- Email: [email protected]
Pre-licensed driving experience and car crash involvement during the learner and restricted, licence stages of graduated driver licensing: Findings from the New Zealand Drivers Study
Accident Analysis and Prevention 62 (2014) 153–160
Contents lists available at ScienceDirect
Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap
Pre-licensed driving experience and car crash involvement during the learner and restricted, licence stages of graduated driver licensing: Findings from the New Zealand Drivers Study Dorothy J. Begg a,∗ , John D. Langley a , Rebecca L. Brookland a , Shanthi Ameratunga b , Pauline Gulliver b a b
University of Otago, New Zealand University of Auckland, New Zealand
a r t i c l e
i n f o
Article history: Received 8 February 2013 Received in revised form 17 July 2013 Accepted 31 August 2013 Keywords: Young drivers Pre-licensed driving Graduated licensing: Motorcycles Cars Crashes
a b s t r a c t Objective: The aim of this study was to determine whether pre-licence driving experiences, that is driving before beginning the licensing process, increased or decreased crash risk as a car driver, during the learner or the restricted licence stages of the graduated driver licensing system (GDLS). Method: Study participants were 15–24 year old members of the New Zealand Drivers Study (NZDS) – a prospective cohort study of newly licensed car drivers. The interview stages of the NZDS are linked to, the three licensing stages of the GDLS: learner, restricted and full. Baseline demographic (age, ethnicity, residential location, deprivation), personality (impulsivity, sensation seeking, aggression) and, behavioural data, (including pre-licensed driving behaviour), were obtained at the learner licence interview. Data on distance driven and crashes that occurred at the learner licence and restricted licence stages, were reported at the restricted and full licence interviews, respectively. Crash data were also obtained from police traffic crash report files and this was combined with the self-reported crash data. The analysis of the learner licence stage crashes, when only supervised driving is allowed, was based on the participants who had passed the restricted licence test and undertaken the NZDS, restricted licence interview (n = 2358). The analysis of the restricted licence stage crashes, when unsupervised driving is first allowed, was based on those who had passed the full licence test and completed the full licence interview (n = 1428). Results: After controlling for a range of demographic, personality, behavioural variables and distance driven, Poisson regression showed that the only pre-licence driving behaviour that showed a consistent relationship with subsequent crashes was on-road car driving which was associated with an increased risk of being the driver in a car crash during the learner licence period. Conclusion: This research showed that pre-licensed driving did not reduce crash risk among learner or restricted licensed drivers, and in some cases (such as on-road car driving) may have increased risk. Young people should be discouraged from the illegal behaviour of driving a car on-road before licensing. © 2013 Elsevier Ltd. All rights reserved.
1. Introduction The over-representation of young drivers in motor vehicle crashes is a serious public health problem world-wide (Organisation for Economic Co-Operation and Development, 2006). The two main factors that contribute to this elevated crash risk are immaturity and driving inexperience (Huang and Winston, 2011; Williams, 2006; Williams et al., 2012). Immaturity is often
∗ Corresponding author. Tel.: +64 3 479 8509; fax: +64 3 479 8337. E-mail address: [email protected] (D.J. Begg). 0001-4575/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.aap.2013.08.027
addressed by imposing a mandatory minimum driver licensing age, which is typically between 16 and 18 years (Australian Government, 2012; Department for Transport, 2012; European Commission, 2012; Insurance Institute for Highway Safety, 2012). Addressing the problem of driving inexperience is more challenging because to gain experience the learner driver must be exposed to the traffic environment, and this has to occur at the time when they are at greatest risk of being involved in a traffic crash, that is in the first few months of unsupervised driving (Begg and Langley, 2009; Lewis-Evans, 2010; Mayhew et al., 2003). In some countries, such as New Zealand (NZ), Australia and North America, graduated driver licensing systems (GDLS) have been introduced to allow the learner
154
D.J. Begg et al. / Accident Analysis and Prevention 62 (2014) 153–160
driver to gain initial driving experience in relatively safe conditions (Williams and Shults, 2010). In NZ, the GDLS has three licensing stages: learner, restricted, and full. A learner licence can be gained at 15 years of age1 by passing a computer-based multi-choice test, and applies for a minimum of six months during which time the learner driver must be supervised by an experienced driver. A restricted licence requires passing a practical on-road driving test, and allows unsupervised driving, except between 10 pm and 5 am, and when carrying passengers. The restricted licence applies for 18 months although this can be reduced to 12 months by completing an approved course (either Defensive Driving or Street Talk) (New Zealand Transport Agency, 2012). A full privilege licence, with no GDLS conditions, requires passing a second practical driving test. In NZ, as well as elsewhere, young drivers have been shown to have low crash rates during the supervised learner licence period but this changes quite dramatically once unsupervised driving is allowed, with the first few months of unsupervised driving being the time when crash risk is at its highest (Lewis-Evans, 2010; Mayhew et al., 2003; McCartt et al., 2003; Williams, 2003). In an effort to lessen this risk some jurisdictions are extending the learner licence period and requiring a minimum number of hours of supervised driving before unsupervised driving is allowed. The minimum number of hours varies considerably by jurisdiction from around 30 h in several US states (Williams et al., 2012) to 120 h in Victoria, Australia (Vicroads, 2012). In NZ the number of hours of supervised driving practice at the learner licence stage is not mandated but in 2011 changes were made to the licensing system to make the restricted licence test more difficult. This change was intended to encourage 120 h of supervised driving practice (Ministry of Transport, 2011). Some young people gain their initial driving experience by driving on-road before they get a car drivers licence. This has been reported in several countries (Begg et al., 2012; Elliott et al., 2008; McDowell et al., 2009; Senserrick et al., 2010). In NZ, it is illegal to drive on a public road before licensing (even if supervised by an experienced driver) yet around 50% of the participants in a large study of newly licensed car drivers (New Zealand Drivers Study, NZDS) reported having done this (Begg et al., 2012). In Australia, in the DRIVE study, (which, like the NZDS, is a prospective cohort study of young drivers), the prevalence of pre-licence driving was 25% (Senserrick et al., 2010). However, crash involvement was not examined in either of these published studies so it is not known if on-road driving before licensing had any effect on crash risk once licensed driving had commenced. Most published studies that have investigated pre-licence driving behaviour among young people have examined factors associated with traffic crashes in relation to the licence status of the drivers involved (Hanna et al., 2010; Hasselberg and Laflamme, 2009; Lam, 2003) or personal and behavioural factors, including crashes, associated with pre-licence driving behaviour (Elliott et al., 2008). None of these studies have examined pre-licence driving and then followed-up the drivers to determine whether pre-licence driving influenced their crash risk, once they were licensed. The one exception to this was a longitudinal study of young drivers in Western Australia (WA) which followed young newly licensed drivers through the first years of licensed driving. One of the key findings from this study was that young drivers who drove daily or weekly before passing their licence test (i.e. gaining their L-plates) were 75–100% more likely to be involved in a crash in the first 12 months as a probationary (P-plate) driver, when unsupervised driving was first allowed, than those who never drove before gaining their L-plates (Stevenson et al., 2001). As the authors note, however,
1 The minimum licence age increased to 16 years in 2011 but was 15 years when the NZDS cohort was recruited.
driving exposure during the P-plate crash period was not included in the multivariate analysis, a factor that may have influenced their results. In NZ, there is a belief that young people who have grown up in a rural area, and had the opportunity to drive vehicles around farm paddocks and on farm roads, are experienced (and presumably safer) drivers when they begin licensing, than their urban counterparts who have not had this experience. This argument is often put forward by the rural sector during discussions on minimum driver licence age, and has been used to try and stop the licence age from being raised (Begg and Langley, 2009). There is no direct evidence in NZ to support, or refute, this argument. It seems rather unlikely that this belief would only apply in New Zealand, but to our knowledge no one has examined pre-licensing off-road driving experience among young people so it is unknown what effect may have on crash risk among newly licensed drivers. The aim of the present study was to examine the relationship between both on-road and off-road pre-licensed driving experience reported by newly learner licensed car drivers, and the traffic crashes experienced as a driver during the learner licence stage, when only supervised driving is allowed, and the restricted licence stage, when unsupervised driving is first allowed. Motorcycle use was also included because it has been suggested, without any supporting evidence, that riding a motorcycle would be good preparation for young people about to start driving a car (Begg and Langley, 2009). In particular, we sought to determine the independent effects for each transport mode (car on-road and off-road, motorcycle on-road and off-road) in relation to a traffic crash as a car driver while on a learner licence, and while on a restricted licence, after controlling for a range of potential confounding socio-demographic, personality, and behavioural factors, including distance driven as a learner licensed driver and as a restricted licensed driver. 2. Method 2.1. New Zealand drivers study (NZDS) NZDS is a prospective cohort study of 3992 newly licensed car drivers in New Zealand. Fig. 1 shows the stages of the NZDS and how they are linked to the stages of the NZ graduated driver licensing system. The cohort was recruited, and the learner licence (baseline) questionnaire was completed, very soon after passing their learner licence test. This took place between 1st February 2006 and 31st January 2008, at various locations throughout New Zealand. Details of the recruitment procedures have been reported previously (Begg et al., 2012; Begg et al., 2009a,b; Langley et al., 2012a). The NZDS restricted licence (first follow-up) and full licence (second follow-up) telephone interviews took place very soon after the study participants passed the respective licence tests. 2.2. Data 2.2.1. Data extraction The present analysis used data from the learner (baseline), restricted, and full licence interviews. Data from the restricted licence interviews were extracted on 17 September, 2010 at which time 71% of participants had progressed to a restricted licence, 88% of them completed the NZDS restricted licence (first follow-up) interview. Data from the full licence interviews were extracted on 30 May 2011 at which time 40% of the full cohort had progressed to a full licence, and 93% of them completed the full licence (second follow-up) interview. For the present study our primary interest was young drivers so only those aged 15–24 years at learner licence (93% of the total cohort) were included (n = 2358 completed the
D.J. Begg et al. / Accident Analysis and Prevention 62 (2014) 153–160
155
obtained from the official New Zealand traffic crash reports (TCR) recorded by the New Zealand Police. The police only record crashes involving injury, so to provide a more complete crash coverage we also included self-reported crashes that were defined as a crash that occurred on a public road where someone was injured and/or there was vehicle or property damage. Crashes that occurred during the learner licence stage were self-reported at the restricted licence interview and restricted licence stage crashes were reported at the full licence interview. The combined (TCR and self-report) crash file was checked and duplicates removed. There were 124 crashes at the learner licence stage and 347 at the restricted licence stage. For the analysis the crash outcome was dichotomised (yes/no).
Fig. 1. Diagram showing the linkage between the interview stages of the New Zealand Drivers Study and the Graduated Driver Licensing System.
restricted licence interview and n = 1428 completed the full licence interview). 2.2.2. Pre-licensed driving experience The pre-licensed driving data, collected at the learner licence (i.e. baseline) interview, included a series of questions relating to driving experience before passing the learner licence test. If responded in the affirmative, the question: “Have you ever driven a car on or off-road?” was followed up with “How many times in total have you driven a car on-road (that is on a public road)?” “How many times in total have you driven a car off-road?” The response options were never, once, 2–5 times, 6–10 times, 11–20 times, >20 times. The same format was used for motorcycle use. For the analysis of the learner licence stage crashes, because of small numbers in some of these driving experience groups, the categories were collapsed to never, 1–20 times, >20 times for pre-licence car driving, and never or ≥1 times for motorcycle use. For the restricted licence stage crash analysis the driving experience responses were grouped into four categories based on the number of times driven: never, 1–5 times, 6–20 times, >20 times. 2.2.3. Motor vehicle traffic crashes Motor vehicle traffic crashes as a car driver (or any vehicle covered by a car driver’s licence such as a van, utility, light truck) were
2.2.4. Potential confounding factors A number of factors associated with pre-licensed driving behaviour were identified in an earlier study by Begg et al. (2012). These potential confounding factors have been included in the present analysis. A brief description of these factors is provided here with further details available elsewhere (Begg et al., 2012, 2009a,b; Langley et al., 2012a). Briefly, age at the time of obtaining a learner licence was calculated from each participants’ date-of-birth and the date their learner licence was issued. Ethnicity was selfidentified and to simplify interpretation of the results, anyone who ¯ (i.e. the indigenous people of New Zealand) self-identified as Maori ¯ ¯ was classified as Maori and the remainder as non-Maori. Residential location was determined using the “urban/rural profile” used by Statistics New Zealand (Statistics New Zealand, 2009). This classified the participants’ residential addresses as either urban or rural. Social deprivation was measured using the NZDep2006 score which combines nine variables from the 2006 census, which reflect eight dimensions of deprivation (Salmond et al., 2007). These variables are: income (2 variables), home ownership, support (e.g. single parent family), employment, qualifications, living space, communication (e.g. access to telephone), and transport (access to a car). The residential address of each participant was used to assign a deprivation score ranging from 1 to 10. For the multivariate analysis these levels were collapsed to create three levels of deprivation: low, medium and high (Adolescent Health Research Group, 2008). Alcohol use was measured using the first three questions of the AUDIT (the Audit-C) (Babor et al., 1989). The responses to these three questions were summed to give a score ranging from 0 to 12. Following the convention established by alcohol researchers (e.g. Bradley et al., 2007) a cut-off of ≥4 was used to indicate hazardous drinking for males and ≥3 for females. Cannabis or marijuana use was measured by the following question: “How often do you use marijuana or cannabis?” The response options were: Never, monthly or less, 2–4 times per month, 2–3 times per week, 4 or more times a week. Reponses were classified at three levels; “never”, “monthly” (monthly or less and 2–4 times per month), “weekly” (2–3 times per week, and 4+ times per week). The use of herbal highs was measured by asking “How often do you use Herbal Highs (party pills) such as “Rapture”, “Bliss”?” Response options were the same as for cannabis, but because of the small number of users these were categorised as yes (have used) or no (never). Personality was measured using the Zuckerman IMP-SS scale (Zuckerman et al., 1993). This gave a measure of impulsivity (8 items), sensation seeking (11 items), and aggression–hostility (17 items). Gender specific cut-off points were selected to classify each of the scales into three categories: low, middle, or high. The low group had scores in the lowest quartile, the high group had scores in the highest quartile, and the remainder were in the middle quartiles (approximately 50%). Time on the learner licence was included to account for the varying time spent on a learner licence. Unsupervised driving while on a learner licence has been shown to be significantly related to crash involvement as a learner licensed driver, and is also related to pre-licence
156
D.J. Begg et al. / Accident Analysis and Prevention 62 (2014) 153–160
driving (Langley et al., under consideration for publication). This was included (yes/no) as a potential confounder.
Table 1 Frequency distribution of pre-licensed driving experience categories by crash as a car driver at the learner and restricted licence stages. Pre-licence driving
2.2.5. Distance driven An estimate of total distance driven was obtained using a series of questions that have been shown to be a reliable method of obtaining such estimates (Lajunen and Özkan, 2011). These questions were administered by trained interviewers who “walked” the young driver through their driving experiences as a learner licensed driver (at the restricted licence interview) and as a restricted licensed driver (at the full licence interview). (a) To obtain an indication of how frequently they had driven each participant was asked “How often did you drive on a public road with or without a supervisor. Was it “daily”, or “at least weekly”, or “fortnightly”, or “monthly”. If they responded “yes” then they were asked. (a) How many times in a typical day (week, fortnight, month)? The interviewer knew the exact dates when the participant had passed their licence tests (obtained from the NZ driver licensing data base) so was able to accurately calculate the number of days (weeks, fortnights or months) from one test to the next. If they answered “No”, that is they had not driven at least monthly, they were asked “How many times all up” they had driven on their learner or restricted licence. (b) To get an indication of the distance they had driven they were asked “On a typical drive how far did you drive?” The distance options were < 11 kms, 11–20 kms, 21–50 kms, 51–100 kms, >100 kms, which for the analysis were recoded to 5, 15, 35, 75, 150kms. The responses to (a) and (b) were multiplied to provide an estimate of the total distance driven while on learner or restricted licence. As appropriate, this series of questions was asked for driving with a professional instructor, driving with a supervisor, or driving unsupervised. The totals for each of these categories were summed to provide an estimate of their overall total driving exposure (distance driven). Acknowledging the imprecision of this measure, for the analysis the total distance driven was recoded into three categories, low (first quartile), mid (second and third quartiles), high (top quartile). This categorisation ensured that those with high and low exposure were separated by the larger middle category, which should help to minimise misclassification bias at the high and low levels of driving exposure.
Learner licence stage Crash (n = 124) n
No Crash (n = 2222) %
n
%
44 46 30
37 38 25
1157 789 203
54 37 9
54 52 15
45 43 12
1187 784 184
55 37 8
Motorcycle on road Never Yes
102 20
83 17
1878 264
88 12
Motorcycle off road Never Yes
91 30
75 25
1591 550
74 26
Car on-road Never 1–20 times >20 times Car off-road Never 1–20 times >20 times
Restricted licence stage Crash (n = 347)
No crash (n = 1077)
n
%
n
%
Car on-road Never 1–5 times 6–20 times >20 times
182 98 30 33
53 29 9 10
583 293 110 73
55 28 10 7
Car off-road Never 1–5 times 6–20 times >20 times
185 96 31 32
54 28 9 9
556 298 107 100
52 28 10 9
Motorcycle on-road Never 1–5 times 6–20 times >20 times
289 29 9 16
84 8 3 5
929 72 25 32
88 7 2 3
Motorcycle off-road Never 1–5 times 6–20 times >20 times
231 56 16 39
68 16 5 11
761 109 57 129
72 10 5 12
2.3. Statistical analysis All statistical analyses were undertaken using SAS version 9.3 (SAS Institute Inc., 2010). First, the frequency distribution for each of the pre-licensed driving variables, by crash status at the learner and restricted licence stage, were examined. Second, separate Poisson regression models with robust error variance (Zou, 2004) were run to examine each of the four pre-licensed driving categories (i.e. car on-road and car off-road, motorcycle on-road and motorcycle off-road with (1) a learner licence driver crash and (2) a restricted licence driver crash as the outcome (unadjusted results). That is, eight separate analyses were undertaken. Each of these models was then re-run including all of the potential confounders (adjusted results).
3.2. Crashes as a learner licensed driver With crash as a learner licensed driver as the outcome, each of the pre-licence driving variables was entered into separate Poisson regression models (unadjusted results Table 2). Compared with someone who had not driven a car on-road before licensing, the unadjusted relative risk (RR) of a crash for someone who had driven a car >20 times on-road before licensing was 3.5. The adjusted results (Table 2) show that, after controlling for the potential confounders, the RR was 1.7 (95% CI 1.0–2.9). The adjusted results show that pre-licence car off-road driving and motorcycle use, both onroad and off-road were not found to have any effect on crash risk as a learner licensed driver.
3. Results
3.3. Crashes as a restricted licensed driver
3.1. Frequency distributions
The same procedure was followed for the analysis of the restricted licence crashes. The results for the pre-licence driving categories from these Poisson regression models are presented in Table 3. Overall there was no consistent relationship between any of the pre-licence driving categories and the restricted licence crashes.
The frequency distributions for the pre-licensed driving variables, by crash status, at the learner licence and restricted licence stages are presented in Table 1.
D.J. Begg et al. / Accident Analysis and Prevention 62 (2014) 153–160
157
Table 2 Results from separate Poisson regression analyses showing the unadjusted and adjusted relative risks (RR), 95% confidence intervals (CI) and p-values for an on-road car crash as a learner licensed driver by each of the pre-licence driving categories* . Pre-licence driving
RR
95% CI
p-Value
Unadjusted
RR
95% CI
p-Value
Adjusted
Car on-road Never 1–20 times >20 times
Reference 1.5 3.5
1.0–2.3 2.3–5.5
.05 .00
Reference 1.2 1.7
0.8–1.8 1.0–2.9
.44 .05
Car off-road Never 1–20 times >20 times
Reference 1.4 1.7
1.0–2.1 1.0–3.0
.06 .05
Reference 1.3 1.2
0.9–1.9 0.7–2.2
.23 .53
Motorcycle on-road Never Yes
Reference 1.4
0.9–2.2
.17
Reference 1.0
0.6–1.5
.86
Motorcycle off-road Never Yes
Reference 1.0
0.6–1.4
.83
Reference 0.8
0.5–1.3
.36
*
Adjusted for gender, age at learner licence, residential location, ethnicity, sensation seeking, impulsivity, aggression, alcohol use, cannabis use, herbal high use, time on learner licence, distance driven on learner licence, and unsupervised driving on a learner licence.
The only significant result was that those who had ridden a motorcycle off-road 1–5 times had a RR of 1.4 (95% CI 1.1–1.8) compared with those who had not ridden a motorcycle off-road before licensing.
such as gender, sensation seeking, alcohol use, drug use, so these factors do not explain this result.
4. Discussion
In the present study the risk of being in a crash at the learner licence stage was related to driving a car on-road before getting a learner licence, and especially for those who had driven a substantial number of times (more than 20). This applied to around 10% of the young learner licence drivers in this study. It is possible that having done this amount of on-road driving before licensing these young drivers were over-confident, a factor which is known to increase crash risk among young drivers (Matthews and Moran, 1986). They may also be less law abiding than the other young drivers, but to some extent this was adjusted for by including unsupervised driving in the analysis. Whether this increase in crash risk at the learner licence stage is due to over-confidence, breaches of
4.1. Main findings The results from this study showed that, overall, pre-licensed driving experience had relatively little influence on crash involvement at the learner or restricted licence stage, when distance driven and other factors associated with crash risk were taken into consideration. The only pre-licence driving experience that showed a consistent relationship with crash risk was on-road car driving which increased risk at the learner licence stage. This result had been adjusted for distance driven as well as a range of other factors
4.2. Learner licence crashes
Table 3 Results from separate Poisson regression analysis showing the unadjusted and adjusted relative risks (RR), 95% confidence intervals (CI) and p-values of an on-road car crash as a restricted licensed driver for each of the pre-licence driving categories. Pre-licensed driving
RR
95% CI
p-Value
RR
95% CI
p-Value
Adjusteda
Unadjusted Car on-road Never 1–5 times 6–20 times >20 times
Reference 1.1 0.9 1.3
0.9–1.3 0.6–1.3 1.0–1.8
.63 .55 .09
Reference 1.0 0.9 1.2
0.8–1.2 0.6–1.3 0.8–1.6
.98 .54 .50
Car off-road Never 1–5 times 6–20 times >20 times
Reference 1.0 0.9 1.0
0.8–1.2 0.6–1.3 0.7–1.3
.82 .54 .86
Reference 1.0 1.0 0.9
0.8–1.2 0.7–1.3 0.6–1.3
.68 .78 .50
Motorcycle on-road Never 1–5 times 6–20 times >20 times
Reference 1.2 1.1 1.4
0.9–1.7 0.6–2.0 0.9–2.1
.25 .71 .11
Reference 1.2 1.2 1.3
0.8–1.6 0.7–2.1 0.9–2.0
.34 .59 .20
Motorcycle off-road Never 1–5 times 6–20 times >20 times
Reference 1.5 0.9 1.0
1.1–1.9 0.6–1.5 0.7–1.3
.00 .79 .90
Reference 1.4 0.9 0.9
1.1–1.8 0.6–1.4 0.7–1.3
.00 .76 .63
a Adjusted for gender, age at learner licence, residential location, ethnicity, sensation seeking, impulsivity, aggression, alcohol use, cannabis use, herbal high use, time on learner licence, driving exposure and unsupervised driving on a learner licence, distance driven on restricted licence.
158
D.J. Begg et al. / Accident Analysis and Prevention 62 (2014) 153–160
the law, or some other reason, it could be countered by the presence of a responsible supervisor who is in charge of the vehicle. Although supervision is a legal requirement at the learner licence stage, it is known that many young people breach this condition (Langley et al., under consideration for publication) but we do not know if this was the case when these crashes occurred. Pre-licence on-road driving (even if supervised by an experienced licensed driver) is illegal in New Zealand and can result in penalties such as a 400 NZD fine, being forbidden to drive, or having the vehicle seized on the roadside and impounded for 28 days. These young people may be unaware of the potential consequences if they are caught engaging in this behaviour, so improving knowledge of the penalties may help deter them from driving on-road before licensing. Also, improving the level of enforcement may help deter this behaviour. In this study, around half of the newly licensed drivers had driven on-road before licensing but only 5% reported that they had ever been stopped by the police when doing so. These figures suggest that there is scope for enforcement to be improved. 4.3. Restricted licence crashes Pre-licence driving behaviour showed no consistent relationship with restricted licence crashes. At the restricted licence stage the young driver is permitted to legally drive unsupervised, and this is the period when crash risk is at its highest (Begg and Langley, 2009; Lewis-Evans, 2010; Mayhew et al., 2003). It is possible, therefore, that factors other than pre-licence driving experience, such as poor hazard recognition, driving at excessive speed, following too closely, or dangerous overtaking, are the more important crash risk factors at this stage. Also, before they reached the restricted licence stage these young drivers had to have spent at least six months on a learner licence. The experience they gained during this time may have negated any effect that pre-licence driving had on restricted licence crashes. 4.4. Comparison with Western Australian study The findings from this study do not fully support those from the Western Australian study by Stevenson et al. (2001) who found that on road pre-licensed car driving was associated with crashes in the first twelve months of probationary (P-plate) licensed driving, which is the equivalent to the restricted licence stage in NZ, when unsupervised driving is first allowed. In the present study pre-licence car driving, either on or off-road, was not found to be associated with crashes at the restricted licence stage but on-road car driving more than twenty times was associated with a crash at the learner licence stage. This effect was strong (RR 3.5) in the unadjusted analysis (Table 2), when distance driven (i.e. driving exposure) was not included in the analysis. The WA study was unable to include a measure of driving exposure in the initial analysis but when this was obtained at a follow-up interview, post hoc analysis showed a strong association between driving exposure and pre-licence driving. The authors suggested that, without a driving exposure measure, the frequency of driving prior to L-plates could be interpreted as an adjustment for driving exposure (Stevenson et al., 2001). 4.5. Methodological considerations The NZDS is a multistage, comprehensive prospective cohort study of a large number of newly licensed car drivers especially designed to provide scientific evidence on a wide range of issues affecting young drivers. The prospective nature of the study ensures that the explanatory factors, such as attitudes and behaviours, were measured before the outcomes (e.g. crashes) had occurred, thereby ensuring temporality. In the present study our measure of crashes
was a combination of self-reported and police traffic crash records. This measure was chosen because, in NZ, only crashes involving injury are required to be recorded in the police traffic crash records and, despite this being a legal requirement, it is known that even crashes involving hospitalised injury are seriously underreported (Alsop and Langley, 2001). The under-reporting of crashes is not unique to NZ. Stevenson and Palamara used police reported crashes in the WA study and considered the under-reporting of crash outcomes to be the most “noticeable” limitation in their study (Stevenson et al., 2001). By including self-reported crashes, we were able to capture injury crashes that had not been recorded by the police, as well as crashes that did not involve injury but did involve property damage, and would otherwise have been missed. The NZDS cohort is not a randomly selected representative sample of all newly licensed drivers in NZ, therefore, prevalence estimates presented in this study may not apply to all young newly licensed drivers in New Zealand, or elsewhere. Attempts were made during recruitment to ensure heterogeneity of the cohort by recruiting in the main geographic regions of the country, and among a wide range of ethnic groups. Recruitment was carried out by local people in each region, and mostly face-to-face, so that people who are often reluctant to participate in such studies were more likely to be recruited. Our recruitment rates of around 80–90% attest to the success of this method (Langley et al., 2012a). The NZDS is an on-going prospective cohort study which is constantly changing as study members progress, at their own pace, through the graduated driver licensing system. At present the NZ GDLS does not have maximum time limits at any licence stage,2 the time spent on any licence can vary considerably. To conduct the analysis for this study it was necessary to set a data extraction date for the follow-up interviews. At that date, those included in the analysis of crashes at the learner licence stage had to have progressed to a restricted licence and undertaken the NZDS restricted interview. Of the 3693 15–24 years olds in the NZDS, 73% had gone on and passed their restricted licence test and 88% of them had completed the NZDS first follow-up interview. Similarly for the analysis at the restricted licence stage, 42% had progressed to a full licence and 93% of them completed the second NZDS follow-up interview. These figures show that progression through to the full graduated licensing system is somewhat tardy, but for those who had progressed the NZDS follow-up rates of 88–93% were extremely high. The lack of progression through the licensing stages of the GDLS can have implications for studies based on the different licence stages. A NZDS paper comparing those who had, and had not, progressed from a learner to restricted licence, after being eligible to do so for at least two years, has been published (Langley et al., 2012b). The “non-progressors” were shown to have a different socio-demographic and behavioural profile to the “progressors”, they drove less, were less likely to be pre-licence car drivers, and they were at reduced risk of being a traffic offender. To take full advantage of the available data, for the present study the analysis of the crashes that occurred at the learner licence stage were based on all those who had progressed to a restricted licence and completed the NZDS restricted licence interview, and the restricted licence stage crash analysis was based on those who had continued on to a full licence and completed the NZDS full licence interview. This meant that the restricted licence crash analysis was based on a subset (n = 1428) of those in the learner licence crash analysis (n = 2358). A comparison of the 1428 who had progressed to a full licence with the 930 who were still on a restricted licence showed that those who had progressed to a full licence were more likely to be male, aged 15 at learner licence, to live in an area of
2 The NZ Government announced in January 2013 that five-year time limits will be introduced at the learner and restricted licence stages.
D.J. Begg et al. / Accident Analysis and Prevention 62 (2014) 153–160
¯ low deprivation, not of Maori ethnicity, not high alcohol or cannabis users but were herbal high users. They were more likely to have driven a car off-road before licensing, but did not differ for any of the other pre-licence driving behaviours. Perhaps not surprisingly, they were also less likely to have spent more than 12 months on their learner licence. To assess the effect these differences may have had on the learner licence stage crash results, a post hoc multivariate analysis was run with only those who had completed the full licence interview, and were therefore the same people as were included in the restricted licence stage crash analysis. These results showed that driving a car both on-road and off-road increased the risk of a learner licence crash. This suggests that those who had progressed to a full licence may have been practising driving before they began licensing but this practice increased rather than decreased their crash risk, once they became licensed, even after controlling for risk-taking (impulsivity, sensation seeking, aggression) distance driven, unsupervised driving, and a range of socio-demographic and other behavioural variables. Virtually all of the data used in this study was self-reported which has the potential to be biased, particularly in the form of social desirability bias. This is especially the case for questions about illegal or dangerous behaviours, which applied to some of the variables in this study. Research has shown, however, that social desirability on self-reported risky behaviours is not necessarily substantial (Lajunen and Summala, 2003; Sullman and Taylor, 2010) and the frequency with which some illegal behaviours (e.g. prelicence on-road car driving, cannabis use) were reported in this study suggest it may not have been a substantial problem. Inaccurate recall is also a perceived weakness of self-reported behaviour and is considered by some to be less desirable than official records for events such as crashes. However, a study validating the accuracy of recall of crash events by young people showed that young people can recall crash details with a high degree of accuracy, even several years after the event (Begg et al., 1999). Also, the Australian DRIVE study, using a very similar study design to the NZDS, found a high level of accuracy in young drivers’ self-reports of police recorded crashes and convictions (Boufous et al., 2010). Validating explanatory data, such as pre-licence driving behaviour, is extremely difficult, if not impossible, as there is no other record of these behaviours with which the self-reported data can be compared. Another factor that may have influenced the results was the categorisation of the explanatory variables which was based on frequencies and therefore the cut-off points were somewhat arbitrary. Where the numbers allowed, the categories were chosen so as to provide a low, moderate and high exposure group, that were compared with a non-exposed group. One particular challenge facing researchers undertaking studies such as the NZDS is obtaining reliable and valid driving exposure data. Some of the newer technologies, such as in-vehicle recording, are limited to smaller studies (Williams et al., 2012) and not practical for a large study such as this. For this study our estimates of driving exposure were obtained by “walking” the participant through a series of questions to get an estimate of driving frequency and time and distance for a typical trip. The interviewers knew exactly how long the driver had been on a particular licence so this part of the equation was not estimated or based on recall. The final total distance was categorised so that the lowest and highest quartiles represented the low and high exposure groups, with the remaining 50% in the middle group. This categorisation was used to minimise misclassification at the high and low levels of exposure.
5. Conclusion The findings from this research suggest that, on the whole, prelicensed driving did not have very much effect on crash risk among
159
learner or restricted licensed drivers, although frequent on-road car driving before licensing may have increased crash risk at the learner licence stage. The popular belief that pre-licensed driving experience makes young drivers more competent and better drivers needs to be dispelled, and young people should be discouraged from the illegal behaviour of driving a car on-road before licensing. Acknowledgments The New Zealand Drivers Study is funded by the Health Research Council of New Zealand (2005–2013), the Road Safety Trust (2005–2013) and was funded by the Accident Compensation Corporation (2005–2009). No funding source had any involvement in the preparation of this paper. The authors wish to acknowledge Dr. Ari Samaranayaka, Injury Prevention Research Unit, University of Otago, for comments on an earlier version of this paper. References Adolescent Health Research Group, 2008. Youth’07: The Health and Wellbeing of Secondary School Students in New Zealand. Technical Report. The University of Auckland, Auckland, pp. 2008. Alsop, J., Langley, J.D., 2001. Under-reporting of motor vehicle traffic crash victims in New Zealand. Accident Analysis and Prevention 33, 353–359. Australian Government, 2012. Living in Australia: Applying for a Driver’s Licence, http://www.immi.gov.au/living-in-australia/settle-in-australia/to-do-first/ apply-drivers.htm (retrieved 30.03.12). Babor, T.F., de la Fuente, J.R., Saunders, J.B., Grant, M., 1989. AUDIT: The Alcohol Use Disorders Identification Test, Guidelines for Use in Primary Health Care. World Health Organisation, Geneva. Begg, D., Langley, J., Williams, S., 1999. Validity of self-reported crashes and injuries in a longitudinal study of young adults. Injury Prevention 5, 142–144. Begg, D., Sullman, M., Samaranayaka, A., 2012. The characteristics of young prelicensed drivers: evidence from the New Zealand Drivers Study. Accident Analysis and Prevention 45, 539–546. Begg, D.J., Langley, J.D., 2009. A critical examination of the arguments against raising the car driver licensing age in New Zealand. Traffic Injury Prevention 10, 1–8. Begg, D.J., Langley, J.D., Brookland, R.L., Ameratunga, S., Broughton, J.R., 2009a. The opinions of newly licensed drivers in New Zealand on the minimum car driver licensing age and reasons for getting a licence. New Zealand Medical Journal 122 http://nzma.org.nz/journal/122-1306/3880/ Begg, D.J., Langley, J.D., Broughton, J.R., Brookland, R.L., Ameratunga, S., McDowell, A.J., 2009b. New Zealand Drivers Study: a follow-up study of newly licensed drivers. Injury Prevention 15, 281, http://dx.doi.org/ 10.1136/ip. 2009.021998a.1-9. Boufous, S., Ivers, R., Senserrick, T., 2010. Accuracy of self-report of on-road crashes and traffic offences in a cohort of young drivers: the DRIVE study. Injury Prevention 16, 275–277. Bradley, K., DeBenedetti, A., Volk, R., Williams, E., Frank, D., Kivlahan, D., 2007. AUDIT-C as a brief screen for alcohol misuse in primary care. Alcoholism Clinical and Experimental Research 7, 1207–1217. Department for Transport, 2012. Learners and New Drivers and Riders, http://www.direct.gov.uk/en/Motoring/LearnerAndNewDrivers/index.htm (retrieved 30.03.12). Elliott, M., Ginsburg, K., Winston, F., 2008. Unlicensed teenaged drivers: who are they, and how do they behave when they are behind the wheel? Pediatrices, Offical Journal of the American Academy of Pediatrics 122, 994–1000. European Commission, 2012. Road Safety: Driving Licence, http://ec.europa.eu/transport/road safety/behavior/driving licence en.htm (retrieved 07.02.13). Hanna, C.L., Hasselberg, M., Laflamme, L., Moller, J., 2010. Road traffic crash circumstances and consequences among young unlicensed drivers: A Swedish cohort study of socioeconomic disparities. BMC Public Health 10, 8. Hasselberg, M., Laflamme, L., 2009. How do car crashes happen among young drivers aged 18–20 years? Typical circumstances in relation to license status, alcohol impairment and injury consequences. Accident Analysis and Prevention 41, 734–738. Huang, P., Winston, F., 2011. Young Drivers. Handbook of Traffic Psychology. Academic Press, Norfolk, VA USA, pp. 315–338. Insurance Institute for Highway Safety, 2012. Summary Table: US Young Driver Licensing Systems, http://www.ihhs.org/laws/GraduatedLicenseCompare.aspx (retrieved 26.03.12). Lajunen, T., Özkan, T. (Eds.), 2011. Self-Report Instruments and Methods. Handbook of Traffic Psychology. Academic Press, Waltham, MA. Lajunen, T., Summala, H., 2003. Can we trust self-reports of driving? Effects of impression management on driver behaviour questionnaire responses. Transportation Research Part F: Traffic Psychology and Behaviour 6, 97–107. Lam, L., 2003. A neglected risky behavior among children and adolescents: Underage driving and injury in New South Wales, Australia. Journal of Safety Research 34, 315–320.
160
D.J. Begg et al. / Accident Analysis and Prevention 62 (2014) 153–160
Langley, J., Begg, D., Brookland, R., Ameratunga, S., McDowell, A., Broughton, J., 2012a. Establishment of the New Zealand Drivers Study. The New Zealand Medical Journal (Online) 125, 98–112. Langley, J., Begg, D., Brookland, R., Samaranayaka, A., Jordan, H., Davie, G., 2012b. Nonprogression through graduated driver licensing: characteristics, traffic offending, and reasons for nonprogression. Traffic Injury Prevention 13, 7–13. Langley, J., Begg, D., Samaranayaka, A., Brookland, R., Weiss, H., 2013. Unsupervised driving by learner licence holders: associated characteristics and crash risk. Accidental Analysis and Prevention 60, 24–30. Lewis-Evans, B., 2010. Crash involvement during the different phases of the New Zealand Graduated Driver Licensing System (GDLS). Journal of Safety Research 41, 359–365. Matthews, M.L., Moran, A.R., 1986. Age differences in male drivers’ perception of accident risk: the role of perceived driving ability. Accident, Analysis and Prevention 18, 299–313. Mayhew, D., Simpson, H., Pak, A., 2003. Changes in collision rates among novice drivers during the first months of driving. Accident Analysis and Prevention 35, 683–691. McCartt, A.T., Shabanova, V.I., Leaf, W.A., 2003. Driving experiences, crashes and traffic citations of teenage beginning drivers. Accident Analysis and Prevention 35, 311–320. McDowell, A.J., Begg, D.J., Connor, J., Broughton, J., 2009. Unlicensed driving among ¯ urban and rural Maori drivers: New Zealand Drivers Study. Traffic Injury Prevention 10, 538–545. Ministry of Transport, 2011. Safer Journeys Action Plan 2011–2012. Ministry of Transport, Wellington. New Zealand Transport Agency, 2012. The Graduated Driver Licensing System (GDLS) has Changed, http://www.nzta.govt.nz/licence/ photo/gdls-changing.html (retrieved 27.09.12). Organisation for Economic Co-Operation and Development, 2006. Young Drivers: The Road to Safety. Policy Brief. Organisation for Economic Co-operation and Development and the European Conference of Ministers of Transport (ECMT) Joint Transport Centre, Paris.
Salmond, C., Crampton, P., Atkinson, J., 2007, August. NZDep2006 Index of Deprivation. Department of Public Health, University of Otago, Wellington. SAS Institute Inc., 2010. Language Reference: Concepts, Second ed. SAS Institute Inc., Cary, NC. Senserrick, T., Chen, H., Boufous, S., Ivers, R., Stevenson, M., Norton, R., 2010. Demographic Factors Associated with Pre-licensed Driving in NSW Young Driver Cohort: The DRIVE Study. Australasian Road Safety Research Policing Education Conference, Canberra. Stevenson, M., Palamara, P., Morrison, D., Ryan, G., 2001. Behavioral factors as predictors of motor vehicle crashes in young drivers. Traffic Injury Prevention 2, 247–254. Statistics New Zealand, 2009. New Zealand: An Urban/Rural Profile Wellington2008, Available from: http://www.stats.govt.nz/ Update. statistics-by-area/urban-rural-profiles/default.htm (accessed 09.02.09). Sullman, M.J.M., Taylor, J.E., 2010. Social desirability and self-reported driving behaviours: Should we be worried? Transportation Research Part F: Traffic Psychology and Behaviour 13, 215–221. Vicroads, 2012. Licences: Get your P’s. http://www.vicroads.vic.gov.au/Home/Licences/ GetYourPs/ Williams, A., 2003. Teenage drivers: patterns of risk. Journal of Safety Research 34, 5–15. Williams, A., 2006. Young driver risk factors: successful and unsuccessful approaches for dealing with them and an agenda for the future. Injury Prevention 12, 4–8. Williams, A., Shults, R., 2010. Graduated driver licensing research, 2007-present: a review and commentary. Journal of Safety Research 41, 77–84. Williams, A.F., Tefft, B.C., Grabowski, J.G., 2012. Graduated driver licensing research, 2010-present. Journal of Safety Research 43, 195–203. Zou, G., 2004. A modified Poisson regression approach to prospective studies with binary data. American Journal of Epidemiology 159, 702–706. Zuckerman, M., Kulman, D.M., Joireman, J., Teta, P., Kraft, M., 1993. A comparison of three structural models for personality: the big three, the big five, and the alternative five. Journal of Personality and Social Psychology 65, 757–768.
Contents lists available at ScienceDirect
Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap
Pre-licensed driving experience and car crash involvement during the learner and restricted, licence stages of graduated driver licensing: Findings from the New Zealand Drivers Study Dorothy J. Begg a,∗ , John D. Langley a , Rebecca L. Brookland a , Shanthi Ameratunga b , Pauline Gulliver b a b
University of Otago, New Zealand University of Auckland, New Zealand
a r t i c l e
i n f o
Article history: Received 8 February 2013 Received in revised form 17 July 2013 Accepted 31 August 2013 Keywords: Young drivers Pre-licensed driving Graduated licensing: Motorcycles Cars Crashes
a b s t r a c t Objective: The aim of this study was to determine whether pre-licence driving experiences, that is driving before beginning the licensing process, increased or decreased crash risk as a car driver, during the learner or the restricted licence stages of the graduated driver licensing system (GDLS). Method: Study participants were 15–24 year old members of the New Zealand Drivers Study (NZDS) – a prospective cohort study of newly licensed car drivers. The interview stages of the NZDS are linked to, the three licensing stages of the GDLS: learner, restricted and full. Baseline demographic (age, ethnicity, residential location, deprivation), personality (impulsivity, sensation seeking, aggression) and, behavioural data, (including pre-licensed driving behaviour), were obtained at the learner licence interview. Data on distance driven and crashes that occurred at the learner licence and restricted licence stages, were reported at the restricted and full licence interviews, respectively. Crash data were also obtained from police traffic crash report files and this was combined with the self-reported crash data. The analysis of the learner licence stage crashes, when only supervised driving is allowed, was based on the participants who had passed the restricted licence test and undertaken the NZDS, restricted licence interview (n = 2358). The analysis of the restricted licence stage crashes, when unsupervised driving is first allowed, was based on those who had passed the full licence test and completed the full licence interview (n = 1428). Results: After controlling for a range of demographic, personality, behavioural variables and distance driven, Poisson regression showed that the only pre-licence driving behaviour that showed a consistent relationship with subsequent crashes was on-road car driving which was associated with an increased risk of being the driver in a car crash during the learner licence period. Conclusion: This research showed that pre-licensed driving did not reduce crash risk among learner or restricted licensed drivers, and in some cases (such as on-road car driving) may have increased risk. Young people should be discouraged from the illegal behaviour of driving a car on-road before licensing. © 2013 Elsevier Ltd. All rights reserved.
1. Introduction The over-representation of young drivers in motor vehicle crashes is a serious public health problem world-wide (Organisation for Economic Co-Operation and Development, 2006). The two main factors that contribute to this elevated crash risk are immaturity and driving inexperience (Huang and Winston, 2011; Williams, 2006; Williams et al., 2012). Immaturity is often
∗ Corresponding author. Tel.: +64 3 479 8509; fax: +64 3 479 8337. E-mail address: [email protected] (D.J. Begg). 0001-4575/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.aap.2013.08.027
addressed by imposing a mandatory minimum driver licensing age, which is typically between 16 and 18 years (Australian Government, 2012; Department for Transport, 2012; European Commission, 2012; Insurance Institute for Highway Safety, 2012). Addressing the problem of driving inexperience is more challenging because to gain experience the learner driver must be exposed to the traffic environment, and this has to occur at the time when they are at greatest risk of being involved in a traffic crash, that is in the first few months of unsupervised driving (Begg and Langley, 2009; Lewis-Evans, 2010; Mayhew et al., 2003). In some countries, such as New Zealand (NZ), Australia and North America, graduated driver licensing systems (GDLS) have been introduced to allow the learner
154
D.J. Begg et al. / Accident Analysis and Prevention 62 (2014) 153–160
driver to gain initial driving experience in relatively safe conditions (Williams and Shults, 2010). In NZ, the GDLS has three licensing stages: learner, restricted, and full. A learner licence can be gained at 15 years of age1 by passing a computer-based multi-choice test, and applies for a minimum of six months during which time the learner driver must be supervised by an experienced driver. A restricted licence requires passing a practical on-road driving test, and allows unsupervised driving, except between 10 pm and 5 am, and when carrying passengers. The restricted licence applies for 18 months although this can be reduced to 12 months by completing an approved course (either Defensive Driving or Street Talk) (New Zealand Transport Agency, 2012). A full privilege licence, with no GDLS conditions, requires passing a second practical driving test. In NZ, as well as elsewhere, young drivers have been shown to have low crash rates during the supervised learner licence period but this changes quite dramatically once unsupervised driving is allowed, with the first few months of unsupervised driving being the time when crash risk is at its highest (Lewis-Evans, 2010; Mayhew et al., 2003; McCartt et al., 2003; Williams, 2003). In an effort to lessen this risk some jurisdictions are extending the learner licence period and requiring a minimum number of hours of supervised driving before unsupervised driving is allowed. The minimum number of hours varies considerably by jurisdiction from around 30 h in several US states (Williams et al., 2012) to 120 h in Victoria, Australia (Vicroads, 2012). In NZ the number of hours of supervised driving practice at the learner licence stage is not mandated but in 2011 changes were made to the licensing system to make the restricted licence test more difficult. This change was intended to encourage 120 h of supervised driving practice (Ministry of Transport, 2011). Some young people gain their initial driving experience by driving on-road before they get a car drivers licence. This has been reported in several countries (Begg et al., 2012; Elliott et al., 2008; McDowell et al., 2009; Senserrick et al., 2010). In NZ, it is illegal to drive on a public road before licensing (even if supervised by an experienced driver) yet around 50% of the participants in a large study of newly licensed car drivers (New Zealand Drivers Study, NZDS) reported having done this (Begg et al., 2012). In Australia, in the DRIVE study, (which, like the NZDS, is a prospective cohort study of young drivers), the prevalence of pre-licence driving was 25% (Senserrick et al., 2010). However, crash involvement was not examined in either of these published studies so it is not known if on-road driving before licensing had any effect on crash risk once licensed driving had commenced. Most published studies that have investigated pre-licence driving behaviour among young people have examined factors associated with traffic crashes in relation to the licence status of the drivers involved (Hanna et al., 2010; Hasselberg and Laflamme, 2009; Lam, 2003) or personal and behavioural factors, including crashes, associated with pre-licence driving behaviour (Elliott et al., 2008). None of these studies have examined pre-licence driving and then followed-up the drivers to determine whether pre-licence driving influenced their crash risk, once they were licensed. The one exception to this was a longitudinal study of young drivers in Western Australia (WA) which followed young newly licensed drivers through the first years of licensed driving. One of the key findings from this study was that young drivers who drove daily or weekly before passing their licence test (i.e. gaining their L-plates) were 75–100% more likely to be involved in a crash in the first 12 months as a probationary (P-plate) driver, when unsupervised driving was first allowed, than those who never drove before gaining their L-plates (Stevenson et al., 2001). As the authors note, however,
1 The minimum licence age increased to 16 years in 2011 but was 15 years when the NZDS cohort was recruited.
driving exposure during the P-plate crash period was not included in the multivariate analysis, a factor that may have influenced their results. In NZ, there is a belief that young people who have grown up in a rural area, and had the opportunity to drive vehicles around farm paddocks and on farm roads, are experienced (and presumably safer) drivers when they begin licensing, than their urban counterparts who have not had this experience. This argument is often put forward by the rural sector during discussions on minimum driver licence age, and has been used to try and stop the licence age from being raised (Begg and Langley, 2009). There is no direct evidence in NZ to support, or refute, this argument. It seems rather unlikely that this belief would only apply in New Zealand, but to our knowledge no one has examined pre-licensing off-road driving experience among young people so it is unknown what effect may have on crash risk among newly licensed drivers. The aim of the present study was to examine the relationship between both on-road and off-road pre-licensed driving experience reported by newly learner licensed car drivers, and the traffic crashes experienced as a driver during the learner licence stage, when only supervised driving is allowed, and the restricted licence stage, when unsupervised driving is first allowed. Motorcycle use was also included because it has been suggested, without any supporting evidence, that riding a motorcycle would be good preparation for young people about to start driving a car (Begg and Langley, 2009). In particular, we sought to determine the independent effects for each transport mode (car on-road and off-road, motorcycle on-road and off-road) in relation to a traffic crash as a car driver while on a learner licence, and while on a restricted licence, after controlling for a range of potential confounding socio-demographic, personality, and behavioural factors, including distance driven as a learner licensed driver and as a restricted licensed driver. 2. Method 2.1. New Zealand drivers study (NZDS) NZDS is a prospective cohort study of 3992 newly licensed car drivers in New Zealand. Fig. 1 shows the stages of the NZDS and how they are linked to the stages of the NZ graduated driver licensing system. The cohort was recruited, and the learner licence (baseline) questionnaire was completed, very soon after passing their learner licence test. This took place between 1st February 2006 and 31st January 2008, at various locations throughout New Zealand. Details of the recruitment procedures have been reported previously (Begg et al., 2012; Begg et al., 2009a,b; Langley et al., 2012a). The NZDS restricted licence (first follow-up) and full licence (second follow-up) telephone interviews took place very soon after the study participants passed the respective licence tests. 2.2. Data 2.2.1. Data extraction The present analysis used data from the learner (baseline), restricted, and full licence interviews. Data from the restricted licence interviews were extracted on 17 September, 2010 at which time 71% of participants had progressed to a restricted licence, 88% of them completed the NZDS restricted licence (first follow-up) interview. Data from the full licence interviews were extracted on 30 May 2011 at which time 40% of the full cohort had progressed to a full licence, and 93% of them completed the full licence (second follow-up) interview. For the present study our primary interest was young drivers so only those aged 15–24 years at learner licence (93% of the total cohort) were included (n = 2358 completed the
D.J. Begg et al. / Accident Analysis and Prevention 62 (2014) 153–160
155
obtained from the official New Zealand traffic crash reports (TCR) recorded by the New Zealand Police. The police only record crashes involving injury, so to provide a more complete crash coverage we also included self-reported crashes that were defined as a crash that occurred on a public road where someone was injured and/or there was vehicle or property damage. Crashes that occurred during the learner licence stage were self-reported at the restricted licence interview and restricted licence stage crashes were reported at the full licence interview. The combined (TCR and self-report) crash file was checked and duplicates removed. There were 124 crashes at the learner licence stage and 347 at the restricted licence stage. For the analysis the crash outcome was dichotomised (yes/no).
Fig. 1. Diagram showing the linkage between the interview stages of the New Zealand Drivers Study and the Graduated Driver Licensing System.
restricted licence interview and n = 1428 completed the full licence interview). 2.2.2. Pre-licensed driving experience The pre-licensed driving data, collected at the learner licence (i.e. baseline) interview, included a series of questions relating to driving experience before passing the learner licence test. If responded in the affirmative, the question: “Have you ever driven a car on or off-road?” was followed up with “How many times in total have you driven a car on-road (that is on a public road)?” “How many times in total have you driven a car off-road?” The response options were never, once, 2–5 times, 6–10 times, 11–20 times, >20 times. The same format was used for motorcycle use. For the analysis of the learner licence stage crashes, because of small numbers in some of these driving experience groups, the categories were collapsed to never, 1–20 times, >20 times for pre-licence car driving, and never or ≥1 times for motorcycle use. For the restricted licence stage crash analysis the driving experience responses were grouped into four categories based on the number of times driven: never, 1–5 times, 6–20 times, >20 times. 2.2.3. Motor vehicle traffic crashes Motor vehicle traffic crashes as a car driver (or any vehicle covered by a car driver’s licence such as a van, utility, light truck) were
2.2.4. Potential confounding factors A number of factors associated with pre-licensed driving behaviour were identified in an earlier study by Begg et al. (2012). These potential confounding factors have been included in the present analysis. A brief description of these factors is provided here with further details available elsewhere (Begg et al., 2012, 2009a,b; Langley et al., 2012a). Briefly, age at the time of obtaining a learner licence was calculated from each participants’ date-of-birth and the date their learner licence was issued. Ethnicity was selfidentified and to simplify interpretation of the results, anyone who ¯ (i.e. the indigenous people of New Zealand) self-identified as Maori ¯ ¯ was classified as Maori and the remainder as non-Maori. Residential location was determined using the “urban/rural profile” used by Statistics New Zealand (Statistics New Zealand, 2009). This classified the participants’ residential addresses as either urban or rural. Social deprivation was measured using the NZDep2006 score which combines nine variables from the 2006 census, which reflect eight dimensions of deprivation (Salmond et al., 2007). These variables are: income (2 variables), home ownership, support (e.g. single parent family), employment, qualifications, living space, communication (e.g. access to telephone), and transport (access to a car). The residential address of each participant was used to assign a deprivation score ranging from 1 to 10. For the multivariate analysis these levels were collapsed to create three levels of deprivation: low, medium and high (Adolescent Health Research Group, 2008). Alcohol use was measured using the first three questions of the AUDIT (the Audit-C) (Babor et al., 1989). The responses to these three questions were summed to give a score ranging from 0 to 12. Following the convention established by alcohol researchers (e.g. Bradley et al., 2007) a cut-off of ≥4 was used to indicate hazardous drinking for males and ≥3 for females. Cannabis or marijuana use was measured by the following question: “How often do you use marijuana or cannabis?” The response options were: Never, monthly or less, 2–4 times per month, 2–3 times per week, 4 or more times a week. Reponses were classified at three levels; “never”, “monthly” (monthly or less and 2–4 times per month), “weekly” (2–3 times per week, and 4+ times per week). The use of herbal highs was measured by asking “How often do you use Herbal Highs (party pills) such as “Rapture”, “Bliss”?” Response options were the same as for cannabis, but because of the small number of users these were categorised as yes (have used) or no (never). Personality was measured using the Zuckerman IMP-SS scale (Zuckerman et al., 1993). This gave a measure of impulsivity (8 items), sensation seeking (11 items), and aggression–hostility (17 items). Gender specific cut-off points were selected to classify each of the scales into three categories: low, middle, or high. The low group had scores in the lowest quartile, the high group had scores in the highest quartile, and the remainder were in the middle quartiles (approximately 50%). Time on the learner licence was included to account for the varying time spent on a learner licence. Unsupervised driving while on a learner licence has been shown to be significantly related to crash involvement as a learner licensed driver, and is also related to pre-licence
156
D.J. Begg et al. / Accident Analysis and Prevention 62 (2014) 153–160
driving (Langley et al., under consideration for publication). This was included (yes/no) as a potential confounder.
Table 1 Frequency distribution of pre-licensed driving experience categories by crash as a car driver at the learner and restricted licence stages. Pre-licence driving
2.2.5. Distance driven An estimate of total distance driven was obtained using a series of questions that have been shown to be a reliable method of obtaining such estimates (Lajunen and Özkan, 2011). These questions were administered by trained interviewers who “walked” the young driver through their driving experiences as a learner licensed driver (at the restricted licence interview) and as a restricted licensed driver (at the full licence interview). (a) To obtain an indication of how frequently they had driven each participant was asked “How often did you drive on a public road with or without a supervisor. Was it “daily”, or “at least weekly”, or “fortnightly”, or “monthly”. If they responded “yes” then they were asked. (a) How many times in a typical day (week, fortnight, month)? The interviewer knew the exact dates when the participant had passed their licence tests (obtained from the NZ driver licensing data base) so was able to accurately calculate the number of days (weeks, fortnights or months) from one test to the next. If they answered “No”, that is they had not driven at least monthly, they were asked “How many times all up” they had driven on their learner or restricted licence. (b) To get an indication of the distance they had driven they were asked “On a typical drive how far did you drive?” The distance options were < 11 kms, 11–20 kms, 21–50 kms, 51–100 kms, >100 kms, which for the analysis were recoded to 5, 15, 35, 75, 150kms. The responses to (a) and (b) were multiplied to provide an estimate of the total distance driven while on learner or restricted licence. As appropriate, this series of questions was asked for driving with a professional instructor, driving with a supervisor, or driving unsupervised. The totals for each of these categories were summed to provide an estimate of their overall total driving exposure (distance driven). Acknowledging the imprecision of this measure, for the analysis the total distance driven was recoded into three categories, low (first quartile), mid (second and third quartiles), high (top quartile). This categorisation ensured that those with high and low exposure were separated by the larger middle category, which should help to minimise misclassification bias at the high and low levels of driving exposure.
Learner licence stage Crash (n = 124) n
No Crash (n = 2222) %
n
%
44 46 30
37 38 25
1157 789 203
54 37 9
54 52 15
45 43 12
1187 784 184
55 37 8
Motorcycle on road Never Yes
102 20
83 17
1878 264
88 12
Motorcycle off road Never Yes
91 30
75 25
1591 550
74 26
Car on-road Never 1–20 times >20 times Car off-road Never 1–20 times >20 times
Restricted licence stage Crash (n = 347)
No crash (n = 1077)
n
%
n
%
Car on-road Never 1–5 times 6–20 times >20 times
182 98 30 33
53 29 9 10
583 293 110 73
55 28 10 7
Car off-road Never 1–5 times 6–20 times >20 times
185 96 31 32
54 28 9 9
556 298 107 100
52 28 10 9
Motorcycle on-road Never 1–5 times 6–20 times >20 times
289 29 9 16
84 8 3 5
929 72 25 32
88 7 2 3
Motorcycle off-road Never 1–5 times 6–20 times >20 times
231 56 16 39
68 16 5 11
761 109 57 129
72 10 5 12
2.3. Statistical analysis All statistical analyses were undertaken using SAS version 9.3 (SAS Institute Inc., 2010). First, the frequency distribution for each of the pre-licensed driving variables, by crash status at the learner and restricted licence stage, were examined. Second, separate Poisson regression models with robust error variance (Zou, 2004) were run to examine each of the four pre-licensed driving categories (i.e. car on-road and car off-road, motorcycle on-road and motorcycle off-road with (1) a learner licence driver crash and (2) a restricted licence driver crash as the outcome (unadjusted results). That is, eight separate analyses were undertaken. Each of these models was then re-run including all of the potential confounders (adjusted results).
3.2. Crashes as a learner licensed driver With crash as a learner licensed driver as the outcome, each of the pre-licence driving variables was entered into separate Poisson regression models (unadjusted results Table 2). Compared with someone who had not driven a car on-road before licensing, the unadjusted relative risk (RR) of a crash for someone who had driven a car >20 times on-road before licensing was 3.5. The adjusted results (Table 2) show that, after controlling for the potential confounders, the RR was 1.7 (95% CI 1.0–2.9). The adjusted results show that pre-licence car off-road driving and motorcycle use, both onroad and off-road were not found to have any effect on crash risk as a learner licensed driver.
3. Results
3.3. Crashes as a restricted licensed driver
3.1. Frequency distributions
The same procedure was followed for the analysis of the restricted licence crashes. The results for the pre-licence driving categories from these Poisson regression models are presented in Table 3. Overall there was no consistent relationship between any of the pre-licence driving categories and the restricted licence crashes.
The frequency distributions for the pre-licensed driving variables, by crash status, at the learner licence and restricted licence stages are presented in Table 1.
D.J. Begg et al. / Accident Analysis and Prevention 62 (2014) 153–160
157
Table 2 Results from separate Poisson regression analyses showing the unadjusted and adjusted relative risks (RR), 95% confidence intervals (CI) and p-values for an on-road car crash as a learner licensed driver by each of the pre-licence driving categories* . Pre-licence driving
RR
95% CI
p-Value
Unadjusted
RR
95% CI
p-Value
Adjusted
Car on-road Never 1–20 times >20 times
Reference 1.5 3.5
1.0–2.3 2.3–5.5
.05 .00
Reference 1.2 1.7
0.8–1.8 1.0–2.9
.44 .05
Car off-road Never 1–20 times >20 times
Reference 1.4 1.7
1.0–2.1 1.0–3.0
.06 .05
Reference 1.3 1.2
0.9–1.9 0.7–2.2
.23 .53
Motorcycle on-road Never Yes
Reference 1.4
0.9–2.2
.17
Reference 1.0
0.6–1.5
.86
Motorcycle off-road Never Yes
Reference 1.0
0.6–1.4
.83
Reference 0.8
0.5–1.3
.36
*
Adjusted for gender, age at learner licence, residential location, ethnicity, sensation seeking, impulsivity, aggression, alcohol use, cannabis use, herbal high use, time on learner licence, distance driven on learner licence, and unsupervised driving on a learner licence.
The only significant result was that those who had ridden a motorcycle off-road 1–5 times had a RR of 1.4 (95% CI 1.1–1.8) compared with those who had not ridden a motorcycle off-road before licensing.
such as gender, sensation seeking, alcohol use, drug use, so these factors do not explain this result.
4. Discussion
In the present study the risk of being in a crash at the learner licence stage was related to driving a car on-road before getting a learner licence, and especially for those who had driven a substantial number of times (more than 20). This applied to around 10% of the young learner licence drivers in this study. It is possible that having done this amount of on-road driving before licensing these young drivers were over-confident, a factor which is known to increase crash risk among young drivers (Matthews and Moran, 1986). They may also be less law abiding than the other young drivers, but to some extent this was adjusted for by including unsupervised driving in the analysis. Whether this increase in crash risk at the learner licence stage is due to over-confidence, breaches of
4.1. Main findings The results from this study showed that, overall, pre-licensed driving experience had relatively little influence on crash involvement at the learner or restricted licence stage, when distance driven and other factors associated with crash risk were taken into consideration. The only pre-licence driving experience that showed a consistent relationship with crash risk was on-road car driving which increased risk at the learner licence stage. This result had been adjusted for distance driven as well as a range of other factors
4.2. Learner licence crashes
Table 3 Results from separate Poisson regression analysis showing the unadjusted and adjusted relative risks (RR), 95% confidence intervals (CI) and p-values of an on-road car crash as a restricted licensed driver for each of the pre-licence driving categories. Pre-licensed driving
RR
95% CI
p-Value
RR
95% CI
p-Value
Adjusteda
Unadjusted Car on-road Never 1–5 times 6–20 times >20 times
Reference 1.1 0.9 1.3
0.9–1.3 0.6–1.3 1.0–1.8
.63 .55 .09
Reference 1.0 0.9 1.2
0.8–1.2 0.6–1.3 0.8–1.6
.98 .54 .50
Car off-road Never 1–5 times 6–20 times >20 times
Reference 1.0 0.9 1.0
0.8–1.2 0.6–1.3 0.7–1.3
.82 .54 .86
Reference 1.0 1.0 0.9
0.8–1.2 0.7–1.3 0.6–1.3
.68 .78 .50
Motorcycle on-road Never 1–5 times 6–20 times >20 times
Reference 1.2 1.1 1.4
0.9–1.7 0.6–2.0 0.9–2.1
.25 .71 .11
Reference 1.2 1.2 1.3
0.8–1.6 0.7–2.1 0.9–2.0
.34 .59 .20
Motorcycle off-road Never 1–5 times 6–20 times >20 times
Reference 1.5 0.9 1.0
1.1–1.9 0.6–1.5 0.7–1.3
.00 .79 .90
Reference 1.4 0.9 0.9
1.1–1.8 0.6–1.4 0.7–1.3
.00 .76 .63
a Adjusted for gender, age at learner licence, residential location, ethnicity, sensation seeking, impulsivity, aggression, alcohol use, cannabis use, herbal high use, time on learner licence, driving exposure and unsupervised driving on a learner licence, distance driven on restricted licence.
158
D.J. Begg et al. / Accident Analysis and Prevention 62 (2014) 153–160
the law, or some other reason, it could be countered by the presence of a responsible supervisor who is in charge of the vehicle. Although supervision is a legal requirement at the learner licence stage, it is known that many young people breach this condition (Langley et al., under consideration for publication) but we do not know if this was the case when these crashes occurred. Pre-licence on-road driving (even if supervised by an experienced licensed driver) is illegal in New Zealand and can result in penalties such as a 400 NZD fine, being forbidden to drive, or having the vehicle seized on the roadside and impounded for 28 days. These young people may be unaware of the potential consequences if they are caught engaging in this behaviour, so improving knowledge of the penalties may help deter them from driving on-road before licensing. Also, improving the level of enforcement may help deter this behaviour. In this study, around half of the newly licensed drivers had driven on-road before licensing but only 5% reported that they had ever been stopped by the police when doing so. These figures suggest that there is scope for enforcement to be improved. 4.3. Restricted licence crashes Pre-licence driving behaviour showed no consistent relationship with restricted licence crashes. At the restricted licence stage the young driver is permitted to legally drive unsupervised, and this is the period when crash risk is at its highest (Begg and Langley, 2009; Lewis-Evans, 2010; Mayhew et al., 2003). It is possible, therefore, that factors other than pre-licence driving experience, such as poor hazard recognition, driving at excessive speed, following too closely, or dangerous overtaking, are the more important crash risk factors at this stage. Also, before they reached the restricted licence stage these young drivers had to have spent at least six months on a learner licence. The experience they gained during this time may have negated any effect that pre-licence driving had on restricted licence crashes. 4.4. Comparison with Western Australian study The findings from this study do not fully support those from the Western Australian study by Stevenson et al. (2001) who found that on road pre-licensed car driving was associated with crashes in the first twelve months of probationary (P-plate) licensed driving, which is the equivalent to the restricted licence stage in NZ, when unsupervised driving is first allowed. In the present study pre-licence car driving, either on or off-road, was not found to be associated with crashes at the restricted licence stage but on-road car driving more than twenty times was associated with a crash at the learner licence stage. This effect was strong (RR 3.5) in the unadjusted analysis (Table 2), when distance driven (i.e. driving exposure) was not included in the analysis. The WA study was unable to include a measure of driving exposure in the initial analysis but when this was obtained at a follow-up interview, post hoc analysis showed a strong association between driving exposure and pre-licence driving. The authors suggested that, without a driving exposure measure, the frequency of driving prior to L-plates could be interpreted as an adjustment for driving exposure (Stevenson et al., 2001). 4.5. Methodological considerations The NZDS is a multistage, comprehensive prospective cohort study of a large number of newly licensed car drivers especially designed to provide scientific evidence on a wide range of issues affecting young drivers. The prospective nature of the study ensures that the explanatory factors, such as attitudes and behaviours, were measured before the outcomes (e.g. crashes) had occurred, thereby ensuring temporality. In the present study our measure of crashes
was a combination of self-reported and police traffic crash records. This measure was chosen because, in NZ, only crashes involving injury are required to be recorded in the police traffic crash records and, despite this being a legal requirement, it is known that even crashes involving hospitalised injury are seriously underreported (Alsop and Langley, 2001). The under-reporting of crashes is not unique to NZ. Stevenson and Palamara used police reported crashes in the WA study and considered the under-reporting of crash outcomes to be the most “noticeable” limitation in their study (Stevenson et al., 2001). By including self-reported crashes, we were able to capture injury crashes that had not been recorded by the police, as well as crashes that did not involve injury but did involve property damage, and would otherwise have been missed. The NZDS cohort is not a randomly selected representative sample of all newly licensed drivers in NZ, therefore, prevalence estimates presented in this study may not apply to all young newly licensed drivers in New Zealand, or elsewhere. Attempts were made during recruitment to ensure heterogeneity of the cohort by recruiting in the main geographic regions of the country, and among a wide range of ethnic groups. Recruitment was carried out by local people in each region, and mostly face-to-face, so that people who are often reluctant to participate in such studies were more likely to be recruited. Our recruitment rates of around 80–90% attest to the success of this method (Langley et al., 2012a). The NZDS is an on-going prospective cohort study which is constantly changing as study members progress, at their own pace, through the graduated driver licensing system. At present the NZ GDLS does not have maximum time limits at any licence stage,2 the time spent on any licence can vary considerably. To conduct the analysis for this study it was necessary to set a data extraction date for the follow-up interviews. At that date, those included in the analysis of crashes at the learner licence stage had to have progressed to a restricted licence and undertaken the NZDS restricted interview. Of the 3693 15–24 years olds in the NZDS, 73% had gone on and passed their restricted licence test and 88% of them had completed the NZDS first follow-up interview. Similarly for the analysis at the restricted licence stage, 42% had progressed to a full licence and 93% of them completed the second NZDS follow-up interview. These figures show that progression through to the full graduated licensing system is somewhat tardy, but for those who had progressed the NZDS follow-up rates of 88–93% were extremely high. The lack of progression through the licensing stages of the GDLS can have implications for studies based on the different licence stages. A NZDS paper comparing those who had, and had not, progressed from a learner to restricted licence, after being eligible to do so for at least two years, has been published (Langley et al., 2012b). The “non-progressors” were shown to have a different socio-demographic and behavioural profile to the “progressors”, they drove less, were less likely to be pre-licence car drivers, and they were at reduced risk of being a traffic offender. To take full advantage of the available data, for the present study the analysis of the crashes that occurred at the learner licence stage were based on all those who had progressed to a restricted licence and completed the NZDS restricted licence interview, and the restricted licence stage crash analysis was based on those who had continued on to a full licence and completed the NZDS full licence interview. This meant that the restricted licence crash analysis was based on a subset (n = 1428) of those in the learner licence crash analysis (n = 2358). A comparison of the 1428 who had progressed to a full licence with the 930 who were still on a restricted licence showed that those who had progressed to a full licence were more likely to be male, aged 15 at learner licence, to live in an area of
2 The NZ Government announced in January 2013 that five-year time limits will be introduced at the learner and restricted licence stages.
D.J. Begg et al. / Accident Analysis and Prevention 62 (2014) 153–160
¯ low deprivation, not of Maori ethnicity, not high alcohol or cannabis users but were herbal high users. They were more likely to have driven a car off-road before licensing, but did not differ for any of the other pre-licence driving behaviours. Perhaps not surprisingly, they were also less likely to have spent more than 12 months on their learner licence. To assess the effect these differences may have had on the learner licence stage crash results, a post hoc multivariate analysis was run with only those who had completed the full licence interview, and were therefore the same people as were included in the restricted licence stage crash analysis. These results showed that driving a car both on-road and off-road increased the risk of a learner licence crash. This suggests that those who had progressed to a full licence may have been practising driving before they began licensing but this practice increased rather than decreased their crash risk, once they became licensed, even after controlling for risk-taking (impulsivity, sensation seeking, aggression) distance driven, unsupervised driving, and a range of socio-demographic and other behavioural variables. Virtually all of the data used in this study was self-reported which has the potential to be biased, particularly in the form of social desirability bias. This is especially the case for questions about illegal or dangerous behaviours, which applied to some of the variables in this study. Research has shown, however, that social desirability on self-reported risky behaviours is not necessarily substantial (Lajunen and Summala, 2003; Sullman and Taylor, 2010) and the frequency with which some illegal behaviours (e.g. prelicence on-road car driving, cannabis use) were reported in this study suggest it may not have been a substantial problem. Inaccurate recall is also a perceived weakness of self-reported behaviour and is considered by some to be less desirable than official records for events such as crashes. However, a study validating the accuracy of recall of crash events by young people showed that young people can recall crash details with a high degree of accuracy, even several years after the event (Begg et al., 1999). Also, the Australian DRIVE study, using a very similar study design to the NZDS, found a high level of accuracy in young drivers’ self-reports of police recorded crashes and convictions (Boufous et al., 2010). Validating explanatory data, such as pre-licence driving behaviour, is extremely difficult, if not impossible, as there is no other record of these behaviours with which the self-reported data can be compared. Another factor that may have influenced the results was the categorisation of the explanatory variables which was based on frequencies and therefore the cut-off points were somewhat arbitrary. Where the numbers allowed, the categories were chosen so as to provide a low, moderate and high exposure group, that were compared with a non-exposed group. One particular challenge facing researchers undertaking studies such as the NZDS is obtaining reliable and valid driving exposure data. Some of the newer technologies, such as in-vehicle recording, are limited to smaller studies (Williams et al., 2012) and not practical for a large study such as this. For this study our estimates of driving exposure were obtained by “walking” the participant through a series of questions to get an estimate of driving frequency and time and distance for a typical trip. The interviewers knew exactly how long the driver had been on a particular licence so this part of the equation was not estimated or based on recall. The final total distance was categorised so that the lowest and highest quartiles represented the low and high exposure groups, with the remaining 50% in the middle group. This categorisation was used to minimise misclassification at the high and low levels of exposure.
5. Conclusion The findings from this research suggest that, on the whole, prelicensed driving did not have very much effect on crash risk among
159
learner or restricted licensed drivers, although frequent on-road car driving before licensing may have increased crash risk at the learner licence stage. The popular belief that pre-licensed driving experience makes young drivers more competent and better drivers needs to be dispelled, and young people should be discouraged from the illegal behaviour of driving a car on-road before licensing. Acknowledgments The New Zealand Drivers Study is funded by the Health Research Council of New Zealand (2005–2013), the Road Safety Trust (2005–2013) and was funded by the Accident Compensation Corporation (2005–2009). No funding source had any involvement in the preparation of this paper. The authors wish to acknowledge Dr. Ari Samaranayaka, Injury Prevention Research Unit, University of Otago, for comments on an earlier version of this paper. References Adolescent Health Research Group, 2008. Youth’07: The Health and Wellbeing of Secondary School Students in New Zealand. Technical Report. The University of Auckland, Auckland, pp. 2008. Alsop, J., Langley, J.D., 2001. Under-reporting of motor vehicle traffic crash victims in New Zealand. Accident Analysis and Prevention 33, 353–359. Australian Government, 2012. Living in Australia: Applying for a Driver’s Licence, http://www.immi.gov.au/living-in-australia/settle-in-australia/to-do-first/ apply-drivers.htm (retrieved 30.03.12). Babor, T.F., de la Fuente, J.R., Saunders, J.B., Grant, M., 1989. AUDIT: The Alcohol Use Disorders Identification Test, Guidelines for Use in Primary Health Care. World Health Organisation, Geneva. Begg, D., Langley, J., Williams, S., 1999. Validity of self-reported crashes and injuries in a longitudinal study of young adults. Injury Prevention 5, 142–144. Begg, D., Sullman, M., Samaranayaka, A., 2012. The characteristics of young prelicensed drivers: evidence from the New Zealand Drivers Study. Accident Analysis and Prevention 45, 539–546. Begg, D.J., Langley, J.D., 2009. A critical examination of the arguments against raising the car driver licensing age in New Zealand. Traffic Injury Prevention 10, 1–8. Begg, D.J., Langley, J.D., Brookland, R.L., Ameratunga, S., Broughton, J.R., 2009a. The opinions of newly licensed drivers in New Zealand on the minimum car driver licensing age and reasons for getting a licence. New Zealand Medical Journal 122 http://nzma.org.nz/journal/122-1306/3880/ Begg, D.J., Langley, J.D., Broughton, J.R., Brookland, R.L., Ameratunga, S., McDowell, A.J., 2009b. New Zealand Drivers Study: a follow-up study of newly licensed drivers. Injury Prevention 15, 281, http://dx.doi.org/ 10.1136/ip. 2009.021998a.1-9. Boufous, S., Ivers, R., Senserrick, T., 2010. Accuracy of self-report of on-road crashes and traffic offences in a cohort of young drivers: the DRIVE study. Injury Prevention 16, 275–277. Bradley, K., DeBenedetti, A., Volk, R., Williams, E., Frank, D., Kivlahan, D., 2007. AUDIT-C as a brief screen for alcohol misuse in primary care. Alcoholism Clinical and Experimental Research 7, 1207–1217. Department for Transport, 2012. Learners and New Drivers and Riders, http://www.direct.gov.uk/en/Motoring/LearnerAndNewDrivers/index.htm (retrieved 30.03.12). Elliott, M., Ginsburg, K., Winston, F., 2008. Unlicensed teenaged drivers: who are they, and how do they behave when they are behind the wheel? Pediatrices, Offical Journal of the American Academy of Pediatrics 122, 994–1000. European Commission, 2012. Road Safety: Driving Licence, http://ec.europa.eu/transport/road safety/behavior/driving licence en.htm (retrieved 07.02.13). Hanna, C.L., Hasselberg, M., Laflamme, L., Moller, J., 2010. Road traffic crash circumstances and consequences among young unlicensed drivers: A Swedish cohort study of socioeconomic disparities. BMC Public Health 10, 8. Hasselberg, M., Laflamme, L., 2009. How do car crashes happen among young drivers aged 18–20 years? Typical circumstances in relation to license status, alcohol impairment and injury consequences. Accident Analysis and Prevention 41, 734–738. Huang, P., Winston, F., 2011. Young Drivers. Handbook of Traffic Psychology. Academic Press, Norfolk, VA USA, pp. 315–338. Insurance Institute for Highway Safety, 2012. Summary Table: US Young Driver Licensing Systems, http://www.ihhs.org/laws/GraduatedLicenseCompare.aspx (retrieved 26.03.12). Lajunen, T., Özkan, T. (Eds.), 2011. Self-Report Instruments and Methods. Handbook of Traffic Psychology. Academic Press, Waltham, MA. Lajunen, T., Summala, H., 2003. Can we trust self-reports of driving? Effects of impression management on driver behaviour questionnaire responses. Transportation Research Part F: Traffic Psychology and Behaviour 6, 97–107. Lam, L., 2003. A neglected risky behavior among children and adolescents: Underage driving and injury in New South Wales, Australia. Journal of Safety Research 34, 315–320.
160
D.J. Begg et al. / Accident Analysis and Prevention 62 (2014) 153–160
Langley, J., Begg, D., Brookland, R., Ameratunga, S., McDowell, A., Broughton, J., 2012a. Establishment of the New Zealand Drivers Study. The New Zealand Medical Journal (Online) 125, 98–112. Langley, J., Begg, D., Brookland, R., Samaranayaka, A., Jordan, H., Davie, G., 2012b. Nonprogression through graduated driver licensing: characteristics, traffic offending, and reasons for nonprogression. Traffic Injury Prevention 13, 7–13. Langley, J., Begg, D., Samaranayaka, A., Brookland, R., Weiss, H., 2013. Unsupervised driving by learner licence holders: associated characteristics and crash risk. Accidental Analysis and Prevention 60, 24–30. Lewis-Evans, B., 2010. Crash involvement during the different phases of the New Zealand Graduated Driver Licensing System (GDLS). Journal of Safety Research 41, 359–365. Matthews, M.L., Moran, A.R., 1986. Age differences in male drivers’ perception of accident risk: the role of perceived driving ability. Accident, Analysis and Prevention 18, 299–313. Mayhew, D., Simpson, H., Pak, A., 2003. Changes in collision rates among novice drivers during the first months of driving. Accident Analysis and Prevention 35, 683–691. McCartt, A.T., Shabanova, V.I., Leaf, W.A., 2003. Driving experiences, crashes and traffic citations of teenage beginning drivers. Accident Analysis and Prevention 35, 311–320. McDowell, A.J., Begg, D.J., Connor, J., Broughton, J., 2009. Unlicensed driving among ¯ urban and rural Maori drivers: New Zealand Drivers Study. Traffic Injury Prevention 10, 538–545. Ministry of Transport, 2011. Safer Journeys Action Plan 2011–2012. Ministry of Transport, Wellington. New Zealand Transport Agency, 2012. The Graduated Driver Licensing System (GDLS) has Changed, http://www.nzta.govt.nz/licence/ photo/gdls-changing.html (retrieved 27.09.12). Organisation for Economic Co-Operation and Development, 2006. Young Drivers: The Road to Safety. Policy Brief. Organisation for Economic Co-operation and Development and the European Conference of Ministers of Transport (ECMT) Joint Transport Centre, Paris.
Salmond, C., Crampton, P., Atkinson, J., 2007, August. NZDep2006 Index of Deprivation. Department of Public Health, University of Otago, Wellington. SAS Institute Inc., 2010. Language Reference: Concepts, Second ed. SAS Institute Inc., Cary, NC. Senserrick, T., Chen, H., Boufous, S., Ivers, R., Stevenson, M., Norton, R., 2010. Demographic Factors Associated with Pre-licensed Driving in NSW Young Driver Cohort: The DRIVE Study. Australasian Road Safety Research Policing Education Conference, Canberra. Stevenson, M., Palamara, P., Morrison, D., Ryan, G., 2001. Behavioral factors as predictors of motor vehicle crashes in young drivers. Traffic Injury Prevention 2, 247–254. Statistics New Zealand, 2009. New Zealand: An Urban/Rural Profile Wellington2008, Available from: http://www.stats.govt.nz/ Update. statistics-by-area/urban-rural-profiles/default.htm (accessed 09.02.09). Sullman, M.J.M., Taylor, J.E., 2010. Social desirability and self-reported driving behaviours: Should we be worried? Transportation Research Part F: Traffic Psychology and Behaviour 13, 215–221. Vicroads, 2012. Licences: Get your P’s. http://www.vicroads.vic.gov.au/Home/Licences/ GetYourPs/ Williams, A., 2003. Teenage drivers: patterns of risk. Journal of Safety Research 34, 5–15. Williams, A., 2006. Young driver risk factors: successful and unsuccessful approaches for dealing with them and an agenda for the future. Injury Prevention 12, 4–8. Williams, A., Shults, R., 2010. Graduated driver licensing research, 2007-present: a review and commentary. Journal of Safety Research 41, 77–84. Williams, A.F., Tefft, B.C., Grabowski, J.G., 2012. Graduated driver licensing research, 2010-present. Journal of Safety Research 43, 195–203. Zou, G., 2004. A modified Poisson regression approach to prospective studies with binary data. American Journal of Epidemiology 159, 702–706. Zuckerman, M., Kulman, D.M., Joireman, J., Teta, P., Kraft, M., 1993. A comparison of three structural models for personality: the big three, the big five, and the alternative five. Journal of Personality and Social Psychology 65, 757–768.