1. Introduction
Meta-analysis of the effect of road lighting across the world as an accident countermeasure has suggested that general overall reductions in fatal and injury collisions can be expected during hours of darkness [
1,
2,
3]. Higher savings in injury collisions have been reported in countrywide studies, particularly in rural areas [
4]. Furthermore, where lighting has been installed, a relationship between increasing luminance levels and decreasing night to day crash ratios has been identified [
5]. An increased collision frequency at night has been shown to occur on longer links in the network, although it has been postulated that this may be down to other factors such as increased speed rather than lighting conditions [
6]. Indeed, the presence of lighting may itself be counterproductive in this respect; drivers may compensate for the presence of road lighting on links in terms of increased speed or reduced concentration [
7]. The impact of weather conditions in combination with lighting has also been shown to have an effect, suggesting collisions may be caused by an inability in drivers to adjust speed in accordance with degraded visual performance [
8].
One road feature where there appears to be significant correlation between safety and lighting is at junctions. Studies have shown that unlit junctions have higher night to day crash ratios and greater injury severity than those that are fully illuminated or feature simple destination lighting [
6,
9,
10,
11,
12]. However, it has been recognised that due to fewer drivers driving during hours of darkness, exposure is lower and the benefits of lighting may be overstated in respect to other non-light related safety countermeasures [
13]. Regardless of the net level of effect of lighting, in such situations older drivers are over represented in collision statistics [
14] due to differences in abilities such as gap acceptance [
15].
Future scenario planning exercises suggest that assistive technologies may alter the way in which people live and travel as the population ages [
16]. However, for the foreseeable future, mobility is a significant contributor to happiness and good health, particularly for older people [
17,
18,
19]. Such effects though, can be tempered by the proximity to urban areas and the spatial planning adopted [
20,
21]. Notably, in rural areas, public transport and active travel options often are limited and access to the car has been shown to be a key determinant of travel needs being met [
22]. Drivers report feeling less comfortable driving in the dark, potentially resulting in self-regulation and avoidance [
23,
24,
25]. On this basis, transport policymakers need to look beyond direct road safety implications when considering providing for drivers at night [
26].
In terms of sustainability, the environmental impact of street lighting due to energy consumption has been substantially reduced in recent years, through the development of LED luminaires, solar power and adaptive control [
27,
28,
29,
30]. However, illumination solutions may result in adverse light pollution impacts in rural areas, the health and environmental effects of which are often overlooked in infrastructure development [
31]. Lighting also has a significant capital and revenue cost implication and it has been recognised its application should be focused on addressing specific collision concerns, as opposed to universal adoption [
32]. Further to this, poor driver confidence does not necessarily translate into a lack of ability [
33,
34]. On this basis, investment priorities to balance truly effective road safety measures with interventions to improve driver confidence need to be carefully considered.
Active Road Studs (also known as Internally Illuminated Raised Pavement Markers) are a development of traditional reflective road studs incorporating powered internal LED illumination. Through eliminating the reliance on reflected light and increasing illumination intensity through an active source, it is suggested they could assist drivers in several applications [
35]. Field trials of the studs suggest they could have positive effects on vehicle positioning and speed [
36]. Simulator studies support these assertions and also suggest increases in confidence from drivers may be expected in conjunction with their use [
37,
38]. However, to date no assessment has been made of whether this theoretical change in confidence translates to a real-world scenario. This work is intended to address this gap in knowledge.
The aim of this study was to measure self-reported levels of confidence felt in safety terms, using real-world rural junctions and links during hours of darkness where active road studs are installed, and how this may affect drivers’ propensity to travel. The objectives were to:
quantify the levels of confidence of rural road users in different lighting conditions;
determine the relative levels of importance of different road features to users in relation to confidence of driving at night;
measure the effect on driver confidence resulting from the installation of active road studs; and
understand the reasons for any changes in safety perception.
The work described here forms part of a larger study investigating the effects of active road studs on driver behaviour, such as lane discipline, speed and gap acceptance. The focus in this element is driver confidence.
2. Materials and Methods
2.1. Study Area
The A1 is a UK trunk road (road of national strategic importance) running east from Edinburgh in Scotland to the border with England and beyond. The road is dual carriageway from the edge of the city for around 50 km. After this, it is mostly single carriageway for the remaining 32 km to the border, with a short (4 km) section of dual carriageway roughly mid-length and a section of dual carriageway crossing the border itself. As a result of various upgrades in the latter decades of the 20th century, it is predominantly built to modern alignment standards. The single carriageway section of the A1 is rural in nature. The road carries a mixture of longer distance strategic traffic and local journeys between the small towns and villages in the surrounding area.
For several years, residents have campaigned for improvements to be made to address perceived issues of safety along the route. Despite this, a review by the national roads authority reported to local elected members that only two slight injury collisions had been recorded [
39]. However, the review also found during hours of darkness, the visibility of junctions on the route could be perceived as poor and an improvement scheme was subsequently developed to address user concerns. The scheme comprised the use of LED Active Road Studs to highlight junctions and selected links with a view to improving the perception of safety and the experience of users. In total, over 4200 studs were installed at nine junctions, including two intermediate stretches of carriageway. The installation process for the studs and examples of their use in the scheme are shown in
Figure 1 and
Figure 2.
2.2. Population and Sample
The survey population of interest were the local users of the route. Longer distance and nonlocal users were not within the scope of this investigation. To focus on the experiences of the target group, a questionnaire survey was developed for distribution in the study area. Introductory text included with the survey provided details of the purpose of the study, along with contact details of the researchers for any further information. An optional prize draw was included to encourage response; however, respondents were informed that the survey was anonymous and that no personal details would be retained, being used only for the draw if they chose to participate. Approval of the survey and study approach was obtained in accordance with Edinburgh Napier University’s Code of Practice on Research Integrity.
The distribution of the survey was non-discriminatory with a goal of sampling the widest cross-section of the population possible. In interpretation of the results, two factors are important to note. The first is that only one questionnaire per household was provided. This means where multiple drivers were resident at the address, the results could reflect the views of only one respondent and variations in confidence between household members may not be represented. The second relates to the age distribution of respondents. A comparison of the 2011 UK census data for the study area population with the sample obtained is shown in
Figure 3. Although the population data represents residents rather than drivers, the proportion of younger respondents in the sample was notably lower. For the purposes of reliable statistical analysis, a single category of those between 18 and 39 years old was formed to represent the younger driver group.
2.3. Survey Content
The main goal of the questionnaire was to assess the importance of road features during hours of darkness and the opinion of active road studs in this context. Respondents were asked to rate on a five-point Likert scale varying from ‘not important’ through to ‘very important’ the following six features: street lighting at junctions; street lighting on curves; road studs at junctions; road studs on curves; solid white lines at edges of road; and white line markings in the centre of the road. They were also asked on a five-point scale from ‘strongly worsened’ through to ‘strongly improved’ whether the introduction of the active road stud had altered their opinion on seven themes: clarity of junctions; clarity of curves; clarity of lanes; their confidence driving in darkness; their opportunities for travel; overall safety; and incidence of speeding on the route.
To permit exploration of potential variation between groups, demographic information was asked regarding the age and gender of respondents. The age categories were defined in accordance with the bands used in UK national government reporting [
40]; as previously described some lower age bands were latterly combined prior to analysis. All respondents were asked to rate their confidence driving on the A1 in daylight hours and during hours of darkness using a five-point Likert scale from ‘not confident’ through to ‘very confident’. An open-ended question invited comments on the views of the active road studs and their general perception of safety on the route. Using a five-point Likert scale (strongly disagree to strongly agree), respondents were asked to state whether active road studs should be more widely used.
A series of secondary questions were asked to investigate contributory factors to driver confidence. This included how many years they had held a driving licence; an estimation of their annual distance driven in miles to the nearest thousand; the number of road accidents they had been involved in during the five years prior to the survey and how many had been during hours of darkness. To validate previous answers, they were also asked if they were aware that active studs had been installed prior to receiving the survey. Finally, respondents were asked to provide the frequency they make specific journey types during hours of darkness, namely: travel to/from a place of work; travelling for work; taking children to school; carer responsibilities; hobbies and recreation; visiting friends or relatives; and shopping.
2.4. Survey Distribution
The target was to sample the widest possible range of respondents within the study area. A hand-delivered survey to all residential addresses in the study area was deemed the most effective option. Distribution of the survey was undertaken by the UK national postal service as part of its normal postal deliveries during the week commencing 20 November 2017. The survey was reported in two local newspaper articles during the week of distribution to raise awareness. A closing date for the survey was 22 December 2017, the day after the winter solstice. This was chosen to ensure respondents had specific familiarity with driving in dark conditions at the time of completion. In total, the survey leaflet was delivered to 10,817 residential postal addresses, comprising the TD113, TD135, TD145 and TD151 postcode sector areas as shown in
Figure 4.
Each paper questionnaire could be returned directly through the UK postal service, with postage prepaid. Respondents were also given a link at which they could complete the survey online if they preferred. The online survey comprised identical questions and descriptive text to those shown on the paper survey and was hosted using the university’s secure online survey software. In total 652 paper questionnaires were returned with a further 46 completed online. The total of 698 returns represents a response rate of 6.5% against the distributed surveys. The paper surveys were manually coded; a numerical system was used for the quantitative questions and qualitative answers were fully transcribed. Every case was given a unique serial number for ease of reference. The electronic and paper responses were combined into a single master file for subsequent statistical analysis.
2.5. Statistical Analysis
Each question in the survey was assessed using appropriate descriptive statistics. For the Likert scale-based questions, percentages of respondents giving answers at the five points were reported. Mileage driven was converted into categorical bands and a median value calculated. Each question was tested against three groups: age, gender and reported confidence driving during hours of darkness. Further statistical analysis was undertaken to investigate potential relationships between driver confidence, mileage driven and journey frequency. For the open-ended questions, content analysis was undertaken and responses grouped by theme. Active road stud themes were identified and grouped into: psychological effects; physiological effects; operation and maintenance; potential for wider use and general opinions. Non-stud related themes were grouped into route upgrading; lighting at junctions; and sign clutter.
Quantitative statistical analysis of the returns was undertaken using Statistical Package for the Social Sciences (SPSS) v23 (IBM). Cronbach’s alpha was determined to validate the internal consistency of the scale questions, with values below 0.5 taken as unacceptable [
41]. Association between variables was identified using Pearson’s chi-squared (χ
2). The null hypothesis was that there is no association between the tested variables with the alternative hypothesis being that an association could be inferred. The significance level was set at 5%, meaning any variable pairs with
p < 0.05 were taken to have associated inferred. Effect size was measured using Cramér’s V (φ
c), giving values between 0 (no association) and 1 (complete association). For the paired ranked ordinal variables, Spearman’s correlation coefficients (ρ) have also been calculated to determine the strength and type of association. Effect size definitions have been taken as +/−0.1 as small, +/−0.3 as moderate and +/−0.5 as large [
42]. Qualitative content analysis of the open-ended question was undertaken using the software package NVivo Pro v12 (QSR International, Melbourne, Australia).
4. Discussion
This investigation aimed to examine changes in the perception of drivers using rural road links and junctions at night, following the installation of active road studs. Of relevance was the importance of active road studs to drivers relative to other road features such as lines, street lighting and traditional reflective studs. The case study used centred around a route where the actual statistical chances of being involved in a collision were average, but the fear of involvement in one was high. This perception of safety is of particular concern for elderly and less-confident users who may self-regulate their driving as result, with the potential consequences being a loss of independence and overall health. The cost and environmental impact of traditional interventions such as street lighting made the active road stud an intervention worthy of investigation.
Significantly, the response to the use of the active road stud was found to be highly positive across all demographic groups, with strong support for wider use on the road network. Most notable was that users believed the safety of the route had been improved with a resulting increase in confidence. Driving at night is a situation where driver confidence is recognised as being lower, often to the point of potential avoidance altogether [
44,
45]. This issue may disproportionately disadvantage specific groups such as elderly drivers and thereby affect their quality of life [
46,
47]. In improving driver confidence, it is suggested that the active road stud may potentially contribute to addressing such concerns, particularly in situations where alternative travel options are limited, such as rural areas.
Both the quantitative and qualitative data presented suggest the first reason for the increase in driver confidence related to the studs is the improvement in clarity of road features, such as junctions and curves. This finding is intuitive; a typical European low beam headlight is at best only able to illuminate up to just under 100 m from the lateral centreline of the vehicle, with coverage at the edges of the road significantly less [
48,
49]. By contrast, an active road stud can potentially be seen to distances of over 600 m [
50]. Difficulty in judging distances to junctions is cited in studies as being a reason for problems with driving at night by users [
51,
52]. In this regard, provision of active road studs would seem to potentially contribute to reducing this concern for rural drivers.
The responses to the survey also suggest that the active road stud not only assists in highlighting the road itself, but also in mitigating difficulties caused by the presence of other vehicles. Up to half of drivers across all demographic groups have previously reported the lights from other vehicles restricting their vision whilst driving at night [
53,
54]. Ultimately this issue can result in driving cessation, particularly in older females, where low-contrast acuity in glare has been found to be a contributory factor [
55]. In the present study, users reported a benefit in being able to perceive the road beyond the headlights of approaching vehicles. By providing additional definition to the road alignment, the presence of active road studs would appear to be of assistance to all drivers, particularly those with higher levels of glare or contrast sensitivity.
Despite these apparent benefits, the survey also found a residual demand for further street lighting to be provided. This would support the suggestion that drivers rely most heavily on their photopic visual acuity, and that situations in which mesopic vision is important such as night driving are more problematic, particularly for elderly users [
56,
57]. Establishment of a fully lit road network would be impractical both in cost and environmental terms and in this regard compromises are inevitable. Provision of active road studs would appear to at the very least narrow the discomfort and anxiety changes experienced between the different night driving conditions. In this particular case study, a negligible number reported some visual distraction, but contrary to what has previously been reported [
58], no significant evidence was found to suggest the studs would exacerbate glare and justify dimming.
On general confidence in driving in darkness, associations found here suggest differences between demographic groups are apparent. Drivers were found to be less confident at night in general, with female respondents being less confident than males. Confidence was also found to increase with mileage driven, echoing findings elsewhere [
59]. No association was found between confidence and age, although this could be explained by the focus on a driver-based sample of the current study and the potential exclusion of some older drivers who are already self-restricting [
46,
53]. Previous involvement in a collision has been found to reduce driver confidence in various situations [
25,
60]; the results here suggest a greater significance on reduced driving confidence at night may be apparent.
There are several road features that could assist with driver confidence at night and this work found that effectiveness varies by situation. Street lighting was found to be important at junctions but less so on curves. This may be associated with the assertion that increased luminance has been shown to potentially reduce concentration and centreline adherence on curved sections of road [
7,
23]. Road studs were rated more highly than lighting at junctions, but particularly so for curves. This could be explained by the benefits that road studs provide in terms of increased preview time of the road ahead, which is of most importance when negotiating curved sections [
61]. White lining at both the centre and edges of carriageway was deemed important by all groups. Overall, there was no association between importance of any feature and age. Female drivers found street lighting more important than male drivers and notably overall, the importance of lighting diminished with increasing confidence.
The diminishing of the importance of lighting with increasing confidence due to other road features, in particular the active road stud, is a key contribution to knowledge made by this study. In practice, what this may suggest is that in situations where lighting brings with it a high cost (either financial or environmental) or potential disbenefit (decreased concentration or increased speeds), the active road stud might be considered as a viable alternative by road engineers. In this way, the active road stud could contribute to increasing driver confidence at night and thereby potentially opening journey opportunities, particularly where alternative modes are limited. Given the issue of confidence driving at night is known to disproportionately affect female and elderly road users, use of such a measure could also address concerns of equality in access to transport, particularly in rural areas.
Whilst the results of this study are believed to be high in ecological validity, two limitations need to be noted. Firstly, the focus of the survey was on those who drive rather than those who do not. The distribution of one survey per postal address also means there is a possibility that the views of other household members, who may be less confident, might not have been captured. Consequently, in relative terms, the results presented here may not comprehensively represent the views of the lowest confidence drivers. The second limitation concerns the assumption that a change in confidence relates to a real change in behaviour. It has been suggested that in some situations, stated self-regulation has a limited relationship with actual objectively measured driving patterns [
62,
63]. In this regard, although this study shows significant potential for the translation of increased confidence into improved journey opportunities, this may be open to challenge. To address this, further work in this area such as longitudinal study of changes in travel patterns where active studs had been installed would seem appropriate.
Finally, it is worth reinforcing that this investigation has focused on the effect of active road studs specifically on the perception of safety of drivers. In this regard, it would appear from the results presented that the studs have a high level of benefit at a relatively low cost, particularly when compared with the installation of street lighting. Potentially, there may be even greater benefits arising from implementation of the studs, but these could only be quantified if the resultant actual changes in road safety are also accounted for. Whilst, for example, this study has suggested drivers may perceive a difference in the level of speeding on a treated route, user opinion alone does not mean there has been a real change. Any collision cost savings from improved speed limit compliance, better lane discipline and gap acceptance can only be ascertained through a trend analysis of night-time casualties before and after installation. Such analysis was beyond the scope of this study but is certainly worthy of further investigation.