of a Modified Latin Hypercube Sampling (MLHS) method in the estimation of a mixed logit model for vehicle choice
Transportation Research Part B
The crash severity impacts of fixed roadside objects
Journal of Safety Research
The influence of alcohol, age and number of passengers on the night-time risk of driver fatal injury in New Zealand
Accident Analysis and Prevention
Driver-injury severity in single-vehicle crashes in California: a mixed logit analysis of heterogeneity due to age and gender
Accident Analysis and Prevention
A note on modeling pedestrian-injury severity in motor-vehicle crashes with the mixed logit model
Accident Analysis and Prevention
Passenger carriage and car crash injury: a comparison between younger and older drivers
Accident Analysis and Prevention
An evaluation of the New Zealand graduated driver licensing system accident
Accident Analysis and Prevention
Crash involvement during the different phases of the New Zealand Graduated Driver Licensing System (GDLS)
Journal of Safety Research
Highway accident severities and the mixed logit model: an exploratory empirical analysis
Accident Analysis and Prevention
Car crashes: the effect of passenger presence and other factors on driver outcome
Safety Science
Using the U.S. National Household Travel Survey to estimate the impact of passenger characteristics on young drivers’ relative risk of fatal crash involvement
Accident Analysis and Prevention
Examining the influence of aggressive driving behavior on driver injury severity in traffic crashes
Accident Analysis and Prevention
Modeling the behavior of novice young drivers during the first year after licensure
Accident Analysis and Prevention
A longitudinal investigation of psychological and social predictors of traffic convictions among young New Zealand drivers
Transportation Research Part F
The statistical analysis of highway crash-injury severities: a review and assessment of methodological alternatives
Accident Analysis and Prevention
Revisiting the concept of the ‘problem young driver’ within the context of the ‘young driver problem’: who are they?
Accident Analysis and Prevention
Statistical analysis of accident severity on rural freeways
Accident Analysis and Prevention
Cited by (81)
An integrated multi-resolution framework for jointly estimating crash type and crash severity
2024, Analytic Methods in Accident ResearchHow heterogeneity has been examined in transportation safety analysis: A review of latent class modeling applications
2023, Analytic Methods in Accident ResearchAn econometric framework for integrating aggregate and disaggregate level crash analysis
2023, Analytic Methods in Accident ResearchFactors associated with driver injury severity of motor vehicle crashes on sealed and unsealed pavements: Random parameter model with heterogeneity in means and variances
2023, International Journal of Transportation Science and TechnologyCitation Excerpt :In other words, 27.02% of the distribution is less than 0, and the remaining 72.98% is more than 0. This outcome is not in promise with previous studies (Kim et al., 2013; Weiss et al., 2014), and it is in line with (Chiou et al., 2013; Zubaidi et al., 2020). Other variables related to different drivers’ age have been found significant and sustain different types of injury severity probabilities.