Abstract
A lack of Hazard Management (HM) and Speed Management (SM) skills have been identified as a leading factor in young novice drivers’ road crash involvement. Despite the success of individual training programmes targeting these two skills, their generalizability to one another remains unclear. Hence, the aim of the present research was to address this limitation. Ninety young novice drivers were randomly divided into five different training groups, and following training, their HM and SM skills were assessed on two different occasions in a fixed-based driving simulator. The result revealed that HM and SM training improved their targeted skill; however, no generalization of training was evident. These findings emphasise the need for a new training technique that could improve these two critical road safety skills among young novice drivers.
Introduction
Young novice drivers under 25 years with less than three years of unsupervised driving experience are overrepresented in road crashes. According to Transport for New South Wales (TfNSW) in Australia, young drivers constitute 14 per cent of all license holders, yet are involved in approximately a quarter of fatal road crashes in Australia (Transport for NSW, 2025). Deficiencies in skills to manage hazards and speed are identified as leading factors in these crashes (Clarke et al., 2006; Horswill et al., 2015).
Road hazards are any feature/s or event/s that can cause harm (Barragán et al., 2021; Benda & Hoyos, 1983; Castro et al., 2021; Haworth et al., 2000). Road hazards are typically divided into three categories based on their predictive ability: behavioral prediction (BP; a potential threat is visible to drivers), environmental prediction (EP; a potential threat obscured by other road users or features), and divided focus and attention (DF; multiple potential hazards requiring drivers to divide their focus and attention to anticipate the actual threats; (see Crundall et al., 2010).
Speed is also considered a road hazard and is reported to be one of the leading causes of road crashes (Transport for NSW, 2023). Speeding behaviors, like exceeding the speed limit or travelling at an inappropriate speed for the road conditions, reduce drivers’ chances of detecting possible threats and the effectiveness of safety measures, such as braking (van Leeuwen et al., 2015; World Health Organization, 2023). Travelling too fast also increases the severity of the impact and associated injuries (Organization for Economic Cooperation and Development, 2006).
Several studies have outlined a deficiency in Hazard Management (HM) and Speed Management (SM) skills among young novice drivers. Moreover, Borowsky and Oron-Gilad (2013) found a lack of HM skills in these drivers led to them ineffectively managing hazards, particularly when the hazard was obscured by environmental features, such as a car suddenly pulling out from a large truck (i.e., EP hazard), and cutting-off the driver’s car. Similarly, Molloy and colleagues found that young novice drivers travel at speeds too fast for the road conditions compared to experienced drivers, and this effect was most pronounced at lower speeds, such as 50 and 60km/hr, as opposed to 70 and 80 km/hr (Molloy, Molesworth, & Li, 2023).
The acquisition of both HM and SM skills can take years to develop. However, research (Deery, 1999; LaVoie et al., 2018) highlights training as central to short-cutting this experiential process for young novice drivers. Several techniques have proven effective in improving these skills. For example, Commentary training (Horswill et al., 2022; Isler et al., 2011; McKenna et al., 2006; Zhang et al., 2022), which encourages the driver to look for possible hazards, and feedback pertaining to the drivers’ errors (Roberts et al., 2021; Vlakveld et al., 2011) showed potential to improve HM skills. For SM skills, feedback about a driver’s speeding behavior, along with information about the financial penalties and safety implications (referred to as combined feedback), has been shown to positively influence young novice drivers’ SM skills (Molloy, Molesworth, Williamson, et al., 2023).
While training targeting HM and SM skills individually has yielded positive results, how skills acquired from such targeted training generalize beyond the training remains unknown. Therefore, the aim of the present study is to investigate whether training interventions targeting one skill set, such as HM training, generalizes to SM skills, and vice versa.
Method
Participants
Ninety participants (24 females) voluntarily participated in the research. Eligible participants were between 18 and 25 years (M = 19.60, SD = 1.59) and held a provisional driver’s license (allowing drivers to drive without supervision) (M = 1.26 years, SD = .90). The sample size was deemed sufficient to detect between-group effects using Cohen’s criteria. Power calculations demonstrated that this sample size was sufficient to reveal a medium to large effect size (effect size = 0.75, α = 0.05, actual power = 0.8, groups = 5, N = 90). This study was approved by the University of New South Wales (UNSW) Ethics Panel (HC230245). All participants who completed the study were reimbursed for their time with a $50 gift voucher.
Design
The experimental design consisted of a 5 x 3 mixed repeated measure design. The between-groups independent variable was ‘training’, containing five levels: Control, Implicit Learning, Combined Feedback, HM Training, and HM Training with HM Feedback. The repeated measures variable was ‘testing occasion’, containing three levels: baseline, immediate post-test, and one week post-test drives. Two dependent variables were employed to measure HM and SM skills, namely, the percentage of travelled distance with EP hazards, and maximum speed in the 60 km/h speed zone. The former demonstrated the distance travelled before the brake is applied (after the release of the accelerator) when the hazard is visually evident, while the latter demonstrated the highest magnitude of speed in a 60 km/h speed zone.
Apparatus and materials
This study employed a fixed-based driving simulator consisting of three 27-inch wide flat-screen displays providing a 150-degree field of view, a Logitech G29 Driving Force Racing Wheel with pedals (accelerator and brake pedals), and a driver’s seat sourced from a Mazda 626. In addition, a Dell model AY410 speaker generated the simulated sound of vehicle engines, brakes, and road noise. The simulator software was the UC-win/Road driving simulation software version 15.1.4 (64-bit) developed by FORUM 8 (https://www.forum8.com).
Two questionnaires were used in the study. A demographic questionnaire (i.e., personal information and driving history) and a motion and simulator sickness checklist questionnaire (i.e., examine drivers’ proneness and symptoms of simulator sickness).
Procedure
Participants were recruited through communication platforms at UNSW Sydney. Eligible individuals were informed that the study would consist of two sessions, one week apart. Before the first session, participants were asked to complete a demographic questionnaire online via the UNSW Qualtrics platform.
In Session 1, each participant was randomly assigned to one of five groups. The Simulator Sickness Checklist was administered before, during, and after the study sessions to ensure participants’ readiness and related simulator sickness. Following this, each participant completed a 5 km practice drive to familiarize themselves with the simulator, followed by a 21 km baseline drive. Based on the New South Wales speed zone standards, the test routes included three distinct speed zones (50, 60, and 80 km/h). For this study, only driving behavior in the 60 km/h zone was examined. In addition, nine on-road hazards were systematically located along the route. These included three BP hazards, three EP hazards, and three DF hazards (see Crundall et al., 2010). For this study, only EP hazards were examined. EP hazards are not directly visible to the driver as environmental features obscure them and require anticipation based on contextual cues (Crundall et al., 2010). After the baseline drive, participants received their respective training: (1) Control (Filler video and Exposure no hazard), (2) Implicit Training (Filler video and Exposure to hazards), (3) Combined Feedback (Filler video, Exposure to hazards and Receive speed feedback: numbers of times exceeded speed limits, maximum speed, speeding penalties, and safety implications), (4) HM Training (Training video, and Commentary Training), and (5) HM Training with HM Feedback (Training video, Commentary Training followed by HM Feedback). Directly following training, participants completed a second 21 km drive (immediate post-test drive).
One week later, participants returned for a follow-up session. In this session, they completed a 5 km practice drive followed by a 21 km (one week post-test drive) test drive. While the structure and length of the three 21 km drives were consistent, surface features, such as the location and appearance of hazards, were altered to reduce route memorization effects.
Result
Analysis
Both data related to HM and SM skills were analysed using IBM SPSS version 29. A 5x3 mixed-repeated measure analysis of variance (ANOVA) was conducted, followed by post-hoc tests (Tukey HSD (Honest Significant Difference)) in the case where statistical significances were identified. With all analyses, α was set at .05 and adjusted to maintain the familywise error (stated when adjusted). Prior to all analyses, violations of homogeneity were examined and adjusted when breaches occurred (stated if occurred).
Percentage of travelled distance before braking with EP hazards
A mixed repeated measures analysis revealed a statistically significant Time by Group interaction, F (8, 170) = 7.23, p < .001, η2 = .25. A main effect for Time and Group was also revealed. Since the main effects for both Time and Group are encompassed in the interaction, only this was analysed. Post hoc analysis (One-way ANOVA) examining differences in travelled distance before braking with EP hazard failed to show a significant difference between groups at the baseline. However, there was a significant difference between groups in the immediate and one week post-test drives.
In the Immediate post-test drive, the analysis indicated a significant difference between groups for travelled distance before braking, F (4,85) =20.53, p < 0.001, η2 = .49 (see Figure 1). The percentage of travelled distance before braking with EP hazards in the HM Training with HM Feedback group was significantly lower than all other groups, (the smallest t, t (34) = 2.89, p < .001, Cohen’s d = .96, was between the HM Training and HM Training with HM Feedback groups). In addition, participants in the HM Training group responded to the hazards earlier than the Control and Combined Feedback groups (the smallest t, t (34) = 3.10, p = .004, Cohen’s d = 1.01, was between the Combined Feedback and HM Training groups).
Percentage of travelled distance before braking with EP hazards.
Similarly, the result revealed significant differences between groups at the one week post-test drive for percentage of travelled distance before braking (F (4,85) = 11.93, p < 0.001, η2 =.36). Tukey post hoc comparison reveals that drivers in the HM Training with HM Feedback group responded to the hazards earlier than participants in any other group (the smallest t, t (34) = 3.26, p = .003, Cohen’s d = 1.09, was between HM Training and HM Training with HM Feedback groups). Furthermore, participants in the HM Training group had a significantly lower percentage of travelled distance before braking than participants in the Control group, t (34) = 2.59, p = .014, Cohen’s d = .86.
Maximum speed in the 60km/h speed zone
A mixed repeated measures analysis identified a statistically significant Time by Group interaction for maximum speed in the 60km/h zone, F (8, 170) = 5.74, p < .001, η2 = .21. A main effect for Group was also present, however, not for Time. Since the interaction incorporates the main effect of Group, only this was analysed. Post hoc analysis (One-way ANOVA) failed to reveal a significant difference between groups at the baseline. This test indicated that the random allocation of participants to each group was successful.
The result in the immediate post-test drive showed significant differences between groups for maximum speed in the 60km/h zone, F (4,85) = 5.66, p < 0.001, η2 = .21 (see Figure 2). Tukey HSD post hoc revealed that maximum speed in the Combined Feedback group was significantly lower than all other groups (the smallest t, t (17.35) = 4.04, p < .001, Cohen’s d = 1.35, was between the Combined Feedback and HM Training groups).
Maximum speed during each testing occasion in the 60km/h speed zone.
At one week post-test drive, there was also a significant difference between groups based on maximum speed in the 60km/h zone, F (4,85) = 6.05, p < 0.001, η2 = .22. The post hoc indicated that trained drivers in the Combined Feedback group had lower maximum speed than drivers in any other group (the smallest t, t (18.05) = 3.57, p = .002, Cohen’s d = 1.19, was between the Combined Feedback and Control groups). No other results revealed a statistically significant difference.
Discussion
The main aim of this study was to investigate the effect of Hazard Management skill training on Speed Management skills, and vice versa, with young novice drivers. The results revealed no generalization of training. The results did illustrate the efficacy of both HM training and SM training in improving their targeted skill. Moreover, HM training was effective in reducing the percentage of travelled distance before braking when negotiating EP hazards across testing occasions. Drivers trained in HM training responded earlier than untrained drivers, a result consistent with previous research (e.g., McKenna et al., 2006; Zhang et al., 2022). Similarly, trained drivers in combined feedback group exhibited better SM skills than untrained drivers. These trained drivers had a lower maximum speed in the 60km/h speed zone than the other groups across testing occasions. These findings are also in line with the previous studies illustrating the benefit of combined feedback (SM training) (e.g., Molloy et al., 2018a, 2018b; Molloy, Molesworth, Williamson, et al., 2023).
Limitations and future research
While the results of this research are positive, they are not without their limitations. The data collected was from a simulated driving environment, which may not fully replicate real-world driving conditions. In addition, the study measured outcomes at two occasions (i.e., immediate post-training and one week post-training), offering limited insight into the long-term retention and generalization of skills. Future research should be conducted in real-world driving settings with extended follow-up periods. Given the lack of generalization of the training effect between the two targeted skills, future research should also investigate alternative training approaches that may yield a different outcome. Such training might include the combination of existing SM and HM training techniques employed in the current research (training groups 3 and 5). Combined feedback capitalizes on reinforcement theories (Molloy, Molesworth, Williamson, et al., 2023), while HM Commentary training with Feedback capitalizes on both information processing and situational awareness theories (Crundall et al., 2010).
Future research should also look at expanding the sample. The current study recruited university students. How university students differ from other young drivers remains unknown, and hence, another area for research.
Conclusion
These findings highlight the benefits of Hazard Management and Speed Management training in improving their respective target skills. These improvements suggest that such training can accelerate the development of HM and SM skills, which typically require years to reach the level of experienced drivers. The results also indicate that the application of these skills is limited to the skill itself and does not generalize to other skills.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research was made possible through the support of the Australian Research Council (Discovery Project # DP220103562).