Editorial

1. Introduction

In 2006 it was the 50th anniversary of the Institute of Advanced Motorists (IAM) and 2007 marks the 50th anniversary of the journal Ergonomics. The journal has a history of publishing research on driving and driver behaviour (for example, 103 papers have been published on this topic in the past 10 years and there was a special issue on Risky Decision Making in Transport Operation in volume 31 (number 4) in 1988, edited by Ivan Brown and Wiel Janssen). Recent papers on driving in Ergonomics have included research into performance effects of feedback (Ho et al. 2006, Walker et al. 2006), vehicle dynamics (Harris et al. 2005), vehicle automation (Stanton and Young 2005), telematics (Hampton and Langham 2005), seat design (Makhsous et al. 2005, Durkin et al. 2006), motion sickness (Lin et al. 2005), age and performance (Dukic et al. 2006) and traffic sign design (Jamson et al. 2005). This range of topics is indicative of the breath of the discipline (Karwowski 2005).

In celebration of both of the IAM and Ergonomics 50th anniversaries, a special issue of Ergonomics was commissioned on Driver Safety. The spirit of the special issue was to encourage researchers to submit papers that can help make driving safer, by informing vehicle and road environment design, law enforcement strategies, driver education programmes and road transportation policy. The special issue began with a Driver Safety conference run jointly between Brunel University and the IAM. The IAM has 112 000 members and was established in 1956 as the UK's leading road safety organization dedicated to raising driving and riding (further details may be found on their website – www.iam.org.uk). In summary, the conference presented a wide range of topics related to driver safety, including crash asymmetry, road risk mapping, driver education and capability, driving skills, visual search and attention, speeding and headways. One of the big take-home messages from the conference was that driver training, education and coaching has a big part to play in improving safety on the roads. To give this some context, it has been estimated that worldwide there are 1.2 million fatalities and as many as 50 million injuries in road accidents each year (World Health Organization 2004). The road accident statistics for Australia, the UK and the USA (to reflect the nations of authors' papers contributing to this special issue) are shown in table 1.

Table 1. Accident statistics for 2005.

	Australia	UK	USA
Deaths	1627	3201	39 189
Population	20.4 million	60.1 million	296.5 million
Death rate∗	7.97	5.33	13.22
∗Numbers of deaths per 100 000 of the population.

The accident statistics show that large numbers of people die on the roads throughout the world on an annual basis, making road and driver safety a high priority for any government. The data in table 1 are for deaths only, the figures for injuries are many times higher, as illustrated in the data from WHO.

Five main themes emerge from the 12 papers in the special issue, namely: driving violations; driver education and training; design of warnings; vehicle automation; and the effects of other tasks on driving. Seven papers were from the UK, four were from the USA and one was from Australia.

2 Driving violations

Three papers address driving violations. Research by Glendon revealed that tailgating (i.e. following a leading vehicle too closely) and speeding (i.e. exceeding the posted speed limit) were the two most frequently observed driving violations in a observational study on motorways. Tailgating is the subject of the paper by Brackstone and McDonald, who show that drivers generally follow too closely. Speeding is the subject of the paper by Stradling, who argues that speed choice is determined by a complex interaction between opportunity, obligations and inclination.

Glendon (Driving violations observed: An Australian study) reports that previous research has shown a positive relationship between driving violations and crash involvement. His study planned to investigate the preponderance of violations on roads, using a direct observation methodology. To do this, a road vehicle was equipped with a small video camera and the car was driven on motorways (cf. freeways) in three states of Australia, over a 10 d period. Data were recorded on almost 3000 cases in 40 h of video tape. Glendon developed a violation taxonomy comprising speeding, phone use, lane position, obstructed view, tailgating, vehicle condition, overloading, pollution and others. By far the greatest frequency of violations were tailgating (23% of the sample) and speeding (45% of the sample). All of the other violation categories accounted for only 12% of the sample. Of the drivers observed in the study, 60% committed some form of violation. No gender differences were observed, but younger drivers were more likely to be observed committing a driving violation. Glendon notes that there are some methodological drawbacks to the video-based data collection, but argues that it is a cost-effective way to gather a large dataset. He also suggests that the video data could be used as part of a driver education system, showing examples of driving violations in order to promote the correct behaviour.

Brackstone and McDonald (Driver headway: How close is too close on a motorway?) use an instrumented vehicle to gain insights into driver headway on motorways. Sensors at the front of the vehicle detect headways for drivers of the host vehicle, whereas sensors at the rear of the vehicle enable the researchers to sample headways used by other road users. The latter approach enables data to be collected on a much larger sample of drivers. Brackstone and McDonald's studies show that, above speeds of 50 mph, drivers are generally following too closely (when compared to distances recommended by the UK Highway Code). In comparison with headways in France and Germany, France had the shortest and the UK had the longest. Brackstone and McDonald suggest that the driver choice of the headway is based on very subtle cues, more than just the gap between the two vehicles. Drivers read the road ahead and consider the behaviour of the vehicles in front of the leading vehicle. This said, the driver's choice of headway may also reflect a misplaced faith in their own driving skills and a belief that they can compensate the reduced headway by increasing their attention to the driving task.

Stradling (Car driver speed choice in Scotland) argues that drivers enjoy the feelings of mastery and control that a vehicle provides. Drivers report that they gain independence, autonomy and freedom from car ownership. This is not without some cost, however, as inappropriate choice of speed is associated with accidents and the greater the speed of the vehicle, the greater the likely severity of a crash. Stradling studied the underlying reasons for speeding. He found three main causes: the situational opportunity (such as the power of the vehicle and the apparent road conditions that make speeding a possibility); the perceived obligations (such as the perceived need to be at a certain place and a certain time); and the individual inclinations (such as the desire of the driver to select a certain vehicle speed). Drivers respond to situational pressures, such as they increase their speed when they are late for appointments. Drivers also respond to social pressures, such as they increase their speed if the traffic around them is faster than they are. These underlying reasons for speeding suggest that any response to reduce chosen speeds is going to have to address the motivations for speeding as well as the traditional road engineering and law enforcement approaches.

3 Driver education and training

Three papers address driver education and training. All three papers are concerned with the changes that occur in drivers, as well as the sustainability of those changes. Stanton et al. evaluate an advanced driver coaching system to identify changes in driver's competencies. Underwood looks at the differences in visual scanning patterns between novice and advanced drivers and to what extent scanning patterns can be taught. Groeger and Banks critique the skill transfer literature, to consider the longer-term benefits of formal driver training.

Stanton, Walker, Young, Kazi and Salmon (Changing drivers' minds: An evaluation of an advanced driver coaching system) present the results of a study on the effects of a driver coaching system used by the IAM. The study examined the effects on the three areas of driver competencies: knowledge, skills and attitude (KSA). The research compared drivers going through the IAM coaching, with drivers who were simply accompanied and drivers who received no intervention. The IAM coaching system has four facets: it is based on the IPSGA system of driving (information, position, speed, gear and acceleration); it comprises individualized instruction; it is based in vehicles on the road; and it uses distributed practice typically over 8–12 weeks. The results of study showed that drivers who were coached improved significantly on all of the KSA measures (i.e. structure of knowledge, observed vehicle control behaviours and external locus of control), whereas the drivers in the two control conditions made little or no improvement. The researchers argued that not all driver education schemes are the same; some programmes are more successful than others. The features that make IAM coaching successful include the focus on a system of driving and the individualization of assessment and feedback.

Underwood (Visual attention in the transition from novice to advanced driver) reports that despite new drivers having to pass a theory test, hazard perception test and an on-road test, they still have much higher accident rates than their more experienced counterparts. He asks whether this experience might be short-circuited, by training better visual scanning patterns to novices, in order to improve visual attention to more salient features of the road environment, as well as reducing distraction.

Underwood argues that the high accident liability of novice drivers is in part due to their failure to search the roadway adequately and in part due to their distractibility from the main driving task. Previous research has shown that nearly 80% of crashes are due to some form of inattention. Underwood's research has shown that novice drivers' visual search patterns are very predictable, mainly focused on a fixed position ahead of the vehicle, whereas expert drivers have no fixed visual search. This is because they are directing their search patterns in response to the changing road conditions and particular situations they encounter. He argued that the more flexible search patterns of expert drivers enables them to be more aware of what is going on around them, whereas the relatively fixed strategies of novice drivers suggests that they have relatively impoverished situation awareness. It is this poorer situation awareness that puts novice drivers at greater risk. Optimistically, Underwood reports on a study that shows that visual scanning patterns can be trained to novice drivers and that this training effect transfers to the real road environment.

Groeger and Banks (Anticipating the context and circumstances of skill transfer) report research on graduated licensing, arguing that it is unreasonable to expect the formal training that drivers receive at the beginning of their driving careers to set them up for life. His research suggests that the basic driving text does discriminate between drivers on a number of key driving skills, which is reassuring. The challenge for future research is to conduct longitudinal studies into driver skill, to see how they develop over time and which interventions are more successful. Using theories of skill acquisition, he argued that the practice of a particular skill is specific to the situation; thus, training and practice for novice drivers needs to be situationally specific. Groeger and Banks contrast generalized transfer (where learning transfers beyond those behaviours trained) and specific transfer (where learning transfers only to those behaviours focused on). They argue that the weight of the research evidence mainly supports specific transfer. This means that the transfer function works best when training is transferred to the same manoeuvre, in the same car, in the same location, in the same session with the same tutor. Groeger and Banks argue for graduated exposure of drivers to risk as their competence increases, such as exposure to night driving and motorway driving should depend on the development of demonstrable skills to drive in those situations.

4 Design of warnings

The potential for the driver to be distracted is mentioned throughout the special issue and the implication for the design of warnings is considered by two papers in this section. The first paper by Lees and Lee investigates driver's responses to a collision avoidance system. This research addressed the appropriateness of the intervention by the driver. Another paper on drivers' intervention to warnings, by Catchpole and McKeown, looked at the design of ambulance sirens.

Lees and Lee (The influence of distraction and driving context on driver response to imperfect collision warning systems) summarize that the distraction of the driver is implicated in the majority of accidents. They suggest that driver distraction is likely to become worse with the increasing prevalence of in-vehicle devices. The likely effects included poorer lane keeping, reduced headways (i.e. increased tailgating behaviour), longer driver response times, poorer judgement, poorer decisions and less sub-optimal responses. Collision warning devices have a potential role in helping drivers to identify situations that require their intervention, in order to avoid potential collisions. These systems can assist in the identification of situations and bring the situations to the driver's attention. Lees and Lee argue that the ability of the driver to understand and assess the warnings may change when the driver is involved in non-driving tasks (such as speaking on a mobile phone whilst driving). They speculate that engaging in non-driving-related activities might lead to inappropriate levels of trust, such as becoming over-reliant on the warnings. Lees and Lee set out to test this idea in a driving simulator. They distinguished between unnecessary alarms (those that were logically consistent with the design of the collision warning system, but did not signify a need for an intervention by the driver) and false alarms (those that were neither consistent with the design of the system nor required intervention). Their research found that unnecessary alarms fostered trust in the collision warning system whereas false alarms fostered distrust. Perhaps surprisingly, the distraction activity did not influence the driver's interpretation of the unnecessary and false alarms. Lees and Lee conclude that design effort should focus on eliminating false alarms rather than unnecessary alarms.

Catchpole and McKeown (A framework for the design of ambulance sirens) note that the rapid transit of the casualty to the hospital greatly influences their survival rate; therefore, the design of siren to get drivers to move out of the path of the ambulance is critical. The use of sirens can help reduce journey time, but drivers of other road vehicles have difficulty in localizing the direction and heading of the ambulance. Catchpole and McKeown propose that it is important to understand the interaction between detection, localization, acoustics and urgency criteria before embarking on the design of the siren. In testing the siren, one also has to consider the possibility of annoyance, crew hearing, health and safety, communications, drivers' responses and any environmental masking. Catchpole and McKeown compared two sirens in an ambulance over a 3 month period. The first was the standard ‘wail and yelp’ siren, which was mounted under the wheel arch of the ambulance. The second was a new two-stage, localizable siren mounted behind the radiator grill. The results of the study showed that the new siren had less penetration into the ambulance (the siren location effect – making the auditory environment more acceptable to the drivers and casualties) and the transit times to the hospital were quicker (the siren auditory design effect – assisting other road uses to identify the direction of the ambulance and thus get out of the way).

5 Vehicle automation

There are two papers on vehicle automation, one on the effects of driver stress and one on the potential effect of surpassing driver skill. Funke et al. propose that, as driving can be a stressful activity, automation could, potentially, reduce the stress of driving. Young and Stanton propose that automation could make novice drivers appear like experts, functionally at least.

Funke, Matthews, Warm and Emo (Vehicle automation: A remedy for driver stress?) argue that driver stress has a potentially positive and negative influence on driving performance. Positively, it could encourage drivers to make a fuller appreciation of the hazards and get them to drive in a more cautious manner. Negatively, it could distract them from the driving task by focusing on the stressful reaction rather than the road environment. Funke et al. propose that vehicle automation technologies could either alleviate or exacerbate the stressful effects of driving. A ‘resources’ theory would predict that vehicle automation would alleviate the effects, because lower workload would result in lower levels of stress. An ‘effort-regulation’ theory would predict that the potential for misallocation of effort (either too much or too little) could result in greater stress with vehicle automation. Although they found that automation did not reduce mental workload, it did reduce driver stress. Funke et al. concluded that their findings support the ‘resources’ theory, although they note that the differential effects of different types of automation might support the ‘effort-regulation’ theory. This still needs to be explored.

Young and Stanton (What's skill go to do with it? Vehicle automation and driver mental workload) suggest that whilst it may be welcomed if vehicle automation relieves drivers of mental workload in periods of high demand, this same effect could lead to mental underload in periods of low demand. Invoking a theory of Malleable Attentional Resources, they explain how performance of the driver can deteriorate in underload conditions. The research explores the relationship between skill, mental workload and vehicle automation, to suggest that automation of the operational driving tasks of a novice may be akin to expert performance. The focus of the paper being on the way in which the lack of skill in novice drivers can be supplanted by automation. Comparison of novice and expert drivers showed there were substantial reductions in workload for the novice drivers when operational tasks were automated. The study also showed that automation effectively closed the skills gap, making the mental workload demands far more manageable for the novices (in normal driving scenarios). Novices were overloaded without the automation present. Young and Stanton point out that although automation may alleviate some of the workload demands for a novice driver in the short term, longer-term benefits might be questionable, as the driver would not have the necessary skills to cope with unusual or emergency conditions.

6 The effects of other tasks on driving

Finally, there are two papers that deal with the effects of the operation of in-vehicle devices on driver performance. These papers are concerned with the potential conflict between operation of the vehicle and operation of the devices. Horrey and Simons study the effects of conducting a concurrent activity (such a speaking on a mobile phone), whereas Ma and Kaber study the effects of interacting with a navigation system. The former is a parallel, unrelated task, whereas the latter is an intermittent, related task.

Horrey and Simons (Examining cognitive interference and adaptive safety behaviours in tactical vehicle control) are concerned that operation of in-vehicle devices could lead to driver distraction and inattention to the primary driving task. They hypothesize that the driver might adapt to the distraction by driving at a slower speed or leaving larger headways. Horrey and Simons distinguish between operational driving tasks and tactical driving tasks. The driving task has been characterized by other researchers as three interacting levels of strategic tasks (such as driving safely and efficiently to the destination and route planning), tactical tasks (such as navigation and taking advantages of opportunities to overtake) and operational tasks (such as vehicle inputs to maintain position in lane and safe distances from other vehicles). Higher tasks have longer time horizons and lower tasks have shorter time horizons. Horrey and Simons investigated the nature of the safety margins for both tactical and operational tasks of driving, with and without the concurrent cognitive task in a driving simulator on a two-lane freeway. The results of the study show that drivers increase their safety margins in the operational driving task (i.e. the vehicle-following task), but not the tactical driving task (i.e. the overtaking task). Horrey and Simons argue that drivers are unable to translate the safety margins from operational to tactical manoeuvres, possibly because it is just too difficult to undertake.

Ma and Kaber (Situation awareness and driving performance in a simulated navigation task) consider the factors affecting navigation, such as the integration of navigation knowledge, environment interaction, spatial orientation and vehicle status. They manipulated the level of information reliability in a navigation system for a simulated driving task. Ma and Kaber expected to find that higher levels of reliability would improve the driver's awareness on the navigation task. Drivers with the completely reliable navigation system had lower level 1 situation awareness (perception of elements) than the lower reliability conditions, whereas they had higher level 2 situation awareness (comprehension of situation). There were no differences in level 3 situation awareness (projection of near future). These findings suggest that the drivers in the highest information reliability conditions were operating at a different level to those in the lower reliability conditions. With high information reliability, drivers did not have to focus on individual features to navigate; rather they could focus on comprehending the overall situation. By contrast, those drivers in the lower information reliability condition were forced to focus on identifying individual features in the navigation task, to verify the navigation task and their position on the route; therefore, they were less able to devote resources of comprehending the situation.

7 Conclusions

In summary, the 12 papers that make up this special issue of driver safety have made important contributions to driver safety, including driver behaviour, driver training, design of warning systems, vehicle automation and driver distraction. Driver violations appear to be widespread (up to 60% of drivers on motorways, based on a sample of three states in Australia). Of these violations, tailgating and speeding were the most prevalent. Younger drivers were more likely to commit driving violations, which may require more effort in driver training programmes (e.g. not only modelling correct behaviour but illustrating what is wrong with widespread practice), to reduce this phenomenon. Drivers tend to drive too closely above speeds of 50 mph, which again has training implications. The pressure to exceed the posted speed limit seems to come from social and situational pressures, which has implications for road and vehicle design, training and societal norms. Driver training needs to be individualized, coaching to a system of driving (such as IPSGA) on the road, with a schedule of distributed practice. Novice driver training should include instruction in flexible visual scanning and search patterns to encourage drivers to focus on salient information for improved situation awareness. Longevity of the effects of driver education and training can be improved with situation-specific transfer. Warning design (including the design of the sound and the placement of the device) is crucial to both attract the attention of the driver and help direct them to the appropriate course of action, as well as reducing any potential distractive effects. Vehicle automation design needs to be approached with caution. Some immediate benefits have been found in terms of reducing driver stress and the performance of the novice driver, but longer-term cost and benefits need to be assessed. The number of distractions within vehicles is increasing, including mobile phones and navigation systems. It seems that drivers cannot adapt their safety margins to cope with more complex tasks when using these systems. Operation of these systems also appears to affect driver situation awareness. To conclude, there is much evidence from the special issue to suggest that Ergonomics has much to offer the design of driver education and training programmes, the design of driver interfaces and driver assistance systems with motor vehicles, the design of vehicle automation and a deeper understanding of why drivers behave as they do. It is to be hoped that these messages will provide fuel for thought for future research as well as informing those who design vehicle and road environments, law enforcement strategies, driver education programmes and road transportation policy.