Research Report AP-R Distraction and Attitudes Towards Safe Pedestrian Behaviour

Transcription

1 Research Report AP-R Distraction and Attitudes Towards Safe Pedestrian Behaviour

2 Prepared by Alexia Lennon, Amy Williamson, Mark King, Ioni Lewis, and Mazharul Haque Project Manager Edward Rose Abstract Pedestrians account for around 14% of Australian road deaths annually. International research suggests that pedestrian distraction from smart phones may lead to greater risk of trauma. Ownership of smart phones in Australia is high, suggesting this may present an emerging road safety challenge. Austroads commissioned research to identify: community attitudes towards, and personal, social and environmental factors influencing, safe pedestrian behaviour; groups most at risk of distracted walking; most likely locations; and effective countermeasures. The project had three components: a literature review; an intercept survey with pedestrians; and an on-line survey of pedestrians. The literature review identified that pedestrian distraction from technology is influenced by age, gender, and type of activity. Nonintersection locations are more likely than intersections for pedestrian crashes generally, though there is no research on distraction-specific crashes. Illegal pedestrian behaviours may be both common and riskier. Many effective countermeasures address general pedestrian crash risk, but few address distraction from mobile phones while crossing, and these have been educational in nature, with unproven effectiveness. Surveys results suggested that use of smart phone while crossing the road was low for the sample overall, but significantly higher among year olds, with 30% indicating they used their smart phones for texting or internet access at risky levels while crossing the road. Risk perception and attitudes towards using a smart phone while crossing were important factors influencing likely behaviour for year olds. Countermeasures should integrate education and engineering interventions within a Safe System approaches, target pedestrians under 30 years, and be located at high pedestrian activity locations. Keywords Publisher Austroads Ltd. Level 9, 287 Elizabeth Street Sydney NSW 2000 Australia Phone: austroads@austroads.com.au About Austroads Austroads is the peak organisation of Australasian road transport and traffic agencies. Austroads purpose is to support our member organisations to deliver an improved Australasian road transport network. To succeed in this task, we undertake leading-edge road and transport research which underpins our input to policy development and published guidance on the design, construction and management of the road network and its associated infrastructure. Austroads provides a collective approach that delivers value for money, encourages shared knowledge and drives consistency for road users. Austroads is governed by a Board consisting of senior executive representatives from each of its eleven member organisations: Roads and Maritime Services New South Wales Roads Corporation Victoria Department of Transport and Main Roads Queensland Main Roads Western Australia Department of Planning, Transport and Infrastructure South Australia Department of State Growth Tasmania Department of Transport Northern Territory Territory and Municipal Services Directorate, Australian Capital Territory Commonwealth Department of Infrastructure and Regional Development Australian Local Government Association New Zealand Transport Agency. Pedestrian distraction, smart phone distraction, pedestrian mobile phone use, pedestrian attitudes, Theory of Planned Behaviour, Health Beliefs Model, Mobile Phone Involvement, hypothetical pedestrian crossing scenarios. ISBN Austroads Project No. SS1957 Austroads Publication No. AP-R Publication date February 2016 Austroads 2016 This work is copyright. Apart from any use as permitted under the Copyright Act 1968, no part may be reproduced by any process without the prior written permission of Austroads. Pages 96 This report has been prepared for Austroads as part of its work to promote improved Australian and New Zealand transport outcomes by providing expert technical input on road and road transport issues. Individual road agencies will determine their response to this report following consideration of their legislative or administrative arrangements, available funding, as well as local circumstances and priorities. Austroads believes this publication to be correct at the time of printing and does not accept responsibility for any consequences arising from the use of information herein. Readers should rely on their own skill and judgement to apply information to particular issues.

3 Summary and Recommendations This research investigated pedestrian distraction due to use of technological devices and alcohol impairment. Pedestrian behaviour and pedestrian safety (what pedestrians should do) were examined. An extensive review of the literature was conducted. In addition, brief interviews with pedestrians (N = 211) and an on-line survey of pedestrians (N = 268) were conducted. Findings as they relate to questions in the project brief are summarised below. What factors influence pedestrian behaviour compared to pedestrian safety? Age: Adolescents and young adults are the most likely to be distracted pedestrians. Mobile phone-related injury has been found to be higher among younger people, especially teens, in US studies. Rates of smart phone ownership in Australia are high, especially among the young and make performance of cognitively or visually demanding interactive use possible while crossing the road. Location: Most pedestrian fatalities occur at non-intersection locations. At signalised intersections, crossing within 20m of a crossing but not at the crossing increases crash risk by factor of eight compared to legal crossing. Odds of distracted crossing are higher when at signalised crossings than at unsignalised. Location of the crossing relative to the origin and destination of the pedestrian, availability of pedestrian signals, the number of lanes and whether the road is one-way versus two-way influence pedestrian behaviour and the likelihood of disobeying a crossing signal. Pedestrian crash risk by type of location appears to be multi-factorial in nature, with different sets of factors associated with risk across different location-time frameworks. Illegal road use Illegal road use (e.g. crossing away from marked crossings; crossing on red signal) by pedestrians is widespread (e.g. ~20% of crossings at signalised intersections, Brisbane). Exposure: Engaging in distracting smart phone tasks while walking and crossing Up to 40% of pedestrians may be distracted by mobile phones when crossing the road Smart phone use for texting or internet access while crossing is widespread for year olds, with approximately 30% reporting high frequency of using smart phones to text or access internet functions while crossing the road. Type of mobile phone activity Texting, talking and cognitively demanding smart phone use is associated with greater reaction times and increased errors for competing tasks. Type of smart phone activity (voice call, text, internet access) is the strongest predictor (compared to presence/absence of signals, presence/absence of median strip) of self-reported use of a mobile phone while crossing for pedestrians aged years. Listening to music through headphones/earpods may represent a lower level of distraction than texting, voice calls or internet access/use. Austroads 2016 page i

4 Risk perception The majority of pedestrians are aware of the risks of injury from crossing while distracted, and self-report that they do not engage in this behaviour. Younger adults are significantly less aware than older adults of their susceptibility to injury from distraction by using a smart phone while crossing year olds regard voice calls as less risky than texting/internet use while crossing the road. Attitudinal research suggests adults year olds with lower risk perceptions, positive attitudes towards drink walking and perceptions that their friends/family also drink walk or approve of this behaviour are more likely to drink walk. A substantial minority of pedestrians may constitute a high risk subgroup. These are the high frequency smart phone users who engage in the more highly distracting smart phone activities (e.g. initiating text messages, initiating internet access) while crossing the road Pedestrians appear to perceive the advantages of not using smart phones while crossing the road as well as regard doing so as not likely to result in boredom, wasting their time or preventing them responding to important messages. They also agreed that stopping using their phones while crossing would be easy to do. More positive attitudes towards smart phone use while crossing the road and greater perceptions that friends/important others would approve were associated with greater intentions to using a smart phone while crossing the road. Mobile phone involvement Around 35% of year olds may be overinvolved with their mobile phones and therefore use them at inappropriate or unsafe times (e.g. when crossing the road; while driving). This group may also be more difficult to influence in relation to safer behaviour. Impairment by alcohol Crash statistics suggest that more than a third of fatally injured pedestrians have a BAC >.05, the majority of these having a BAC >.15. Drink walking is prevalent among young adult (18-24 years old) patrons of licensed venues. Most alcohol-involved pedestrian casualties are men, struck at night-time, on Friday or Saturday, are walking home, have been drinking for several hours, and are struck within two to three hours of finishing drinking. Around 20% of the pedestrians surveyed in this research indicated they had walked after drinking two or more standard drinks in the previous hour at least once per week during the previous 3 months, and the majority of these (2/3) thought they were affected by the alcohol at the time. Drink walkers may be aware that they are impaired but choose to drink walk regardless. In the pedestrians surveyed for this report, drink walking-related injuries and near misses were rare, affecting only 2% of the sample. In what ways do these factors affect pedestrian behaviour especially in relation to pedestrian safety? Distraction while crossing and walking increases erratic behaviour and decreases safety-related behaviour Pedestrians distracted by mobile phones walk more slowly, change directions more often, acknowledge others less, look left and right less, are less likely to look at traffic before starting to cross, miss more safe opportunities to cross, take longer to initiate crossing, are more likely to cross unsafely into oncoming traffic, spend more time looking away from the road, and make more errors than pedestrians who are not distracted. Austroads 2016 page ii

5 Cognitive and visual distraction by mobile phones is associated with riskier crossing decisions and less safe behaviour. Texting, talking and cognitively demanding smart phone use appears to compete for attentional resources and is associated with greater reaction times and increased errors for competing tasks Risk of injury is increased for pedestrians using mobile phones while crossing Males and younger pedestrians (under 31 years old) are at greater risk of injury from distracted crossing. Lower perceptions of risk, or problematic mobile phone involvement may elevate exposure (more frequent engagement in distracted crossing) Around 35% of year olds may be overinvolved with their mobile phones and therefore use them at inappropriate or unsafe times (e.g. when crossing the road; while driving). This group may also be more difficult to influence in relation to safer behaviour. Greater tendency to use mobile phones while crossing, or problematic levels of mobile phone involvement may lower perceptions of the risks of distraction while crossing the road, or may increase willingness to ignore these risks. Pedestrians generally appear aware of the risk of distraction while crossing and are also receptive to safety interventions that promote not using smart phones while crossing In relation to drink walking, attitudinal research suggests adults years old with lower risk perceptions, positive attitudes towards drink walking and perceptions that their friends/family also drink walk or approve of this behaviour are more likely to drink walk Impairment by alcohol increases crash risk and decreases safe road use Crash data analysis and simulator-based studies verify the association between alcohol impairment and less safe crossing behaviours and increased crash involvement. Drink walking-related injuries and near misses were rare in the sample surveyed for this report, affecting only 2%. However, if such involvement levels generalise to the general population, it suggests that large numbers of young people may be at risk (as they were more likely to report frequent drink walking). In which locations are these behaviours most relevant when considering pedestrian road safety? Crash data shows that mid-block pedestrian crashes are high, and although pedestrians perceive smartphone use as more dangerous at these locations and therefore avoid it, their compensation for the risk appears to be inadequate. Most smartphone use while crossing is reported for signalised crossings, where there may be a protective effect due to both the extra degree of control afforded by the signals and the advantage of crossing in a group where not everyone needs to monitor safety while crossing. What measures have been used to influence and change pedestrian behaviours? How appropriate would these measures be in an Australian and New Zealand context? Most interventions focus on single approaches (e.g. education, engineering, enforcement). Education and awareness intervention programs and campaigns are common but evaluation of these is rare and so effectiveness is unknown. Interventions using enforcement for illegal pedestrian behaviour are uncommon, with logistical difficulties (such as generally low concentrations of pedestrians behaving illegally or wide dispersal of locations for such behaviours, affecting detection) and the costs involved (such as use of police time to detect and issue infringement notices) generally making these rare. Engineering countermeasures to pedestrian behaviour are relatively common, and many have been evaluated and found to be effective. Austroads 2016 page iii

6 Separating pedestrians from motorised traffic in either time or space improves pedestrian safety generally. On-road separation in space, such as footpaths, has been found to increase pedestrian safety by as much as double, and provision of footpaths addresses the issue of distracted walking (though not distracted crossing). Thus safety benefits are likely if footpath infrastructure were to be installed where it does not currently exist on the transport network. However, for pedestrian crossing, separation in space is impractical in most situations on the transport network in Australia and New Zealand, especially outside central urban areas and areas of high pedestrian activity. Separation in time may be more applicable as a countermeasure to distracted pedestrian crossing. Signalisation affects pedestrian decisions about locations for crossing and safety. Raised median strips on multi-lane roads have lowered crash rates. Lowering general vehicle speed limits across high pedestrian concentration areas and local roads is effective at reducing pedestrian crash rates and injury severity. Novel approaches to entertaining pedestrians while they wait to cross at signals has been found to reduce illegal crossing behaviours but may provide effectiveness only in the short term, while novelty value is high. Pedestrian countdown timers have not been shown to be effective. Technology used to warn pedestrians (e.g. audio beacons) is effective and low cost. Smart technologies (e.g. intelligent road furniture/vehicles) that warn motorists of pedestrians are growing in use but as yet are unevaluated. However, these may offer benefits that are worth exploring, particularly if used in combination with education. Real time intersection traffic analysis (including pedestrian traffic) via video technology may provide an effective method of identifying where traffic and engineering countermeasures for distracted pedestrians should be focused. Drink walking Alcohol impairment in road safety is addressed by a number of countermeasures such as Safe Night Out, lock-out programs, designated driver programs, and responsible service of alcohol (RSA). While they do not specifically address drink walking, strategies which are effective at reducing high levels of alcohol consumption, or pedestrian intoxication (e.g. RSA), or that minimise pedestrian activity while intoxicated/impaired are likely to have the greatest benefit. Education strategies have been used but effectiveness is unknown. Effective engineering countermeasures to intoxicated pedestrian crashes include Dwell on red, and pedestrian fences in late night licensed precincts. Lowering vehicle speeds and altering traffic signal timings at high alcohol times and locations (e.g. latenight entertainment precincts) and pedestrian fencing in proximity to licensed venues has been implemented in Australian cities (but not evaluated). Enforcement countermeasures have included implementation of lock-out laws/policies to reduce movement between venues and greater police activity and these have been found to provide some effectiveness in reducing traffic incidents generally and potentially of benefit for pedestrian crash involvement. The Drink Safe Walk Safe project, which had a combination of educational, engineering and enforcement measures, was evaluated as effective and may provide an approach for wider implementation. Many of the countermeasures identified have been used in the Australian and/or New Zealand context and thus widespread implementation would be suitable if/as required. Austroads 2016 page iv

7 Recommended measures to address road safety issues, gaps in existing research, and recommended areas for future research An integrated approach to pedestrian safety, in keeping with the principles of a safe system and incorporating educational, enforcement and engineering measures should be considered and adopted where possible. In addition, countermeasures to smart phone use when crossing should: Focus on young people (under 31 years old) because the behaviour is most common in this group Be located in high pedestrian volume areas (e.g. intersections and city CBD areas, around main tripgenerating locations such as universities) because the behaviour is likely to be more common in such locations, and in order to reach as many pedestrians as possible. Take the form of signs on signal posts as a low cost option. Be included in the road safety education content (e.g. school-based) that targets adolescents, pre-licence and learner drivers. Such materials should attempt to influence attitudes and perceptions of the risk in relation to smart phone/mobile phone use while engaged in other demanding and safety related tasks Public education should seek to: Increase pedestrian awareness of crash risk associated with distraction from phones when crossing (especially texting/internet access) Challenge perceived disadvantages of not using a smart phone while crossing Challenge perceptions that it is possible to cross safely while texting/using the internet Encourage young people to look out for, and model safe behaviour for, their friends Countermeasures to drink walking should: Take an integrated approach, incorporating a combination of educational, enforcement and engineering measures and involving proprietors of licensed premises, staff of licensed premises, and community representatives in the planning and implementation of the measures. Incorporate responsible service of alcohol (RSA). Jurisdictions could consider mandating this and enforcing its operation. Incorporate educational approaches (e.g. display of education based posters, screening of drink walkingrelated television commercials within licensed premises). Jurisdictions could consider mandating a minimum level of such. Incorporate those engineering countermeasures that have proven or promising effectiveness such as: Dwell on red pedestrian fences in late night licensed precincts localised lower vehicle speed limits (using variable signage if required) altering traffic signal timings at high alcohol times and locations Incorporate targeted enforcement, especially at high alcohol times and locations and target motorist behaviour as well as pedestrian behaviour. Future research could investigate: The evidence for, and size of, the crash-risk from distracted crossing and from drink walking. The technical feasibility, user acceptance and effectiveness of countermeasures that provide feedback to smart phone users (ranging from an auditory or visual warning to blanking the screen) identified as about to cross the road (using GPS coordinates or sensors/detectors of mobile phones-in-use close to the crossing/intersection). Austroads 2016 page v

8 Feasibility and effectiveness of pavement/intersection treatments such as no mobile phone symbols painted at the road edge where pedestrians gazing downwards at their phones are likely to see them. The use of video monitoring methods of determining high risk locations for smart phone use while crossing the road. The feasibility of electronic enforcement of smart phone use while crossing the road. The feasibility (including assessing the level of community acceptance) of introducing and enforcing legislation governing distracted or alcohol impaired walking and crossing. The identification and development of countermeasures specifically targeting the small but higher risk young pedestrian group (estimated as comprising 10% of year olds in this report) who are less aware of their susceptibility to injury as pedestrians, more likely to engage in risky pedestrian behaviours generally and also frequently use smart phones while crossing the road. Austroads 2016 page vi

9 Contents 1. Introduction Background and aims Structure of the report Literature Review Background and Scope General factors that influence pedestrian behaviour and safe road use Pedestrian distraction from use of hand-held technological devices Mobile phone use Personal music devices Internet use Locations where pedestrian crashes are most likely Influence of the built environment on likelihood of pedestrian crashes Land use Road factors and speed limits Environmental factors Walking after having consumed alcohol ( drink walking ) Summary of the research on pedestrian distraction from mobile phones or alcohol impairment Countermeasures to distracted or risky pedestrian behaviour Education and Awareness Campaigns Enforcement countermeasures Engineering Summary of countermeasures and their effectiveness Investigating Pedestrian Use of Potentially Distracting Technology While Walking and Crossing the Road Intercept interviews Method Results Discussion Online Survey Theoretical underpinnings Participants and data collection Survey Design Materials Results Discussion Conclusions and Recommendations Recommended measures to address road safety issues Recommended areas for future research References Appendix A Review of Research Relating to Factors Influencing Pedestrian Behaviour and Safe Road Use Austroads 2016 page vii

10 Appendix B Intercept Interview Schedule Appendix C Summary of Results for Intercept Interview Responses to Drink Walking Questions. 96 Tables Table 3.1: Proportions of pedestrians who use their smart phones while walking and crossing the road by activity and categories of frequency (self-report) Table 3.2: Walking while using a smart phone by level of exposure and age group Table 3.3: Crossing the road while using a smart phone by age group and level of exposure Table 3.4: Proportions of pedestrians (%) engaging in different smart phone activities while crossing the road (N=268) Table 3.5: Crossing the road and smart phone use (N=268) by age and exposure category Table 3.6: Pedestrian Behaviour Scale question wording and Mean (SD) by age group Table 3.7: Comparison of high and low self-reported usual risky pedestrian behaviour for high and low levels of frequency (high, low) use of smart phone while crossing (n = 247) Table 3.8: Mobile phone involvement level (high, low) by age group and gender Table 3.9: Comparison of high and low mobile phone involvement for self-reported usual risky pedestrian behaviour (high, low) and frequency of smart phone use while crossing the road (high, low) for both all age (Full sample, n = 247) and young people only (n = 168) Table 3.10: Pedestrian responses (mean rating) to susceptibility, severity, benefits, barriers and self-efficacy measures (Health Belief Model) Table 3.11: Health Beliefs Model variable scores by age group Table 3.12: Questions and mean responses (SD) for measures of attitudes, group norms, perceptions of control and intentions (TPB) in relation to using smart phones while crossing Table 3.13: Summary of hierarchical regression analysis for Theory of Planned Behaviour standard (Attitude, Subjective Norms, Perceived Behavioural Control) and additional (mobile phone involvement, Group norm) predictors of intentions to cross the road while using a smart phone for text/internet access (entire sample, year olds, N=268) Table 3.14: Summary of hierarchical regression analysis for Theory of Planned Behaviour standard (Attitude, Subjective Norms, Perceived Behavioural Control) and additional (mobile phone involvement, Group norm) predictors of intentions to cross the road while using a smart phone for text/internet access (18-30 year olds only) Table 3.15: Mean ratings of the likelihood of crossing in scenarios varying by i) type of smart phone activity (voice call, text/internet), ii) presence or absence of pedestrians signals, and iii) median strips (present, absent) by age group Figures Figure 2.1: Newspaper advertisement from the Pedestrian Council of Australia s Lambs to the slaughter campaign Figure 2.2: Advertisements from the Pedestrian Council of Australia s Don t Tune Out campaign Figure 2.3: Join the Drive s Share the Road campaign Figure 2.4: City of Melbourne s Share the Road Campaign: Red Man, Green Man Figure 2.5: Auckland City Council s Don t Step into Danger: Fire campaign Figure 2.6: Auckland City Council s Don t Step into Danger: Snake campaign Figure 2.7: Wellington City Council Cross the road with a clear head campaign Figure 2.8: Word Cloud illustrating the range of potential distractions that people nominated when walking around the city Figure 2.9: Christchurch City Council s Could they stop? Cross safely campaign Figure 2.11: Never let a mate walk home drunk campaign Figure 2.11: The Dancing Red Man campaign Figure 2.12: Examples of components of countermeasures Figure 2.13: An illustration of Volvo s pedestrian and cyclist detection technology Austroads 2016 page viii

11 1. Introduction 1.1 Background and aims This report reflects the results of CARRS-Q research in response to an Austroads tender to investigate pedestrian distraction specifically due to use of technological devices or due to alcohol impairment while interacting with the road environment. This report documents the methods and findings in relation to each of the nine key project objectives: 1. Identify community and pedestrian attitudes to safe pedestrian behaviour 2. Identify factors that influence pedestrian behaviour and safety 3. Identify areas of road safety concern 4. Identify those pedestrian groups most at risk of distracted walking and the locations where risk is increased 5. Ascertain current levels of knowledge of the contribution to distracted walking or impairment by alcohol, drugs or fatigue 6. Summarise current measures to influence pedestrian behaviour 7. Explore personal, social and environmental factors that influence pedestrian behaviour and particularly use of potentially distracting technology while walking 8. Identify or propose countermeasures that have the potential to address community and pedestrian attitudes and perceptions of risk in relation to distracted walking and crossing, or that encourage safe pedestrian behaviour, especially in the context of Australia and New Zealand 9. Identify areas for future research in the distracted walking/pedestrian area 1.2 Structure of the report To address Objectives 1-3, 5 and 6, the report begins with a critical review of the Australian, New Zealand and international literature in relation to pedestrian distraction. This is presented in Chapter 2. The background to the problem is firstly presented as a way of introducing the nature of pedestrian safety research. The scope of the review is also outlined in this section. An overview of factors that influence pedestrian behaviour and safe road use is given and evidence for the extent or prevalence of pedestrian distraction from technology and alcohol impairment as well as known or suspected impact on safety is summarised. In particular, evidence relating to the types of locations where distracted walking is most likely is presented. Finally, countermeasures to distracted or risky pedestrian behaviour that have been trialled in Australia or New Zealand and evidence of their effectiveness is discussed. These includes engineering countermeasures to problem pedestrian behaviours and recent developments that have taken place in detailed real time intersection traffic analysis (including pedestrian traffic) via video technology. Other technological advancements and the issue of pedestrian visibility and conspicuity have also been briefly summarised. Austroads 2016 page 1

12 In order to meet Objectives 1, 2, 4 and 7, for this project, two empirical studies were conducted: an intercept interview and an online survey. A description of the rationale, approach, method, results and findings from these studies are presented in Chapter 3. The focus in the intercept interview was on estimating the extent to which pedestrians are exposed to potential crash risk as a result of using technology, that is, the size of the problem. An additional aim was to identify factors that might be influential in relation to distracted pedestrian behaviour so that these might inform the design of the subsequent study (the online survey). The online survey was intended to obtain a larger sample of pedestrians drawn from a more diverse, broader and larger population of pedestrians than was possible in the intercept interviews. Lastly, objectives 8 and 9 are addressed in the conclusions and recommendations in Chapter 4. Austroads 2016 page 2

13 2. Literature Review 2.1 Background and Scope Unlike driving, where public activity is completely regulated and there are access restrictions (i.e. driver licensing), walking is a natural and largely unregulated activity undertaken from childhood onwards as an incidental part of everyday life. Crossing or walking along roads forms a minor part of total walking, but presents the highest risk because of the interaction with motor vehicles. Pedestrians are therefore an important vulnerable road user group and represent, globally, 22% of all road deaths (WHO 2013). In Australia, pedestrians represent approximately 14% of road fatalities, accounting for 2,022 deaths in the ten years (calculated from data reported in BITRE 2013). In New Zealand the proportion of pedestrian deaths is around 11% of the annual road deaths, with 33 pedestrians being killed in 2012 and more than 370 pedestrians killed over the period (Ministry of Transport 2013). In addition to their relatively low mass (compared to motorised traffic), pedestrians are also rendered vulnerable by their inherent lack of protection in a crash. This is exacerbated by factors which increase the likelihood of pedestrian interaction with motorised traffic. For these reasons, use of roads by pedestrians is regulated to some extent, and certain regulations protective of pedestrians apply to drivers. However, illegal use of the road by pedestrians is widespread (e.g. 20% of crossings at signalised intersections at a sample of sites in Brisbane: King, Soole & Ghafourian 2009) and enforcement is rare for logistical reasons. In addition, most road crossing requires pedestrians to integrate visual and auditory information, make judgements of speed and driver intention, and decide when it is safe to cross within the constraints of their walking speed and ability to vary it. Even for pedestrians who can successfully integrate this information under normal circumstances, distraction or temporary impairment (e.g. from alcohol) can interfere with the decision making process at a range of points pedestrians may fail to notice important auditory or visual information, or make incorrect judgements of speed (especially where multiple lanes or vehicles are involved), or incorrectly make an attribution of driver intention, or misjudge their own ability to get across in a given gap. The challenges for pedestrians also vary by location, with signalised crossings requiring the least decision-making, while mid-block crossing of busy multi-lane roads with no centre median arguably presenting the greatest challenge. Additional contextual factors such as night-time, lack of street lighting or rain are also contributing factors. There is little evidence that pedestrians lack knowledge of the rules for crossing roads (King, Soole & Ghafourian 2009). A particular group most likely to be ignorant of the rules, children, receive education about crossing roads in school. The reasons for unsafe road use are therefore more likely to lie with pedestrian motives, their general walking behaviours (given that road crossing is a small element of walking activity) and other activities or practices to which walking is incidental. Some examples are: 1. engaging in a conversation while walking and focusing on the personal interaction so that attention to environmental changes is reduced; 2. listening to music or a podcast through headphones so that ability to hear vehicles is reduced; 3. leaving a venue at which alcohol has been consumed to walk somewhere else and having impaired ability to make decisions about safe crossing; 4. hurrying to meet a time-critical deadline (e.g. a bus departure) such that there is increased willingness to accept potential risks in accuracy of judgements about driver intention or available gaps. Austroads 2016 page 3

14 These examples illustrate that distraction and impairment both have the potential to exacerbate crash risk for pedestrians. With the growth in use of mobile technology for entertainment and communication, pedestrians are increasingly likely to multitask while walking, elevating the risk of distraction during road use. The area of pedestrian distraction is not well researched, and focuses on crash outcomes (whether pedestrians who were distracted generally by mobile phones have a higher risk) or observed behaviour (proportion of pedestrians distracted and association with unsafe behaviour). There is little research into motivations or attitudes that might influence or inform countermeasure development. There is even less research into the impairment of pedestrians by drugs or fatigue (Tulu et al. 2013). In contrast, studies into the factors affecting drink walking are growing. Due to the nature of this research, many of these studies have adopted observational methodologies. Because this relies purely on observed behaviour, it is somewhat limiting in terms of what we can infer about underlying reasons for the behaviour. Such studies do, however, provide useful insights into how people actually behave in a real world setting. More recently, not without their own limitations, through advancements in simulation technology there has been an increase in the number of studies that have examined pedestrian behaviour using a virtual environment in which various aspects of the experimental design can be manipulated, such as vehicle speed and gaps in which people choose to cross. Analyses of crash data also provide useful information about the demographic and geographic characteristics of pedestrian incidents. Advances in statistical and geospatial modelling techniques also offer novel insight into where and when pedestrian crashes occur and where countermeasures should therefore be focused. For the purpose of this review, we have excluded literature that relates specifically to primary-school aged child pedestrians, brain injured pedestrians, disabled pedestrians, those with a specific visual impairment, and pedestrian behaviour at railway level crossings. We have drawn from national and international literature, with a focus on what countermeasures have been trialled in Australia and New Zealand, though material from elsewhere around the world has been included where appropriate. Peer-reviewed and grey literature has been covered. 2.2 General factors that influence pedestrian behaviour and safe road use Pedestrian behaviour is influenced by a wide range of factors, including demographic factors such as gender (Gannon, Rosta, Reeve, Hyde & Lewis 2014; Rosenbloom, Nemrodov & Barkan 2004; Holland & Hill 2010; Tom & Granie 2011) and age (Oxley Fildes, Ihsen, Charlton & Day 1997; Dommes, Granie, Cloutier, Coquelet & Huguenin-Richard 2014¹; Cavallo & Dommes 2014; Dommes, Cavallo, Dubuisson, Tournier & Vienne 2014²; Dunbar 2012; Lobjois & Cavallo 2007). A range of other factors have also been examined. These include personality and attitudes (Schwebel, Stavrinos & Kongable 2009), self-identity (Holland, Hill & Cooke, 2009), conformity and group norms (McGhie, Lewis & Hyde 2012), socioeconomic factors (Zhuang & Wu, 2011; Tulu, Washington, King & Haque, 2013), cultural factors (Nordfjaern & Simsekoglu 2013; Rosenbloom, Shahar & Perlman 2008; Rosenbloom 2009) and other factors such as trip length and purpose (Kothuri, Clifton & Monsere 2014). The research into the influence of these factors is presented in Appendix A. As seen in the evidence presented in Appendix A, pedestrian safety is influenced by multiple factors, including demographics such as gender and age, social conformity and group identity, and cultural or religious factors. Age-related declines in performance of crossing tasks have been widely documented, with perceptual and cognitive deficits arguably playing a substantial role in many of the older pedestrian crashes (Oxley et al 1997). The adoption of compensating strategies as well as the apparent improvement seen in the simulator-based training of older pedestrians (Dommes et al 2012) offers promise when considering what can be done to improve the safety of this at-risk group. Gender differences in visual search strategies and decision making is another well researched influence on pedestrian safety (Tom & Granie 2011). Understanding how gender roles influence the internalisation of traffic rules is important when considering how to reduce pedestrian crashes (Granie 2009). Austroads 2016 page 4

15 The potential influence of group differences in terms of attitudes, beliefs, perceived risk and intended behaviour also has important implications for road safety (Holland & Hill 2007; Yagil 2000). Understanding, for example, potential motives or intentions among young adults is important when considering that individuals with lower perceptions of risk of negative outcomes from impaired walking, who hold positive attitudes towards this behaviour or who perceive it as something that their friends/family engage in or approve of, are more likely to drink walk (McGhie et al 2012, Gannon et al 2014, discussed in detail in the next section). Personality or temperament factors may provide a way of targeting at-risk pedestrians for intervention purposes (Schwebel et al 2009). Finally, though less relevant to Australia, our understanding of pedestrian behaviour in developing countries is in need of development in order to have a positive impact on the high burden of injury and death in such countries, especially where rapid motorisation is occurring. Learning more about these less well understood contributing factors will be an important way forward for these regions where pedestrian risk is so high (Tulu et al 2013). The consideration of cultural factors in different regions will also be important for future research and countermeasure development. The main focus of this report is distraction from the use of technology (Nasar, Hecht & Wener 2008; Hatfield & Murphy 2007; Stavrinos, Byington & Schwebel 2009; Neider, McCarley, Crowell, Kaczmarski & Kramer 2010) and impairment resulting from alcohol use (Gannon, Rosta, Reeve, Hyde & Lewis 2014; Lang, Tay, Watson, Edmonston & O Connor 2003; McGhie, Lewis & Hyde 2012). The literature regarding these factors is reviewed in the sections which follow. 2.3 Pedestrian distraction from use of hand-held technological devices In Australia, there are 11.9 million adult smart phone users (Australian Communications and Media Authority 2013) and 94% of young adults aged years old use a mobile phone (Department of Broadband Communications and Digital Economy 2008). Madden et al (2013) showed that 78% of American teenagers (aged years old) now have a mobile phone, with almost half of these owning smart phones. In the Netherlands, nearly every young person has a mobile phone; including nearly one quarter of 8 year olds, 45% of nine years olds, 60% of 10 year olds, and 69% of 11 year olds (SWOV 2013). Given these high mobile phone ownership levels, it is not surprising that recent years have seen the publication of several studies that have investigated the effects of such technological devices on pedestrian injury risk. Prevalence of pedestrian use of mobile phones has been examined in terms of texting, having a conversation while holding the phone, and using the internet. Many of these studies involve the observation of pedestrians at intersections or midblock crossings. More recently other studies have used virtual environments to manipulate aspects of the pedestrian task experimentally. Analysis of hospital or policereported crash data has provided some insight into pedestrian injury according to demographic as well as geographic and socioeconomic factors. Literature in these areas is reviewed next. Austroads 2016 page 5

16 2.3.1 Mobile phone use Several studies report the deleterious effects of mobile phone use on crossing behaviour and safety, especially among younger people. US hospital emergency data shows an increase in the proportion of pedestrian injuries that involved distraction from mobile phones in recent years (Nasar & Troyer, 2013). Mobile phone use while crossing the road and mobile phone-related injuries have been found to be higher for people under the age of 31 years (Nasar & Troyer, 2013; Nieuwesteeg & McIntyre, 2010) and particularly high among teens (Ferguson, Green & Rosenthal, In research on teenage pedestrian distraction from mobile phones, a US report published by Safe Kids Worldwide reported observations of 34,000 students at road crossings in front of schools and findings from discussion groups with over 2,400 students. Twenty percent of high school students and 12% of middle school students were observed crossing the street while distracted. Students were most often texting on a phone (39%), or using headphones (39%). A further 20% were talking on their mobile phone. Girls were 1.2 times more likely than boys to be walking while distracted, with 17% of girls observed to be distracted and 14% of boys. The odds of being distracted were found to be 26% higher if there was a traffic light present, suggesting that teens may be more willing to use technology when they perceive their surroundings to be safe. Results from the discussion groups revealed that half of students (49%) reported using a mobile phone while walking to school. Four out of ten reported listening to music while walking to school. Seventy eight percent of students reported that it s a problem for children of other ages, not children their own age (Ferguson et al 2013). Earlier observational studies have shown that between 8-33% of pedestrians perform a distracting activing while crossing the road and that mobile phone users cross unsafely into oncoming traffic more often than other observed pedestrians (Nasar, Hecht & Wener, 2008; Thompson, Rivara, Ayyagari & Ebel, 2013; Bungum, Day & Henry, 2005; Hatfield & Murphy, 2007; Cooper, Schneider, Ryan & Cox, 2012; Basch, Ethan, Rajan & Basch, 2014; Brumfield & Pulugurtha, 2011). Thompson et al (2013) observed 1,101 pedestrians and found that nearly one third (29.8%) performed a distracting activity while crossing (6.2% talking on a handheld phone (phone to ear), 11.2% listening to music and 7.3% texting). Text messaging, talking on a mobile phone, and talking with a companion increased crossing time. Bungum et al. (2005) observed 866 individuals and found that approximately 20% of pedestrians were distracted in some way as they crossed the street. Further, only 13.5% of walkers looked left and right while crossing the street and waited on the curb until the light had turned green before stepping into the intersection. Using a case-control design to match for time and demographics (gender and approximate age), Hatfield and Murphy (2007) observed 546 pedestrians at signalised and unsignalised pedestrian crossings in three Sydney suburbs (reflecting low, medium and high socio-economic status). Results showed that, of the 182 pedestrians who were using their mobile phones while crossing, 140 were talking on a handheld mobile, six were talking on a hands-free mobile and 36 were texting. They found that among females, pedestrians who crossed while talking on a mobile phone crossed more slowly, and were less likely to look at traffic before starting to cross, to wait for traffic to stop or to look at traffic while crossing, compared to matched controls. For males, pedestrians who crossed while talking on a mobile phone crossed more slowly at unsignalised intersections. Cooper et al (2012) observed 12 intersections in San Francisco and found that 8% of pedestrians used their mobile devices while crossing. Female pedestrians were more likely than males to talk on their mobile phone while crossing the street, but males were more likely to violate traffic signals while walking (or bicycling). Basch et al (2014) observed the 10 intersections with the highest frequency of pedestrian-motor vehicle collisions in Manhattan, New York. More than one in four of the 3784 pedestrians observed was distracted by mobile electronic devices while crossing during the walk (28.8%) and don t walk (26.3%). In a study conducted at the University of North Carolina (UNC), Brumfield and Pulugurtha (2011) observed seven midblock crossings at the University campus. Results showed that 29% of pedestrians were noticeably distracted while crossing the road (16% talking on a mobile phone and 7% texting). While they also found that the chance of conflict was the same for distracted and attentive pedestrians, the study also showed that drivers were 40% more likely to give way to distracted pedestrians than to those who appeared attentive. In an experimental movement study that compared normal walking with reading or writing text messages while walking, Schabrun, Van den Hoorn, Moorcraft, Greenland and Hodges (2014) found that, among other posturerelated variables, when participants read or wrote a text, they walked at a slower speed. Austroads 2016 page 6

17 Experimental studies have shown the effects of distraction and inattention caused by mobile phone use when crossing the road (Stavrinos, Byington & Schwebel, 2009; 2011; Masuda, Sekine, Sato & Haga, 2014; Hyman, Boss, Wise, McKenzie & Caggiano, 2010). Stavrinos et al (2009) examined mobile phone conversations among college students, using two experiments in a simulated task. Their first experiment examined whether pedestrians would display riskier behaviour when distracted by a naturalistic mobile phone conversation. Results showed that, when distracted, participants exhibited significantly riskier behaviour for three of the four variables measured. That is, they left significantly less time to spare, missed more opportunities to cross, and were hit or almost hit significantly more times than when not distracted. Interestingly, the fourth variable, attention to traffic, was not affected by distraction; participants made the appropriate motions to look left and right before crossing, but at some point during the decision making process, failed to actually capture/process the information adequately. In a second experiment, Stavrinos et al 2011 examined the impact of three types of distraction on pedestrian safety: engaging in a cell phone conversation, engaging in a cognitively challenging spatial task by phone, and engaging in a cognitively challenging mental arithmetic task by phone. Results of this experiment showed that all forms of distraction resulted in significantly riskier pedestrian behaviour across all four variables measured (time left to spare, missed opportunities, attention to traffic, hits/close calls). Results indicated that a naturalistic cell phone conversation was generally just as detrimental as more cognitively demanding tasks such as counting backwards by threes or engaging in a spatially focused conversation (Stavrinos et al 2011). Masuda et al (2014) investigated the inattention of pedestrians due to using a numerical or touch screen mobile phone for texting. Twenty-four university students participated in a manipulation of mobile phone task and signal detection talk while walking around a 3x3 meter square. They were instructed to walk at a faster pace than usual whilst also maintaining as straight a line as possible, and turn right at the corners. In four conditions (texting, talking, a more interactive cognitive task, and a control), auditory and visual signals were presented to the participants as they walked and they were asked to press a mouse key when the visual signal changed colour or when the auditory pitch changed. Reaction time to either visual or auditory signal was significantly longer under the three experimental conditions than in the control. The number of errors was also larger in the mobile phone use conditions. Finally, Hyman et al. (2010) investigated the effects of divided attention during walking. In the first of two studies, they observed the walking behaviour of individuals talking on a mobile phone, individuals walking with no electronic devices, individuals walking and listening to a music device (as a different type of divided attention), and individuals walking in pairs (conversational divided attention). Results showed that mobile phone users walked more slowly, changed directions more frequently, and were less likely to acknowledge other people than those in the other conditions. In a second study, these researchers investigated the possibility that talking on a mobile phone leads to inattention blindness, as measured by whether individuals noticed an unusual stimulus a unicycling clown. Mobile phone users were less likely to notice this stimulus along their walking route Personal music devices The effect of portable personal music devices on pedestrian behaviour has also been investigated with mixed results that suggest that walking while listening to music with headphones represents a different type of distraction from that of using a mobile phone for talking or texting (Walker, Lanthier, Risko & Kingstone, 2012). Contrary to results in relation to distraction from talking on a mobile phone, an observational study of 347 pedestrians at a university campus showed that among males, pedestrians listening to personal music devices (not selecting a song or playing with the actual device) displayed more looking behaviour (as measured by number of head movements left and right) than those not listening to such devices. Females showed no differences in looking behaviour for the two conditions (Walker et al., 2012). Austroads 2016 page 7

18 Simulator studies have also produced contrasting results, finding that distracted participants are more likely to look away from the street environment (and look toward other places such as their telephone or music device) than undistracted participants (Schwebel, Stavrinos, Byington, Davis, O Neal & de Jong, 2012). After controlling for demographics, walking frequency and frequency of media use, Schwebel et al (2012) examined the effect of talking or texting on a mobile phone with that of listening to music through headphones among college students and showed that distracted participants in all conditions were more likely to be hit by a vehicle than those who were not distracted. Neider, McCarley, Crowell, Kaczmarski and Kramer (2010) compared the effects of talking on a mobile phone with that of using portable music devices. Using a virtual environment and a within-subjects design, 36 pedestrians navigated through a series of unsigned intersections in three conditions: no distraction, listening to music through headphones, and conversing on a mobile phone using a hands free device. Participants were less likely to successfully cross the road when conversing on a mobile phone than when listening to music, even though they took more time to initiate their crossing than when conversing on a mobile phone. Neider, Gaspar, McCarley, Crowell, Kaczmarski and Kramer (2011) went on to assess these two behaviours among older adults using another simulated task. Results showed that older adults were more vulnerable to dual-task impairments than younger adults when the crossing task conditions were more difficult Internet use With the increasing popularity of smart phones, the effect of mobile internet use on pedestrian injury has also received some attention. In their national survey of 802 teens aged years, Madden et al (2013) showed that 37% of American teens have a smart phone, up from just 23% in Findings also showed that 74% of teens aged reported accessing the internet on mobile phones, tablets, and other mobile devices at least occasionally. One in four teens are mobile-mostly internet users and older girls are especially likely to report this (34% compared to 24% of boys). Byington and Schwebel (2013) conducted a study to investigate crossing behaviour while accessing the internet among 92 college students. In a virtual environment, participants crossed a street 20 times, half the time while undistracted and half the time while conducting a mobile internet task. When distracted, participants waited longer to cross the street, missed more safe opportunities to cross, took longer to initiate crossing when a safe gap was available, looked left and right less often, spent more time looking away from the road and were more likely to be hit or almost hit by an oncoming vehicle. Furthermore, participants reported using mobile internet with great frequency in daily life, including while crossing the road. Results were controlled for gender, age, ethnicity, and pedestrian and mobile internet experience. 2.4 Locations where pedestrian crashes are most likely Pedestrian crossings can be grouped according to two main locations: at an intersection (signalised or unsignalised) or in the middle of a block ( midblock ). Most of the research into the locations of pedestrian crashes is centred on these locations, with the majority of pedestrian fatalities shown to occur at nonintersection locations (Balk, Bertola, Shurbutt & Do 2014; Kim, Brunner & Yamashita, 2008; Gitelman, Balasha, Carmel, Hendel & Pesahov, 2012). Other locations include car parks, driveways, footpaths, underpasses or overpasses, and standing or sitting on the road. While signalised intersections appear to be safer for pedestrians than uncontrolled intersections, they are still dangerous situations for pedestrians. A recent UK report identified that 25% of pedestrian deaths and serious injuries in London occur at pedestrian crossing facilities (e.g. pelican, puffin or zebra crossings) of which half (13%) were signalised (Greater London Authority, 2014). A recent examination of Victorian policereported casualty pedestrian collisions in Melbourne CBD (for the period ) showed that almost half of pedestrian-involved crashes occurred at signalised intersections (Oxley, Yuen, Corben, Hoareau & Logan, 2013). Being struck on either the near side (that is, the beginning of crossing from the pedestrian perspective) or far side (that is, the end of the crossing task from the pedestrian perspective) of the road is another location variable that has been examined. The Transport Accident Commission (TAC) in Victoria reported that between 2009 and 2013, of the 209 pedestrian fatalities (in Victoria), 32% involved the pedestrian crossing the road and being struck from the near side, 19% from the far side and 13% while playing, working, lying or standing on a carriageway (TAC 2014). Austroads 2016 page 8

19 From a safety perspective, understanding the characteristics of where pedestrians prefer to cross a road, conditions under which they decide to cross and compliance with traffic control are critical (Sisiopiku & Akin 2003). This section will focus on those studies that have investigated the location of pedestrian crashes. The following section will cover those studies that have investigated built environment, situational and environmental factors. In a Victorian study that conducted 200 telephone interviews with injured pedestrians, Nieuwesteeg and McIntyre (2010) found that pedestrians were usually not at fault when crossing at intersections, but were usually at fault when crossing midblock. Right turns at an intersection were found to be particularly problematic, with vehicles not giving way to pedestrians. Near-side crashes were common in midblock cases with parked cars highlighted as a risk factor. Respondents were aged years (n=110) and 60 years and over (n=90). Based on observations in Brisbane, King, Soole and Ghafourian (2009) showed that most pedestrians waited for the green man signal before crossing and the most common illegal behaviour was crossing away from the signals but within 20m, followed by crossing against a flashing red man, then against a steady red man. Relative risk ratios were calculated using crash data for the observation sites over an 11 year period. This showed that crossing against the lights and crossing close to the lights both had a crash risk per crossing event of approximately eight times that of legal crossing at signalised intersections. Using observations and survey data, pedestrian behaviour at, and perceptions towards, various facilities (including signalised and unsignalised intersections and midblock crossings) was examined by Sisiopiku and Akin (2003). Video footage was collected at the footpaths of the study site, and perceptions and preference information was collected through surveying users of the site. Results showed that most pedestrians crossed at a designated location, and compliance at crossings throughout the study site was 71.4%. Marked midblock crossings were found to be the most influential pedestrian facility. It was also evident that the crossing location, relative to the origin and destination of the pedestrian, was influential on pedestrian decisions to cross at a designated location. The effect of the availability of a pedestrian signal on pedestrian decisions to cross at a specific location was quite high (74%) (Sisiopiku & Akin 2003). Alhajyaseen, Asano and Nakamura (2013) recorded video footage at several signalised intersections in Japan to analyse the gap acceptance behaviour of left-turning traffic and their interactions with pedestrians. They found that drivers accepted shorter lags in the presence ofsingle pedestrians while being more conservative in relation to the gaps between multiple pedestrians. Findings suggest that drivers pay more attention as the number of pedestrians increases. Drivers also accepted shorter gaps between pedestrians crossing the near-side for the driver when compared to gaps between pedestrians on the far side (relative to the driver). The authors attributed this to the lower visibility of near-side pedestrians and their relative position to the driver s line of sight while turning, whereas far-side pedestrians can be easily seen by leftturners. The relationship between marked crossings (i.e. zebra crossings) and pedestrian crash risk is unclear. In the presence of a marked crossing it can be argued that pedestrians may falsely believe they are safer and thus attempt to cross the road without due caution (WHO 2013). Results from a before and after, observational and questionnaire-based evaluation of the installation of a marked crossing in Edinburgh (Havard & Willis, 2012) suggested that pedestrians were more likely to use the location to cross the road, waited less time to cross and walked more slowly after the zebra crossing had been installed than before. Pedestrians reported that they felt safer, and less vulnerable to traffic, and were more confident. A large matched-case study on pedestrian crossings provides evidence that the infrastructure factors influencing crash risk may be complex. This study involved 1000 marked crossings and 1000 matched unmarked comparison sites in the USA (US Department of Transportation, Zegeer, Stewart, Huang & Lagerwey, 2002). Legal crossings exist at all public intersections in the USA wherever there is a footpath on at least one side of the street. Midblock crossings can only exist if specifically marked. The study sites were intersection or midblock locations with no traffic signals or stop signs on the main road approach (uncontrolled locations). The marked crossings had one of six marking patterns (i.e. zebra, dashed lines or a ladder pattern). Very few had any type of supplementary warning signs and none had traffic calming measures or special pedestrian devices. The matched unmarked intersections sites were typically the opposite leg of the same intersection as the selected marked crossing site or a midblock location on the same street, usually a block or two away. Results showed that on two-lane roads, the presence of a marked crossing alone at an uncontrolled location was associated with no difference in pedestrian crash rate, Austroads 2016 page 9

20 compared to an unmarked crossing. On multi-lane roads with high traffic volume, having a marked crossing alone was associated with a higher pedestrian crash rate compared to an unmarked crossing. Findings indicate that marked crossings should not be implemented without additional safety measures (Zegeer, et al., 2002). 2.5 Influence of the built environment on likelihood of pedestrian crashes Although research on pedestrian distraction by smart phones is growing, the topic is relatively underdeveloped compared to other aspects of pedestrian safety, and the authors did not find any that related specifically to how features of the built environment influence likelihood of pedestrian distraction. Research on drink walking is also relatively scant, though blood alcohol levels for pedestrians is included in official crash data where testing has been carried out. Accordingly, this section summarises research on the impact of features of the built environment on pedestrian crash frequency rather than tendency for pedestrians to be distracted. Aside from the location of crossings in terms of whether they are at an intersection or midblock, various aspects of the built environment can affect pedestrian behaviour and safety. These include whether footpaths or off road walking areas are provided, land use characteristics (i.e. employment density) demographics (i.e. population density), transit supply (i.e. presence of metro stations) and road network characteristics (i.e. number of intersections and speed limits) (Miranda-Moreno, Morency & El-Geneidy 2011). Pedestrian collisions occur more frequently in urban areas than rural settings in high income countries, and the opposite is true in some low and middle income countries (WHO 2013) Land use Alavi (2013) found that pedestrian crash risk is multi-factorial in nature, with different sets of factors associated with risk across different space-time frameworks. In his analysis of pedestrian safety in the Melbourne CBD, the three most powerful predictors of pedestrian collision rate (during daytime hours) were land uses surrounding the intersection (floor space area of entertainment areas and number of legs with shops), road characteristics (major versus minor intersections, type of division of roads, hook turn possibility and percentage of left-turn movements) and public transport (number of bus stops and distance from nearest railway station). During hours of darkness, collision rate was highly correlated with the land use characteristics (floor space area of entertainment areas, capacity of amusement and gaming centres, capacity of cinemas, theatre, concert halls and stadiums, capacity of accommodation) and road characteristics (major versus minor intersections and type of division of roads). For midblock locations, the most powerful predictor of pedestrian collisions during the day was public transport (distance from nearest railway station, number of tram stops and routes, number of bus stops and routes) followed by the land use (density of shops, floor space area of offices, capacity of non-commercial accommodation and floor space area of entertainment areas) and road characteristics (number of driveways and length of midblock). At night though, midblock collisions were mainly predicted by the length of the midblock, followed by land use (floor space areas of non-commercial accommodation, capacity of cinemas, theatres, concert halls and stadiums, floor space area of entertainment areas and capacity of bars, taverns and pubs) and public transport characteristics (number of tram stops). Various factors were associated with injury severity. At an intersection, the correlates of major trauma were time of day, vehicle movement, pedestrian age, vehicle colour, and land use and public transport characteristics. At midblock, correlates were time of day, land use characteristics, and the interaction of day or week and speed zone. Austroads 2016 page 10

21 In earlier work, Oxley et al (2003) found that in Melbourne CBD during night hours, particularly on weekends, pedestrian collisions were clustered around night clubs and bars and involved a higher proportion of young adult males crossing at intersections. In contrast, collisions occurring during business hours were evenly distributed throughout weekdays, across multiple locations on streets, more prevalent around public transport facilities and with less severe injury outcomes than those occurring during night hours. Nieuwesteeg and McIntyre (2010) showed that pedestrians are usually injured in familiar locations that they are likely to be confident with, and are most likely to be struck in shopping, residential and business areas, on routine type trips such as going to work, school or shopping. Schneider, Ryznar and Khattak (2004) combined pedestrian crash data with survey data from drivers and pedestrians on their perceptions of safety/danger around a university campus and showed that certain locations on the campus were perceived as dangerous, though pedestrian crashes had not yet occurred there, while the actual locations of policereported crashes were not perceived to be dangerous by pedestrians or drivers Road factors and speed limits The impact of road factors such as number of traffic lanes, speed and location characteristics on pedestrian crashes has been investigated in several studies (Hanson, Noland & Brown, 2014; Wang, Haque, Chin & Yun, 2013; Balk et al 2014; Garder, 2004; Islam et al 2014; Luoma & Peltola, 2013). Many of these studies have analysed crash data to identify which road factors have the biggest impact on pedestrian safety. These include an analysis of pedestrian crash data from the State of Maine for the five year period For this study, analyses showed that 68% of all crashes occurred in clear weather, 75% occurred on dry roads, 61% happened in daylight and 71% on level, straight roads with adequate sight distance. Most crashes happened away from an intersection, and in the absence of any traffic control device or signage. However, 19% occurred at a three-leg intersection, and a further 17% at a four-leg intersection. In a prediction model, high speeds and wide roads were found to lead to more crashes (Garder, 2004). Islam et al (2014) analysed pedestrian crash data from Connecticut and found that greater crossing distance and small building setbacks (distance at which buildings are located from the edge of the road) were associated with larger numbers of pedestrian-vehicle crashes. The latter finding was counterintuitive in that vehicle speed is usually lower in areas where setback is small. Greater pedestrian activity and more complex interactions however may account for this. Luoma and Peltola (2013) investigated the effect of walking direction along rural two-lane roads on pedestrian risk. They analysed police-reported road crashes from Finland between 2006 and 2010 in which a vehicle had struck a pedestrian walking along the road. They showed that the mean effect of facing traffic compared to walking along the direction of vehicle traffic was a 77% decrease in pedestrian accidents. Wang, Haque, Chin and Yun (2013) examined pedestrian injuries in Singapore using crash data from 2003 to 2008, containing approximately 4000 pedestrian crashes. Severity of injury was found to be higher during night time, with the likelihood of fatal or serious injury being higher for crashes on roads with high speed limits, roads with multiple lanes, school zones, and roads with two-way traffic. Lack of footpaths is common in suburban and regional areas in Australia, however there is little literature that examines the role this plays in pedestrian crashes. A small number of US studies have shown that the presence or absence of footpaths plays a role in the probability of a crash occurring and the severity of the resulting injuries. A study using Police Accident Reports from 1993 to 1997 in New Hampshire found that the probability of a crash is two times more likely at a site without a footpath than at a site with one (Ossenbruggen, Pendharkar & Ivan, 2001). Hanson, Noland and Brown (2014) found that severity of pedestrian casualties has been shown to be associated with lack of footpaths and street buffers (i.e. planted areas, bike lanes, on-street parking) on one or both sides of the street, high speed roads, roads with six or more lanes and a median, and lack of traffic lighting when it is dark. This study used data derived from Google Street View as well as pedestrian casualty police record data in New Jersey between 2007 and 2009 (Hanson et al 2014). Similarly, Yu (2015) used crash data from 2008 to 2012 in Texas and found that the probability of a pedestrian being severely injured or killed in a crash decreased in areas with more footpaths. Austroads 2016 page 11

22 High density of licensed restaurants and venues for alcohol consumption may be associated with increased potential for vehicle accidents and possibly pedestrian injury in the US (Gruenewald et al 2002, cited in Palk, Davey, & Freeman 2009). In a spatial analysis of pedestrian injury hot spots in San Francisco, LaScala, Gerber and Gruenewald (2000) showed a positive correlation between reported had been drinking pedestrian collisions and number of bars per kilometre of roadway. In Australia, other studies suggest that when alcohol has been made more available through the introduction of extended liquor trading hours, there is a positive relationship between alcohol availability, licenced premises and traffic accidents across a number of Australian cities (Palk et al 2009). Palk et al (2009), also highlight that a number of cross sectional studies have demonstrated a similar close and positive relationship between the density of licensed premises and motor vehicle crashes. Balk et al (2014) analysed the characteristics of more than 70,000 crossings at 20 locations in the Washington DC metropolitan area. Among other findings, they showed that the longer the distance that pedestrians were required to travel to cross the road, the more likely they were to cross entirely during the walk phase of the light cycle. Pedestrians were more likely to cross during the don t walk phase on one-way streets than on two-way streets. When physical barriers like guardrails and fences were present between the roadway and footpath, pedestrians were less likely to cross at unmarked non-intersection areas. Finally, in an observational study of pedestrians in Florida (which has the highest pedestrian fatality rate in the U.S), Gawade (2014), identified various site characteristics, demographics and pedestrian behaviours. Results showed that as intersection size increased, pedestrians were more likely to cross away from crossings. Users of a T-intersection were less likely to violate the pedestrian signal and jaywalk than users on four-segment intersections. Those who did not use the footpath also tended to cross on red and jaywalk rather than crossing on green; indicating that those who violate one pedestrian law may also tend to violate others Environmental factors Several studies have also investigated the effect of environmental conditions such as adverse weather conditions (Li & Fernie 2010), low light conditions (King, Wood, Lacherez & Marszalek 2012) and night time conspicuity (Tyrell, Wood & Carberry 2004; Yagil, 2000) on pedestrian crashes. The effect of winter weather on pedestrian compliance rates has also been investigated with results showing that road crossing behaviour was less safe in inclement weather conditions than in fine weather (Li & Fernie,2010). Most pedestrian fatalities are found to occur during the evening or at night time (Stimpson, Wilson & Muelleman 2013; Cairney & Coutts 2003; Eichelberger, Cicchino & McCartt 2013; Harruff, Avery & Alter-Pandya 1998). A recent UK report that was launched at the 2014 National Road Safety Conference analysed the injury data from 30,000 adult pedestrians who were injured in road collisions between 6pm and 6am during the period Results showed that males were at greatest risk of being injured as a pedestrian at night; that casualties often lived in similar types of communities; and that their actions often contributed to the collision through alcohol impairment, wearing dark clothing and/or dangerous actions in the carriageway (Road Safety Great Britain 2015). Yagil (2000) found that darkness was perceived to be the most influential physical factor for pedestrian safety and increased the tendency to wait for the walk sign. The influence of optimism and group-serving interpretations of safety was investigated among pedestrians (and cyclists) in terms of road use in general and under low light conditions with results supporting the existence of a group-serving bias, both overall and under the low light conditions (King, Wood & Marszalek, 2012). The authors presented the example that pedestrians and cyclists show a lower level of agreement than drivers with the statement that it is dangerous for pedestrians and cyclists to use the road in low lighting conditions. An implication of this finding is that pedestrians and cyclists overestimate their visibility to others at night. Tyrrell, Wood and Carberry (2004) examined pedestrian estimates of their own visibility in an experimental closed-road circuit task and confirmed that pedestrians do indeed overestimate their conspicuity to approaching drivers at night. Some authors have investigated the effect of daylight saving time which shifts an hour of daylight to the busier evening traffic hours on pedestrian fatalities (Ferguson, Preusser, Lund, Zador & Ulmer 1995). Austroads 2016 page 12

23 2.6 Walking after having consumed alcohol ( drink walking ) In 2010, approximately two in five (39.8%) Australians aged 14 years and older drank at least once in the last 12 months in a way that placed them at risk of an alcohol-related injury from a single drinking occasion. Males were far more likely to do so than females, with 64% aged years, and 54.9% aged years placing themselves at risk of an alcohol-related injury at least once a month (AIHW 2011). Alcohol remains a major cause of road death among pedestrians in Australia. Evidence suggests that males and young adults are more likely to engage in the behaviour of walking after having consumed alcohol, or drink walking and that the impairment caused by alcohol may affect safe crossing decisions (Gannon, Rosta, Reeve, Hyde & Lewis 2014; Lang, Tay, Watson, Edmonston & O Connor 2003; McGhie, Lewis & Hyde 2012; Lascala, Gerber & Gruenewald 2000; Oxley, Lenné & Corben 2006; Cairney & Coutts 2003; Tulu et al 2013; O Connor, Tay, Watson, Edmonston & Lang 2004; Fontaine & Gourlet 1997; Holubowycz 1995; Harruff, Avery & Alter-Pandya 1998). Several studies have examined Australian crash data in an attempt to quantify the proportion of pedestrian crashes involving alcohol (Austroads, 2004; Cairney & Coutts, 2003; Hutchinson, Kloeden & Lindsay, 2009; Eichelberger, Cicchino & McCartt, 2013; Garder, 2004; Nieuwesteeg & McIntyre, 2010). An analysis of Australian Transport Safety Bureau (ATSB) fatalities from is reported in Austroads (2004) and showed that of the 863 pedestrian deaths (aged 16 years and over) that occurred during that period, 35% had a blood alcohol concentration (BAC) of 50mg/100 ml or greater, 240 of them having a BAC of 150 mg/100 ml or greater. These figures represent 35% and 28% of all pedestrian fatalities respectively. In a subsequent study that analysed the characteristics of 240 fatalities of intoxicated pedestrians between 1999 and 2001 in South Australia, it was found that 84% were male, with most under the age of 35. Collisions occurred almost always at night, with Friday and Saturday nights over-represented. Most had been drinking for long periods of time, most often six hours or more, and most frequently at hotels or other people s homes. In 25% of the cases, the fatality occurred while walking home (30% when walking along the road, 20% when lying on the road, 2% when sitting on the road and 5% when standing on or near the edge of the road). Almost all pedestrians were struck within two hours of finishing drinking (Cairney & Coutts 2003). Hutchinson et al (2009) reported on crashes involving intoxicated pedestrians in South Australia for the period As found in the ATSB data (Austroads 2004), there were a disproportionate number of pedestrian casualties on Friday and Saturday nights. Most were male and aged between years. Most crashes occurred midblock rather than at an intersection, with no form of traffic control, and with the speed limit being 50 or 60km/hour. Nieuwesteeg and McIntyre (2010) analysed Australian pedestrian injuries and found that one quarter of the younger pedestrians had consumed alcohol in the three hours before the crash (and 3% of the older group). In the United States, Eichelberger et al (2013) used crash data to investigate the proportion of alcoholimpaired fatally injured pedestrians (and bicyclists) from Results were that although the number of pedestrians killed in motor vehicle crashes declined overall, the percentages with high BACs did not change much. Of those fatally injured, the percentage with BACs at or over 0.08% was 39% in 1992 and 37% in During , 20,326 pedestrians were fatally injured, and 37% of them had BACs at or over 0.08%. Similarly to the Australian characteristics, alcohol involvement in pedestrian crashes was higher among males (43%), those aged (47-50%), those killed on weekends (49%) and at night (60%). Garder (2004) analysed fatal pedestrian crashes in Maine and showed that 17% of those fatally injured were under the influence of alcohol. Austroads 2016 page 13

24 Some researchers have attempted to verify exactly how much alcohol impairs pedestrian safe performance of behaviours related to crossing the road. In an experimental design using a simulated environment, Oxley, Lenne and Corben (2006) compared sober and alcohol impaired participants on performance of road crossing behaviour. A total of 41 participants (aged years) took part in the study with 17 participants (nine males and eight females) forming the no alcohol group and 24 participants (fifteen males and nine females) forming the alcohol group. Participants in the alcohol group were given a dose of 0.75 ml of alcohol per kg of body weight (mixed with 480ml of orange juice) which was expected to produce a BAC in the range of %. A series of 15 traffic scenes in a simulated road environment were presented from the perspective of a pedestrian waiting to cross at the kerb, with two approaching vehicles traveling in the near side lane. Time gap and vehicle speed were manipulated. Participants sat two metres in front of the screen and, using a keyboard to select yes or no, responded to the traffic scenes as if they were to cross the road immediately behind a trigger vehicle and before the second vehicle. Results were analysed in terms of no alcohol, low BAC and high BAC. Results showed some trends that suggested that alcohol impairment affects pedestrian crossing behaviour, especially in the selection of safe gaps amongst adults with higher BAC levels. Alcohol did not affect walking speed. Decision times however were significantly shorter in the no alcohol group compared to those in the high BAC group, which, interestingly, were significantly shorter than those in the low BAC group. It was suggested that those in the low BAC group may have been more aware of their impairment and adjusted their decisions appropriately. Other studies have examined the drink walking behaviour and attitudes of young people by surveying and breath testing patrons as they leave licenced venues (Lang, Tay, Watson, Edmonston & O Connor, 2003; O Connor, Tay, Watson, Edmonston & Lang, 2004). In one study, seventy-eight young people under the age of 30 years (45 males) were surveyed as they exited a Queensland suburban hotel. Survey items examined demographic information, travel plans for the night, factors influencing travel choice, usual drinking patterns, understanding of drink walking and risk perceptions. At the end of the survey, respondents were offered the opportunity to provide a breath sample for alcohol content analysis. Results showed that over half of respondents (55%) recalled being impaired by alcohol while walking on at least one occasion (mainly falling over, stumbling or getting lost). Twenty-three percent reported sustaining an injury themselves (mostly minor) and nearly half (47%) knew someone who had been hurt while drink walking. For most respondents, the next intended venue was their own home (65%) or a friend s or partner s home (15%). Of those who were travelling home for the night and who gave a BAC reading (n=46), 50% had a BAC over or equal to 0.05mg/ml. Only eight participants (10%) reported that walking was their mode of travel for the night. On the night of the interview, convenience (40%) was the factor most often nominated as important when making travel arrangements for getting home after drinking, followed by drinking arrangements (22%). When asked how many drinks an average person could have before not being able to walk safely, the average estimated number was 7.22 drinks. However, participants also thought that this would depend on gender of the person and the type of drink (Lang et al 2003). In a further study, O Connor et al (2004) went on to survey and breath test 534 people as they left 12 licenced venues in Brisbane, Queensland. These included 328 participants aged years. Survey items related to past experiences of drink walking, knowledge and perceptions of the dangers and legal consequences of both drink driving and drink walking, trip planning and factors influencing mode of transport choice. Walking was the most commonly reported mode of transport (46% for those who had already consumed alcohol) for arriving at the survey venue. Walking was also the most commonly reported mode of transport (46%) for getting to their next intended venue (either another pub 42% or home 39%). Most young participants (65.6%) said they make travel arrangements before going out drinking at least half of the time, but 20.7% reported never planning the homeward journey before going out drinking. Commitments for the following day, cost of taxis, where they were going and the arranged designated driver were the most influential factors in terms of how participants planned their transport on the night of the survey. Compared to those over the age of 25 years, the younger participants reported more instances of drink walking in the previous three months. Only 13.1% of the younger participants reported that they had not drink walked in the last three months. Younger people viewed drink walking on the road as less dangerous than the older group and were also more likely to have been hurt whilst drink walking. Austroads 2016 page 14

25 The influence of conformity, group identity and norms on individuals drink walking intentions has been examined (McGhie, Lewis & Hyde, 2012; Gannon et al 2014; Haque, Clapoudis, King, Lewis, Hyde & Obst, 2012). Intention to drink walk was assessed in one study using a scenario-based questionnaire that measured the influence of low group identity (i.e. strangers), high group identity (i.e. friends), high conformity (i.e. friends or strangers crossing against the red man) and low conformity (i.e. friends or strangers waiting at the red man) using a repeated-measures design. A self-report questionnaire was administered to 151 university students (62 males, 83 females) aged years (mean age years). Participants responded to each scenario and indicated their intention to drink walk using a 7-point likert scale ( strongly disagree to strongly agree ). Individuals reported the highest drink walking intentions when in the presence of friends (high identity) and when their friends were also crossing against the red man (high group conformity) (McGhie, Lewis & Hyde, 2012). Haque, Clapoudis, King, Lewis, Hyde and Obst (2012) used the Theory of Planned Behaviour to investigate the factors which influence individuals drink walking intentions. A total of 215 young people aged years (169 female) completed a self-report questionnaire which assessed the standard TPB constructs (attitude, subjective norm and perceived behavioural control) and the additional constructs of risk perception, anticipated regret and past behaviour. Results showed that 27.7% of respondents reported never drink walking in the past month, with 19.2% reporting rarely in the past month and 11.7% reporting not often in the past month. Participants reported having moderately high intentions to drink walk (on 7-point scale, strongly disagree to strongly agree ). The standard TPB variables explained 63.2% of the variance in individuals reported intentions to drink walk, and the additional variables explained a further 6.1% of the variance. Gannon et al (2014) further investigated which normative influences could be used to predict young pedestrians decisions to walk while intoxicated. Using the Theory of Planned Behaviour, a self-report questionnaire was administered to 118 young pedestrians (aged 17-25) in relation to intentions to drink walk. The questionnaire included demographic characteristics and past drink walking frequency, TPB-related items related to intention to drink walk, attitude, subjective norm, perceived behavioural control, risk perception and group norms (friendship and parent. Overall, the extended TPB explained 72.8% of the variance in young people s intention to drink walk within the next six months, with attitude, perceived behavioural control, friendship group norm, and gender (male) as significant predictors. Males had higher intentions to drink walk and lower perceptions of risk regarding drink walking. It is possible that these findings regarding drink-walking may be relevant to research on, and understanding of, the use of distracting technology while walking. For instance, each of the factors above, that is, perceptions of risk, personal attitudes and beliefs about walking, attitudes and behaviour of important referent groups, modelling from important others, and perceptions of both likelihood and severity of negative outcomes, represent meaningful aspects of the determinants of behaviour. As such, greater knowledge about their influence over distracted walking is useful to informing and developing effective countermeasures Summary of the research on pedestrian distraction from mobile phones or alcohol impairment Prevalence of pedestrian distraction by hand held technological devices appears to be increasing; up to 40% of observed pedestrians are distracted when crossing the road. Pedestrians distracted by mobile phones may be at increased risk of collisions due to more erratic behaviour while distracted (compared to non-distracted pedestrians). Mobile phone users walk more slowly, change directions more frequently, are less likely to acknowledge other people, look left and right fewer times, are less likely to look at traffic before starting to cross, and make more errors than pedestrians who are not distracted Most pedestrian fatalities occur at non-intersection locations, and may be more likely to be at fault in these locations than at intersections. However there is little data in relation to identifying the extent to which pedestrian distraction by mobile phones or portable music devices is associated with pedestrian crashes. Illegal behaviours of crossing against the red man (entering on flashing and steady red man) and crossing within 20 m of the crossing but not at the crossing has been shown to have a crash risk approximately eight times that of legal crossing at signalised intersections Austroads 2016 page 15

26 Location of the crossing relative to the origin and destination of the pedestrian, availability of pedestrian signals the number of lanes and whether the road is one-way versus two-way influence pedestrian behaviour and the likelihood of disobeying a crossing signal. Pedestrian crash risk is multi-factorial in nature, with different sets of factors associated with risk across different space-time frameworks. Impairment from alcohol of pedestrian performance of crossing-related tasks has been verified by various different methods including: self-report crash data and simulator Crash risk elevation from alcohol impairment is also supported by evidence from simulator studies and analysis of crash data Overall it seems that alcohol impairment affects pedestrian behaviour especially in the selection of safe gaps and crossing decision times. 2.7 Countermeasures to distracted or risky pedestrian behaviour As stated by the World Health Organisation: Pedestrian collisions, like all road traffic crashes, should not be accepted as inevitable because they are, in fact, both predictable and preventable. Key risk factors for pedestrian road traffic injury are vehicle speed, alcohol use by drivers and pedestrians, lack of safe infrastructure for pedestrians and inadequate visibility of pedestrians. Reduction or elimination of the risks faced by pedestrians is an important and achievable goal (WHO 2013, p. ix). The final section of this literature review will summarise countermeasures to distracted or risky pedestrian behaviour with a focus on those that have been trialled in Australia or New Zealand, though ideas from other countries will be mentioned where these appear to have relevance. In addition, any evidence of the effectiveness of specific countermeasures will also be given. Countermeasures will be discussed in terms of three categories: Education, Enforcement and Engineering. There is also a section that addresses pedestrian friendly smart technology in interventions. When considering pedestrian safety more broadly, each of these plays a critical role. When considering distraction specifically, it s expected that education and enforcement will take precedence. It is important however to acknowledge that pedestrian safety is a multi-disciplinary problem and when considering the Safe System approach to road safety, risk factors and interventions related to road users, vehicles and the road environment are addressed in an integrated manner (WHO 2013). Recent developments in detailed real time intersection traffic analysis (including pedestrian traffic) via video technology will also be reviewed for potential countermeasure proposal/development Education and Awareness Campaigns Changing the attitudes and behaviour of drivers and pedestrians is a complex, long term undertaking that requires a multi-faceted intervention approach. It relies not only on knowledge and skills, but also on community support, perceptions of vulnerability and risk, social norms, engineering measures and law enforcement (WHO 2013) a Targeting pedestrian use of technology while walking and crossing the road The communication of public health messages through mass media has seen the release of various safety awareness campaigns to target pedestrian use of technology when crossing the road. Across Australia and New Zealand, these include: Austroads 2016 page 16

27 In New South Wales, the Local Government Road Safety program and Sydney City Council launched a twopart central-city pedestrian safety campaign, Watch Out, in The first part targeted motorists and professional drivers while the second part was aimed more at pedestrians. Both consisted of media advertising and print-based road safety messages to raise awareness of risks of pedestrian crashes in the inner city area and advise drivers and pedestrians of the numbers of people killed in the areas immediately local to the sites of the posters and advertising (see data/assets/pdf_file/0006/137418/woca-wynyard-cross-track.pdf; data/assets/pdf_file/0005/137417/woca-town-hall-cross-track.pdf; In December 2010, the Pedestrian Council of Australia launched the Lambs to the slaughter wait for the green pedestrian safety campaign (see Figure 2.1). The awareness campaign targeted conformity behaviour of many pedestrians, invoking the metaphor of acting like sheep when crossing the road, i.e. when one or a few people cross, others may tend to follow without necessarily checking that it is safe to do so. Many of these may also be listening to ipods and MP3 players, texting or using mobile phones. The Pedestrian Council of Australia has also launched its latest pedestrian safety awareness campaign targeting pedestrians using electronic mobile devices when crossing the road Don t tune out. Stop look listen think ( (see Figure 2.2). Figure 2.1: Newspaper advertisement from the Pedestrian Council of Australia s Lambs to the slaughter campaign Source: Pedestrian Council of Australia Austroads 2016 page 17

28 Figure 2.2: Advertisements from the Pedestrian Council of Australia s Don t Tune Out campaign Source: Pedestrian Council of Australia Queensland s Join the Drive road safety initiative includes a Share the Road campaign asking Queenslanders to say thanks to other courteous road users and to model safe behaviour. The campaign covers courteous road use by drivers, cyclists, truck drivers, motorcyclists and pedestrians. In the pedestrian video clip (see Figure 2.3) there is a woman exercising while listening to music through ear phones. As she approaches the road, she removes her ear phones and looks both ways. An approaching car stops and gives way for the pedestrian to cross the road (The State of Queensland 2015). Figure 2.3: Join the Drive s Share the Road campaign Source: The State of Queensland 2015 Austroads 2016 page 18

29 In December 2007, the City of Melbourne commenced a Share the Road campaign. The first phase involved actors dressed as pedestrian signal red man and green man talking to pedestrians and showing them how to cross safely at the Swanston Street and Flinders Street intersection in the CBD (see Figure 2.4). The campaign highlighted the fact that headphones and mobile phones cause distractions and can impose a crash risk and that the highest numbers of crashes happen between 8am and 6pm. It also highlighted that over 30% of pedestrians killed had a BAC level above The Red Man, Green Man campaign was followed by an educational and community services announcement in the New Year. spx and Figure 2.4: City of Melbourne s Share the Road Campaign: Red Man, Green Man Source: City of Melbourne Auckland City Council has launched various outdoor adverts at danger spots around the CBD, including the Shark, Snake and Fire footpath graphic designs by Ogilvy New Zealand (see Figure 2.5). The murals were accompanied by pedestrian crash statistics and the line Don t step into danger. In 2011, Auckland Transport launched a six week pedestrian safety campaign targeting both pedestrians and drivers. The campaign was called Check before you step and aimed to address the increased risk to teen and adult pedestrians due to changing demographics of people living in Auckland. Harre and Wrapson (2004) conducted an evaluation of a central city pedestrian safety campaign in Auckland. Five intersections were targeted for use of visual media, interactive footpath mime and rewards for safe behaviour. Results showed that the proportion of pedestrians crossing on a red light almost halved from 6.61% before the campaign to 3.63% after the campaign. Crossing during the red light phase was significantly lower during the footpath mime intervention period (2.42%) compared to prior to the intervention (4.73%). However, no improvements were seen in driver behaviour. A survey of attitudes showed high awareness of the campaign and some improvement in knowledge amongst those surveyed, but no change in general attitudes. Austroads 2016 page 19

30 Figure 2.5: Auckland City Council s Don t Step into Danger: Fire campaign Source: Figure 2.6: Auckland City Council s Don t Step into Danger: Snake campaign Source: Austroads 2016 page 20

31 Wellington City Council launched a road safety campaign in the CBD encouraging people to cross the road safely: Cross the road with a clear head (see Figure 2.7). It aimed to make people more aware of traffic and their environment: to think, and look, before crossing. The message was displayed on posters, buses and bus stop signage, as well as road safety billboards. Other measures implemented at the same time included wider footpaths, street furniture, more crossing points, and consultations with the community on lowering the speed limit in the city centre. Figure 2.7: Wellington City Council Cross the road with a clear head campaign Source: In addition, Wellington City Council also attempted to highlight the dangers of distraction when crossing the road, discouraging pedestrians from using mobile phones or headphones. In a recent campaign that focused on raising awareness about distraction, respondents to a Facebook quiz won prizes for entering and telling the Council what distracted them when walking around the city and crossing the road. A Word Cloud (see Figure 2.8) was created to illustrate the range of potential distractions that people nominated when walking around the city. and Austroads 2016 page 21

32 Figure 2.8: Word Cloud illustrating the range of potential distractions that people nominated when walking around the city Source: Wellington City Council Christchurch City Council launched a pedestrian safety campaign Could they stop? Cross safely aimed at jaywalking pedestrians who take risks to get to their destinations (see Figure 2.9). Figure 2.9: Christchurch City Council s Could they stop? Cross safely campaign Source: Christchurch City Council b Targeting pedestrian drink walking Countermeasures in the drink walking space have previously been classified as prevention of high levels of intoxication, minimising pedestrian activity when intoxicated, and minimising risk of injury among intoxicated pedestrians (Brooks 1998, cited in CASR 2009). The effectiveness of these strategies however is largely unknown, with difficulties in both implementation and measurement. In the USA, the NHTSA report on safety countermeasures outlined that communication and outreach can be directed at a variety of audiences, including law enforcement, drivers, alcohol servers, community leaders and friends and family of likely impaired pedestrians. Targeting impaired pedestrians themselves is another obvious measure, though arguably more difficult and that reaching others who are in a position to help prevent such crashes may be the most effective (NHTSA 2011). Austroads 2016 page 22

33 In relation to the effectiveness of countermeasures, Miller, Curtis, Chikritzhs and Toumbourou 2015 recently summarised a range of interventions for reducing alcohol supply, alcohol demand and alcohol related harm, using a literature review and Delphi study. Ratings were given to the interventions according to their impact or level of effectiveness across a range of outcome measures. In terms of reducing alcohol-related harm, the literature review identified a paucity of well-researched and strong evidence-based interventions internationally (Miller et al 2015). However, Miller et al, 2015 provided a rating for the different approaches and countermeasures they reviewed. From this, four approaches reached the second highest rating for effectiveness: targeted police interventions; Safer Bars; mandatory plastic glasses; and community interventions. Targeted police interventions involved highly visible enforcement of drinking laws in and around licensed premises, and have been found to reduce alcohol-related assaults, as well as harm more generally. Safer Bars is a comprehensive strategy for training staff to deal with problem behaviour by patrons, and has been found to be effective in changing staff attitudes and reducing bar room aggression (Graham, Jelley, & Purcell 2005, cited in Miller et al 2015). Use of plastic glassware in venues is an effective and easy to implement strategy for reducing glass related injuries and assaults. Finally, community interventions included publicity campaigns, local task force activities, and community forums and discussion groups, including responsible service of alcohol interventions. While this strategy appears promising, previous implementation was judged to be in a context of low level (if any) regulation or enforcement and was thus regarded as unlikely to have further impact in Australia (Miller et al 2015). The Delphi study showed that 43 other interventions were moderately effective, with those rated as most effective being sections of legal acts relating to reducing supply. Reducing the alcohol content of beverages served after midnight and banning the sales of shots and similar alcoholic beverages also rated well (Miller et al 2015). For drink walking, the importance of promoting awareness not only among pedestrians but also among drivers to alert them to the possibility that pedestrians might be impaired has been recognised. Queensland s Safe Night Out Strategy outlines detailed action points relating to alcohol-affected patrons to ensure a safe night out for all. The strategy covers penalties for anti-social or violent behaviour, policing of alcohol and drug-related violence and anti-social behaviour, ways to sober people up if drunk and disorderly in public, as well as the responsible service of alcohol. A communication strategy was implemented to provide clear and targeted messages to people across Queensland, and safe night out precincts for key entertainment areas in the State (Queensland Government, 2014) were also established. As yet, this strategy has not been evaluated. In December 2011 the Pedestrian Council of Australia launched the Never let a mate walk home drunk campaign (see Figure 2.10). Figure 2.10: Never let a mate walk home drunk campaign. Source: Pedestrian Council of Australia Austroads 2016 page 23

34 State-based initiatives South Sydney developed the Drink Safe Walk Safe project in July 1996 which ran until November The project targeted those inner city streets with high numbers of hotels and night venues and was aimed at people aged years. While licenced premises operators in the area were apparently aware of the effect of alcohol on pedestrian safety, with most having observed dangerous pedestrian behaviour outside their hotels late at night, they regarded themselves as not responsible for the safety of patrons once they had left the premises. Licensees saw their responsibility as limited to responsible service of alcohol rather than infrastructure changes around venues. The campaign used the promotional slogan Don t be another staggering statistic and involved promotional material for entertainment venues, railway stations, bus rear walls, as well as cinema advertisements, lobbying of the RTA for changes to pedestrian walk phases at busy intersections in entertainment areas during non-peak times, development and use of Pedestrian Accident Area Warning Signs, Road Safety Audits, ongoing media promotions, promotions at the Sydney Football Stadium and Sydney Cricket Ground during major sporting events, liaison with the deliverers of Responsible Service of Alcohol courses and development of a promotional video. Over the two year campaign period there was a 45% reduction in the number of alcohol-affected pedestrian accidents that were recorded; however this figure should be interpreted with some caution. Following the campaign, recommendations included encouraging the use of more aggressive advertising, development of pedestrian-oriented road infrastructure, targeting motorist behaviour in entertainment venues, vehicle speed control, exploring the possibility of using enforcement to deter behaviour, involving management of hotel venues in the planning of environmental infrastructure changes, and incorporating pedestrian safety education into staff training courses (Elliot 2002). Using the same slogan as the previously implemented campaign in South Sydney, Drink Safe Walk Safe was initiated to address the issue of pedestrian safety, particularly targeting those affected by alcohol in the Wollongong City area. The initiative used a major event that the region was hosting during 2003 to raise the profile of drink walking and place it on the local road safety agenda. Strategies included responsible service of alcohol, with information and posters sent to venues in the area to remind them of their responsibility and encourage them to promote the campaign slogan. Through development of a Pedestrian Management Plan, marshalling techniques were used to effectively direct people to the event venues from the main transport stations. Promotional material was used as widely as possible, including 2000 glow sticks displaying the campaign message, 250 posters distributed to venues, with 20 volunteers recruited from local service groups. General pedestrian and driver education strategies were also adopted using promotional information and 10 radio advertisements to reinforce the safety message. A series of telephone surveys conducted after the event found that 87% of venues in the CBD had participated in the strategy as part of their ongoing responsible service of alcohol policy. In addition, there was a 40% increase in the proportion of people who reported that they had been exposed to a drink walking strategy in the Wollongong area after the implementation of the strategy, and a 10% increase in the proportion of people who could identify drink walking as a road safety issue (Campbell 2005). South Australia, through the Motor Accident Commission in SA, currently operates a drink walking campaign. Downloadable campaign print materials and information in relation to the statistics for pedestrian drink walking-related crashes are provided on a semi-interactive website hosted by the Motor Accident Commission: This campaign is not yet evaluated. While Victoria hasn t developed a campaign specifically targeting intoxicated pedestrians, other more general campaigns have the potential to include pedestrian safety issues. These include the drink driving resource Looking after our mates, and the Party Safe program. Walk Safe is a multi-action program developed and implemented by the City of Stonnington, and is now also used by City of Port Philip. The program recommends the inclusion of intoxicated pedestrians as one of the key high risk groups (Elliot 2002). In partnership with VicRoads, the Victorian Policy and the TAC, the Walk Safe program involved the installation of various treatments at key activity locations in the area (e.g. painted strips between tram tracks, coloured and textured pedestrian crossings, pedestrian fencing and some timing changes to traffic signals) Walk Safe was supported by a major publicity campaign which included street banners, an effort to encourage all schools and pre-schools to undertake road safety education, and a short period of enforcement of traffic regulations relating to pedestrians. A trial of these measures was undertaken at 17 locations around the main entertainment area in Stonnington. Through a series of intercept surveys with 106 members of the public during weekend evenings between 6-9pm, it was found that most participants Austroads 2016 page 24

35 considered pedestrian safety at least a slight problem in the City of Stonnington, with 18% considering it a major problem. Over half those surveyed (58%) said that they considered their safety as a pedestrian when they drank alcohol, with Stonnington residents more likely than non-residents to report they do so and residents significantly more able to recall Walk Safe activities than non-residents. The most powerful message was found to be the chalk body outline. However, most believed that the Walk Safe activities didn t influence their behaviour and many said that they were sensible drinkers to begin with. One quarter of participants reported that advertising would be the best method to make them think about their safety as a pedestrian and most reported that they would support a program to promote the responsible serving and consumption of alcohol (Cairney, Stephenson & Macaulay 2004). In 2002 Queensland Transport developed a public education campaign aimed at intoxicated pedestrians. The message was Drink walking is a killer on our roads and incorporated low level advertising on the rear walls of buses, posters and coasters. The visual image for the campaign was two beer glasses representing legs walking (Elliot 2002). In 2010, the Queensland Government commenced a two year trial of Drink Safe Precincts (DSPs) to specifically reduce alcoholrelated violence in the key entertainment precincts of Surfers Paradise, Townsville and Fortitude Valley (Queensland Government 2013). The Queensland trial was evaluated after 14 months and again at 18 months. The key components of the trial were increased and high-visibility policing, enforcement of liquor licensing laws, providing support, rest and recovery services, and coordinating venues, police, ambulance, community support services and transport providers. Additional strategies included improved taxi zones and supervision, improved transport information, improved lighting and other crime prevention initiatives, such as measures to reduce crowding and footpath queuing. Both evaluations concluded that there were some aspects of improved community safety and reduced alcohol-related violence. However these positive conclusions were not supported by the data, and as reported by the Queensland Audit Office (2013), the trial was poorly planned, implemented and evaluated (Miller et al 2015). Another Queensland measure was the Sunshine Coast Drink Safe Coalition Project. This was conducted during and involved coordinated and facilitated multi-component strategies to promote safe drinking and reduce binge drinking particularly in young people. The Project delivered more than 40 Drink Safe events and activities and partnered and supported a further 25, with an estimated outreach to 2000 young people. No evaluation has been conducted (Miller et al 2015). Some communities in Western Australia have also targeted alcohol-affected pedestrians using educational material. The Keep Light while Walking at Night campaign targeted Aboriginal pedestrians in Broome. Information relating to pedestrian safety was also made available in a problem area in Kalgoorlie, with regular police enforcement undertaken. A program was also undertaken in Cottesloe, with involvement from local businesses, implementation of engineering measures such as kerb extensions, painted medians and upgraded street lighting, and extra police presence during busy times (Cairney, Stephenson & Macaulay 2004). In previous years, the Northern Territory Government used television commercials to help raise awareness about sharing the road environment and the safety responsibilities of both pedestrians and drivers. Currently, the Northern Territory addresses the issue of pedestrian distraction through road safety education in schools. Alcohol management plans were developed in 2004 as part of the Northern Territory Alcohol Framework. Although the plans address supply, demand, and harm reduction, so far this measure has mainly focused on reducing supply through a set of alcohol restrictions (Senior et al 2009 cited in Miller et al 2015). Margolis et al (2008 cited in Miller et al 2015) evaluated the plans in four remote communities in the Northern Territory and found them to be effective in reducing serious injury in the communities. More recently, a partnership between the Northern Territory Government, the City of Darwin and the Australian Hotels Association (NT Branch) has resulted in the implementation of an alcohol-related safety initiative, Darwin Safe, to improve safety in the CBD. Components of this initiative include several of the alcohol service measures reported as effective (Miller et al 2015) such as banning shots/shooter style drinks after midnight, use of plastic or tempered glassware, and Lock-Outs. There are also enforcement measures Territory-wide (such as Alcohol Protection Orders, Declared Precinct Areas, increased fines for failure to refuse to leave a licensed premises) which work in conjunction with responsible service of alcohol. Though the initiative primarily targets the prevention of alcohol-fuelled violence, crime, and disorderly conduct, the approach of reducing intoxication has the potential to improve road safety generally and pedestrian safety specifically. In particular, the initiative includes continued funding for the employment of security staff and a concierge to provide Austroads 2016 page 25

36 security on Friday and Saturday nights at a taxi rank located near late-night licenced venues ( A US program in the City of Baltimore, where 40% of pedestrian crashes were found to involve alcohol, outlined a set of best practice guidelines for the development and implementation of countermeasures to reduce alcohol involved pedestrian crashes. A comprehensive countermeasure program was implemented in the city centre to cover the locations of 73% of the previous pedestrian alcohol crashes in 21% of the land area (Blomberg & Cleven 2000). Walk Smart Baltimore was adopted as the program slogan with a footprint as its logo. The program involved a range of media announcements and advertisements across multiple local organisations. A video was produced and used to train police officers about appropriate responses to the pedestrian alcohol problem. Other measures were distribution of retro-reflective caps to pedestrians, installation of road way signs warning of high pedestrian traffic, display of rotating message banners. Engineering measures included improvements to lighting and crossings, and removal of roadway items that might restrict sight distances of both drivers and pedestrians. The authors suggested five steps to a community intervention to reduce pedestrian alcohol-related crashes: 1. Assessing the local problem determining the magnitude of the problem and contributing factors, when and where the crashes occur in the community; 2. Establishing a community coalition obtaining widespread local support; 3. Designing the program select or develop a variety of problem-specific and culturally appropriate remedial actions; 4. Implementing the program create and carry out a plan for implementation; and 5. Evaluating the program collect and analyse data to determine program effectiveness (Blomberg & Cleven 2000). In his review of possible strategies for reducing intoxicated pedestrian casualties, Elliot (2000) recommends that the most likely effective countermeasures will not be standalone, highly targeted strategies. The top priority should focus on reducing alcohol consumption through enforcing existing laws in relation to serving intoxicated patrons. Such enforcement should be accompanied by mandating training in Responsible Service of Alcohol for all staff who come into contact with drinking patrons and a mass media campaign aimed at encouraging drinkers to accept the legal requirement for a server to say no to already intoxicated patrons. Elliot argues that there is also a need to raise pedestrian awareness about the dangers of alcohol while walking so that being intoxicated is perceived as equally dangerous when walking as when driving. Educating young people at school about these dangers was one proposed option. (Elliot 2000). While driver training has not yet been shown to reduce overall crash rates, adding more pedestrian information to such training could increase knowledge and desire to share the road safely with pedestrians, especially in relation to how to avoid common types of crashes with pedestrians and to improve drivers anticipation of and interactions with pedestrians (as well as also helping to make them safer pedestrians) (NHTSA 2011) Enforcement countermeasures General (non alcohol-impaired) pedestrian behaviour is regulated through road safety laws in most jurisdictions. In the USA, violations include crossing with undue care and attention and illegal crossing such as jay walking. While locations for pedestrians to cross the road, and the manner in which crossing the road should occur are regulated in both Australia and New Zealand, other pedestrian behaviours, such as distracted walking or drink walking are not, with the exception of South Australia where there is a penalty of $106 in relation to must not walk without due care for pedestrians. In addition, because pedestrian behaviour tends to be less constrained by infrastructure, and illegal behaviours may be dispersed geographically or temporally, enforcement is problematic and may lead to public uncertainty about the legal implications. However, lack of enforcement arguably increases complacency around walking behaviour. Enforcement of general road traffic rules for motorists, such as speeding and giving way to pedestrians, is regarded as critically important for the safety of pedestrians. Little research reporting the trial or subsequent effect/evaluation of enforcement of pedestrian behaviour was found for this review. Austroads 2016 page 26

37 The NHTSA has previously argued that enforcement campaigns for pedestrian safety should be aimed at both drivers and pedestrians and should start with the necessary communications and outreach educational efforts to announce and describe the campaign. In keeping with principles of deterrence, enforcement activities are more effective when highly visible and publicised. Enforcement messages then reinforce the safety/legal behaviour requirements and raise the expectation that failure to comply may result in legal consequences. Recommendations have included that enforcement campaigns should involve coordinated programs that include training for law enforcement officers, prosecutors and judges so that everyone understands the purposes of the campaign (NHTSA 2011) a Enforcement approaches to drink walking When considering enforcement countermeasures specifically for drink walking, training programs for responsible service of alcohol are available in different Australian states. Better enforcement of the law around serving alcohol to an intoxicated person is arguably a point at which intervention could take place (Hutchinson et al 2009). More vigorous enforcement, training of bar staff, licenced venue staff encouraging the friends of intoxicated people to look after them, provision of minibus services and other ways of organising transport, provision of bus stops and taxi ranks close to pubs and clubs as well as considerations around civil liability have been suggested. Loxley et al 2004 (cited in Hutchinson et al 2009) highlighted several other enforcement interventions that could be considered for the general prevention of occasional alcohol abuse and which might then have flow-on effects for pedestrian safety. These were: 1. Increased alcohol taxation; 2. Hypothecated taxes on alcohol to fund treatment and prevention programs; 3. Restriction trading hours; 4. Responsible service of alcohol, provided such programs are accompanied by visible enforcement; 5. Restrictions on alcohol price discounting; 6. Licensee codes of conduct, provided these are accompanied by enforcement; 7. Licensing restrictions in indigenous communities; and 8. Declaration of indigenous communities as dry. Increased policing of general public drunkenness to the extent possible, through increased police presence and lock out laws such as those recently introduced in NSW is another level of enforcement. Sweeper programs that effectively sweep alcohol impaired pedestrians from the streets until they no longer have high BACs are intended to reduce pedestrian exposure to traffic risk. Problems lie in identifying at-risk pedestrians, determining responsibility for picking up the targets, having somewhere they are kept until they are sober, deciding whether friends or family are notified, having methods for returning people home after the intervention, costs involved and the legal rights of those being swept (NHTSA 2011). Queensland has also trialled a lock out strategy on the Gold Coast in 2004 and Brisbane City and Fortitude Valley in 2005 to prevent patrons from entering late-night venues between the hours of 3-5am. In their evaluation of the strategy, Palk, Davey, and Freeman (2008) concluded that the lockout strategy may have been marginally effective for some kinds of offences in certain areas. Results showed that reported alcoholrelated incidents on the Gold Coast were proportionally reduced after the introduction of the strategy for all time periods included in the study, and these reductions were significant during the critical time period of 3-6am. The major categories of offences for incidents were alcohol-related disturbances/disputes, alcoholrelated sexual offences, and alcohol-related street disturbances. The only significant variation to occur in the Brisbane City area involved traffic offences during the 3-6am time period, with a reduction of 12.9% following the introduction of the lock out (Palk et al 2008). Most recently, lockouts have been evaluated in Newcastle, NSW. Others have investigated the impact of lockout laws on alcohol-related assaults around the country, with patterns of results across studies being unclear (Miller et al 2015). Austroads 2016 page 27

38 In South Australia, under the Road Traffic Act, a person must not walk without due care or attention or without reasonable consideration for other persons using the road or face a $105 fine. In one enforcement attempt, South Australian Police were reported to have fined and cautioned 62 pedestrians for walking when distracted in the 18 months to July distracted-by-their-mobile-phones-face-105-fine/story-fnii5yv ?nk= d67c0ec6519fa77c4fe Engineering In terms of road traffic and environment factors, various engineering options have been designed to assist pedestrian accessibility and safety. Corben and Duarte (2006) acknowledge that in locations where there is a high level of pedestrian activity, opportunities to reduce vehicle or pedestrian volume or to reduce road widths may be limited, or indeed impracticable. In such situations, the two alternatives are to actually separate pedestrians from vehicles in time or space, and secondly, to moderate vehicle speeds. When considering distraction or impairment specifically, aside from physically preventing pedestrians from interacting with vehicles (i.e. grade separation) there is limited capacity in this space to actually prevent collisions related to distraction or inattention. It is for this reason that education and enforcement are expected to be more important in addressing the problem of pedestrian distraction. However, there are promising intelligent transport and technology-based interventions that are being developed, trialled or evaluated for other aspects of road safety and which may be applicable and appropriate to pedestrian safety. Section c summarises some of these. For completeness, an overview of general pedestrian safety engineering options will be provided in this section a Traffic engineering measures A range of traffic engineering measures have been developed and widely implemented to improve the movement and safety of pedestrians. An overview of these measures is provided here with supporting evidence in relation to each. Arguably, an effective starting point for pedestrian safety is to reduce pedestrian exposure to vehicles through separating pedestrians from road traffic. Effective measures include use of sidewalks/footpaths, marked crossings in conjunction with other measures such as traffic signals and appropriately located overpasses and underpasses. In a pedestrian crash reduction study that focused on safe roads and roadsides, Wellington City Council reviewed pedestrian casualties within the CBD area identified as the Golden Mile. The review identified a suite of possible route-wide treatments. These included a margin treatment between the footpath and road, a clear zone for sight distance improvement, site clearings at pedestrian crossing points, implementation of pedestrian crossings and pedestrian detection systems, lane and carriageway consistency and uniformity of treatments, improved street lighting, use of barriers, use of technology to warn pedestrians of an approaching vehicle such as a bus, speed management measures and consideration of all road users, not just pedestrians (GHD, 2012). A second stage of this work involved an observational study of pedestrian and vehicle behaviour and a human factors assessment at specific locations of interest in the Wellington CBD (Thomas, Rive, Frith & Charlton 2011). Signalisation of pedestrian crossings plays an important role in the management of vehicle-pedestrian interaction. As highlighted earlier, in their road user survey, Sisiopiku and Akin (2003) found that the effect of the availability of pedestrian signals influenced pedestrians decisions to cross at a specific location was high (74% indicated their decisions were affected). In a study that examined the effectiveness of countermeasures in New York City, Chen, Chen, Ewing, McKnight, Srinivasan and Roe (2013) showed that those related to signals such as split phase timing, signal installations, an all pedestrian phase and increased pedestrian crossing duration, reduce crashes. It should be noted that results for such interventions in lowerincome countries have found more mixed results. Quistberg, Koepsell, Boyle, Miranda, Johnston and Ebel (2014) investigated the presence of visible traffic signals, pedestrian signals, and signal timing to determine their effectiveness for pedestrian safety in Lima, Peru. A matched case-control design was used for 97 collisions at intersections. Collisions were more common where a phased pedestrian signal (green or redlight) was present compared to unsignalised crossings, and a longer pedestrian-specific signal duration was negatively associated with collision risk. Collisions were more common in the presence of any signalisation visible to pedestrians. The study has implications for pedestrian compliance with traffic signals in lower income countries. Austroads 2016 page 28

39 There are several types of pedestrian crossing signals. Pelican (Pedestrian Light Control) crossings, where pedestrians can use a push button to activate the pedestrian phase of a signalised crossing, have a walk/green man phase and a don t walk/red man phase. This type of crossing has been found to increase pedestrian safety as well as being perceived positively by the public (Malkhamah, Tight and Montgomery 2005) and is used widely in Australia. Puffin (Pedestrian User Friendly Intelligent) crossings are similar to pelican crossings, also having green and red pedestrian phases, with the addition of sensors that allow the adjustment of the length of the pedestrian phase to allow for quicker and slower pedestrian crossing times. These crossings also allow for a flashing amber vehicle signal that allows for early vehicle travel where there are no longer any pedestrians in the crossing. Lenne, Corben and Stephan (2007) examined traffic signal phasing at intersections to improve safety for alcohol-affected pedestrians specifically. They investigated a Dwell-on-red treatment whereby a red traffic signal is displayed to all vehicle directions during the period when no traffic is detected, so that drivers approach high-risk intersections at a lower speed. The treatment was associated with a reduction in mean vehicle speeds and substantial reductions in the number of vehicles travelling at threatening speeds in regard to the severity of pedestrian injury. Wall and Bauer (2012) investigated methods and technologies to make signalised intersections safer for pedestrians. Their study focused on acoustic beacons (audible tones generated at the destination curb), passive pedestrian detection that doesn t require pedestrians to make physical contact with the pedestrian button, and pre-emption that disrupts the normal traffic signal timing (i.e. rail road and emergency vehicle). The technologies were tested and evaluated. They concluded that audio beaconing is a low cost but highly effective method, passive pedestrian detection is simple to implement but equipment costs are high and that information conveyed through pre-emption must have a low cognitive load. A recent video was created showing a novel approach to try to get pedestrians to wait at the red man by making it more entertaining. A dancing red man is simulated in real time by pedestrians who can enter an enclosed space and stand there and dance. Instead of a stationary red man, the red man dances, capturing the attention of passing pedestrians who stop, wait and watch (see Figure 2.11). The initiative was conducted by Smart car manufacturer as part of a campaign promoting inspiration to think outside of the box. The result of the social experiment that took place in Lisbon, Portugal showed that 81% more people stopped at the red light and waited for the green man (The Dancing Man 2014). Long term benefits are unknown and would warrant further investigation. It is possible that pedestrians may become used to such measures, which might then lose their effectiveness. Similarly, touchscreen StreetPong was designed by students in Germany in an attempt to make waiting at the lights a more enjoyable experience. After overwhelming interest in this on social media, the students installed a small number of prototypes around the country and are now crowd funding to raise more money for large scale production. The multi-player game of touch screen ping pong named ActiWait is played wirelessly with a person on the other side of the crossing (Daily Mail 2015). Marked crossings indicate the optimal or preferred location for pedestrians to cross the road. They help to indicate a pedestrian right-of-way and motorists need to give way to pedestrians at these points. They are commonly installed at signalised intersections and other high volume pedestrian locations such as school zones (WHO 2013). It is recommended that they be installed in conjunction with other physical enhancements such as traffic signals, that they are not appropriate where traffic speed is high or on roads with more than two lanes, that locations should be convenient for pedestrians and accessible for those with a disability, they should guide pedestrians to cross at locations where there is street lighting at night and with adequate visibility between vehicles and pedestrians (WHO 2013). Austroads 2016 page 29

40 Figure 2.11: The Dancing Red Man campaign Source: WCv15qsmSM%253A%253B1B1pRgCVV87rAM%253Bhttp%25253A%25252F%25252Fwww.psfk.com%25252F2014% 25252F09%25252Fsmart-dancing-traffic-light-mancampaign.html&source=iu&pf=m&fir=rE1GWCv15qsmSM%253A%252C1B1pRgCVV87rAM%252C_&usg= kje5kwzn XQdkOBTiaPpS17oLDsA%3D&ved=0ahUKEwiV1fHIn5LLAhUY1mMKHS6gDQkQyjcIKA&ei=uJnOVpWZMpisjwOuwLZI #imgrc=re1gwcv15qsmsm%3a The use of countdown timers has also been widely implemented in many jurisdictions. Information is typically displayed as a numeric countdown that informs pedestrians how many seconds they have remaining to cross safely (McTiernan, Levasseur & Brisbane 2012). Two independent trials of pedestrian countdown timer (PCT) technology were undertaken in Australia during 2010 and The first was undertaken by VicRoads in 2010 at three signalised intersections in Melbourne CBD and the second by the New South Wales Centre for Road Safety in late 2010-early 2011 at four signalised intersections in Sydney CBD. Both used video technology for observations of pedestrian behaviour and intercept surveys to explore understanding of, and attitude towards, the PCTs as well as any self-reported changes in behaviour. Incident and near miss actions were observed through several conflict parameters, as well as pedestrian body language while waiting, actions while crossing, pedestrian approximate age, gender and visible signs of Austroads 2016 page 30

41 physical impairment. In the Victorian trial there were no reliable indications of an improvement in pedestrian behaviour or a reduction in risk after the installation of the PCT and results varied from crossing to crossing. A small improvement in compliance was observed at the busiest crossing, but less than half reported having noticed the PCT. In the Sydney trial there were also minimal and mixed results in terms of safety. The influence of PCTs was greatest at wider crossings and during periods with the highest number of pedestrians crossing per cycle (McTiernan, Levasseur & Brisbane 2012; Levasseur & Brisbane 2011). Count down timers were introduced in Brisbane in 2012 but to our knowledge have not yet been formally evaluated. London has recently announced a Pedestrian SCOOT technology that is currently in a trial phase across the network. It uses cameras to detect the number of people waiting to cross and alters the timing of the Green Man invitation to cross accordingly. That is, reducing the minimum time for the green man from 6 seconds to 4 seconds when only a few people are present but increasing the green man when there are lots of pedestrians present, to allow for all those waiting to get across the road. Questions remain about the minimum number of pedestrians needed to trigger the system, how this will be balanced with the number of vehicles waiting for pedestrians to cross, how many locations could benefit from the technology and whether it will replace existing countdown timers (Greater London Authority 2014). The impact of medians and centre refuges has also received attention with reductions found in speed and pedestrian exposure risk by allowing the pedestrian to cross one direction of traffic at a time (King 2003) and simplifying the decision making task for alcohol affected pedestrians (Cairney et al 2004). In their analysis of 1,000 marked crossings and 1,000 matched unmarked crossings, Zegeer et al (2002) found that raised medians provided significantly lower pedestrian crash rates on multi-lane roads compared to those with no raised median. It is argued that such measures increase perceived control over the crossing task (Stacey et al 2014). The impacts of physical barriers, even small barriers such as flower planters that prevent pedestrians from easily crossing have also shown to impact willingness to cross (Stacey et al 2014). The influence of guardrails on pedestrian crossing behaviour at roundabouts under various conditions was examined by Cohen, Bar-Gera, Parmet & Ronen (2013). Using 60 hours of video footage, 11,116 pedestrian crossings were observed (25% of which did not occur at a crossing and thus deemed a violation). Results showed that the rate of violations without a guardrail is 20-30% higher than with a full guardrail. Compliance was also higher when traffic volumes were higher. Pedestrian overpasses and underpasses allow for uninterrupted flow that is separate from vehicle traffic and are used primarily in areas with high pedestrian volume (WHO 2013) b Traffic calming measures to reduce vehicle speed Higher vehicle speeds increase crash and injury risk for pedestrians (Corben & Duarte 2006). The higher the speed of a vehicle, the shorter the time a driver has to stop and avoid a crash, including hitting a pedestrian. Higher speeds also mean that pedestrians may misjudge the speed of an approaching vehicle and mistakenly assume it is safe to cross the road (WHO 2013). Corben and Duarte (2006) concluded that to reduce pedestrian crash risk, an innovative and comprehensive approach is required and that moderating excessive vehicle speeds to uniformly lower levels in locations with high pedestrian concentrations offers a potentially cost-effective solution. Three options for moderating vehicle speeds in high pedestrian concentration areas were discussed: 50km/hour speeds zones supported by enforcement and publicity; gateway treatments such as road narrowing, changes in pavement texture and roundabouts; and streetscape improvements such as lane narrowing, tree planting, use of medians that signify changed traffic conditions. Generally, such traffic calming measures either require motorists to change their direction by moving to the left or right, or change elevation by either going up or down, and vary from minor changes to major rebuilds (WHO 2013). It is suggested that a combination of traffic-calming measures provides the greatest benefit and that they should be context-specific so that they are appropriate for the type of road (WHO 2013). NHTSA reports that speed limit reductions can be most effective when introduced for a limited area as part of a visible area-wide change, i.e. through identifying an area as a special pedestrian friendly zone with new landscaping, or streetscaping (NHTSA 2011). In their review of intoxicated pedestrian crashes, Cairney et al (2004) explain that by reducing speed limits, gaps in traffic are easier to judge for alcohol impaired pedestrians and drivers have greater capacity to avoid a collision (and in an event the severity would be less). Sydney has recently (as of 27 September 2014) rolled out a 40km/h speed limit zone in the CBD in an effort to improve pedestrian safety. Austroads 2016 page 31

42 Figure 2.12: Examples of components of pedestrian countermeasures Source: NRMA Source: City of Sydney While 50km/h speed limits in urban areas in Australia have been shown to improve pedestrian safety and reduce crashes (Haworth, Unger, Corben and Vulcan 2002), speeds above 30 km/h markedly increase the risk of severe injury or fatality for pedestrians in a crash (WHO 2013). Thus speed limits lower than 50 km/h are desirable especially in areas where pedestrian traffic is more likely. In Victoria the City of Yarra and City of Melbourne have instituted 40km/h area-wide speed limits to improve pedestrian safety. Speeds lower than 30 km/h would provide a much safer environment for pedestrians. Similarly, Canberra has also approached pedestrian safety through lowering of vehicle speeds in key locations. After trial and review of 40 km/h speed limits in the town centres of Gungahlin and Woden, Canberra, in 2011, these were made permanent in those centres. Implementation of a 40 km/h limit was introduced into three other identified precincts with high levels of pedestrian movement and minimum retail/commercial development on local roads in 2013: Belconnen, Civic and Tuggeranong. In addition, the ACT government plan for the modification of a major central city area into a shared zone with reduced speed limits of 20km/h were announced, with work expected to commence in late 2015 ( Many European countries (Germany, Sweden, Holland, United Kingdom and others) have already lowered speed limits to 30km/h in local areas and areas of high pedestrian use. All Australian states and territories have reduced speed limits for school zones (usually 40 km/h), though there is variation in the limits and the times at which they apply, with the lowest speed limit being 25 km/h in South Australia on roads where the speed limit is less than 80km/h at other times c Pedestrian-friendly smart technologies Advancements in vehicle technology such as brake assistance and collision avoidance also offer promise for pedestrian safety in the future. A French evaluation concluded that cars equipped with Brake Assist had a 10% lower involvement in pedestrian fatalities than cars without (cited in WHO 2013). Volvo utilise radar and camera technology that stops the car when a pedestrian is in the way (se Figure 2.13). This technology has recently been updated to apply when cyclists are too close (Volvo 2015). While focussed on cyclist detection, CARRS-Q researchers are developing mobile phone technology to enable cars to talk to bikes, with potential implications for pedestrians. The smart phone application can locate, track and communicate between devices and warn both rider and driver if there is a reasonable chance of a collision (Demmel 2014). Among other recent engineering-based measures in Singapore, use of smart technologies has seen the installation of intelligent road studs at pedestrian crossings to warn motorists about the presence of pedestrians at night; personal electronic devices for elderly pedestrians to allow more crossing time; flashing beacons at zebra crossings; and countdown timers and audio signals to aid hearing impaired and vision impaired pedestrians (Haque, Chin & Debnath 2013). Austroads 2016 page 32

43 Figure 2.13: An illustration of Volvo s pedestrian and cyclist detection technology Source: Volvo 2015 The use of real time intersection traffic analysis (including pedestrian traffic) via video technology provides a novel method for identification of where traffic and engineering countermeasures should be focused. Such advancements allow traffic authorities to monitor transport facilities, traffic flow and identify traffic violations (Haque et al 2013). The use of automated video analysis has been demonstrated for the detection of road users in a traffic scene and their classification as either a pedestrian or vehicle, identification of important events that may lead to collisions and the calculation of several conflict indicators like time to collision and post encroachment time (Ismail, Sayed, Saunier & Lim, 2009). The feasibility of before and after evaluations of safety treatments using vision based road user tracking has also been demonstrated (Ismail, Sayed & Saunier 2010). This technology has potential implications for understanding and evaluating the safety of distracted pedestrians d Countermeasures to improve visibility of pedestrians Other countermeasure options relate to increasing the visibility of pedestrians. Generally, providing crossing enhancements such as traffic signals and raised crossing islands can make pedestrians more visible to motorists, especially during dusk, dawn and at night (WHO 2013). Improved roadway lighting increases the visibility of pedestrians at night, especially at pedestrian crossings and has been associated with significant reductions in night time pedestrian crashes (WHO 2013). This is likely to have an impact on the number of alcohol-affected pedestrian crashes, specifically, since most of these occur at night (Cairney et al 2004). Removing physical objects such as trees and billboards that obstruct drivers from seeing pedestrians, implementing kerb extensions, installing signals to alert motorists that pedestrians might be crossing are other measures for improving pedestrian visibility (WHO 2013). Improving pedestrian conspicuity through the use of retro-reflective strips has also been shown to improve pedestrian safety at night, particularly when they are positioned on eight moveable joints (Wood, Tyrrell, Marszalek, Lacherez, Chaparro & Britt 2010). Other opportunities for improving pedestrian conspicuity include provision of devices for parents to fasten semi-permanently to children s clothing, light sticks and reflective bands which could be supplied with new cars or distributed through insurance companies and automobile clubs for use during breakdowns or emergencies, and retro-reflective materials used in shoes, backpacks and jackets (NHTSA 2011). Austroads 2016 page 33

44 2.7.4 Summary of countermeasures and their effectiveness This section has summarised the known countermeasures relating to pedestrian distraction and impairment across Australia and New Zealand. These are broadly categorised in terms of education, enforcement and engineering-based measures. General pedestrian safety and distracted crossing Most interventions focus on single approach (e.g. education, engineering, enforcement). Since distracted and impaired walking are not covered by legislation in Australia or New Zealand (except in SA), only education and engineering are currently applicable, and only education/awareness measures were found in the literature. Education and awareness intervention programs and campaigns are common, but evaluation of these is rare, and so effectiveness is unknown. One evaluated intervention targeting safe/legal use of pedestrian crossings and signals in central city Auckland showed improvement in pedestrian behaviour. Interventions using enforcement for illegal pedestrian behaviour are uncommon, with logistical difficulties (such as generally low concentrations of pedestrians behaving illegally or wide dispersal of locations for such behaviours, affecting detection) and the costs involved (such as use of police time to detect and issue infringement notices) generally making these rare. Engineering countermeasures to pedestrian behaviour are relatively common, and many have been evaluated and found to be effective. Separating pedestrians from motorised traffic in either time or space improves pedestrian safety generally. On-road separation in space, such as footpaths, has been found to increase pedestrian safety by as much as double, and provision of footpaths addresses the issue of distracted walking (though not distracted crossing). Thus safety benefits are likely if footpath infrastructure were to be installed where it does not currently exist on the transport network. However, for pedestrian crossing, separation in space is impractical in most situations on the transport network in Australia and New Zealand, especially outside central urban areas and areas of high pedestrian activity. Separation in time may be more applicable as a countermeasure to distracted pedestrian crossing. Signalisation affects pedestrian decisions about locations for crossing and safety. Raised median strips on multi-lane roads have lowered crash rates. Lowering general vehicle speed limits across high pedestrian concentration areas and local roads is effective at reducing pedestrian crash rates and injury severity. Novel approaches to entertaining pedestrians while they wait to cross at signals has been found to reduce illegal crossing behaviours but may provide effectiveness only in the short term, while there is high novelty value. Pedestrian countdown timers have not been shown to be effective. Technology used to warn pedestrians (e.g. audio beacons) is effective and low cost. Smart technologies (e.g. intelligent road furniture/vehicles) that warn motorists of pedestrians are growing in use but as yet are unevaluated. However, these may offer benefits that are worth exploring, particularly if used in combination with education. Real time intersection traffic analysis (including pedestrian traffic) via video technology may provide an effective method of identifying where traffic and engineering countermeasures for distracted pedestrians should be focused. Intelligent transportation developments may have application to the area of distraction from smart phones. Austroads 2016 page 34

45 Drink walking Alcohol impairment in road safety is addressed by a number of countermeasures such as Safe Night Out, lock-out programs, designated driver, responsible service of alcohol (RSA). While they do not specifically address drink walking, strategies which are effective at reducing high levels of alcohol consumption, or pedestrian intoxication (e.g. RSA), or that minimise pedestrian activity while intoxicated/impaired are likely to have the greatest benefit. Education strategies have been used but effectiveness is unknown. Effective engineering countermeasures to intoxicated pedestrian crashes include Dwell on red, and pedestrian fences in late night licensed precincts. Lowering vehicle speeds and altering traffic signal timings at high alcohol times and locations (e.g. latenight entertainment precincts) and pedestrian fencing in proximity to licensed venues has been implemented in Australian cities (but not evaluated). Enforcement countermeasures have included implementation of lock out laws/policies to reduce movement between venues and greater police activity and these have been found to provide some effectiveness in reducing traffic incidents generally and potentially of benefit for pedestrian crash involvement. The Drink Safe Walk Safe project, which had a combination of educational, engineering and enforcement measures, was evaluated as effective and may provide an approach for wider implementation. Austroads 2016 page 35

46 3. Investigating Pedestrian Use of Potentially Distracting Technology While Walking and Crossing the Road As highlighted in the literature reviewed above, motivations and perceptions are important factors influencing pedestrian behaviour. Pedestrian perception of risk appears to be related to choices about when, where and how to cross the road. Distraction by mobile phone use, especially for complex activities such as texting or internet access has been demonstrated to affect the safety of decisions about crossing, and the level of risk taking. Walking and crossing while impaired by alcohol has also been found to increase crash risk. As the use of potentially distracting technologies is volitional, attempting to reduce risky pedestrian use or to increase awareness of the hazards involved or perceptions of injury risk are arguably more likely to be effective if based on an understanding of the motives or perceptions that underpin them. The research reported below was aimed at investigating pedestrian behaviour and their attitudes, beliefs and intentions in relation to mobile phone use while walking and crossing the road. In order to meet the second of the objectives for this project, two studies were conducted: an intercept interview and an online survey. The intercept interview was designed to allow the research team to estimate the extent to which pedestrians are exposed to potential crash risk as a result of using technology, that is, the size of the problem. An additional aim was to identify factors that might be influential in relation to distracted pedestrian behaviour so that these might inform the design of the subsequent study (the online survey). The online survey was intended to obtain a more diverse, broader and larger sample of pedestrians than the intercept interviews and to investigate self-reported exposure to distraction while crossing the road, attitudes, beliefs, perceptions of facilitators and barriers, and intentions in relation to this behaviour. Details of these two studies are given below. 3.1 Intercept interviews Intercept interviews were chosen as a method for this part of the project. These are brief (usually) interviews with people going about their normal activities in selected locations. Locations are typically selected on the basis that they are likely to have high numbers of potential participants who will be rich sources of information because they have high exposure to the target or interest areas of the research. Participation is normally invited by approaching (intercepting) people who appear to meet the relevant selection criteria for the research sample. An interviewer then presents the questions verbally to those people who agree to take part, and notes the responses. In the current study, the interest was in estimating the prevalence of walking and crossing the road while simultaneously using technology that is likely to be distracting. Drink walking was also of interest. In order to try to maximise the usefulness of the information, and the probability of high response rates, it was decided to sample from Brisbane city centre for three reasons: 1. The majority of pedestrian crashes occur in the CBD 2. Pedestrian traffic density is likely to be high due to the tendency to use public transport to access the city 3. Many pedestrians on city-centre streets would be frequent users of intersections that have relatively high crash occurrence (thus a road user group with high exposure). Austroads 2016 page 36

47 For the purposes of this study, two distracting/impairing behaviours were chosen: using a smart phone/mp3 player; and alcohol at a level that is likely to exceed.05 mg/ml. Smart phone use was chosen because it was deemed to be the most common source of technological distraction amongst pedestrians, and recent observational studies (as cited in Section 1) suggest that such distraction is at high levels and likely to increase. Walking after drinking alcohol, though a much less common behaviour, has been shown to compromise a person s ability to make safe crossing decisions in simulated crossing studies (Oxley et al 2006). It is also a behaviour disproportionately associated with crash involvement (Austroads 2004; Cairney & Coutts 2003; Hutchinson et al 2009; Nieuwesteeg & McIntyre 2010) and thus likely to place a pedestrian at increased risk of injury. Collecting information on both behaviours allows for examining whether pedestrians who engage in distracted walking and crossing are also likely to walk while impaired by alcohol. Previous research in the road safety has suggested that distraction can be of different types with differing levels of influence over ability to carry out driving or pedestrian tasks (Stavrinos, Byington & Schwebel 2009). Accordingly, three levels of distraction from smart phone use were distinguished: visual and cognitive (text, internet activities); cognitive only (phone calls); and audio only (using a device for listening to music). For walking and crossing after consuming alcohol, the research team wished to select a sufficient level of alcohol consumption that might represent a blood alcohol concentration of around.05. Thus, the target behaviour (drink walking) was defined as walking/crossing after having consumed more than two standard alcoholic drinks within an hour prior to walking/crossing. Although drink walking was of high interest, it was expected that few pedestrians in these locations at this time of day and day of the week would actually be impaired by alcohol, or that they would report drink walking regularly. This, coupled with time limitations imposed by this method of interviewing led to the decision to focus primarily on distraction by smart phone/mp3 player use, and to collect drink walking information as a secondary focus. This component of the research was approved by the QUT Human Research Ethics Committee (Approval number ) Method a Participants and data collection The Web Crash online database was used to identify the Brisbane CBD intersections that had the highest number of pedestrian crashes between 2001 and The two study intersections were chosen because they accounted for a high number of pedestrian crashes during that period and were busy intersections with a high pedestrian volume. The intersections were wide and therefore safe locations to conduct the interviews without obstructing the footpath. Questions focused on the use of technology while walking and crossing the road as well as walking after consuming alcohol. Examples of questions are: About how many times a week do you have to cross a road while you are walking How often in the last week have you initiated a call while walking? How often in the last week have you monitored the internet while crossing the road? and In the last 3 months, about how often have you walked after consuming at least two standard drinks in the previous hour? Demographic information was also collected. Responses were anonymous and no personal information was requested. Flip cards were used to assist the delivery of, and response to, interview questions. The interview schedule is included in Appendix B. Interviews were approximately 10 minutes in duration. Participants were offered a $5 Coffee Club voucher in acknowledgement of their time. Austroads 2016 page 37

48 Responses were obtained from a convenience sample of 211 pedestrians aged between 17 and 65 years (mean age 32 years), with 53% women, 47% men. Of these, over half were aged years (56%), with a further 23% aged years as might be expected given the choice of location for recruitment (central business district of metropolitan city). The majority indicated their occupation as full time employed (45%), with a further 17% employed part time, 22% full time students and 12% neither employed nor studying, and 4% either full time carers or part time students. Purpose of trip was primarily going to or from work (25%) or study (19%), with a further 25% shopping (31% other ). Almost half of the participants (47%) used public transport at least once a day, and 25% reported using public transport several times a week, suggesting the sample was commuter-based. Only 12% indicated they never used public transport. Average time spent walking for transport or on a public roadway was high, equivalent to 3 ½ hours per week, with 30% of the sample walking minutes per week, and 23% walking 2 ½ to 7 hours per week. Only 22 % walked less than 60 minutes per week. Most of the sample (79%) reported crossing a road while walking for transport at least 10 times per week, and a further 9% crossed 6-10 roads each week while walking for transport. Response rates were not calculated due to the likely unreliability of this as a measure of refusal for the intercept method of approach: it is not possible to determine whether those people who actively avoid contact with the researcher are aware of the purpose of the study or mistake the approach for a marketing, sales or begging activity. In addition, since the times of day chosen for sampling encompassed the peak travel periods for city workers, many refusals were on the basis of not having enough time to respond to the survey b Procedure Four experienced Research Officers worked in pairs and conducted the intercept interviews on the footpaths of two busy intersections in Brisbane CBD (corner of Albert and Elizabeth Streets and corner of Edward and Ann Streets). All pedestrians who appeared to be between the ages of 17 and 65 were approached and asked if they would like to participate. A screening criterion was that participants had to own a smart phone. Only a few people were ineligible on this basis. While the focus was on people who were using smart phones/mp3 players, pedestrians were also invited to participate even if they did not appear to be distracted or using smart phones/mp3 players at the time of intercept. This was an attempt to capture individuals who potentially walk while distracted or impaired at other times than those of the survey, and also to gain information about potentially protective factors from pedestrians who identify as never walking while distracted or impaired. The interviews took place during the first week of November 2014 (Monday, Wednesday and Thursday) between 8am and 5pm. It was our intention to capture regular pedestrian and commuting patterns Results Section a c detail the analytical techniques and the results of these in relation to each of the research questions of interest. Readers with less interest in these details may turn to the discussion that appears on in Chapter 4 where the findings are summarised and their implications are discussed a Pedestrian self-reported frequency of using a mobile phone/mp3 player while walking and crossing the road For this part of the study, the focal interest was on the extent to which pedestrians use distracting technology while walking or crossing and on identifying the types of use and the most typical use. Accordingly, questions asked about using a smart phone for three main types of use (text, voice call, access the internet) and three interactions (initiate, monitor, respond to) while walking and crossing. An additional item in relation to audioonly device use (with headphones) was also asked for walking and crossing. Table 3.1 displays the wording of the questions, response options and proportions of the sample giving each response. Austroads 2016 page 38

49 Examining walking first, overall use of smart phones while walking was high, with 28-50% of the pedestrians indicating that they used their smart phones for one or more of the activities at least daily (responses of more than once per day or once per day ). A further 11-25% did so several times per week, suggesting that this sample of pedestrians were relatively frequent users of smart phones while walking. Together 43-65% responded several times per week, once per day or more than once per day to the activities. Monitoring or responding to the internet while walking appeared to be less common activities, with 48% and 43% respectively giving responses of several times per week, once per day or more than once per day. It is notable that for each of the text, internet and voice call activities; there were around 20% of the sample who said they never did this while walking (see Table 3.1). For crossing the road, the proportion of pedestrians giving higher use responses was much lower. Proportions of the sample who never used their smart phones while crossing ranged from 50% (answering a call) up to 72% (monitoring internet). However, around 16% indicated they used their smart phones for texting activities (initiate, monitor, respond to) or voice calls (initiate, answer) while crossing (that is, responses of more than once per day or once per day ). A somewhat lower proportion, 12%, used their phones for internet access activities (initiate, monitor, respond) this frequently. Audio-only devices appear to be an exception to this level of use, with almost 30% of the sample using these at least daily while crossing. Table 3.1: Proportions of pedestrians who use their smart phones while walking and crossing the road by activity and categories of frequency (self-report) Question wording and target activity How often do you use your smart phone to while walking More than once per day (%) Once a day (%) Several times a week (%) Once a week or less (%) Never (%) Initiate a text Monitor text messages Respond to a text Initiate a call Answer a call Initiate an internet search or interaction Monitor internet (e.g. Facebook) Respond to internet (e.g. ) Use an audio-only device to listen to music/radio with earphones/buds How often do you use your smart phone to while crossing a road? More than Once per day Once a day Several times a week Once a week or less Initiate a text Monitor text messages Respond to a text Initiate a call Answer a call Initiate an internet search or interaction Monitor internet (e.g. Facebook) Respond to internet (e.g. ) Use an audio-only device to listen to music/radio with earphones/buds Never Austroads 2016 page 39

50 3.1.2.b Age group and exposure in relation to smart phone use while walking and crossing the road Responses were next separated on the basis of age group in order to examine whether there were any patterns in the results that might highlight particular issues relevant to future interventions or applications of the results. Inspection of the distribution of the responses to the behavioural measures of the higher risk behaviours (cognitive, visual and physical distraction while crossing the road- initiating texts or internet activities while crossing) suggested that engaging in these activities decreases after age 30 years. Accordingly, the youngest age group was set at years. The older age group categories were somewhat arbitrarily determined on the reasoning that people under age 45 years are likely to be more familiar with and therefore heavier users of, smart phone text and internet-based features than older people. Participants were initially grouped into three age categories: years (n = 118); years (n = 48); and years (n = 44) (total n = 210 for which age was stated). Two categories of exposure were distinguished: High exposure responses of more than once per day, once per day and once per week ; Low exposure responses of less than once per week or never. The inclusion of several times per week in the high exposure category was based on the patterns of distribution in the responses: individuals giving these responses to some activities were more likely to give high frequency of use responses to the other behaviours than to give the lower frequency responses (that is once per week or less or never ). Walking and smart phone use When categorised according to age and frequency of the behaviour, it was clear that pedestrians aged years reported the greatest engagement in each type of smart phone use while walking, with the exception of audio-only devices. Table 3.2 displays the responses to the walking items according to age and exposure category. Table 3.2: Walking while using a smart phone by level of exposure and age group Age group How often do you use your smart phone to while walking? Low exposure (% of age group) (N=118) High exposure (% of age group) Low exposure (% of age group) (N=48) High exposure (% of age group) Low exposure (% of age group) (N=44) High exposure (% of age group) Initiate a text Monitor text messages Respond to a text Initiate a call Answer a call Initiate an internet search or interaction Monitor internet (e.g. Facebook) Respond to internet (e.g. ) Use an audio-only device to listen to music/radio with earphones/buds Austroads 2016 page 40

51 Two way analysis of variance (ANOVA) was carried out in order to determine if the differences in reported behaviours for the different age groups were statistically significant, and whether there were differences between men and women (overall or in the specific age groups). Total exposure to distracted walking was deemed to be the sum of responses to all questions on walking (nine items in Table 3.2). Mean scores for each person were used as an average of each person s exposure for distracted walking (low scores indicate responses equivalent to more frequent use). In addition, we reasoned that use of headphones only (for listening to music) because it provides an auditory only distraction, might represent a lower level of risk when walking. To examine this, a second mean score was calculated for the walking items, excluding the audio only item (eight items). Accordingly, two separate two way ANOVAs were conducted (age group, three levels; gender, two levels), with the mean score for responses to walking items (with and without the audio only item) as the dependent variable. A main effect of age group only was found for mean responses to the full set of items on use of smart phones while walking (including walking with audio-only) F (5, 203) = , p <.001, η 2 (adj) =.291, suggesting a large sized effect. There was no significant main effect of gender, nor a significant interaction effect for age group by gender. Post hoc testing (Tukey) revealed that, pedestrians aged years (M = 2.66; SD = 1.05) were significantly more likely than the year olds (M=3.34, SD 1.22) to indicate that they walked while using their smart phones (p <.001). The year olds were in turn significantly more likely than the year olds (M = 4.29, SD =.74) to report doing so (p <.001). Similarly, excluding the item for walking while using audio-only, a main effect of age was detected for mean responses to use of smart phones while walking F (5, 203) = , p <.001, η 2 (adj) =.256, which is a large sized effect. There was no significant main effect of gender, nor a significant interaction effect for age group by gender. Post hoc testing (Tukey) revealed that pedestrians aged years (M = 2.63; SD = 1.13) were significantly more likely than the year olds (M 3.29, SD 1.38) to indicate that they walked while using their smart phones (p <.01). The year olds were in turn significantly more likely than the year olds (M = 4.24, SD =.79, p <.001) to report doing so. Crossing the road and smart phone use Age-based differences were also evident for smart phone use while crossing the road. Pedestrians aged under 30 years old were more likely to report using their smart phones while crossing than those aged in or those aged years. Table 3.3 below displays the frequencies of reporting the nine different risky crossing behaviours by age group. The same categorisation for high and low exposure/risk responses was used in relation to responses to the crossing behaviours as those for walking (as described above). Two way analysis of variance (ANOVA) was carried out in order to determine if the differences in reported behaviours for the different age groups were significant, and whether there were differences between men and women (overall or in the specific age groups). To determine average exposure across the different smart phone activities, responses to the items on crossing were summed (nine items) and the mean score for each person determined. In addition, we reasoned that use of headphones only (for listening to music) because it provides an auditory only distraction, might represent a lower level of risk when crossing. To examine this, a second mean score was calculated for the crossing items, excluding the audio only item (eight items). Accordingly, two separate two way ANOVAs were conducted (age group, 3 levels; gender, 2 levels), with the mean score for responses to crossing items (with and without the audio only item) as the dependent variable. A main effect of age group only was found for mean responses to the full set of items on use of smart phones while crossing (including crossing with audio only) F (5, 199) = 8.378, p <.001, η 2 (adj) =.153, suggesting a medium sized effect. There was no significant main effect of gender, nor a significant interaction effect for age group by gender. Post hoc testing (Tukey) revealed that the pedestrians aged years (M = 3.77; SD = 1.14) were significantly more likely than either the years (M = 4.26, SD = 1.01) or years (M = 4.82, SD =.48) pedestrians to indicate that they crossed a road while using their smart phones. Although there was also a difference between the mean responses for the year olds and the year olds at p =.022, Levine s test of equality of error variances was significant, suggesting that a more conservative significance level be used in this test, and thus this difference did not meet the more stringent 1% significance level. Austroads 2016 page 41

52 Similarly, excluding the item for crossing while using audio only, a main effect of age only was detected for mean responses to use of smart phones while crossing F (5, 199) = 6.518, p <.001, η 2 (adj) =.119 suggesting a small to medium sized effect. Posthoc testing (Tukey) revealed that year olds were significantly more likely to report using their smart phones while crossing (M = 3.84, SD = 1.14) than were either the year olds (M = 4.30, SD 1.01) or the year olds (M = 4.84, SD = 1.08). The difference between the two older age groups was not statistically significant (using the more stringent 1% significance level due to the inequality of error variances). Table 3.3: Crossing the road while using a smart phone by age group and level of exposure Age group How often do you use your smart phone to. while crossing a road? Low exposure (% of age group) (N=118) High exposure (% of age group) Low exposure (% of age group) (N=48) High exposure (% of age group) Low exposure (% of age group) (N=44) High exposure (% of age group) Initiate a text Monitor text messages Respond to a text Initiate a call Answer a call Initiate an internet search or interaction Monitor internet (e.g. Facebook) Respond to internet (e.g. ) Use an audio-only device to listen to music/radio with earphones/buds c Drink walking In addition to smart phone use, we were also interested in the prevalence and nature of walking after having consumed alcohol. As drink walking was not the primary focus of the study, it was anticipated that this section of the interview would be limited in terms of the information that could be provided. Accordingly, questions were fairly broad and asked about how often people had walked after consuming two or more alcoholic drinks in the previous three months. Participants were also asked to identify the types of locations where they typically drink walked and under what circumstances. Their perceptions about whether they were affected by the alcohol they had consumed was also sought as a measure of whether drink walking was undertaken knowingly. In addition, we were interested in whether social groups might be an influential factor on tendency to drink walk, so one question asked participants about their perceptions as to whether their friends also walked after drinking. Lastly, an exploratory question asked about the injury outcomes from drink walking. For this, the wording was whether they had experienced a near miss or been injured after consuming alcohol and how likely they felt it would be that they would be injured (susceptibility). Austroads 2016 page 42

53 As described above, data was collected from pedestrians on a Monday and a Wednesday, at times when the majority were walking to or from work, so it was not expected that anyone in the sample would be impaired by alcohol at the time of the interview. Self-reported incidence of drink walking was relatively low, with 45% of the pedestrians in this sample indicating that they had never drink walked in the previous three months (see Appendix C). Thus only the other 55% of the sample who had indicated that they had walked after drinking received the additional interview questions about drink walking. Of those who responded, the majority said they had drink walked once per month or less often (61% of the drink walkers; 34% of entire sample). However, the remaining 39% of the individuals (45 people) indicated they had drink walked once per week (16 people, 8% of entire sample) or more often (29 people, 14% of entire sample). Moreover, the majority of those who said they had drink walked (69%) also indicated that they thought they were affected by the alcohol at the time. These findings suggest a small but important subgroup of pedestrians who have a high level of exposure to drink walking and who are likely to be placing themselves at high risk of a crash. It also appears that this group may be aware that they are impaired but choose to drink walk regardless. Reported injuries and near misses were rare, as might be expected in such a small and opportunistic sample. Only five people said they had sustained an injury after drink walking, though one of these people had been injured multiple times. One injury had required emergency department treatment and so can be regarded as serious Discussion Patterns of results suggest that young people aged years are more likely to engage in smart phone use when walking and when crossing than either the year olds or the year olds. However, smart phone use while walking was high for both year olds and for year olds (40-80% categorised as high exposure). While walking, text related activities were the most common use of a smart phone by year olds. For years olds, internet-related activities and audio-only devices were the most common while walking. For year olds, monitoring texts or answering a call were the most common uses of a smart phone while walking. These patterns potentially reflect age-group related familiarity and facility with smart phone technology. Proportions of the sample who indicated that they used their smart phones while crossing the road were much lower than the rates for use while walking, and this result was found for all age-groups. This finding is encouraging to the extent that it suggests that there is some acknowledgment of the increased dangers that such use constitutes when attempting to cross the road as opposed to walking. However, excluding the voice call and audio-only activities, around 30% of the year olds indicated they used their smart phones at least once per week for a cognitively and visually distracting activity (that is, texting or internet access) while crossing the road. Many of them did so daily or more often. Moreover, it appeared that those year olds who reported engaging in one activity also tended to report that they engaged in most of the others too, suggesting a subgroup that crosses the road frequently while using a smart phone (higher exposure) and is also more distracted (more complex activity level). This finding is of concern given it suggests that almost one in three young adult pedestrians are at relatively high risk of a crash while crossing the road. Overall, the results of the intercept interviews suggest that there is a particular subgroup of pedestrians who should be targeted in interventions to address distracted crossing. These are the year olds, high frequency (characterised by high use of various activities on their phones) smart phone users. As anticipated, only a small proportion of the sample reported having engaged in drink walking. However, the responses to the questions on this item suggest that there is a small group of pedestrians who have high exposure for this behaviour. Intervening with this group may be warranted. Austroads 2016 page 43

54 Limitations of the study should be borne in mind when considering these results. Firstly, we cannot be certain that pedestrians in the sample interpreted our questions in relation to crossing to mean once you have stepped off the kerb and so their reported frequency of using their smart phones while crossing may not be as high as we have reported. However, there was no feedback from participants in the sample to suggest that they were in doubt as to whether we were asking about while they were waiting to cross or while they were actually in the roadway, so we assume here that they interpreted the questions as we had intended. A second limitation is the self-report method used, which is subject to potential bias from socially desirable responding or errors of recall. We are unable to determine the extent to which our results may have been affected by either of these biases. However, the interviews were carried out in places where pedestrians were likely to actually be engaged in smart phone use while walking or crossing the road, or where they might typically be doing so as part of their normal activity. Thus it is hoped that the intercept approach, where pedestrians were asked about their typical behaviour in context, would have encouraged answers that were less subject to errors of recall or social desirability. An additional limitation is that the drink walking questions did not include asking about whether participants had crossed a road on any of the occasions on which they had drink walked, which might limit the extent to which they can be assumed to have been exposed to interaction with traffic. However, the researchers are aware that injuries to alcohol impaired patrons of late night licenced venues often occur immediately outside the venues, which is one reason for the installation of pedestrian barriers in these locations. Finally, the study results are based on a relatively small sample drawn from Brisbane CBD, which may limit the generalizability of the results to other locations. However, as described, locations for the interviews were carefully chosen intersections based on evidence of pedestrian crashes, and therefore highly relevant locations for the purposes of the research. We have no reason to expect that pedestrian behaviour in other Australian or New Zealand cities is substantially different from Brisbane, and so would expect that the results can be useful when considering other city centres. 3.2 Online Survey An online survey was conducted to obtain a broader and more diverse sample of pedestrians as well as to allow more detailed investigation of smart phone distraction while crossing the road. The aims were to: 1. help obtain prevalence information regarding the extent to which people walk and cross the road while simultaneously using technology which is likely to be distracting; 2. identify whether particular sub-groups of pedestrians are more exposed/at risk of injury than others; 3. explore whether psycho-social models of behaviour (the Health Beliefs Model and the Theory of Planned Behaviour) might provide insight into the factors influencing the target behaviours; and 4. examine pedestrian attitudes and beliefs as well as behaviour in more depth with a view to being able to inform future interventions design. The same two distracting/impairing behaviours were used in the online survey as those that were used in the intercept interviews: using a smart phone/mp3 player; and having consumed alcohol at a level that is likely to lead to a blood alcohol concentration of.05mg/ml. Similarly, the same three levels of distraction from smart phone use were distinguished as in the earlier study: visual and cognitive (text, internet activities); cognitive only (phone calls); and audio only (using a device for listening to music) Theoretical underpinnings As identified in the literature review, there are several public health and social psychology models that have been used to investigate road safety or pedestrian behaviour. Two of these were selected for this study in order to seek greater insight into the factors influencing, or motivations underpinning, pedestrian perceptions of factors that might be related to their walking and crossing behaviours. Austroads 2016 page 44

55 The first of these frameworks, the Health Beliefs Model (HBM) (Becker 1974; Rosenstock 1966) was chosen in order to explore the extent to which participants perceived the risks of distraction from phone use while crossing as well as potential relationships between these risks and reported behaviours. HBM has long been used in the public health area to design and target effective large scale interventions in areas such as stroke prevention (Sullivan, White, Young, Chang, Roos & Scott 2008); nutrition education (Abood, Black & Feral 2006) and osteoporosis prevention (Wallace 2002). This model suggests that people will be more motivated to undertake action such as adopting a healthy behaviour (e.g. eating fruit and vegetables; being more physically active; getting screened for a disease) or to ceasing an unhealthy one (e.g. smoking) if the person thinks that his/her personal likelihood of some negative outcome from not doing so is high, and if the negative outcome is perceived as sufficiently undesirable. That is, perceptions of high personal susceptibility and high severity are more motivating. The model also acknowledges the influence of barriers and facilitators to the desired/suggested behaviour and level of self-efficacy. Where there are fewer barriers, more facilitators and high self-efficacy, people are more likely to act. Accordingly, the questions related to the HBM in the survey were intended to assess pedestrian perceptions of their personal susceptibility to injury from distracted crossing as well as the extent to which they thought having an injury would affect them (severity). Approaches that use the HBM specify a target behaviour which improves the predictive and exploratory value of the model. Discussion within the research team resulted in operationalisation of this behaviour for the purposes of this study as being stopping using my smart phone for texts or internet access while crossing the road. The second theoretical approach underpinning the research was the Theory of Planned Behaviour (TPB) (Ajzen 1991). The TPB is a well-validated decision-making model that has been used successfully to explain and predict people s intentions and subsequent behaviours in regards to an extensive and varied array of social and health-related behaviours such as healthy eating (Connor, Norman & Bell 2002); smoking (Godin, Valois, Lepage & Desharnais 1992) speeding (Lewis, Watson & White 2013); blood donation (Giles, McClenahan, Cairns & Mallet 2004) and hand hygiene practice (Jenner, Watson, Miller, Jones & Scott 2002). TPB has been widely used in the road safety domain to understand behaviours such as speeding by young males (Lewis, Watson & White 2013), predicting intentions not to drink and drive (Moan & Rise 2011), and identifying the factors that influence texting while driving (Gauld, Lewis & White 2013). The central tenet of TPB is that intentions are the most proximal determinant of behaviour. Intentions in turn, are influenced or determined by a person s attitudes, subjective norms and perceptions of behavioural control (PBC) in relation to the particular target behaviour (Ajzen 1991). PBC has been conceptualised as incorporating two elements: the control a person has over the behaviour (controllability) and the difficulty of performing the behaviour (self-efficacy; Ajzen 2002). PBC items are therefore often split according to these two elements as demonstrated in previous studies (Horvath, Lewis & Watson 2012). Thus the model consists of three standard constructs, attitude, subjective norm (SN), and PBC, the strength of which determine the strength of one s Intentions towards engaging in the behaviour, which are then regarded as proxies for whether people will actually engage in the behaviour. Effective use of the model often attempts to obtain a baseline (past or current) level of the target behaviour and a prospective measure of actual behaviour over a specific time period (e.g. the next two weeks ), to allow for analysis of the relationship between reported intention and behaviour, though measures are usually self-reported. Several studies have used TPB constructs, and extended versions of it, to explain pedestrian behaviour (Zhou, Horrey & Yu 2009; Zhou & Horrey 2010; Diaz 2002; Gannon et al 2014). The TPB is often used with additional variables when such additional variables are expected to enhance the predictive ability of the model for investigating a behaviour. A review of the literature regarding TPB-based pedestrian behaviour-related studies reveals that some often included additional variables have been group norms, moral norms, anticipated affect, anticipated regret, past behaviours and perceived risk in relation to drink walking (Gannon et al 2014; Haque et al 2012). Austroads 2016 page 45

56 For the purpose of this study, group norms (friends) and mobile phone involvement were included as extended TPB variables. Mobile phone involvement has been found to be an important influence on young people s intentions to use a mobile phone while driving (White, Walsh, Hyde & Watson 2012). Previous research in the driving area has suggested that people can develop problematic levels of involvement with their mobile phones in a manner similar to other behavioural addictions. Drawing on the concepts within the behavioural addiction area, problem mobile phone involvement is characterised by high levels of cognitive involvement (such as the extent to which the individual is preoccupied by thoughts of his/her mobile phone while not actually using it) and high levels of unnecessary phone-related behaviours (such as constantly checking for missed messages and calls) (Walsh, White, & Young 2011). White and colleagues (2012) concluded that the ease and convenience of using modern mobile phone technology leads to excessive involvement for some young people and that this in turn impairs their decision making in relation to whether to use mobile phones while driving. For this study, it was reasoned that greater involvement with one s mobile phone might also negatively affect decision-making while walking or crossing the road. A person who is highly involved with his or her mobile phone might wish to continue using it even while performing relatively demanding or risky pedestrian tasks, such as those associated with crossing the road. To assess the level of involvement, the 8-item Mobile Phone Involvement Questionnaire (MPIQ) was included in the survey. Further detail in relation to this instrument are given below. The rationale for the inclusion of group norms as an additional TPB variable was two-fold. Firstly, subjective norm has been found to be the weakest determinant of intentions within the TPB in studies which have used the TPB, and thus, this particular variable may benefit from bolstering. In this case by the use of group norm. The second reason was that findings of previous studies in the area of drink walking (e.g., Gannon et al) have shown that normative influences are important in understanding the motivations and perceptions that underlie this risky pedestrian activity. As such, the influence of friends on a pedestrian behaviour, particularly among young people, were reasoned to be potentially influencing factors. Therefore an additional variable related to group norm was included Participants and data collection It was regarded as desirable to obtain responses from adolescents as well as adults in this study since mobile phone use is likely to be high among adolescents due to the high levels of ownership, and they are also more likely to be pedestrians because of their age. Accordingly, adolescents represent an important sub-group of pedestrians who may be at greater risk of distracted crossing. However, because we did not want to present alcohol-related questions to minors, the questions relevant to these two groups were different. Therefore, two versions of the survey were created: one for those aged years, which did not contain any reference to drink walking, and one for those aged years, with drink walking questions at the end of the survey. Ethical approval for the adolescent sample was given as an approved variation to the original application. The information and consent process differed for the two age groups in that the adolescent sample was recruited only through parents, who firstly had to supply the link to the survey to the child, and also had to give consent for the child to participate. This parental consent was indicated by a parent ticking a check box on the information page prior to being able to access the questions. The year old participants were recruited using a broad sampling strategy. Recruitment flyers and s were distributed through the CARRS-Q website, QUT subscription lists and noticeboards and social media (Facebook and Twitter). The QUT psychology participant pool was also utilised. Potential participants were directed to a link (a separate link was set up for the adolescent version) where the online survey could be accessed. By clicking on the link, participants were presented with the Participant Information Sheet and prompted to tick a check box on the bottom of the page to indicate their consent prior to being able to access the questions. After pilot testing, the adolescent and adult versions of the survey were made available for 6 weeks during March/April 2015 via the QUT online survey portal which uses a version of the Key Survey software. Completion of the survey was deemed to be an individual having provided their consent to participate. To increase response rates, a Research Officer also approached students in person on two inner-city campuses of Queensland University of Technology over the course of one week, inviting participants to complete a hard copy version of the survey. These participants signed a written consent form. Austroads 2016 page 46

57 All participants who completed the survey were eligible to enter into a draw for one of eight $50 gift vouchers. Complete responses were obtained from 268 people in the adult sample. Unfortunately, the level of adolescent response to the on-line survey was very poor, with only five people providing a response despite multiple attempts at distributing recruitment flyers. This small sample precluded meaningful analysis and so results are reported below for the adult sample only. Of those who supplied age in the adult sample, over half were aged years (63%), with a further 15% aged years and the remaining 22% aged years. The majority indicated their occupation as full time student (41%), with a further 27% indicating they were employed full time. Around 15% were employed part time and also studying part-time, while 12% were part-time employed only. As might be expected, given the recruitment process, levels of education were relatively high, with 22% of the sample indicating they had postgraduate levels of education, and a further 34% having some undergraduate education. Only 32% had high school certificate level education only. Of those who responded to the question, 21% had incomes under $40,000pa, while 24% indicated incomes between 40-60,000pa, and 40% had incomes of $80,000 or more (though there were higher rates of refusal for this question, at 16% of the sample). Just under one third (32)% of the participants used public transport at least once a day, and 32% reported using public transport several times a week, suggesting the sample were frequent commuters. A quarter were less frequent users of public transport (25% once per week or less ) and 11% indicated they never used public transport. Average time spent walking for transport or on a public roadway was high, equivalent to 3 ½ hours per week, and similar to the sample for the intercept interview with 30% of the sample walking minutes per week, and 23% walking 2 ½ to 7 hours per week. Only 22 % walked less than 60 minutes per week. Most of the sample (79%) reported crossing a road while walking for transport at least 10 times per week, and a further 9% crossed 6-10 roads each week while walking for transport Survey Design An anonymous, cross sectional, self-report online survey design was chosen. Online surveys have the benefit of allowing data collection from a large number of individuals relatively quickly and at low cost. In addition, it is possible to gain a more in-depth understanding of key variables because a greater number of questions can be posed relative to the number that can be asked in an intercept interview. Such surveys also have the advantage that participants can respond when it is convenient for them, which can boost response rates. There were several variables of interest in this study: frequency of smart phone use while crossing the road; type of smart phone use; usual engagement in risky pedestrian behaviours; safety beliefs in relation to pedestrian injury; Questions examined contextual (i.e. environmental), individual (i.e. age and gender) and psychosocial factors (e.g. risk perceptions, attitudes, subjective norms, self-efficacy) likely to influence behaviour. These constructs are important for understanding the influences contributing to pedestrian decision making about, for example, drink walking, walking while using a mobile phone and crossing against a red man. Demographic information was also collected. The survey included two standardised instruments (the Pedestrian Behaviour Scale and the Mobile Phone Involvement scale) as well as purpose-designed items. Each of the instruments used and the purpose for its inclusion is described below Materials a The Pedestrian Behaviour Scale (PBS) In order to assess the extent to which participants engage in risky pedestrian behaviours generally and not distraction-related, a measure of usual behaviour was sought. The Pedestrian Behaviour Scale (PBS) (Granie, Pannetier, Gueho 2013) appeared suitable for this purpose. The 47-item PBS is based on existing validated versions of the Pedestrian Behaviour Questionnaire and other self-reported pedestrian behaviour scales. A shorter 20-item version was used in this study. This version has been validated for use with pedestrians of all ages and shown to have good internal reliability (Granie, Pannetier, Gueho 2013). Items ask about the frequency ( Never, Rarely, Sometimes, Often, Always ) of engaging in different types of pedestrian behaviours. Five subscales distinguish behaviours of offences, errors, lapses, positive behaviour Austroads 2016 page 47

58 and aggressive or antisocial behaviours. Examples of items are I start to cross on a pedestrian crossing and I finish crossing diagonally to save time, I cross without looking, following other people who are crossing and I run across the street without looking because I am in a hurry b Mobile Phone Involvement Questionnaire (MPIQ) As described above, for this study it was reasoned that greater involvement with one s mobile phone might influence the extent to which someone might wish to continue using it while performing the more demanding and risky pedestrian tasks associated with crossing the road. Accordingly, a measure of mobile phone involvement, the 8-item Mobile Phone Involvement Questionnaire (MPIQ) was included in the survey. The questionnaire, developed and validated by Walsh, et al., (2010), is based on Brown s behavioural addiction framework, which includes symptoms such as cognitive and behavioural salience, withdrawal and loss of control. Higher levels of agreement in relation to each symptom indicate more problematic levels of involvement. Examples of MPIQ items are The thought of being without my mobile phone makes me feel distressed and I interrupt whatever else I am doing when I am contacted on my mobile phone c Health Beliefs Model (HBM) Prior to designing the questions for this section of the survey, a small pilot study using six CARRS-Q staff members was conducted to elicit the relevant perceived benefits of, and barriers to, the target behaviour ( stopping using my smart phone for texts or internet access while crossing the road). Only the more visually and cognitively distracting behaviours were used in these items in order to keep the length of the questionnaire at a reasonable level. People who were smart phone users while walking were asked to identify the benefits and costs if they stopped using their phones while crossing. From the responses, the three most common benefits and costs were selected for use in the main survey. All response scales were seven point with 1 = strongly disagree to 7 = strongly agree. Examples of HBM items included in the survey are: There is a good chance that I would be injured if I use my smart phone for texts or internet access when I am crossing the road ; Being injured because I was using my smart phone for texts or internet access while crossing the road would affect my life a lot ; and I could pay more attention to the people with me if I stopped using my smart phone for texts or internet access while crossing the road d Theory of Planned Behaviour As noted previously, to enhance the predictive and explanatory utility of the frameworks used in this study, the target behaviour was identified with considerable specificity in accordance with the TPB s TACT principle (see Ajzen 1991). The target behaviour was operationalised (within all questions) as using a smart phone for texting or internet functions while crossing the road. Participants were asked to think about texting and internet-access types of use and were also told the following: when we say 'for texting or internet' we mean initiate, monitor or respond to (e.g. checking Facebook, writing etc.) in order to make the behaviour as close as possible to behaviours from the earlier sections of the survey. The standard constructs of the TPB were included: Attitudes towards the target behaviour were assessed with 4 items (e.g. For me, using a smart phone for texting or internet functions while crossing the road would be unenjoyable/enjoyable ; Subjective Norms were assessed with four items (e.g. People who are important to me would want me to use a smart phone for texting or internet functions while crossing the road); and PBC was assessed with three items (e.g. I have complete control over whether I use a smart phone for texting or internet functions while crossing the road ). In addition, based on the findings from other road safety applications of the TPB, four items were included to assess the influence of Group Norms specific to friends (e.g. My friends would approve of me using a smart phone for texting or internet functions while crossing the road ). Intentions were assessed with three items (e.g. It is likely that I will use a smart phone for texting or internet functions while crossing the road ). All response options were on a seven point rating scale (where 1 = strongly disagree to 7 = strongly agree ) and the instructions were to respond in terms of how you would feel in the next week. Austroads 2016 page 48

59 3.2.4.e Alcohol affected walking Several items relating to drink walking were also included. These items were based on those used in previous studies that have examined the effects of alcohol on pedestrian behaviour (Rosta 2011; McGhie, 2010; King 2010). Questions asked how often in the last three months people had walked after consuming two or more standard alcoholic drinks in the hour immediately prior to walking, what types of locations these instances occurred in and under what circumstances. Participants were also asked to think about other people who walk after consuming alcohol, and to indicate the types of situations people might be most likely to do this. In addition, participants were asked how they thought their friends might perceive such behaviour. Whether people had been injured or experienced a near miss when walking after consuming alcohol was also included f Factors influencing crossing behaviour In order to explore the environmental factors that might influence pedestrian tendency to use a smart phone while crossing, it was decided to use hypothetical crossing situations. Although discussions within the research team identified a wide range of potentially influential and important influences on pedestrian decision making in crossing situations, for practical reasons (i.e., length of survey and time taken to complete it), only a limited number could be explored in this part of the survey. Three factors were identified as potentially having the highest priority for the survey: type of distraction (cognitive versus cognitive plus visual); type of control on the crossing (pedestrian signals, versus midblock-uncontrolled); and median strip (present or absent). These factors generated eight hypothetical scenarios, which were described in text form. Digital photographs of intersections and potential crossing points with the required features were taken in the local streets around the Kelvin Grove campus of QUT (located in inner-city Brisbane) and used to supplement the text for each scenario. The descriptions for each scenario specified that the participant was approaching the road (with or without signals, with or without a median strip) and needing to cross. Instructions indicated that as they were approaching the road, they were also using their smart phone to either talk to someone (cognitive-only distraction) or to text/access the internet (cognitive plus visual distraction). Participants were then asked to indicate the likelihood that they would cross the road in each of the scenarios (i.e., likelihood of intention to cross was used as a proxy behavioural measure). All participants received and responded to questions regarding all of the scenarios (i.e., the study used a fully repeated measures design). The details which appeared in the survey regarding Scenario 1 are presented below, to provide an example of the nature of information that was provided to participants. The italicised text highlights the manipulation of the independent variables. Austroads 2016 page 49

60 Scenario 1: It is mid-afternoon on a fine day in the middle of the week. You are walking in an urban setting towards what is normally a busy road with a single lane of traffic each way and a median strip in the centre. Traffic seems usual. You need to cross the road. You arrive at the road edge and there is a set of traffic and pedestrian signals. You decide to cross at this point. While you are crossing you are also using your smart phone to talk on a voice call. How likely is it that you would cross the road in this scenario while continuing to use your smart phone to take on a voice call? (Extremely unlikely, very unlikely, somewhat unlikely, not sure, somewhat likely, very likely, extremely likely). Figure 1: Photograph accompanying Scenario Results a Self-reported frequency of using smart phones and MP3 players while crossing the road This section asked about the use of smart phones to carry our potentially distracting activities while crossing the road as well as the frequency of different the types of use. As with the intercept interviews, questions targeted three main types of smart phone use (i.e., text, voice call, access the internet) and three interactions (initiate, monitor, respond to). However, the online survey focussed on crossing behaviour only rather than walking since much of pedestrian walking might be conducted on walking paths with little interaction with traffic. This focus is in keeping with the primary interest of this research on those situations which might present elevated risk of a crash and injury for pedestrians. Table 3.4 displays a summary of the frequencies of responses for smart phone use while crossing. Austroads 2016 page 50

61 As can be seen in Table 3.4, reported frequency of smart phone use while crossing the road was relatively low for the sample as a whole, with the majority of pedestrians indicating that they never used their smart phones while crossing for texting activities ( %) or internet activities ( %), or to initiate a voice call (45.7%). Reported frequency of answering a voice call while crossing was greater, with almost half the pedestrians indicating they did this once a week or less (48.9%). However, around 15% of the sample reported using their smart phones while crossing for text activities several times per week. Using a device to listen to music while crossing appeared to be qualitatively different from the other behaviours. For this activity, over one fifth (21.7%) reported doing this several times per day, and a similar proportion (20.9%) reported listening to music while crossing several times a week to daily. Table 3.4: Proportions of pedestrians (%) engaging in different smart phone activities while crossing the road (N=268) How often do you use your smart phone to. while crossing a road? Crossing and type of smart phone use More than once per day (%) Once a day (%) Several times a week (%) Once a week or less (%) Never (%) Initiate a text Monitor text messages Respond to a text Initiate a call Answer a call Initiate an internet search or interaction Monitor internet (e.g. Facebook) Respond to internet (e.g. ) Use an audio-only device to listen to music/radio with earphones/buds In order to examine whether there were any patterns in the responses that might highlight particular issues relevant to future interventions or applications of the results, responses were further analysed according to age group and exposure. Age groups were defined as for the earlier intercept survey study: years (n = 169); years (n = 40); years (n = 59). Exposure was based on frequency of reported engagement: High exposure responses of more than once per day, once per day and once per week ; low exposure responses of less than once per week or never. Table 3.5 displays the results. As can be seen in Table 3.5, pedestrians aged 30 years and under were more likely to report using their smart phones while crossing than those aged in either of the other two age groups. Moreover, there were relatively high levels reported for each of these behaviours in this age group (26-55% weekly or more often, that is, high exposure), with the exception of responding to an internet search (18%). Two way analysis of variance (ANOVA) was carried out in order to determine if the differences in reported behaviours for the different age groups were statistically significant, and whether there were differences between men and women (overall or in the specific age groups). To generate an average level of exposure, first, pedestrian responses to the items on crossing were summed (nine items) and the mean item score for each person determined. Higher mean scores indicate lower frequency of the behaviour. The second analysis excluded the audio-only behaviour (on the basis that it is likely to represent a lower level of risk when crossing since only hearing and not vision or cognition are engaged in the distracting task). To examine this, a second item mean score was calculated, excluding the audio only item (eight items). Accordingly, two separate two way ANOVAs were conducted (age group, 3 levels; gender, 2 levels), with the item mean score for responses to crossing items (with and without the audio only item) as the dependent variable. Austroads 2016 page 51

62 Table 3.5: Crossing the road and smart phone use (N=268) by age and exposure category How often do you use your smart phone to while crossing a road? Low Exposure (% of age group) Age group High Exposure (% of age group) Low Exposure (% of age group) High Exposure (% of age group) Low Exposure (% of age group) High Exposure (% of age group) Chi square statistic value, p, effect size (df = 2) Initiate a text a a Χ 2 = 14.01, p <.006, Cramer s V =.238 Monitor text messages a a Χ 2 = , p <.001, Cramer s V =.277 Respond to a text a a Χ 2 = , p <.001, Cramer s V =.271 Initiate a call Χ 2 = 8.359, p =.02, ns Answer a call a a Χ 2 = , p <.006, Cramer s V =.233 Initiate an internet search or interaction Monitor internet (e.g. Facebook) Respond to internet (e.g. ) Use an audio-only device to listen to music/radio with earphones/buds a a Χ 2 = , p <.001, Cramer s V = a Χ 2 = , p <.006, Cramer s V = Χ 2 = 8.174, p =.02, ns a a 14.6 a Χ 2 = , p <.001, φ =.327 a Superscripts indicate cells where standardised residuals are greater than +/- 1.8 standard deviations. These cells contribute disproportionately to the effect. A main effect of age group only was found for item mean responses to the full set of items (including crossing with audio only) F (3, 228) = 8.848, p <.001, η 2 (adj) =.147, suggesting a large sized effect. There was no significant main effect of gender, nor a significant interaction effect for age group by gender. Post hoc testing (Tukey) revealed that the pedestrians aged years (M = 3.81; SD =.95) were significantly more likely than either the years (M = 4.30, SD =.65) or years (M = 4.67, SD =.45) pedestrians to indicate that they crossed a road while using their smart phones. Statistically, responses for the year olds did not differ from those of the year olds. Similarly, excluding the item for crossing while using audio only, a significant main effect was detected only for age group F (5, 228) = 8.074, p <.001. The η 2 (adj) =.135 suggests this is a large effect. Posthoc testing (Tukey) revealed that year olds were significantly more likely to report using their smart phones while crossing (M = 3.90, SD =.95, p <.05) than were either the year olds (M = 4.34, SD=.65) or the year olds (M = 4.72, SD =.45). Differences between the two older age groups were not statistically significant. Chi square analysis was used to determine whether there were significant age-based differences in the proportions of pedestrians categorised into high and low exposure levels. A series of eight separate chi square tests were run to test for association between age group and exposure level for the different smart phone behaviour. Results are reported in the right hand column of Table 3.5. Significance level was set at alpha =.006 to compensate for multiple tests. Results indicated that year olds were significantly more likely to be at high exposure for initiating, monitoring and responding to text messages, and for initiating an internet interaction while crossing than the other age groups. For initiating a text, 30% of Austroads 2016 page 52

63 year olds were categorised as high exposure compared to 13% of year olds and 5% of year olds. For monitoring texts while crossing, 43% of this age group were categorised as high exposure compared to only 21% of year olds and 10% of year olds. Similarly, for responding to texts 35% of year olds compared to 15% of year olds and 5% of year olds were categorised as high exposure. Lastly, for initiating an internet interaction, 30% of year olds were categorised as high exposure compared to 13% of year olds and 5% of year olds. For each of these activities, pedestrians aged 45-65% were significantly less likely to be in the high exposure category than the other age groups. Pedestrians aged were less likely than the other age groups to be categorised as high exposure for monitoring of internet activities while crossing, with only 2% of this age group, compared to 27% of year olds and 23% of year olds. Cramer s V values indicate that these effects were all medium sized b Pedestrian Behaviour scale An overall measure of frequency of risky pedestrian behaviours was calculated from the responses to the 20 items in the Pedestrian Behaviour Scale. The reliability of the scale was high, α =.87 so a total score was calculated for each participant, and means of this scale were used in the analyses. Means for the separate items are displayed in Table 3.6. Overall self-reported frequency of risky crossing behaviour was low, with M = 2.12, suggesting that most pedestrians regarded themselves as rarely engaging in the types of behaviours assessed in this measure. However, reported frequency of finishing crossing diagonally (M = 3.14) and of crossing between parked cars (M = 3.13) were at a level equivalent to sometimes and suggests that these behaviours may warrant further attention. A two-way, between groups ANOVA was conducted to assess whether there were age group (18-30; 31-44; years) or gender differences in reported frequency of risky pedestrian behaviours. Table 3.6: Pedestrian Behaviour Scale question wording and Mean (SD) by age group Question wording (response options 1 = Never, Rarely, Sometimes, Often, 5 = Always) Entire sample Mean (SD) I start to cross on a pedestrian crossing and I finish crossing diagonally to save time 3.14 (.97) I cross between vehicles stopped on the roadway in traffic jams 2.87 (.97) I cross the street between parked cars 3.13 (.93) I look at the traffic light and start crossing as soon as it turns red 2.31 (1.12) I cross the street even though the pedestrian light is red 2.31 (.98) I cross diagonally to save time 2.89 (.99) I cross away from the pedestrian crossing even if there is one less than 50m away 2.63 (.95) I cross while talking on my mobile phone or listening to music on my headphones 2.94 (1.29) I cross even though the light is still green for vehicle 2.08 (.92) I start walking across the street, but I have to run the rest of the way to avoid oncoming vehicles 2.06 (.88) I walk on the curb 2.42 (1.07) I avoid using pedestrian bridges or underpasses, even if one is located nearby 1.86 (.94) I cross even though obstacles (parked vehicles, buildings, trees, trash bins, etc.) obstruct visibility 2.25 (.94) I cross even if vehicles are coming because I think they will stop for me 1.20 (.46) I cross without looking, following other people who are crossing 1.50 (.78) I realize that I have crossed several streets and intersections without paying attention to traffic 1.48 (.74) I forget to look before crossing because I am thinking about something else 1.35 (.56) I cross without looking because I am talking with someone 1.41 (.65) I forget to look before crossing because I want to join someone on the other side of the road 1.24 (.47) I run across the street without looking because I am in a hurry 1.25 (.59) Total 2.13 (.47) Austroads 2016 page 53

64 No significant interaction between age group and gender was found. A statistically significant main effect for age group was detected (F (2,240) = 6.524, p <.01). This effect was medium sized (η 2 =.052). Post-hoc comparisons (Tukey) revealed a difference between the mean scores for year olds (M = 2.19, SD =.47) and the year olds (M = 1.93, SD =.42) only, such that year old pedestrians reported significantly less tendency towards risky crossing behaviours than the year olds. A significant main effect was also obtained for gender (F (1,240) = 8.267, p <.01). However, despite being significant, differences for gender were small (η 2 for gender =.03), suggesting little real difference between men (M = 2.23, SD =.45) and women (M = 2.08, SD =.47) in self-reported tendency towards risky crossing behaviours in the measure. Further analyses were carried out to determine whether pedestrians who reported usually engaging in risky pedestrian behaviours (as measured by item mean scores on the PBS) differed from others in their reported frequency of distracted crossing (as measured by high versus low exposure to smart phone use while crossing, as determined in Section a). Participants were first categorised into low versus high risky pedestrian behaviour on the basis of their item mean scores on the PBS. The cut off point for the high risk category was set at 3.6 (equivalent to engaging in most of the risky behaviours more frequently than Sometimes ). We believe this cut off errs on the side of under estimating risk and is therefore a conservative point. A total of 219 participants were categorised as low risk while 47 fell into the high risk pedestrian behaviour categories. Chi square testing of pedestrian behaviour risk category (low, high) against exposure to distracted crossing categories (low, high) revealed a significant difference between groups such that participants who reported more frequent engagement in riskier pedestrian behaviours were also more likely to report high frequency of using their smart phones while crossing. This was the case for the full set of smart phone behaviours (including when listening to music on headphones, χ 2 (1) = , p <.001, φ =.303) and when listening to music on headphones was excluded (χ 2 (1) = , p <.001, φ =.339). Phi values suggest that effect sizes were medium. These results are displayed in Table 3.7 below. This pattern of results suggests that around 10% of pedestrians may engage in relatively high levels of risk-taking across a range of behaviours. Table 3.7: Comparison of high and low self-reported usual risky pedestrian behaviour for high and low levels of frequency (high, low) use of smart phone while crossing (n = 247) Exposure level for crossing while using smart phone (for text, internet, calls) or listening to music on headphones Level of usual risky pedestrian behaviour (item mean score PBS) Low High Totals Low n (% rows) 167 (80.7) 40 (52.0) 207 High n (% rows) 18 (45.0) 1 22 (55.0) 1 40 Totals Exposure level for crossing while using smart phone (for text, internet, calls) Level of usual risky pedestrian behaviour (item mean score PBS) Low High Totals Low n (% rows) 173 (83.3) 34 (16.4) 207 High n (% rows) 18 (45.0) 1 22 (55.0) 1 40 Totals These cells had a major contribution to the chi-square result (standardised residuals +/- 1.96) c Mobile phone involvement As described earlier, we were interested in the extent to which the pedestrians surveyed were attached to using their mobile phones and the potential relationship between this tendency and the reported use during activities that present a greater risk of crash involvement: that is, crossing the road. Austroads 2016 page 54

65 Almost all participants provided complete data on the Mobile Phone Involvement Scale (n = 266 out of 268). Reliability of this scale was calculated as.89 suggesting that the items can be treated as measuring a single underlying dimension for this sample and therefore a measure was formed based on the sum of responses to all items. Mean item scores (i.e. rating of agreement, 1 = strongly disagree to 7 = strongly agree, with the statements in the MPIS) were then calculated and used for the analyses. High mean scores indicate high involvement. For the entire sample, overall mean item score on the Mobile Phone Involvement (MPI) measure was 3.56 (95% CI , SD = 1.39). Mean item scores were also calculated by age group and by gender (see Table 3.8). For the current study, a new variable that distinguished high (and potentially problematic) and low mobile phone involvement was created. Participants were categorised into high (MPI mean item score 4.5 or more, equivalent to responses of agree or strongly agree to most items) and low (MPI mean item score < 4.5) involvement on the basis of scale mean score. For the entire sample, 75 (28%) participants were categorised as high mobile phone involvement, and 191 (72%) were categorised as low mobile phone involvement (see Table 3.8). High and low mobile phone involvement by age and by gender are displayed in Table 3.8. Next, in order to examine whether there were any age or gender differences in mobile phone involvement scores, two one-way ANOVAs were conducted with item mean scores on the MPI scale as the dependent variable (and treated as a continuous measure in the analysis). Results revealed a significant and large effect for age group, with F (2, 246) = , p <.001, η 2 =.163. Post hoc tests (Tukey) showed that year olds had significantly higher scores in relation to mobile phone involvement than year olds or year olds (p <.01 in each case). Scores for the year olds were also significantly higher than those for the year olds (p <.05). Moreover, a chi square test revealed a similarly large significant effect for age in relation to the new variable, high and low mobile phone involvement, (χ 2 = , df = 2, p <.001, Cramer s V =.281). The year olds were significantly more likely to be categorised as having high mobile phone involvement than either of the other two age groups, and more than a third of participants in this age group (36%) had high scores for MPI. In contrast, year olds were significantly more likely to be categorised as having low mobile phone involvement than either of the other two groups (see Table 3.8). In relation to gender, no significant effect was found for the mean MPI scores overall. However, when considered according to high and low scores, chi square testing revealed a significant, though modest, effect of gender, (χ 2 = 4.525, df = 2, p =.034, Fisher s Exact Test, φ =.033). Men were significantly less likely than women to be categorised as high scorers on mobile phone involvement. Table 3.8: Mobile phone involvement level (high, low) by age group and gender Re-categorised level of mobile phone involvement Age MPI Item Mean Score (SD) Low (% of age group) High (% of age group) years n = years n = years n = 41 Gender Male N = 60 Female N = (1.27) ( (1.15) MPI Item Mean Score (SD) Low (%) High (%) 2.23 (.45) (.46) Austroads 2016 page 55

66 Taken together, the results from this measure suggest that young people years are the group most likely to have high, and potentially problematic, mobile phone involvement, and this finding may be more marked for young women in particular. Additional analyses were conducted to examine whether there was any effect of mobile phone involvement and usual level of pedestrian risk taking behaviour (PBS as above, categorised into low and high level of riskier behaviours, as described previously) or frequency of using a smart phone while crossing the road (self-reported frequency of use, high and low, as determined above). Chi square analyses were conducted using the full sample (all ages) and separately for the participants aged years. Significant differences between groups were detected for each analysis (see Table 3.9). For the full sample (n =266), the majority of participants (n = 165, 62%) were categorised as both low mobile phone involvement and low risk usual pedestrian behaviour (low PBS). However, though a much smaller group (n =29), there was a significantly greater proportion of participants who were categorised as having high mobile phone involvement who had also reported higher levels of risky pedestrian behaviour (χ 2 = 7.663, df = 1, p =.007, Fisher s Exact Test, φ =.170). This effect size is regarded as small. Similarly, the majority of pedestrians were categorised into low mobile phone involvement and low exposure for use of smart phones while crossing the road including listening to music only (n = 145, 59%) or excluding listening to music only (n = 150, 56%). However, although much smaller groups (n =29, n = 28 respectively), participants categorised as having high mobile phone involvement were more likely to also report high frequency of smart phone use while crossing the road (χ 2 = , df = 1, p =.000, Fisher s Exact Test, φ =.243 including listening to music only; χ 2 = , df = 1, p =.000, Fisher s Exact Test, φ =.266 excluding listening to music only). These effect sizes are small to medium. Analyses were re-run for the year olds only (n = 168). Proportions of participants categorised as lower levels of risky pedestrian behaviour and lower exposure for mobile phone use while crossing were smaller than for the whole sample (52%- 46%). Results of chi square tests were non significant for the PBS (χ 2 = 3.458, df = 1, p =.080, Fisher s Exact Test, n.s.) suggesting that there is no relationship between younger participants general risky pedestrian behaviours and their mobile phone involvement level. Results for frequency of smart phone use while crossing the road were significant both for inclusion of listening to music only and excluding this behaviour (χ 2 = 4.278, df = 1, p =.049, Fisher s Exact Test, φ =.168; χ 2 = 5.838, df = 1, p =.026, Fisher s Exact Test, φ =.197 respectively) though effect sizes were small. This suggests that over involvement with one s mobile phone may be associated with greater propensity to use it while crossing the road. As mentioned previously, cell sizes are somewhat small for the higher risk categories and so results must be interpreted with some caution as they may not generalise to a larger population. However, the patterns of results suggest that mobile phone involvement may be worthy of further attention in future research. Austroads 2016 page 56

67 Table 3.9: Comparison of high and low mobile phone involvement for self-reported usual risky pedestrian behaviour (high, low) and frequency of smart phone use while crossing the road (high, low) for both all age (Full sample, n = 247) and young people only (n = 168). Level of usual risky pedestrian behaviour (item mean score PBS) Mobile phone involvement (MPI scores) categorised into high and low Full sample, all ages (n = 266) year olds only (n = 168) Low High Totals Low High Totals Low n (% rows) 165 (86.4) 26 (13.6) (84.2) 19 (23.8) 108 High n (% rows) 54 (72.0) 21 (28.0) (46.8) 18 (13.2) 60 Totals Exposure level for crossing while using smart phone (for text, internet, calls) or listening to music on headphones Mobile phone involvement (MPI scores) categorised into high and low Full sample, all ages (n = 266) year olds only (n = 151) Low High Totals Low High Totals Low n (% rows) 145 (81.5) 33 (18.5) (64.2) 27 (32.8) 97 High n (% rows) 40 (58) 29 (42) (35.8) 24 (18.3) 54 Totals Exposure level for crossing while using smart phone (for text, internet, calls) Mobile phone involvement (MPI scores) categorised into high and low Full sample, all ages (n = 266) year olds only (n = 151) Low High Totals Low High Totals Low n (% rows) 150 (84.3) 28 (40.4) (67.5) 23 (29.5) 97 High n (% rows) 41 (53.4) 28 (15.6) (57.4) 23 (16.5) 54 Totals d Beliefs in relation to susceptibility to injury from distracted crossing and benefits and barriers to safer behaviour (application of the Health Beliefs Model) As described earlier, two theoretical models used to explain health and social behaviour were used in exploring pedestrian attitudes and behaviours in relation to smart phones and crossing. The Health Beliefs Model (HBM), was examined through 15 questions related the constructs underlying the HBM. Table 3.8 shows the wording of the items used in this section and the variables/constructs with which they were associated. Response options were on a scale of 1= strongly disagree to 7 = strongly agree, with high scores thus indicating greater degree of agreement. For the analysis, subscales for perceived susceptibility to injury as a result of crossing the road while using a smart phone and perceptions of severity of such injury if it were to happen were formed by taking the mean rating on the relevant items (3 items each, see Table 3.10). Similarly, subscales were formed for the remaining variables: benefits of the safe behaviour, barriers, and self-efficacy in relation to the safe behaviour, with the appropriate items reverse scored (as indicated in Table 3.10). Results suggested high levels of perception of susceptibility to injury as a result of distraction by smart phone use while crossing, and similarly high perceptions that injuries would be non-trivial and have an impact on one s life. Responses to the items relating to the advantages of the safe behaviour suggested high levels of perception of the benefits of the safe behaviour, and self-efficacy scores suggested that this behaviour was not seen as difficult to engage in. Scores for the barriers to the safe behaviour, suggested disagreement that the specified disadvantages would result from stopping using a smart phone while crossing the road. Austroads 2016 page 57

68 A series of one way ANOVAs was conducted on the mean scores for each subscale by age and by gender in order to investigate whether there were age or gender-related differences for any of the variables. Results are displayed in Table Age-related differences were detected for Susceptibility, Severity, and Barriers in relation to the safe behaviour, but not for Benefits. Posthoc testing (Tukey) revealed that the and year olds perceived significantly lower levels of Susceptibility to injury from smart phone use while crossing than did the year olds (see Table 3.11). Patterns were somewhat different for Severity, where the year olds perceived significantly lower levels severity of injury from smart phone use while crossing than or year olds. The and year old groups were not statistically significantly different from one another. For Barriers to the safe behaviour, year olds and year olds were not statistically significantly different. However both and year age groups had significantly higher levels of agreement than the year olds that stopping using their smart phone while crossing the road would result in disadvantages. Values of η 2 suggest that the significant effects are all towards medium sized. This suggests that that in a real situation, even though younger pedestrians (18-44 years) are aware that stopping using their smart phones while crossing is safer and would make them more attentive and aware of what is going on around them, they may simultaneous be less likely to see the risks (disadvantages) of being using their smart phones at these times. Thus younger pedestrians may have competing interests in relation to smart phone use while crossing the road. Table 3.10: Pedestrian responses (mean rating) to susceptibility, severity, benefits, barriers and self-efficacy measures (Health Belief Model) Question wording and variable/subscale (responses are given on a scale of 1-7 with higher scores indicating stronger agreement with the statement) Mean (SD) Susceptibility 5.23 (1.33) There is a good chance that I would be injured if I use my smart phone for texts or internet access when I am crossing the road It is safe to use my smart phone for texts or internet access when I cross the road (Reverse scored) I am likely to be injured if I use my smart phone for texts or internet access when I am crossing the road Severity 5.53 (1.04) Being injured because I was using my smart phone for texts or internet access while crossing the road would affect my life a lot If I had an injury because I was using my smart phone for texts or internet access while crossing the road, it would only be minor (Reverse scored) Getting over an injury I had because I was using my smart phone for texts or internet access while crossing the road would take a long time Benefits of safe behaviour 5.51 (1.32) I would be more aware of traffic if I stopped using my smart phone for texts or internet access while crossing the road I would be much safer if I stopped using my smart phone for texts or internet access while crossing the road I could pay more attention to the people with me if I stopped using my smart phone for texts or internet access while crossing the road Barriers to safe behaviour 2.63 (1.22) If I stopped using my smart phone for texts or internet access while crossing the road it would prevent me from responding to urgent or important messages It would waste my time if I stopped using my smart phone for texts or internet access while crossing the road I would get bored if I stopped using my smart phone for texts or internet access while crossing the road Self-efficacy in relation to safe behaviour 4.65 (.80) I am confident that I could stop using my smart phone for texts or internet access while crossing the road if I wanted to It would be easy for me to stop using my smart phone for texts or internet access while crossing the road It is too hard to stop using my smart phone for texts or internet access while crossing the road (Reverse scored) Austroads 2016 page 58

69 Table 3.11: Health Beliefs Model variable scores by age group Age group years years years F statistic, p-value, effect size (partial eta sq) Mean SD Mean SD Mean SD df 2, 246 Susceptibility 4.89 a a b 1.14 Severity 5.20 a b b 1.06 F = p <.01 η 2 =.056 F = p <.01 η 2 =.048 Benefits of safe Behaviour Barriers to safe behaviour Self-efficacy for safe behaviour a a b a a b.85 F =.582 p =.559 ns F = p <.01 η 2 =.049 F = p =.165 ns ab Rows means with different superscripts are significantly different at p <.01 level No difference for gender on any of the subscales were detected when scores for the entire sample were considered. However, a significant difference between men and women was found for the year olds on the Benefits of the safe behaviour variable, F (1, 164) = 9.634, p <.01. Young women had significantly higher scores on this subscale (M = 5.67, SD =.108) compared to young men (M = 5.02, SD =.182), suggesting that they had greater agreement with the specified benefits of not using a smart phone while crossing, though scores were high for both e Extended TPB constructs predicting intentions to use smart phones while crossing (application of the Theory of Planned Behaviour) The target behaviour was defined as using your smart phone while crossing the road. In relation to the use of smart phones, participants were asked to think about texting and internet access types of use and were also told the following: when we say 'for texting or internet' we mean initiate, monitor or respond to (e.g. checking Facebook, writing etc.) in order to make the behaviour as close as possible to that in the earlier sections of the survey. Table 3.12 lists the items used in this section of the survey and the constructs they were designed to measure. All response scales were 7 point, with 1 = strongly disagree to 7 = strongly agree. Attitudes towards using a smart phone while crossing the road were assessed with four items (see Table 3.102) that assessed attitudes in relation to whether using a smart phone while crossing was enjoyable, satisfying, safe and wise. Ratings for the four items were summed, and the average taken to obtain a mean rating for attitude to using a smart phone while crossing the road. Reliability for the 4-item scale was α =.84. High scores indicate more positive attitudes towards using a smart phone while crossing the road. The subjective norms subscale (SN) was formed by first summing responses to the four items for perceived approval from important others for using a smart phone while crossing and then calculating the mean. Cronbach s alpha for the four item measure was α =.91. Higher scores indicate greater perception that important others would support using a smart phone while crossing the road. Austroads 2016 page 59

70 Three items were designed to measure perceptions about the level of control (PBC) over using a smart phone while crossing the road. It was originally intended that these would form a single scale. However, reliability analysis revealed an unacceptably low value for the three item scale, α =.487. Removal of the item Using a smart phone for texting or internet functions while crossing the road would be easy to do improved the reliability to α =.787. To resolve this situation, and consistent with Horvath, Lewis and Watson (2012), the PBC measure was separated into two: PBC Control (2 items) which considers the perceived control a person has over the behaviour and PBC Efficacy (1 item) which considers the perceived ease (or difficulty) of performing the behaviour (Horvath, Lewis & Watson 2012). Higher scores on either measure indicate greater perception of control over smart phone use while crossing the road. Additional measures were included to improve the predictive power. As described above, group norm and mobile phone involvement were regarded as likely to add to the explanatory power of the TPB. Four items that related to whether my friends would approve and support using a smart phone while crossing (Group Norm) were added. Cronbach s α =.91 for this measure. A mean score for the sum of the responses to Group Norm was used in the analyses. Participant mean scores on the Mobile Phone Involvement Scale was included as the measure of mobile phone involvement. Intention to use a smart phone while crossing the road (Intention) was measured with three items. Mean responses to the three item scale had very high reliability, with Cronbach s α =.96. Higher scores indicate greater intention to use a smart phone while crossing the road. Table 3.12: Questions and mean responses (SD) for measures of attitudes, group norms, perceptions of control and intentions (TPB) in relation to using smart phones while crossing Question wording and variable/subscale Mean (SD) Attitude M=2.61 (1.32) For me, using a smart phone for texting or internet functions while crossing the road would be: unenjoyable enjoyable For me, using a smart phone for texting or internet functions while crossing the road would be: safe unsafe For me, using a smart phone for texting or internet functions while crossing the road would be: unwise wise For me, using a smart phone for texting or internet functions while crossing the road would be: satisfying unsatisfying Subjective Norm (SN) M=2.01 (1.08) Those people who are important to me think that I should use a smart phone for texting or internet functions (initiate, monitor or respond to) while crossing the road Those people who are important to me think using a smart phone for texting or internet functions while crossing the road would be a good thing to do Most people whose opinions I value would approve of me using a smart phone for texting or internet functions while crossing the road People who are important to me would want me to use a smart phone for texting or internet functions while crossing the road Group Norm M=3.61 (1.34) My friends would approve of me using a smart phone for texting or internet functions while crossing the road My friends think that using a smart phone for texting or internet functions while crossing the road would be a good thing to do Most of my friends would use a smart phone for texting or internet functions while crossing the road My friends would regularly use a smart phone for texting or internet functions while crossing the road How many of your friends would use a smart phone for texting or internet functions while crossing the road? Austroads 2016 page 60

71 Question wording and variable/subscale Mean (SD) Perceived Behavioural Control (Control) M=6.33 (1.06) I have complete control over whether I use a smart phone for texting or internet functions while crossing the road It is entirely up to me whether I use a smart phone for texting or internet functions while crossing the road Perceived Behavioural Control (Efficacy) M=4.52 (1.89) Using a smart phone for texting or internet functions while crossing the road would be easy to do Intentions M=2.96 (1.79) I intend to use a smart phone for texting or internet functions while crossing the road I would be willing to use a smart phone for texting or internet functions while crossing the road It is likely that I will use a smart phone for texting or internet functions while crossing the road Overall mean scores on the various TPB measures suggested negative attitudes (M = 2.61), low perceptions of favourability with important others (SNI = 2.01), moderately low perceptions of favourability with friends (SNF = 3.61), high control (PBC Control = 6.33), moderate self-efficacy (PBC self-efficacy = 4.52) and low intentions (2.96) towards using a smart phone while crossing the road. To examine the relationships between the TPB variables and pedestrian intention to use a smart phone while crossing the road, hierarchical regression, with intention (Intention) as the dependent/outcome variable of interest, was conducted. The standard TPB construct of attitude subjective norm (SNI) and perceived behavioural control (PBC, separated into two measures, as described above, PBC Control and PBC Selfefficacy) were entered at the first step. To assess whether the additional predictors of Mobile Phone Involvement and Group (friends) Norms provided additional predictive power over and above that from the standard TPB predictors, these two variables were entered at the second step. Given the differences shown for age group in the earlier analyses, especially that for mobile phone involvement, it was anticipated that either the specific variables that were predictors, or the strength of important predictors, might vary with age. Thus, two separate regressions were performed, the first for the entire sample, and the second for the year olds only. Table 3.13 displays the results for the entire sample and Table 3.14 for the year olds. The regression model for the entire sample significantly predicted intentions to use a smart phone while crossing the road, F (4, 256) = , p <.001. Standard TPB variables were entered at Step 1 and accounted for 61.6% variance in intentions (R 2 adj =.616). Attitude, SN and PBC Self-efficacy were all significant predictors. After entry of the additional variables, mobile phone involvement (MPI) and Group Norm, provided an additional significant 3.0% of the variance explained (R 2 change =.030). In the final model, there were five statistically significant predictors: Attitude, SN, PBC Efficacy, and the mobile phone involvement measure. Attitude was by far the strongest predictor of intention in terms of beta weights (beta =.477, p <.001), followed by SN (beta =.191, p <.001) and Group Norm (beta =.173 p <.001). All beta weights were positive, suggesting that more positive attitudes, greater perceptions that others would approve, and that friends would approve of using a smart phone while crossing the road are associated with stronger intentions to do so. Austroads 2016 page 61

72 Patterns of important predictors were somewhat different for the year olds. The regression model significantly predicted intentions to use a smart phone while crossing the road, F (4, 160) = , p <.001. In Step 1, the standard TPB measures together explained 58.8% of the variance in intentions (R 2 adj =.588). Of these, Attitude and subjective norms (SNI, important others) were significant, but neither of the PBC measures was significant. Additional variables, mobile phone involvement and group norm in Step 2 explained an additional 3.0% of the variance in intention. Both of these additional predictors were significant. In the final model, there were only four statistically significant predictors: Attitude, SN, Group Norm, and the mobile phone involvement measure. Attitude was by far the strongest predictor of intention (beta =.545, p <.001). For this age group, group norm (beta =. 192 p <.001, was the next strongest predictor. SN (beta =.122, p <.001) and mobile phone involvement were weaker predictors (beta =.112), while neither of the PBC measures was a significant predictor for this age group. This pattern of results suggests that, for the younger aged pedestrians (18-30 years), more positive attitudes towards using a smart phone while crossing the road are associated with greater intentions to do so. For year olds, greater perceptions that friends would approve of or support using smart phones while crossing the road were also associated with stronger intentions to use a smart phone while crossing the road. Table 3.13: Summary of hierarchical regression analysis for Theory of Planned Behaviour standard (Attitude, Subjective Norms, Perceived Behavioural Control) and additional (mobile phone involvement, Group norm) predictors of intentions to cross the road while using a smart phone for text/internet access (entire sample, year olds, N=268) Block 1 B SE Beta R 2 Adj R 2 Δ R 2 Attitude *** Subjective Norm ** PBC Control PBC Self-efficacy *.622 *** *** Block 2 MPI * Group Norm (friends) ***.652 *** *** *p <.05, ** p <.01, *** p<.001 Table 3.14: Summary of hierarchical regression analysis for Theory of Planned Behaviour standard (Attitude, Subjective Norms, Perceived Behavioural Control) and additional (mobile phone involvement, Group norm) predictors of intentions to cross the road while using a smart phone for text/internet access (18-30 year olds only) Block 1 B SE Beta R 2 Adj R 2 Δ R 2 Attitude *** Subjective Norm * PBC Control PBC Efficacy Block 2 MPI *.598 *** *** Group norm (friends) **.639 *** *** *p <.05, ** p <.01, *** p<.001 Austroads 2016 page 62

73 3.2.5.f Self-reported likelihood of crossing the road under different hypothetical scenarios As described earlier, eight different scenarios were presented to participants to find out which of three factors (median strips, traffic signals, type of smart phone use), if any, have the most influence over their likelihood of using their smart phones while crossing the road. Participants read the text descriptions, which were accompanied by a photograph of a road with the characteristics in the text description. Immediately following each scenario, participants were asked to rate (on a scale of 1= Very unlikely to 7 = Very likely) how likely they were to cross the road while using their smart phones in the circumstances in the scenario. Mean likelihood ratings for each hypothetical scenario are displayed in Table As can be seen in Table 3.15, self-rated likelihood of crossings varied by the type of smart phone activity described. It appears that, overall, the entire sample of pedestrians were likely to say they would cross if they were making a voice call (mean ratings all above 4) and would not be likely to cross if texting or using the internet (mean ratings all below 4). Ratings of likelihood of crossing were highest in scenarios where smart phones were described as used for a voice call and there were traffic signals (regardless of median presence or absence). Ratings of likelihood of crossing was lowest for scenarios where the type of smart phone use was for texts or internet access and there were no signals (midblock) or median strip. Table 3.15: Mean ratings of the likelihood of crossing in scenarios varying by i) type of smart phone activity (voice call, text/internet), ii) presence or absence of pedestrians signals, and iii) median strips (present, absent) by age group. Scenario characteristic All M (SD) (N = 268) Mean likelihood would cross M (SD) (N = 168) M (SD) (N = 40) M (SD) (N = 41) Voice call, signals, median (Sc1) 4.6 (1.9) 4.8 (1.8) 4.6 (2.1) 3.5 (2.0) Voice call, signals, no median (Sc 5) 4.6 (1.9) 4.7 (1.8) 4.8 (2.0) 4.1 (2.0) Voice call, no signals, median (Sc 2) 4.5 (1.8) 4.5 (1.8) 4.9 (1.8) 3.8 (2.0) Voice call, no signals, no median (Sc 6) 4.7 (1.9) 4.8 (1.8) 4.8 (2.0) 3.8 (2.0) Text/internet, signals, median (Sc 7) 3.4 (1.9) 3.5 (1.9) 3.6 (1.9) 2.5 (1.8) Text/internet, signals, no median (Sc 4) 3.5 (2.0) 3.7 (1.9) 3.8 (2.0) 2.3 (1.7) Text/internet, no signals, median (Sc 8) 3.1 (1.8) 3.3 (1.9) 3.0 (1.7) 2.3 (1.7) Text/internet, no signals, no median (Sc 3) 2.8 (1.8) 2.8 (1.8) 3.0 (1.9) 2.1 (2.0) To see if there were significant differences in the influence of the type of distraction, signals or median strip presence on likelihood of crossing, ANOVA procedures were used. To permit this investigation, a three way ANOVA (three factors- median strips, traffic signals, type of smart phone use, with two levels for each factor) of the responses to scenarios, a series of eight dummy variables was first created, one for the participant s response to each scenario. Results revealed a significant two way interaction between traffic signals and type of smart phone use (F (1, 2120) = 9.961, p <.01). However, the effect of this interaction was very small (η 2 =.005) and hence this effect may be disregarded. Two significant main effects were found. The first was for type of smart phone use (F (1, 2120) = , p <.001), which was a large effect (η 2 =.125). The second main effect was for traffic signals (F (1, 2120) = , p <.01), but this was a very small effect (η 2 =.005). There was no significant effect for median strips. Thus, it appears that for the entire sample, pedestrians were influenced by what type of smart phone activity was described. They were more likely to say they would cross the road while using their smart phones if they were on a voice call (M = 4.6, SD = 1.9) rather than if they were using the phone for texts or internet access (M = 3.2, SD = 1.9). The presence or absence of pedestrian signals appeared to have a very small influence on their ratings of likelihood of crossing, and likelihood of crossing was not related to median strips. Austroads 2016 page 63

74 Given the differences in patterns of responses for younger people and for people high in mobile phone involvement, two further analyses were carried out. To see if the patterns of responses to the scenarios were different for younger people, a second three way ANOVA (three factor- median strips, traffic signals, type of smart phone use, with two levels for each factor) analysis was carried out for the year olds only. A significant two way interaction between type of smart phone use and traffic signals (F (1, 1336) = 7.001, p <.01) and a significant three way interaction for type of smart phone use, traffic signals, and median strip (F (1, 1336) = 8.48, p <.01 were found. However, effect sizes for these interactions were very small (η 2 =.005 and.006 respectively) and so will be disregarded. Significant main effects for type of smart phone use (F (1, 1336) = , p <.001) and for traffic signals (F (1, 1336) = 8.828, p <.01), were found, but there was no significant effect for median strips. Effect of type of smart phone use was large (η 2 =.124) while effect of traffic signals was very small (η 2 =.006), suggesting that year olds were influenced by the type of smart phone activity depicted in the scenario. They were more likely to say they would cross the road while using their smart phones if they were on a voice call (M = 4.7, SD = 1.8) rather than if they were using the phone for texts or internet access (M = 3.3, SD = 1.9). The presence or absence of pedestrian signals was a very small influence on their ratings of likelihood of crossing, and likelihood was not related to median strips g Drink walking To assess the level of drink walking participants were asked to indicate how often they had walked within an hour of consuming two or more standard alcoholic drinks on a seven point scale (1 = never to 7 = very often ). Around 38% of the sample responded never to this question, leaving 165 sets of responses to the remaining questions on drink walking. In relation to the types of locations drink walkers were likely to walk to, around half the sample indicated that they were very likely (responses of very likely or extremely likely ) to walk to four of the five specified locations: to their homes (48%), between licensed premises (50%), to a public transport site (47%) and to areas known or familiar to them (53%). A smaller proportion said they were very likely to drink walk to a friend/partner s house (28%). Drink walkers were asked their perceptions of how likely it was that others would drink walk in four situations. Results indicated the majority of the drink walkers thought others would drink walk home from a licenced venue (68% responded very likely or extremely likely ), between licensed venues on a night out (92% very likely or extremely likely ), home from a private party (52% very likely or extremely likely ) and between private parties (40% very likely or extremely likely ). These patterns of responses suggest that the journey home, between licensed venues or to a public transport site are potentially the most important targets in relation to drink walking Discussion The majority of the pedestrians in this sample indicated that they did not use their smart phones while crossing the road. A large proportion of them reported that they never did this, which is an encouraging finding. However, around 20% of the entire sample indicated that they did use their smart phone for activities such as texting and internet access while crossing the road. Moreover, when the frequency of these activities was categorised into high and low exposure to distraction while crossing, 20-32% of participants were categorised into the high exposure group. Age-based analysis of self-reported behaviours revealed that younger pedestrians aged years were more likely to be categorised as at high exposure. Austroads 2016 page 64

75 Overall responses to the measure of other risky pedestrian behaviours (PBS)suggested that this sample perceived themselves as generally low risk takers as pedestrians. However there were two behaviours which may represent opportunities for interventions. These were finishing crossing the road by walking diagonally, and crossing the road between parked cars. Participants in the year age group were significantly less likely than the year olds to report engaging in risky pedestrian behaviours, suggesting that it is younger adults who should be targeted in any interventions. However, around 10% of the pedestrians in this study appeared to both engage in riskier general pedestrian behaviour and also greater use of their smart phones while crossing the road. Though this is a relatively small proportion of pedestrians overall, the size of the effect obtained in the analyses suggests that this may be an important high risk group to target in interventions. As anticipated, and consistent with their greater likelihood of familiarity and comfort with smart phone technology, or being digital natives, younger adults years had significantly higher scores on the measure of mobile phone involvement (MPIS) than the older adults in the sample. Furthermore, it was the year olds who comprised the majority of the potentially problematic high involvement group: over one third of year olds were categorised as being highly involved with their smart phone. Moreover, the patterns of results from the more detailed analyses (multiple measures) suggested the existence of a small but high risk sub-group characterised by problematic mobile phone involvement coupled with higher frequency of smart phone use while crossing the road. Results for the application of the constructs from the Health Beliefs Model suggested that, as a whole, the pedestrians in this sample were aware of their own susceptibility to injury from distracted crossing and thought that an injury sustained from this cause would be non-trivial. These findings were especially the case for the year olds who appeared to regard potential injury as likely to seriously affect them. However, year olds had significantly less agreement in relation to susceptibility and severity than the other aged adults, though it was still relatively high, suggesting that some year olds were less convinced of the risk of injury from distracted crossing. Over the entire sample, agreement was relatively high that the safe behaviour, stopping using my smart phone while crossing the road would produce benefits of greater safety, greater awareness of traffic, and would allow them to pay more attention to what was going on around them. Similarly, there was high disagreement overall that the safe behaviour would result in any of the negative outcomes (barriers) such as being prevented from responding to urgent messages, wasting their time or getting bored. It appeared that self-efficacy in relation to stopping using their phones while crossing was also high for this sample. Together this pattern of responses suggests that pedestrians would be receptive to interventions that attempt to promote stopping using a smart phone for texts and internet while crossing the road. However, although they also had high disagreement in relation to the barriers to the safe behaviour, year olds and year olds had significantly less disagreement than year olds. It may be that for some of the younger aged pedestrians, this perception coupled with a lower agreement that they are at risk of injury, or lower perceptions of the likely impact of a pedestrian crash might make them somewhat more resistant to attempts to persuade them to alter their smart phone behaviour. The results here support the use of constructs from the Health Beliefs Model as useful in informing or guiding interventions to reduce smart phone use while crossing the road. For all aged pedestrians, public education messages could promote the benefits of deliberately choosing not to use a smart phone while crossing, since this was a behaviour that the participants in this sample regarded as easy to engage in, producing benefits to safety and attention, and not likely to lead to disadvantages such as boredom or being unable to respond to important messages. However, for pedestrians aged years, the results above suggest that there may be a subgroup of pedestrians who are less likely to respond to this kind of messaging and who may require more careful tailoring of interventions. These are the young smart phone users who hold perceptions that they are not very susceptible to injury as a result of using their smart phones while crossing, or that if injured, it would not affect them very much, and who are more likely to see disadvantages to not using their smart phones while crossing (that is, think that they will be bored, will waste their time or will be unable to respond to an important message if they stopped using their phones while crossing). For this subgroup, educational messages may be more effective if they seek to raise awareness of the risk of injury and to challenge perceptions that boredom, wasting time or lack of responsiveness will result. Austroads 2016 page 65

76 Responses for this sample on the questions related to the Theory of Planned Behaviour suggest that overall, pedestrians did not intend to use their smart phones while crossing the road. The overall mean score for the outcome measure of intentions ranged between 1.7 (45-65 year olds) to 3.26 (18-30 year olds) for the different age groups suggesting that for the entire sample intentions to use a smart phone while crossing the road were low, but younger pedestrians had stronger intentions to do so than older pedestrians. In regards to factors influencing intentions to use a smart phone while crossing, the results for the regressions provided a relatively high and significant level of explanation of the variance in intentions, at 65%. Unexpectedly, the additional measures that were included, that is mobile phone involvement and group norms (friends) only contributed a very modest significant 3% of additional explanation of the variance in intentions over and above that from the standard variables (attitude, SN, PBC). According to the beta weights, attitudes were a much stronger predictor than any of the other significant predictors. The results indicated, as expected, that the association between attitudes and intentions was positive thus the more favourable one s attitudes towards using one s phone while crossing the stronger one s intentions to use a smart phone while crossing. This finding suggests that interventions such as public education and advertising campaigns may need to strengthen individuals negative attitudes towards using smart phones while crossing as that would be associated with lower or weaker intentions to engage in the behaviour. In particular, and consistent with the indirect beliefs underpinning the construct of attitude, namely, behavioural beliefs, it may be necessary to challenge the perceived advantages of smart phone use while crossing and/or emphasise the perceived disadvantages of distracted crossing. Either or some combination of approach to message development would strengthen behavioural beliefs against the behaviour and thus increase the extent to which individuals hold unfavourable views of the behaviour. The other significant predictors within the final step of the model for year olds included (in order of relative beta weight) group norm, subjective norm, and mobile phone involvement. For all of these additional predictors, the association between each construct and intentions were positive, which was as expected. Specifically, the more an individual regards important others, and their friends in particular, as important normative referent groups who would approve of their engaging in the behaviour (and who engage in the behaviour themselves), the more he or she would intend to use a smart phone while crossing the road. Similarly, for mobile phone involvement, the more one feels a sense of involvement with a mobile phone, the stronger the intentions to engage in use while crossing. These findings suggest that normative influences are particularly important in this context and with this age group. In turn, this suggests that countermeasures such as public education messages may function well to influence individuals by either challenging this perception that important others approve or by emphasising that there are important others who do not approve of the behaviour. In regards to group norm, a specific example of an advertising message could be one which promotes the notion that if one wants to be sure to keep one s friends safe then a key thing is to model not engaging in smart phone use while crossing the road, given that friends are likely to do what they see other friends doing (or to behave in a manner consistent with their perceptions of what their friends would approve of). The important role of group norm and, in particular, friends in regards to influencing young people s pedestrian safety was found in a study relating to young people s drink walking intentions (Gannon et al 2014) and therefore suggests that there may be a broader, important role of friends in promoting pedestrian safety among young adults. Interestingly, despite theoretical expectations, neither PBC-efficacy nor PBC-control were found to be significant predictors of intentions for this age group. This finding is surprising to the extent that PBC is a relatively strong and consistent predictor of intentions (and behaviours) across a range of health and socialrelated behaviours. The finding is also of further interest to the extent that it indicates a point of departure between the regression results for the younger adults and the overall sample including older adults. In the model for the overall sample, in addition to attitude, subjective norm, group norm, and mobile phone involvement, PBC-efficacy was also a significant, positive predictor of intentions. Together, the findings for both regression models largely support the value of the TPB for predicting and explaining pedestrians intentions to use their smart phone while crossing. TPB-based interventions have been shown to produce beneficial changes in intentions and subsequent behaviours in regards to a range of health and social-related behaviours and therefore the current findings suggest that public education messages guided by the TPB may be of value in regards to efforts to improve pedestrian safety. Austroads 2016 page 66

77 Pedestrian responses to the hypothetical crossings suggest that decisions to use smart phones while crossing the road are strongly influenced by the type of activity (text/internet versus voice calls) rather than the road environment per se. One interpretation of this finding is that pedestrians are well aware of and sensitive to their own level of likely distraction from complex smart phone interaction represented by texting or internet access. Countermeasures could attempt to capitalise on this, reinforcing and supporting pedestrian decisions to not cross while distracted. Taken together the results above suggest that receptivity to interventions that promote not using a smart phone while crossing the road appears likely to be high. These might have greatest effectiveness if they attempt to influence attitudes towards smart phone use by emphasising the risks (especially for the more distracting activities), and challenging the perceived potential benefits or approval from friends for using a smart phone while crossing the road. In addition, interventions could promote modelling of not using smart phones while crossing the road among young people as a way of protecting their friends as well as themselves. Austroads 2016 page 67

78 4. Conclusions and Recommendations The current research was commissioned by Austroads to investigate the area of pedestrian distraction due to use of technological devices alcohol impairment, while interacting with the road environment. The review of the literature concluded that pedestrian safety is influenced by multiple factors, including demographics such as gender and age, social conformity and group identity, and cultural or religious factors. The potential influence of group differences in terms of attitudes, beliefs, perceived risk and intended behaviour also has important implications for road safety (Holland & Hill 2007; Yagil 2000). Understanding, for example, potential motives or intentions among young adults is important when considering that individuals with lower perceptions of risk of negative outcomes from impaired walking, who hold positive attitudes towards this behaviour or who perceive it as something that their friends/family engage in or approve of are more likely to drink walk (McGhie et al 2012, Gannon et al 2014). Personality or temperament factors may provide a way of targeting at-risk pedestrians for intervention purposes (Schwebel et al 2009). The prevalence of pedestrian distraction by hand held technological devices appears to be increasing, with studies showing that up to 40% of observed pedestrians are distracted when crossing the road (Ferguson et al 2013; Thompson et al 2013; Bungum et al 2005). Pedestrians distracted by mobile phones may be at increased risk of crashing with results showing that mobile phone users walk more slowly, change directions more frequently, are less likely to acknowledge other people, look left and right fewer times, are less likely to look at traffic before starting to cross, and make more errors than pedestrians who are not distracted (Thompson et al 2013; Bungum et al 2005; Hatfield & Murphy 2007). In terms of locations, most pedestrian fatalities have been shown to occur at non-intersection locations (Balk et al 2014; Gitelman et al 2012), where they are more often at fault than at intersections (Nieuwesteeg & McIntyre 2010). Crossing against the red man (entering on flashing and steady red man) and crossing within 20 m of the crossing but not at the crossing has been shown to have a crash risk approximately eight times that of legal crossing at signalised intersections (King et al 2009). A range of factors can influence pedestrians decisions about where and when to cross, including the location of the cross walk relative to the origin and destination of the pedestrian, and the availability of pedestrian signals (Sisiopiku & Akin 2003). The number of lanes and whether the road is one-way versus two-way also influences pedestrian behaviour and the likelihood of disobeying a crossing signal (Cambon de Lavalette et al 2009). Aside from the location of crossings, there are multiple other factors relating to the built environment that can impact pedestrian behaviour and safety. These include land use characteristics such as floor space of bars, cinemas and theatres, and building set back (how far the buildings are from the edge of the road), public transport facilities such as train or tram stops, and road network characteristics such as speed limits, number of intersections and road design features such as the presence of hook turns in Melbourne CBD. Finally, several environmental factors have been examined. These include pedestrian risk at night time, or in low light conditions and seasonal variation such as whether people behave differently in winter months. The brief for this project specified a number of questions to which answers were sought. These are addressed below. What factors influence pedestrian behaviour compared to pedestrian safety? Age: Adolescents and young adults are the most likely to be distracted pedestrians. Mobile phone-related injury has been found to be higher among younger people, especially teens, in US studies. Rates of smart phone ownership in Australia are high, especially among the young and make performance of cognitively or visually demanding interactive use possible while crossing the road. Location: Austroads 2016 page 68

79 Most pedestrian fatalities occur at non-intersection locations. At signalised intersections, crossing within 20m of a crossing but not at the crossing increases crash risk by factor of 8 compared to legal crossing. Odds of distracted crossing are higher when at signalised crossings than at unsignalised ones. Location of the crossing relative to the origin and destination of the pedestrian, availability of pedestrian signals the number of lanes and whether the road is one-way versus two-way influence pedestrian behaviour and the likelihood of disobeying a crossing signal. Pedestrian crash risk from type of location appears to be multi-factorial in nature, with different sets of factors associated with risk across different space-time frameworks. Illegal road use Illegal road use (e.g. crossing away from marked crossings; crossing on red signal) by pedestrians is widespread (e.g. ~20% of crossings at signalised intersections, Brisbane). Illegal crossing (e.g. crossing within 20m of a crossing but not at the crossing) increases crash risk. Frequency of illegal crossing appears related to characteristics of the road/location. Exposure: Engaging in distracting smart phone tasks while walking and crossing Up to 40% of pedestrians may be distracted by mobile phones when crossing the road Smart phone use for texting or internet access while crossing is widespread for year olds, with approximately 30% reporting high frequency of using smart phones to text or access internet functions while crossing the road. Type of mobile phone activity Distraction from cognitively demanding voice calls are associated with riskier pedestrian behaviour Texting, talking and cognitively demanding smart phone use is associated with greater reaction times and increased errors for competing tasks Type of smart phone activity (voice call, text, internet access) is the strongest predictor (compared to presence/absence of signals, presence/absence of median strip) of self-reported use of a mobile phone while crossing for pedestrians aged years. Listening to music through headphones/earpods may represent a lower level of distraction than texting, voice calls or internet access/use. Risk perception and attitudes The majority of pedestrians are aware of the risks of injury from crossing while distracted, and self-report that they do not engage in this behaviour. Younger adults are significantly less aware than older adults of their susceptibility to injury from distraction by using a smart phone while crossing year olds regard voice calls as less risky than texting/internet use while crossing the road. Attitudinal research suggests adults years old with lower risk perceptions, positive attitudes towards drink walking and perceptions that their friends/family also drink walk or approve of this behaviour are more likely to drink walk. A substantial minority of pedestrians may constitute a high risk subgroup. These are the high frequency smart phone users who engage in the more highly distracting smart phone activities (e.g. initiating text messages, initiating internet access) while crossing the road Pedestrians appear to perceive the advantages of not using smart phones while crossing the road as well as regard doing so as not likely to result in boredom, wasting their time or preventing them responding to important messages. They also agreed that stopping using their phones while crossing would be easy to do. Austroads 2016 page 69

80 More positive attitudes towards smart phone use while crossing the road and greater perceptions that friends/important others would approve were associated with greater intentions to using a smart phone while crossing the road. Mobile phone involvement Around 35% of year olds may be overinvolved with their mobile phones and therefore use them at inappropriate or unsafe times (e.g. when crossing the road; while driving). This group may also be more difficult to influence in relation to safer behaviour. Impairment by alcohol Crash statistics suggest that more than a third of fatally injured pedestrians have a BAC >.05, the majority of these having a BAC >.15. Drink walking is prevalent among young adult (18-24 years old) patrons of licensed venues. Most alcohol-involved pedestrian casualties are men, struck at night-time, on Friday or Saturday, are walking home, have been drinking for several hours, and are struck within two to three hours of finishing drinking. Around 20% of the pedestrians surveyed in this research indicated they had walked after drinking two or more standard drinks in the previous hour at least once per week during the previous three months, and the majority of these (two thirds) thought they were affected by the alcohol at the time. Drink walkers may be aware that they are impaired but choose to drink walk regardless. In the pedestrians surveyed for this report, drink walking-related injuries and near misses were rare, affecting only 2% of the sample. In what ways do these factors affect pedestrian behaviour especially in relation to pedestrian safety? Distraction while crossing and walking increases erratic behaviour and decreases safety-related behaviour Pedestrians distracted by mobile phones walk more slowly, change directions more often, acknowledge others less, look left and right less, are less likely to look at traffic before starting to cross, miss more safe opportunities to cross, take longer to initiate crossing, are more likely to cross unsafely into oncoming traffic, spend more time looking away from the road, and make more errors than pedestrians who are not distracted. Cognitive and visual distraction by mobile phones is associated with riskier crossing decisions and less safe behaviour. Texting, talking and cognitively demanding smart phone use appears to compete for attentional resources and is associated with greater reaction times and increased errors for competing tasks Risk of injury is increased for pedestrians using mobile phones while crossing Males and younger pedestrians (under 31 years old) are at greater risk of injury from distracted crossing. Lower perceptions of risk, or problematic mobile phone involvement may elevate exposure (more frequent engagement in distracted crossing) Around 35% of year olds may be overinvolved with their mobile phones and therefore use them at inappropriate or unsafe times (e.g. when crossing the road; while driving). This group may also be more difficult to influence in relation to safer behaviour. Greater tendency to use mobile phones while crossing, or problematic levels of mobile phone involvement may lower perceptions of the risks of distraction while crossing the road, or may increase willingness to ignore these risks. Pedestrians generally appear aware of the risk of distraction while crossing and are also receptive to safety interventions that promote not using smart phones while crossing Austroads 2016 page 70

81 In relation to drink walking, attitudinal research suggests adults years old with lower risk perceptions, positive attitudes towards drink walking and perceptions that their friends/family also drink walk or approve of this behaviour are more likely to drink walk. Impairment by alcohol decreases safe road use and is associated with increased crash risk Crash data analysis and simulator-based studies verify the association between alcohol impairment and less safe crossing behaviours and increased crash involvement. Drink walking-related injuries and near misses were rare in the sample surveyed for this report, affecting only 2%. However, if such involvement levels generalise to the general population, it suggests that large numbers of young people may be at risk (as they were more likely to report frequent drink walking). In which locations are these behaviours most relevant when considering pedestrian road safety? The methods used in the research reported here asked about locations characterised by the presence or absence of pedestrian traffic signals and the presence or absence of median strips. It appears that in relation to these locations, pedestrians regarded midblock, unsignalised crossing locations as more risky than those with signals. However, it should be borne in mind that it was the type of smart phone use while crossing the road that had the largest effect on self-reported likelihood of actually using a smart phone while crossing the road. Location type did not appear to have much effect. While participants reported that they were less likely to cross mid-block while using a smartphone, compared with an intersection, this appears to conflict with national crash data that shows a higher level of mid-block pedestrian crashes. It is possible that this is a result of the combination of different levels of exposure and risk. At intersections (especially signalised intersections) there is relatively low risk because groups of people tend to cross together, so for a pedestrian a tactical approach is to assume that other people in a group of crossers are ensuring that it is safe to cross. In addition, signals provide a degree of legal protection through driver compliance (even though there are some offenders) and lower average speeds. At mid-blocks, crossing is often undertaken opportunistically by individuals and there is a lower degree of legal protection and higher than average vehicle speeds. In this situation the risks for a distracted pedestrian are much higher, so that even if they compensate by reducing their likelihood of attempting to cross, the increased risk may outweigh their reduced exposure What measures have been used to influence and change pedestrian behaviours? Most interventions focus on single approach (e.g. education, engineering, enforcement). Since distracted and impaired walking are not covered by legislation in Australia or New Zealand (except in SA), only education and engineering are currently applicable, and only education/awareness measures were found in the literature. Education and awareness intervention programs and campaigns are common, but evaluation of these is rare, and so effectiveness is unknown. One evaluated intervention targeting safe/legal use of pedestrian crossings and signals in central city Auckland showed improvement in pedestrian behaviour. Interventions using enforcement for illegal pedestrian behaviour are uncommon, with logistical difficulties (such as generally low concentrations of pedestrians behaving illegally or wide dispersal of locations for such behaviours, affecting detection) and the costs involved (such as use of police time to detect and issue infringement notices) generally making these rare. Engineering countermeasures to pedestrian behaviour are relatively common, and many have been evaluated and found to be effective. Separating pedestrians from motorised traffic in either time or space improves pedestrian safety generally. On-road separation in space, such as footpaths, has been found to increase pedestrian safety by as much as double, and provision of footpaths addresses the issue of distracted walking (though not distracted crossing). Thus safety benefits are likely if footpath infrastructure were to be installed where it does not currently exist on the transport network. However, for pedestrian crossing, separation in space is impractical in most situations on the transport network in Australia and New Zealand, especially outside central urban areas and areas of high pedestrian activity. Separation in time may be more applicable as a countermeasure to distracted pedestrian crossing. Austroads 2016 page 71

82 Signalisation affects pedestrian decisions about locations for crossing and safety. Raised median strips on multi-lane roads have lowered crash rates. Lowering general vehicle speed limits across high pedestrian concentration areas and local roads is effective at reducing pedestrian crash rates and injury severity. Novel approaches to entertaining pedestrians while they wait to cross at signals has been found to reduce illegal crossing behaviours but may provide effectiveness only in the short term, while there is high novelty value. Pedestrian countdown timers have not been shown to be effective. Technology used to warn pedestrians (e.g. audio beacons) is effective and low cost. Smart technologies (e.g. intelligent road furniture/vehicles) that warn motorists of pedestrians are growing in use but as yet are unevaluated. However, these may offer benefits that are worth exploring, particularly if used in combination with education. Real time intersection traffic analysis (including pedestrian traffic) via video technology may provide an effective method of identifying where traffic and engineering countermeasures for distracted pedestrians should be focused. Drink walking Alcohol impairment in road safety is addressed by a number of countermeasures such as Safe Night Out, lock-out programs, designated driver, responsible service of alcohol (RSA). While they do not specifically address drink walking, strategies which are effective at reducing high levels of alcohol consumption, or pedestrian intoxication (e.g. RSA), or that minimise pedestrian activity while intoxicated/impaired are likely to have the greatest benefit. Education strategies have been used but effectiveness is unknown. Effective engineering countermeasures to intoxicated pedestrian crashes include Dwell on red, and pedestrian fences in late night licensed precincts. Lowering vehicle speeds and altering traffic signal timings at high alcohol times and locations (e.g. latenight entertainment precincts) and pedestrian fencing in proximity to licensed venues has been implemented in Australian cities (but not evaluated). Enforcement countermeasures have included implementation of lock out laws/policies to reduce movement between venues and greater police activity and these have been found to provide some effectiveness in reducing traffic incidents generally and potentially of benefit for pedestrian crash involvement. The Drink Safe Walk Safe project, which had a combination of educational, engineering and enforcement measures, was evaluated as effective and may provide an approach for wider implementation. How appropriate would these measures be in an Australian and New Zealand context? Many of the countermeasures reviewed for this project have already been used in the Australian and/or New Zealand context and therefore appear suitable for more widespread implementation should this be desired. There are promising developments in the pedestrian-friendly mobile technologies area that may provide avenues to prevent crashes by alerting pedestrians and vehicles to each other s presence/proximity. 4.1 Recommended measures to address road safety issues An integrated approach to pedestrian safety, in keeping with the principles of a safe system and incorporating educational, enforcement and engineering measures should be considered and adopted where possible. Austroads 2016 page 72

83 Below, we recommend that there should be exploration of intelligent transport and technology-based methods to warn pedestrians of proximity to crossings or to restrict mobile phone use in such areas as potential interventions. This should be part of a multidisciplinary, multimethod approach that includes education, enforcement and engineering approaches as well since there is the potential for pedestrians to attempt to avoid crossing where such technology is in place, and thus place themselves at greater risk (e.g. by crossing at mid-blocks, potentially also while distracted). Long term changes in risk perception or attitudes towards such behaviours are the eventual goals while increasing safety during the interim period. However, such changes are usually very slow and can require decades to achieve (as happened with seat belt use, for example). Changes to legislation in relation to pedestrian behaviour may support or lead the required changes, providing there is high community support for the legislation. To date, while the results of the surveys conducted for this report are encouraging in relation to attitudes towards safe crossing behaviours, there is no evidence of support for legislation in relation to distraction by mobile phone use or for drink walking. In addition, specific countermeasures to smart phone use when crossing should: Focus on young people (under 31 years old) because the behaviour is most common in this group Be located in high pedestrian volume areas (e.g. intersections and city CBD areas) because the behaviour is likely to be more common in such locations and in order to reach as many pedestrians as possible. Take the form of signs on signal posts as a low cost option. Be included in the road safety education content (e.g. school-based) that targets adolescents, pre-licence and learner drivers. Such materials should attempt to influence attitudes and perceptions of the risk in relation to smart phone/mobile phone use while engaged in other demanding and safety related tasks Public education should seek to: Increase pedestrian awareness of the crash risk associated with distraction from phones when crossing (especially texting/internet access) Challenge perceived disadvantages of not using a smart phone while crossing Challenge perceptions that it is possible to cross safely while texting/using the internet Encourage young people to look out for, and model safe behaviour for, their friends Countermeasures to drink walking should: Take an integrated approach, incorporating a combination of educational, enforcement and engineering measures and involving proprietors of licensed premises, staff of licensed premises, and community representatives in the planning and implementation of the measures. Incorporate responsible service of alcohol (RSA). Jurisdictions could consider mandating this and enforcing its operation. Incorporate educational approaches (e.g. display of education based posters, screening of drink walkingrelated televisions commercials within licensed premises). Jurisdictions could consider mandating a minimum level of such advertising. Incorporate those engineering countermeasures that have proven or promising effectiveness such as: Dwell on red pedestrian fences in late night licensed precincts localised lower vehicle speed limits (using variable signage if required) altering traffic signal timings at high alcohol times and locations. Incorporate targeted enforcement, especially at high alcohol times and locations and target motorist behaviour as well as pedestrian behaviour. Austroads 2016 page 73

84 4.2 Recommended areas for future research Future research could: Investigate the evidence for, and size of, the crash-risk from distracted crossing and from drink walking. Investigate the technical feasibility, user acceptance and effectiveness of countermeasures that providing feedback to smart phone users (ranging from an auditory or visual warning to blanking the screen) identified as about to cross the road (using GPS coordinates or sensors/detectors of mobile phones-inuse close to the crossing/intersection). Investigate the technical feasibility, user acceptance and effectiveness of countermeasures that send a message/signal to smartphones within a certain proximity of the intersection/crossing. Investigate the feasibility and effectiveness of pavement/intersection treatments such as no mobile phone symbols painted at the road edge (where pedestrians gazing downwards at their phones are likely to see them). Investigate the use of video monitoring methods of determining high risk locations for smart phone use while crossing the road. Investigate the feasibility of electronic enforcement of smart phone use while crossing the road. Investigate the feasibility (including assessing the level of community acceptance) of introducing and enforcing legislation governing distracted or alcohol impaired walking and crossing. Consider identifying and developing countermeasures specifically targeting the important minority higher risk young pedestrian group (estimated as comprising 10% of year olds in this report) who are less aware of their susceptibility to injury as pedestrians, more likely to engage in risky pedestrian behaviours generally, and also frequently use smart phones while crossing the road. Austroads 2016 page 74

85 References Abood, D A., Black, D.R., & Feral, D 2003, Nutritional education worksite intervention for university staff: application of the health belief model. Journal of Nutrition Education and Behaviour, 35(5), Australian Institute of Health and Welfare 2011, 2010 National Drug Strategy Household Survey Report. Australian Institute of Health and Welfare, Canberra, Australia. Ajzen, I 1991, The theory of planned behaviour. Organiz. Behav. Human Dec. Process. 50, Ajzen, I 2002, Perceived behavioral control, self-efficacy, locus of control, and theory of planned behavior. Journal of Applied Social Psychology, 32, Alavi, H 2013, Assessing pedestrian crash risk and injury severity in concentrated urban environments. PhD Thesis, Monash University. Available at: Accessed 09 September Austroads 2004, Preventing crashes involving intoxicated pedestrians: Stage 1 Report The extent and nature of the problem. Stage 2 Report An analysis of Australian Coronial Records, Report prepared by ARRB, Australia. Australian Transport Safety Bureau 2001, Alcohol and Road Fatalities, Monograph 5. ATSB, Australia. Australian Communications and Media Authority 2013, Communications report ACMA: Melbourne. Balk, S. A., Bertola, M. A., Shurbutt, J., & Do, A 2014, Human Factors Assessment of Pedestrian Roadway Crossing Behaviour. Report prepared by the U.S Department of Transportation, Federal Highway Administration, McLean, VA. Basch, C.H., Ethan, D., Rajan, S., & Basch, C.E 2014, Technology-related distracted walking behaviours in Manhattan s most dangerous intersections. Injury Prevention, 20, Becker, H.M 1974, The health belief model and personal health behavior. NJ: Slack Blomberg, R.D., & Cleven, A.M 2000, Development, implementation and evaluation of a countermeasure program for alcohol-involved pedestrian crashes. Reported prepared by the US Department of Transportation, National Highway Traffic Safety Administration, Washington DC. Brumfield, R., & Pulugurtha, S.S 2011, When distracted road users cross paths. Article available at Accessed September Bungum, T.J., Day, C., & Henry, J 2005, The association of distraction and caution displayed by pedestrians at a lighted crossing. Journal of Community Health, 30(4) DOI: /s Bureau of Infrastructure, Transport and Regional Economics 2013, Road safety statistics available from Byington, K.W., & Schwebel, D.C 2013, Effects of mobile internet use on college student pedestrian injury risk. Accident Analysis and Prevention, 51, Cairney, P., & Coutts, M 2003, Behaviour patterns associated with fatal crashes involving intoxicated pedestrians an analysis of the ATSB Coronial Database, In: 2003 Australasian Road Safety, Research, Policing and Education Conference, Sydney, New South Wales. Austroads 2016 page 75

86 Cairney, P., Stephenson, W., & Macaulay, J 2004, Preventing crashes involving intoxicated pedestrians, Stage 1 Report The extent and nature of the problem, Stage 2- An analysis of Australian Coronial Records, Report prepared by Austroads, Sydney. Cambon de Lavalette, B., Tijus, C., Poitrenaud, S., Leproux, C., Bergeron, J., & Thouez, J-P 2009, Pedestrian crossing decision-making: a situational and behavioural approach. Safety Science, 47, Campbell, C 2005, Drink safe walk safe. IPWEA NSW Division Annual Conference, New South Wales, Australia. Cavallo, V., & Dommes, A 2014, Can older pedestrians be trained to make safer street-crossing decisions? Presented at the 28 th International Congress of Applied Psychology, 8-13 th July 2014, Paris, France. Chen, L., Chen, C., Ewing, R., McKnight, C.E., Srinivasan, R., & Roe, M 2013, Safety countermeasures and crash reduction in New York City Experience and lessons learned. Accident Analysis and Prevention, 50, Chimba, D., Emaasit, D., Cherry, C.R., & Pannell, Z 2014, Patterning demographic and socioeconomic characteristics affecting pedestrian and bicycle crash frequency. In: 93 rd Annual Meeting of the Transportation Research Board, th January 2014, Washington DC. Conner, M., Norman, P., & Bell, R 2002, The theory of planned behaviour and healthy eating. Health Psychology, 21(2), Cooper, J.F., Schneider, R.J., Ryan, S., & Co, S 2012, Documenting targeted behaviours associated with pedestrian safety. Journal of the Transportation Research Board, 2299, Corben, B., & Duarte, A 2006, Injury reduction measures in areas hazardous to pedestrians. Stage 1: Countermeasure options. Report prepared by the Monash University Accident Research Centre, Melbourne, Australia. Cottrill, C.D., & Thakuriah, P 2010, Evaluating pedestrian crashes in areas with high low-income or minority populations. Accident Analysis and Prevention, 42, Davis, A.L., Avis, K.T., & Schwebel, D.C 2013, The effects of acute sleep restriction on adolescents pedestrian safety in a virtual environment. Journal of Adolescent Health, 53, Demmel, S 2014, Warning system to stop car and cycle crashes. Retrieved 26 February Department of Broadband Communications and Digital Economy 2008, Online statistics. Retrieved 27 September 2012 from Diaz, E.M 2002, Theory of planned behaviour and pedestrians intentions to violate traffic regulations. Transportation Research Part F, 5, Dommes, A., & Cavallo, V 2012, Can simulator-based training improve street-crossing safety for elderly pedestrians? Transportation Research Part F, 15, Dommes, A., Cavallo, V., Vienne, F., & Aillerie, I 2012, Age-related differences in street-crossing safety before and after training of older pedestrians. Accident Analysis and Prevention, 44, Dommes, A., Cavallo, V., & Oxley, J 2013, Functional declines as predictors of risky street-crossing decisions in older pedestrians. Accident Analysis and Prevention, 59, Dommes, A., Cavallo, V., Dubuisson, J-B., Tournier, I., & Vienne, F 2014a, Crossing a two-way street: comparison of young and old pedestrians. Journal of Safety Research, 50, Austroads 2016 page 76

87 Dommes, A., Granie, M., Cloutier, M., Coquelet, C., & Huguenin-Richard, F 2014b, Young and older pedestrians behaviour observed on marked crossways. Presented at the 28 th International Congress of Applied Psychology, 8-13 th July 2014, Paris, France. Dunbar, G 2012, The relative risk of nearside accidents is high for the youngest and oldest pedestrians. Accident Analysis and Prevention, 45, Eichelberger, A.H., Cicchino, J.B., & McCartt, A.T 2013, Profile of fatally injured pedestrians and bicyclists in the United States with high blood alcohol concentrations. In: International Council on Alcohol, Drugs and Traffic Safety Conference, th August 2013, Brisbane, Queensland. Elvik, R., Sorensen, M.W.J., & Naevestad, T 2013, Factors influencing safety in a sample of marked pedestrian crossings selected for safety inspections in the city of Oslo. Accident Analysis and Prevention, 59, Elliot, B.J 2002, A range of possible strategies for reducing intoxicated pedestrian casualties. Report prepared for VicRoads. Ferguson, S.A., Preusser, D.F., Lund, A.K., Zador, P.L., & Ulmer, R.G 1995, Daylight saving time and motor vehicle crashes: the reduction in pedestrian and vehicle occupant fatalities. American Journal of Public Health, 85(1), Ferguson, R.W., Xu, Z., Green, A., & Rosenthal, K.M 2013, Teens and Distraction: An In-Depth Look at Teens Walking Behaviors. Washington, DC: Safe Kids Worldwide. Fontaine, H., & Gourlet, Y 1997, Fatal pedestrian accidents in France: a typological analysis. Accident Analysis and Prevention, 29(3), Gannon, B., Rosta, L., Reeve, M., Hyde, M.K., & Lewis, I.M 2014, Does it matter whether friends, parents, or peers drink walk? Identifying which normative influences predict young pedestrian's decisions to walk while intoxicated. Transportation Research Part F: Traffic Psychology and Behaviour, 22, Gauld, C.S., Lewis, I., & White, K.M 2014¹, Concealing texting while driving: what are young people s beliefs about this risky behaviour? Safety Science, 65, Gauld, C.S., Lewis, I., & White, K.M 2014², Concealing their communication: exploring psychosocial predictors of young drivers intentions and engagement in concealing texting. Accident Analysis and Prevention, 62, Gauld, C., Lewis, I., & White, K.M. 2013, Identifying the determinants of concealed and obvious texting while driving: are they distinct behaviours? In: 2013 Australasian College of Road Safety National Conference, 6-8 th November 2013, National Wine Centre of Australia, Adelaide, SA. Gawade, M., Kourtellis, A., & Lin, P-S 2014, Multivariate analysis on factors influencing pedestrian and bicycle safety. In: 93 rd Annual Meeting of the Transportation Research Board, th January 2014, Washington DC. Giles, M., McClenahan, C., Cairns, E., & Mallet, J 2004, An application of the theory of planned behaviour to blood donation: the importance of self-efficacy. Health Education Research, 19(4), Gitelman, V., Balasha, D., Carmel, R., Hendel, L., & Pesahov, F 2012, Characterization of pedestrian accidents and an examination of infrastructure measures to improve pedestrian safety in Israel. Accident Analysis and Prevention, 44, GHD 2012, Report for Wellington CBD Golden Mile, Pedestrian Crash Reduction Study, May GHD Limited, Tauranga, New Zealand. University of New South Wales, Sydney. Godin, G., Valois, P., Lepage, L., & Desharnais, R 1992, Predictors of smoking behaviour: an application of Ajzen s theory of planned behaviour. British Journal of Addiction, 87(9), Austroads 2016 page 77

88 Gorrie, C.A., & Waite, P.M.E 2005, A survey of older pedestrians in metropolitan Sydney: walking patterns, perceptions and risk exposure. Report prepared for the Motor Accident Authority NSW, November Granie, M 2009, Effects of gender, sex-stereotype conformity, age and internalisation on risk-taking among adolescent pedestrians. Safety Science, 47, Granie, M-A., Pannetier, M & Gueho, L 2013, Developing a self-reporting method to measure pedestrian behaviors at all ages. Accident Analysis and Prevention, 50(1), Granie, M., Brenac, T., Montel, M., Millot, M., & Coquelet, C 2014, Influence of built environment on pedestrian s crossing decision. Accident Analysis and Prevention, 67, Greater London Authority 2014, Feet First: Improving Pedestrian Safety in London, March London Assembly Transport Committee. Hanson, C.S., Noland, R.B., & Brown, C 2014, The severity of pedestrian crashes: an analysis using Google Street View imagery. In: 93 rd Annual Meeting of the Transportation Research Board, th January 2014, Washington DC. Harre, N., & Wrapson, W 2004, The evaluation of a central city pedestrian safety campaign. Transportation Research Part F, 7, Havard, C., & Willis, A 2012, Effects of installing a marked crossing on road crossing behaviour and perceptions of the environment. Transportation Research Part F, 15(3), Harruff, R.C., Avery, A., & Alter-Pandya, A.S 1998, Analysis of circumstances and injuries in 217 pedestrian traffic fatalities. Accident Analysis and Prevention, 30(1), Hatfield, J., & Murphy, S 2007, The effects of mobile phone use on pedestrian crossing behaviour at signalised and unsignalised intersections. Accident Analysis and Prevention, 39, Hatfield, J., & Job, S 2004, Perceptions of priority rules at pedestrian crossings by pedestrians and drivers. University of Sydney, New South Wales. Haque, R., Clapoudis, N., King, M., Lewis, I.M., Hyde, M.K., & Obst, P.L 2012, Walking when intoxicated: an investigation of the factors which influence individuals' drink walking intentions. Safety Science, 50(3), Haque, M.M., Chin, H.C., & Debnath, A.K 2013, Sustainable, safe, smart three key elements of Singapore s evolving transport policies. Transport Policy, 27, Haworth, Corben & Vulcan Holubowycz, O.T 1995, Age, sex, and blood alcohol concentration of killed and injured pedestrians. Accident Analysis and Prevention, 27(3), Holland, C.A., Hill, R., & Cooke, R 2009, Understanding the role of self-identity in habitual risky behaviour: pedestrian road-crossing decisions across the lifespan. Health Education Research, 24(4), Holland, C., & Hill, R 2007, The effect of age, gender and driver status on pedestrians intentions to cross the road in risky situations. Accident Analysis and Prevention, 39, Holland, C., & Hill, R 2010, Gender differences in factors predicting unsafe crossing decisions in adult pedestrians across the lifespan: a simulations study. Accident Analysis and Prevention, 42, Horvath, C., Lewis, I. M., & Watson, B. C 2012, Peer passenger identity and passenger pressure on young drivers speeding intentions. Transportation Research Part F: Traffic Psychology and Behaviour, 15(1), Austroads 2016 page 78

89 Hoque, M.M., Hossain, S., Islam, S., & Rahman, M.A 2013, Safe system for sustainable pedestrian safety in Bangladesh. In: 2013 Australasian Road Safety Research, Policing & Education Conference, th August, Brisbane, Queensland. Hutchinson, T.P., Kloeden, C.N., & Lindsay, V.L 2009, Accidents to intoxicated pedestrians in South Australia. Centre for Automotive Safety Research, Report number CASR061, Adelaide, South Australia. Hyman, I.E., Boss, S.M., Wise, B.M., McKenzie, K.E., & Caggiano, J.M 2010, Did you see the unicycling clown? Inattentional blindness while walking and talking on a cell phone. Applied Cognitive Psychology, 24, Ibrahim, J.M., Day, H., Hirshon, J.M., & El-Setouhy, M 2012, Road risk-perception and pedestrian injuries among students at Ain Shams University, Cairo, Egypt. Journal of Injury and Violence Research, 4(2), Islam, M.S., Serhiyenko, V., Ivan, J.N., Ravishanker, N., & Garder, P.E 2014, Explaining pedestrian safety experience at urban and suburban street crossings considering observed conflicts and pedestrian counts. Journal of Transportation Safety and Security, 6, Ismail, K., Sayed, T., & Saunier, N 2009, Automated analysis of pedestrian-vehicle conflicts. Transportation Research Record: Journal of the Transportation Research Board 2198, Ismail, K., Sayed, T., Saunier, N., & Lim, C 2009, Automated analysis of pedestrian-vehicle conflicts using video data. In Transportation Research Record: Journal of the Transportation Research Board, No. 2140, Transportation Research Board of the National Academies, Washington, D.C., pp Jenner, E.A., Watson, P.W.B., Miller, L., Jones, F & Scott, G.M 2002, Explaining hand hygiene practice: an extended application of the theory of planned behaviour. Psychology, Health & Medicine, 7(3), Kim, K., Brunner, M., & Yamashita, E 2008, Modeling fault among accident-involved pedestrians and motorists in Hawaii. Accident Analysis and Prevention, 40, King, M.J., Wood, J.M., Lacherez, P.F., & Marszalek, R.P 2012, Optimism about safety and group-serving interpretations of safety among pedestrians and cyclists in relation to road use in general and under low light conditions. Accident Analysis and Prevention, 44, King, M 2003, Pedestrian safety through a raised median and redesigned intersections. Transportation Research Board Annual Meeting, Washington DC. King, M. J., Soole, D.W., & Ghafourian, A 2009, Illegal pedestrian crossing at signalised intersections: Incidence and relative risk. Accident Analysis and Prevention, 41(3), King, M. J 2004, Older road user safety: Policy overview and development opportunities. In: Road Safety Issues for Older Road Users. Centre for Accident Research & Road Safety Queensland (CARRS-Q), QUT, Carseldine, Queensland, pp King, M 2010, Understanding young peoples intention for drink walking: applying an extended theory of planned behaviour. Thesis submitted for the Postgraduate Diploma in Psychology: QUT. King, M.J., Wood, J.M., Lacherez, P.F., & Marszalek, R.P 2012, Optimism about safety and group-serving interpretations of safety among pedestrians and cyclists in relation to road use in general and under low light conditions. Accident Analysis & Prevention, 44(1), Kothuri, S., Clifton, K., & Monsere, C 2014, Insights into pedestrian attitudes and perceptions of delay, safety and crossing decisions at signalised intersections. In: 93 rd Annual Meeting of the Transportation Research Board, th January 2014, Washington DC. Austroads 2016 page 79

90 Lang, CP, Tay, RS, Watson, BC, Edmonston, CJ, & O'Connor, EL 2003, Drink walking: An examination of the related behaviour and attitudes of young people. In: Australasian Road Safety Research, Policing and Education Conference - From Research to Action. Sydney, New South Wales. Lascala, E.A., Gerber, D., & Gruenewald, P.J 2000, Demographic and environmental correlates of pedestrian injury collisions: a spatial analysis. Accident Analysis and Prevention, 32(5), Lenne, M, G., Corben, B.F., & Stephan, K 2007, Traffic signal phasing at intersections to improve safety for alcohol-affected pedestrians. Accident Analysis and Prevention, 39, Lewis, I.M., Watson, B.C., & White, K 2013, Psychology and Marketing, 30(9), Levasseur, M., & Brisbane, G 2011, Trial of pedestrian countdown timers in the Sydney CBD Final Report. Report prepared by ARRB for the Roads and Traffic Authority of NSW. Lichenstein, R., Smith, D.C., Ambrose, J.L., & Moody, L.A 2012, Headphone use and pedestrian injury and death in the United States: Injury Prevention, doi: /injuryprev Li, Y., & Fernie, G 2010, Pedestrian behaviour and safety on a two-stage crossing with a center refuge island and the effect of winter weather on pedestrian compliance rate. Accident Analysis and Prevention, 42, Lobjois,R., & Cavallo, V 2007, Age-related differences in street-crossing decisions: the effects of vehicle speed and time constraints on gap selection in an estimation task. Accident Analysis and Prevention, 39, Lobjois, R., & Cavallo, V 2009, The effects of aging on street-crossing behaviour: from estimation to actual crossing. Accident Analysis and Prevention, 41, Luoma, J., & Peltola, H 2013, Does facing traffic improve pedestrian safety? Accident Analysis and Prevention, 50, Mabunda, M.M., Swart, L-A., & Seedat, M 2008, Magnitude and categories of pedestrian fatalities in South Africa. Accident Analysis and Prevention, 40, Madden, M., Lenhart, A., Duggan, M., Cortesi, S., & Gasser, U 2013, Teens and Technology Pew Research Center. Available at Accessed September 04, 2014 Malkhamah, S., Tight, M. & Montgomery, F. 2005, The development of an automatic method of safety monitoring at Pelican crossings. Accident Analysis and Prevention, 37, Masuda, K., Sekine, Y., Sato, H., & Haga, S 2014, Laboratory experiment on visual and auditory inattention of pedestrians using cell phones. Presented at the 28 th International Congress of Applied Psychology, 8-13 th July 2014, Paris, France. McGhie, A., Lewis, I, M., & Hyde, M, K 2012, The influence of conformity and group identity on drink walking intentions: comparing intentions to drink walk across risky pedestrian crossing scenarios. Accident Analysis and Prevention, 45, McGhie, A.C 2010, The influence of conformity and group identity on intentions to drink walk: an application of the theory of planned behaviour. Thesis submitted for the Bachelor of Behavioural Science (Honours): QUT. McTiernan, D., Levasseur, M., & Brisbane, G 2012, Pedestrian countdown timers two Australian trial methods and results. Presented at the 25 th ARRB Conference, Perth, Australia. Ministry of Transport 2013, Pedestrians: Crash statistics for the year ended 31 December Crash factsheet prepared by Ministry of Transport, New Zealand. Austroads 2016 page 80

91 Miranda-Moreno, L.F., Morency, P., & El-Geneidy, A.M 2011, The link between built environment, pedestrian activity and pedestrian-vehicle collision occurrence at signalised intersections. Accident Analysis and Prevention, 43, Moan, I.S., & Rise, J 2011, Predicting intentions not to drink and drive using an extended version of the theory of planned behaviour. Accident Analysis and Prevention, 43, Nasar, J., Hecht, P., & Wener, R 2008, Mobile telephones, distracted attention, and pedestrian safety. Accident Analysis and Prevention, 40, Nasar, J.L., & Troyer, D 2013, Pedestrian injuries due to mobile phone use in public places. Accident Analysis and Prevention, 57, National Highway Traffic Safety Administration 2011, Countermeasures that work: A highway safety countermeasures guide for state highway safety offices sixth edition, US Department of Transportation, Washington DC. Neider, M.B., McCarley, J.S., Crowell, J.A., Kaczmarski, J., & Kramer, A.F 2010, Pedestrians, vehicles, and cell phones. Accident Analysis and Prevention, 42, Neider, M.B., Gaspar, J.G., McCarley, J.S., Crowell, J.A., Kaczmarski, H., & Kramer, A.F 2011, Walking and talking: dual-task effects on street crossing behaviour in older adults. Psychology and Ageing, 26(2), Nieuwesteeg, M., & McIntyre, A 2010, Exploring the pedestrian crash problem from the perspective of injured pedestrians. In: Australasian Road Safety Research, Policing and Education Conference, 31 August-3 September 2010, Canberra, Australian Capital Territory. Nordfjaern, T., & Simsekoglu, O 2013, The role of cultural factors and attitudes for pedestrian behaviour in an urban Turkish sample. Transportation Research Part F, 21, O Connor, E.L., Tay, R. S-T., Watson, B.C., Edmonston, C.J., & Lang, C.P 2004, Drink-walking: an examination of the related behaviour and attitudes of young people. In: The NSW Local Government Road Safety Conference. Ossenbruggen, P. J., Pendharkar, J., Ivan, J. 2001, Roadway safety in rural and small urbanized areas. Accident Analysis & Prevention, 33(4), Oxley, J., Fildes, B., Ihsen, E., Charlton, J., & Day, R. 1997, Differences in traffic judgements between young and old adult pedestrians. Accident Analysis and Prevention, 29(6), Oxley, J., Lenne, M., & Corben, B 2006, The effect of alcohol impairment on road-crossing behaviour. Transportation Research Part F, 9, Oxley, J., Yuen, J., Corben, B., Hoareau, E., & Logan, D 2013, Reducing pedestrian collisions in Melbourne s Central Business District. In: 2013 Australasian Road Safety Research, Policing & Education Conference, th August, Brisbane, Queensland. Palamara, P., & Broughton, M 2013, An investigation of pedestrian crashes at traffic intersections in the Perth Central Business District Report prepared by Curtin-Monash Accident Research Centre, Perth, Australia. Papadimitriou, E., Theofilatos, A., & Yannis, G 2013, Patterns of pedestrian attitudes, perceptions and behaviour in Europe. Safety Science, 53, Prato, C.G., Gitelman, V., & Bekhor, S 2012, Mapping patterns of pedestrian fatal accidents in Israel. Accident Analysis and Prevention, 44, Austroads 2016 page 81

92 Quistberg, A., Koepsell, T.D., Boyle, L.N., Miranda, J.J., Johnston, B.D., & Ebel, B.E 2014, Pedestrian signalisation and the risk of pedestrian-motor vehicle collisions in Lima, Peru. Accident Analysis and Prevention, 70, Road Safety Great Britain 2015, Report examines things that go bump in the night. Accessed 26 February Queensland Government 2014, Safe Night Out Strategy. June Queensland Government. Rosenbloom, T., Nemrodov, D., & Barkan, H 2004, For heaven s sake follow the rules: pedestrians behaviour in an ultra-orthodox and a non-orthodox city. Transportation Research Part F, 7, Rosenbloom, T., Shahar, A., & Perlman, A 2008, Compliance of ultra-orthodox and secular pedestrians with traffic lights in ultra-orthodox and secular locations. Accident Analysis and Prevention, 40, Rosenbloom, T 2009, Crossing at a red light: behaviour of individuals and groups. Transportation Research Part F, 12, Rosenstock, I.M 1966, Why people use health services. Milbank Memorial Fund Quarterly, 44, 94. Rosta, L 2011, Friendship group normative effects on intentions to drink walk: an extension of the theory of planned behaviour. Thesis: QUT. Safe Kids Worldwide 2013, Teens and Distraction: An in-depth look at teens walking behaviour. Safe Kids Worldwide: Washington DC. Shaw, L., Lennon, A.J., & King, M.J 2012, A qualitative investigation of older pedestrian views of influences on their road crossing safety. In: 2012 Australasian Road Safety Research, Policing and Education Conference, Wellington, New Zealand. Schneider, R.J., Ryznar, R.M., & Khattak, A.J 2004, An accident waiting to happen: a spatial approach to proactive pedestrian planning. Accident Analysis and Prevention, 36, Schwebel, D.C., Stavrinos, D., Byington, K.W., Davis, T., O Neal, E.E., & de Jong, D 2012, Distraction and pedestrian safety: how talking on the phone, texting, and listening to music impact crossing the street. Accident Analysis and Prevention, 45, Schwebel, D.C., Stavrinos, D., & Kongable, E.M 2009, Attentional control, high intensity pleasure, and risky pedestrian behaviour in college students. Accident Analysis and Prevention, 41, Schwebel, D.C., Pitts, D.D., & Stavrinos, D 2009, The influence of carrying a backpack on college student pedestrian safety. Accident Analysis and Prevention, 41, Schabrun, S, M., Van den Hoorn, W., Moorcraft, A., Greenland, C., & Hodges, P.W 2014, Texting and walking: strategies for postural control and implications for safety. PLoS ONE 9(1): e doi: /journal.pone Sisiopiku, V.P., & Akin, D 2003, Pedestrian behaviours at and perceptions towards various pedestrian facilities: an examination based on observational and survey data. Transportation Research Part F, 6, Stavrinos, D., Byington, K.W., & Schwebel, D.C 2009, Effect of cell phone distraction on pediatric pedestrian injury risk. Pediatrics, 123, e179. DOI: /peds Stavrinos, S., Byington, K.W., & Schwebel, D.C 2011, Distracted walking: cell phones increase injury risk for college pedestrians. Journal of Safety Research, 42, Austroads 2016 page 82

93 Stimpson, J.P., Wilson, F.A., & Muelleman, R.L 2013, Fatalities of pedestrians, bicycle riders, and motorists due to distracted driving motor vehicle crashes in the U.S., Public Health Reports, 128, Sullivan, K.A., White, K.M., Young, R.McD., Chang, A., Roos, C., & Clinton, S 2008, Predictors of intention to reduce stroke risk among people at risk of stroke: an application of an extended health belief model. Rehabilitation Psychology, 53(4), SWOV 2013, Factsheet: Use of media devices by cyclists and pedestrians. Available at Accessed 04 September 2014 TAC 2014, Accessed 23 September The Dancing Man 2014, Accessed 7 October The State of Queensland 2015, Join the Drive. Accessed 26 February 2015 Thompson, L.T., Rivara, F.P., Ayyagari, R.C., & Ebel, B.E 2013, Impact of social and technological distraction on pedestrian crossing behaviour: an observational study. Injury Prevention, 19, doi: /injuryprev Thomas, J., Rive, G., Frith, B., & Charlton, S 2011, Pedestrian-vehicle accident review: Human factors behavioural assessment. Report prepared by Opus International Consultants Limited, Lower Hutt, New Zealand. Tom, A., & Granie, M 2011, Gender differences in pedestrian rule compliance and visual search at signalised and unsignalised crossroads. Accident Analysis and Prevention, 43, Tyrell, R.A., Wood, J.M., & Carberry, T.P 2004, On-road measures of pedestrians estimates of their own night time conspicuity. Journal of Safety Research, 35, Tulu, G.S., Washington, S., King, M.J., & Haque, M.M 2013, Why are pedestrian crashes so different in developing countries? A review of relevant factors in relation to their impact in Ethiopia. In: 36th Australasian Transport Research Forum (ATRF): Transport and the New World City, 2-4 October 2013, QUT Gardens Point, Brisbane, Australia. Villaveces, A., Nieto, L.A., Ortega, D., Rios, J.F., Medina, J.J., Gutierrez, M.I., & Rodriguez, D. 2012, Pedestrians perceptions of walkability and safety in relation to the built environment in Cali, Columbia, Injury Prevention, 18, Volvo 2015, Accessed 26 February Wall, R., & Bauer, D 2012, Improving pedestrian safety at signalised intersections. Report prepared by the National Institute for Advanced Transportation Technology, University of Idaho. Wallace, L.S 2002, Osteoporosis prevention in college women: application of the expanded health belief model. American Journal of Health Behaviour, 26(3), Walsh, S., White, K.M., & Young, R 2010, Needing to connect: the effect of self and others on young people s involvement with their mobile phones. Australian Journal of Psychology, 62(4), Walker, E.J., Lanthier, S.N., Risko, E.F., & Kingstone, A 2012, The effects of personal music devices on pedestrian behaviour. Safety Science, 50, Austroads 2016 page 83

94 Wang, Y.Y., Haque, M.M., Chin, H.C., & Yun, J.G.J 2013, Injury Severity of Pedestrian Crashes in Singapore. In: 36th Australasian Transport Research Forum (ATRF) : Transport and the New World City, 2-4 October 2013, QUT Gardens Point, Brisbane, Australia. Wang, Y.Y., Haque, M.M., Chin, H.C 2014, A random parameters ordered Probit analysis of elderly pedestrian injuries in Singapore, Accident Analysis & Prevention, Under review. World Health Organisation 2013, Pedestrian Safety: A road safety manual for decision-makers and practitioners. WHO, Geneva, Switzerland. Xiaoli, X., Pingsheng, L., Shuyun, W., & Hongqi, L 2014, Pedestrian fatal crashes at rural intersections in China: characteristics, patterns in contributing factors and potential countermeasures. In: 93 rd Annual Meeting of the Transportation Research Board, th January 2014, Washington DC. Yagil, D 2000, Beliefs, motives and situational factors related to pedestrians self-reported behaviour at signal-controlled crossings. Transportation Research Part F, 3, Yu, C 2015, Built environmental designs in promoting pedestrian safety. Sustainability, 7(7), Zegeer, C.V., Stewart, J.R., Huang, H.H., & Lagerway, P.A 2002, Safety effects of marked vs unmarked crossings at uncontrolled locations: executive summary and recommended guidelines. Report prepared by the US Department of Transportation, Federal Highway Administration. Zhou, Z., Wang, W., Hu, Q., & Li, H 2014, An experimental study of pedestrian behaviour and safety perception. In: 93 rd Annual Meeting of the Transportation Research Board, th January 2014, Washington DC, McLean, VA. Zhou, R., Horrey, W.J., & Yu, R 2009, The effect of conformity tendency on pedestrians road-crossing intentions in China: an application of the theory of planned behaviour. Accident Analysis and Prevention, 41, Zhou, R., & Horrey, W.J 2010, Predicting adolescent pedestrians behavioural intentions to follow the masses in risky crossing situations. Transportation Research Part F, 13, Zhuang, X., & Wu, C 2011, Pedestrians crossing behaviours and safety at unmarked roadway in China. Accident Analysis and Prevention, 43, Zivotofsky, A.Z., Eldror, E., Mandel, R., & Rosenbloom, T 2012, Misjudging their own steps: why elderly people have trouble crossing the road. Human Factors, 54(4), Austroads 2016 page 84

95 Appendix A Review of Research Relating to Factors Influencing Pedestrian Behaviour and Safe Road Use A.1 Factors that influence pedestrian behaviour and safe road use Pedestrian behaviour is influenced by a wide range of factors, including demographic factors such as gender (Gannon, Rosta, Reeve, Hyde & Lewis, 2014; Rosenbloom, Nemrodov & Barkan, 2004; Holland & Hill, 2010; Tom & Granie, 2011) and age (Oxley, Fildes, Ihsen, Charlton & Day, 1997; Dommes, Granie, Cloutier, Coquelet & Huguenin-Richard, 2014b; Cavallo & Dommes, 2014; Dommes, Cavallo, Dubuisson, Tournier & Vienne, 2014a; Dunbar, 2012; Lobjois & Cavallo, 2007). A range of other factors have also been examined, these include personality and attitudes (Schwebel, Stavrinos & Kongable, 2009), self-identity (Holland, Hill & Cooke, 2009), conformity and group norms (McGhie, Lewis & Hyde, 2012), socioeconomic factors (Zhuang & Wu, 2011; Tulu, Washington, King & Haque, 2013), cultural factors (Nordfjaern & Simsekoglu, 2013; Rosenbloom, Shahar & Perlman, 2008; Rosenbloom, 2009) and other factors such as trip length and purpose (Kothuri, Clifton & Monsere, 2014). Impairment from alcohol (Gannon, Rosta, Reeve, Hyde & Lewis, 2014; Lang, Tay, Watson, Edmonston & O Connor, 2003; McGhie, Lewis & Hyde, 2012) and distraction from the use of technology (Nasar, Hecht & Wener, 2008; Hatfield & Murphy, 2007; Stavrinos, Byington & Schwebel, 2009; Neider, McCarley, Crowell, Kaczmarski & Kramer, 2010) are also discussed. Age Older pedestrians are overrepresented in crash statistics and have been shown to adopt less safe crossing strategies than younger pedestrians (Oxley, Fildes, Ihsen, Charlton & Day, 1997; Dommes, Cavallo, Dubuisson, Tournier & Vienne, 2014a; Dommes, Granie, Cloutier, Coquelet & Huguenin-Richard, 2014b; Cavallo & Dommes, 2014; Dommes, Cavallo, Vienne & Aillerie, 2012; Cavallo & Dommes, 2014; Dommes, Cavallo & Oxley, 2013; Dunbar, 2012; Shaw, Lennon & King, 2012; Zivotofsky, Eldror, Mandel & Rosenbloom, 2012; Lobjois & Cavallo, 2007; 2009; Garder, 2004). In a recent French study, Dommes et al (2014b) observed 682 pedestrians from two age groups in the city of Lille in France, on marked crossings, with and without lights. Results showed that older pedestrians were more cautious in terms of compliance with traffic and pedestrian lights and waiting positions on the footpath. Interestingly, when looking at behaviour before and during the crossings, older pedestrians more often focused on the ground and less on vehicles when compared to the younger pedestrians. Oxley et al (1997) also investigated differences in traffic judgements between young and older adult pedestrians through an observational study at two strip shopping centres on arterial roads in Melbourne. Pedestrians were filmed as they crossed the road mid-block without formal crossings. On two-way undivided roads, older pedestrians crossed more frequently when there was closer moving traffic and generally adopted less safe road crossing strategies than younger pedestrians (delay in making their first step forward from the kerb, crossed more slowly and with greater variability, and interacted with moving traffic in a risker way). On one-way divided roads, older pedestrians behaved considerably more safely and in a manner similar to that of the younger pedestrians. Dommes et al (2014a) demonstrated this effect of age in a simulated environment that assessed street crossing behaviour. The study involved 18 young (aged 19-35), 28 younger-old (aged 62-71) and 38 olderold (aged 72-85) participants. A street-crossing simulator was used in which participants are able to perform a real walk across a 5.7m wide two-way street with vehicle traffic (for which speed and available time gaps were varied). Results showed that the older-old pedestrians made street-crossing decisions that led to collisions more often than younger-old or young participants. Younger-old and older-old participants were also found to cross more slowly and use smaller safety margins than young participants. These difficulties seen in the older participant groups mostly occurred in the far lane of the two-way street, in line with previous research. Dommes, Cavallo and Oxley (2013) found similar results using a simulated two-way road environment to assess street crossing behaviour of young (aged 20-35), younger-old (aged 60-67) and older- Austroads 2016 page 85

96 old (aged 70-84) pedestrians. Older-old participants were more likely than young pedestrians to make less safe street-crossing decisions; however, interestingly, this difference was not found between the younger-old and young groups. However, the researchers argued that this finding was partly explained by French crash statistics in terms of age proportions and their alignment with the participant groups in the study. In contrast to Dommes et al (2014a) who found that older pedestrians had most difficulty in the far side of two-way roads, this study showed that many of the oldest group s virtual collisions actually occurred from the nearside lane. Also using a simulated real-walk street-crossing task, Dommes et al (2012a) assessed the effectiveness of a training program for older pedestrians and the extent to which age-related differences in crossing safety could be reduced through such training. Twenty elderly participants (aged 65-83) enrolled in a simulatorbased training program that used behavioural and educational interventions. Twenty younger participants (aged 20-30) were also recruited but did not receive any training. The older participants performed a streetcrossing task three times (one week before the training, one week after and six months later) while the younger participants performed the task once to serve as a baseline condition. The task involved vehicles passing at varying speeds and time gap opportunities for crossing. Results were promising with the older participants showing improvements in crossing behaviour after the training (higher safety margins, more safe decisions, fewer tight fits and fewer unsafe decisions). Effects were greater immediately after training and still present 6 months later. When comparing their performance to that of the younger participants, significant age-related differences were seen prior to training (with older participants adopting smaller safety margins, made fewer safe decisions and more unsafe decisions). Six months after the training there were no significant differences between the two groups; however, issues around risk exposure and safety margins make this finding somewhat inconclusive. Cavallo and Dommes (2014) conducted a further study in which they trained older pedestrians to make safe decisions in simulated street-crossing situations. In a first experiment that focused on one-way traffic conditions, an educational and behavioural intervention program was assessed using the street-crossing simulator. Older pedestrians showed improvements in overall street-crossing safety; however, their ability to take into account oncoming vehicle speed did not improve. The second experiment tested the effectiveness of three intervention programs in two-way traffic situations: cognitive (UFOV-training), behavioural (simulatortraining) and educational (class-room training). No improvements were found from any of these programs. Zivotofsky, Eldror, Mandel and Rosenbloom (2012) used a naturalistic method to compare older pedestrians self-estimations about the length of time it took them to cross the road with actual crossing times. Nineteen older pedestrians (12 women with a mean age of 75 years and seven men with a mean age of 79 years) and 17 undergraduate students participated in the study. Participants were taken to a seven metre wide road (free of obstacles and traffic) and asked to estimate how long it would take them to cross it using a motor imagery process. Participants were then asked to cross the street, and then estimate the time again once they had completed the crossing. Results showed that younger and older pedestrians made similar estimations of how long they would take to cross. However, for the older pedestrians, actual crossing time was significantly longer than both the pre-crossing estimation and the post-crossing estimation (and the pre and post estimations did not differ significantly from each other). In contrast, there were no significant differences in estimated versus actual crossing times for the younger participants. In addition, older participants took significantly longer to cross the road compared to the younger participants. These results confirm that older pedestrians misjudge how long it takes for them to cross the road, in the absence of other traffic or road users. Using a street-crossing simulator, Jobjois and Cavallo (2007) used an estimation task to assess age-related differences in street crossing behaviour in relation to the effects of vehicle speed and time constraints. Seventy-eight participants in three age groups (20-30, 60-70, and 70-80) took part in a simulated crossing task that involved traffic consisting of three vehicles (a motorcycle followed by two cars). Participants crossed the road in two counterbalanced conditions: at their preferred pace and at their fastest possible pace (i.e. under a time constraint), with varying vehicle speeds and inter-vehicle distance. Results showed that older pedestrians selected larger time gaps than younger pedestrians. This compensatory behaviour led to similar safety margins and unsafe-decision rates across the three age groups. In discussing the results the researchers acknowledged that in a more complex situation, such as two-way roads, it was unlikely that this would be the case. Exposure was highlighted as a further issue for consideration. The impact of vehicle speed was found to depend both on age and the time constraint imposed. Under a time constraint, all participants took more risks as speed increased. When there was no time constraint, speed influenced the two older groups only. In a similar experiment, Lobjois and Cavallo (2009) also found that older participants Austroads 2016 page 86

97 selected a greater mean time gap and initiated their crossing sooner than young participants. However, older participants accepted shorter time gaps as speed increased. Qualitative studies have attempted to understand the perspectives of older pedestrians. Gorrie and Waite (2005) conducted 740 interviews with older pedestrians in metropolitan Sydney to investigate walking patterns, perceptions and risk exposure. Half of those interviewed reported difficulty crossing the road, and this was higher for women and those with mobility or medical problems. Traffic lights and zebra crossings were regarded as safe places to cross. Respondents were less confident about crossing at pedestrian refuges. Factors reported as of most concern to older pedestrians were turning cars, cars not stopping, condition of footpath, visibility, speed and volume of traffic, and timing of walk signals at traffic lights. In a more recent qualitative investigation of older pedestrian views of influences on their road crossing safety, Shaw, Lennon and King (2012) identified four key themes in relation to concerns about crossing in a series of semi-structured interviews with 18 pedestrians over the age of 55 years. In terms of physical characteristics of crossings and roads, respondents expressed difficulties negotiating sloped kerbs using designated crossings. The issue of confidence was also raised, particularly in terms of health and physical problems (i.e. fear of falling). Respondents perceived difficulties in crossing to be partially the result of the behaviour of other road users, particularly drivers and cyclists. Informal rules and strategies for safe crossing were discussed including mitigation strategies such as seeking a stretch of road allowing clear vision both ways and trying to make eye contact with drivers to ensure that they ve been seen. Finally, crash data patterns have also been explored. The relative risk of nearside pedestrian accidents, or those occurring in the first half of a road crossing, was examined by Dunbar (2012). Using police road accident data from Great Britain for , it was shown that risk of being killed or seriously injured is highest for the youngest pedestrians (between 10 and 14 years) and declines until the age of 85 years (with an increase seen for those over the age of 85). The relative risk was the same at junctions as when crossing at non-junctions, but lower when the road is wet and when the speed limit is higher. Gender Differences between men and women in terms of pedestrian crossing behaviour have been widely reported, with men generally displaying less safe behaviours than women (Gannon, Rosta, Reeve, Hyde & Lewis, 2014; Rosenbloom, Nemrodov & Barkan, 2004; Rosenbloom, Shahar & Perlman, 2008; Rosenbloom, 2009; Holland & Hill, 2010; Holland & Hill, 2007; Diaz, 2002; Tom & Granie, 2011; Garder, 2004). For instance, Rosenbloom (2009) conducted an observational study of 1392 pedestrians at a street crossing in Israel. Observations were taken in an urban setting at a pedestrian crossing with traffic lights. Results showed that males crossed on red more frequently than females, and the tendency to cross on red was greater when there were fewer people waiting at the curb. Thirteen percent of the pedestrians who arrived at the crossing in the red-light phase crossed the street on a red light. Traffic volume and the red-light crossing of another pedestrian did not significantly predict pedestrian behaviour. Holland and Hill (2010) further explored gender differences in the prediction of unsafe crossing decisions in adult pedestrians of older ages. In a simulation study using filmed real traffic, 218 participants indicated when they would cross the road in a series of gap selection scenarios. Head movement was filmed and a walking time estimate was produced for each participant. Results showed some interesting differences between men and women. As age increased, women were shown to make more unsafe crossing decisions, to leave smaller safety margins and to become poorer at estimating their walking speed. However, driving experience had a positive effect on these agerelated findings. For men, age was not a major factor in predicting unsafe crossing decisions, but reduced mobility was the key factor for men s unsafe crossing decisions. This led them to make more unsafe crossings and delay longer in leaving the curb. The effects of gender and gender-stereotype among adolescents were explored in a self-report study of adolescent pedestrians aged years (n = 278). Respondents (130 boys and 148 girls) were asked to complete a questionnaire that included items about risk and danger perception, gender identity and internalisation of traffic rules in relation to social knowledge. Results showed that masculine sex-stereotype conformity led to a weaker internalisation of traffic rules and that masculinity and internalisation were both good predictors of the risky behaviours reported by participants. Self-reported risk taking was greater in the year olds than the years olds and danger perception was found to drop significantly between the 7 th grade and the 10 th grade. Furthermore, masculine stereotype conformity was a better predictor of risk taking than biological sex, confirming other research; that is, being male or female does not predict the self- Austroads 2016 page 87

98 reported level of risk taking; rather, recognising oneself as masculine, predicts risky pedestrian behaviour (Granie, 2009). The effect of gender on pedestrian rule compliance and visual search at signalised and unsignalised cross roads was examined in an observational study. Four hundred pedestrians were observed on four selected sites in France (three in Paris and one in Rouen). The sites had zebra crossings on each side of the crossing and two had traffic lights while the other two did not. Gender differences in gaze patterns were seen before and during crossing. Men focused on moving vehicles more than women, whereas women turned their heads towards traffic lights and other pedestrians more than men. Females gazes also varied according to the type of crossing, but males gazes did not. Compliance with traffic light rules (temporal crossing compliance) was lower among male pedestrians but compliance with marked crossings (spatial crossing compliance) did not differ between genders (Tom & Granie, 2011). Psychosocial and personality factors The influence of psychosocial factors such as conformity and group identity and personality factors such as sensation seeking are critical to our understanding of human behaviour, including that of safe pedestrian behaviour. These constructs are important for understanding the influences contributing to pedestrian decision making about, for example, drink walking, walking while using a mobile phone and crossing against a red man. The Theory of Planned Behaviour (TPB) is a well validated model that is used to predict people s intentions across various behaviours. It uses the constructs of attitude, subjective norm and perceived behavioural control to best predict a particular behaviour (Ajzen, 1991). Several studies have used these TPB constructs, and extended versions of it, to explain pedestrian behaviour (Zhou, Horrey & Yu, 2009; Zhou & Horrey, 2010; Diaz, 2002; Gannon et al 2014). Others have used social identity theory (McGhie et al 2012) and the health belief model (Yagil, 2000). Recent research into the underlying beliefs of young drivers intentions towards, and engagement in, concealed texting used an extended TPB and provided insight into young people s beliefs about engaging in such behaviour (Gauld, Lewis & White, 2014¹; Gauld, Lewis & White, 2014²; Gauld, Lewis & White, 2013). Diaz (2002) explored pedestrian attitudes towards traffic violations and self-ratings of violations, errors and lapses. Based on TPB, a 16-item questionnaire about pedestrian and traffic behaviour was used to uncover respondent attitudes, subjective norms, and perceived behavioural control towards aberrant behaviour. A total of 146 participants completed the questionnaire. Reported violations, errors and lapses appeared causally related to the intention to violate regulations. Young people had a more positive attitude towards committing violations as pedestrians than adults, perceived subjective norms to be less inhibitory, had less control over violations, had a more positive intention to commit violations, and reported more violations, errors and lapses than the adults in the study. Men reported more frequent traffic rule violations than women, but they didn t differ in terms of errors or lapses. Holland and Hill (2007) used the components of the TPB as predictors of intention to cross in a selection of situations. Participants (N = 293) were divided into four different groups of adults (men, women, drivers and non-drivers) and different age groups. Intention to cross was found to decrease with increasing age and women perceived a greater risk of being run over than men. Zhou, Horrey and Yu (2009) used the Theory of Planned Behaviour in an investigation of the effect of conformity on pedestrian road-crossing intentions in China. A total of 426 respondents completed a selfreport survey that included demographic information, items to measure social conformity, and a questionnaire based on TPB that measured pedestrian intention to cross the road in two different situations. Results showed that pedestrians reported a greater likelihood of crossing the road when other pedestrians were also crossing. In their predictive model, attitude, subjective norm, perceived behavioural control and perceived risk emerged as the most common predictors. Zhou and Horrey (2010) examined social conformity among 510 adolescents aged years in China. They also used a questionnaire that included demographic information, items relating to social conformity and sensation seeking, and TPB-related items about two traffic scenarios. As in their previous study, adolescents reported greater likelihood of crossing the road when others were crossing. Those in middle school were more likely to cross than those in high school. Attitude, subjective norm, perceived behavioural control as well as anticipated affect emerged as common predictors of intention. Elvik, Sorensen and Naevestad (2013) also confirmed this safety in numbers effect with an increase in the number of pedestrians associated with a lower risk of accident for each pedestrian. Austroads 2016 page 88

99 The issue of self-identity as a careful pedestrian was considered by Holland, Hill and Cooke (2009). Using a self-report questionnaire, their study compared the contributions of self-identity and attitudes to the prediction of intentions and crossing behaviour. Participants also completed a simulated task where they were asked to decide when they would choose to cross the road in two scenarios. A total of 362 participants completed the questionnaire (and were assigned to one of four age groups) and of these, 205 also volunteered to participate in the simulation task. Self-identity and attitude were both shown to be significant independent predictors of intention to cross, which in turn predicted behaviour; however, self-identity was a less effective predictor in the riskier of the two scenarios. It was suggested that self-identity may have a greater role in more habitual behaviour. Self-identity was also strongly predicted by age, with only 12% of the variance explained in the oldest group (75+), compared to 23% for the youngest group, 39% for the middle group and 31% for the young-old group. Schwebel, Stavrinos and Kongable (2009) investigated attentional control, high intensity pleasure (the tendency to desire novel, complex, and varied stimuli, one facet of sensation-seeking), and risky pedestrian behaviour in college students. The study used a virtual environment task in which participants completed a series of street crossing trials and also a self-report questionnaire about demographic and individual differences. Individuals scoring high on attentional control (the capacity to focus and shift attention) waited longer to choose gaps to cross and showed some tendency to choose larger gaps after waiting. Individuals scoring high in high intensity pleasure were more likely to experience collisions with traffic in the virtual environment. Using the Health Belief Model, Yagil (2000) examined pedestrian self-reported crossing behaviour in relation to beliefs about the consequences of the behaviour, instrumental and normative motives for compliance with safety rules, and situational factors. A total of 205 students at two higher education institutions in Israel completed a self-report questionnaire. Results showed that crossing against a `Don t walk' sign was predicted by perceived consequences of the behaviour, as well as by normative motives. The normative motives contributed to the prediction of the behaviour more than the instrumental motives. Women s perceptions of their susceptibility to a crash resulting from an unsafe crossing was found to be greater than that of men; women also believed more that crossing against a don t walk sign would annoy drivers. Papadimitriou, Theofilatos and Yannis (2013) identified patterns of pedestrian attitudes and perceptions using the results of the Social Attitudes to Road Traffic Risk in Europe survey. This involved interviews in 19 European countries with 4290 pedestrians. In a principal component analysis, 33 pedestrian behaviours, attitudes and perception variables were explained by eight components (satisfaction with the pedestrian environment, attitude towards penalties, attitude towards in-vehicle devices, attitude towards speed limitations and surveillance, pedestrian behaviour and distraction, attitude towards pedestrian safety design, annoyance with other road users and lack of accessibility). A cluster analysis then revealed three types of pedestrians. The first type generally consisted of pedestrians with positive behaviour and positive attitudes (44.4% of pedestrians). The second type generally consisted of pedestrians with negative behaviour and negative attitudes (30.7% of pedestrians; a group in which male and young pedestrians were overrepresented). The third type generally consisted of neutral pedestrians with positive behaviour but mixed attitudes, (24.9% of pedestrians). The authors conclude that these proportions should be taken into account by policy makers in each country and at a European level. Evidence for the contribution of socioeconomic factors to pedestrian crashes (developing countries) Socioeconomic factors have appeared in various studies that have attempted to characterise pedestrian crashes according to demographic and other characteristics. Recent research on pedestrian crash risk in developing countries offers useful insight into the factors at play in developing, or high poverty level regions. Tulu, Washington, King and Haque (2013) reviewed the factors known to influence pedestrian crash risk in developing countries, drawing on data from Ethiopia (where available). Factors included walking at night, a factor known to be relevant in other countries, but of more concern in developing countries where reflective, high visibility clothing is not common, nor affordable, and street lighting is also less common. Higher exposure, inadequate or absent headlight illumination due to poor compliance with rules and standards, alcohol impairment (of both drivers and pedestrians) were further night time related safety issues. Fatigue is a second key issue for developing countries, with walking being the principle mode of transportation for both Austroads 2016 page 89

100 short and long distance travel. Crossing behaviour, walking along roads where there is a lack of separation between vulnerable road users and motorised traffic, poor enforcement of traffic regulation, poor transportation and land use planning, lack of road safety education and institutional capacity, population growth rate and high annual growth of motorisation are further issues that are negatively affecting the safety of pedestrians in developing countries (Tulu et al 2013). Zhuang and Wu (2011) investigated pedestrian crossing behaviours and safety at unmarked crossways in China, which, as a developing country, has a high number of traffic accidents and pedestrian deaths. The study involved field observations using video footage of 254 pedestrians at a busy unmarked roadway in Hangzhou, China. The site was chosen because, although only 82m from a zebra crossing, most people chose to cross there for convenience. Results showed that 65.7% of pedestrians did not look for vehicles after arriving at the kerb. Pedestrians preferred checking during crossing (and did so for nearly 70% of the crossing time) rather than immediately prior to stepping into the roadway. Pedestrians preferred crossing actively in tentative ways (stopping, stepping back, or running if they had the chance) rather than waiting passively. Furthermore, 41.7% of pedestrians were distracted in some way (including using mobile phone, talking with others, carrying luggage and grooming). Crash data analysis has been used in several studies in different countries to develop profiles of factors influencing crash risk. In China, Xiaoli, Pingsheng, Shuyun and Hongqi (2014) also examined pedestrian fatal crashes, with a focus on low volume rural roads. Twenty-eight fatal pedestrian crashes at intersections were analysed in terms of crash characteristics and causal patterns. The researchers concluded that the main issues for pedestrian safety at these intersections were limited safety design and facilities, inadequate driver training and inadequate pedestrian safety education. In an analysis of crash statistics in Bangladesh, it was found that approximately 80% of pedestrian accidents occurred in urban areas, approximately 50% of the pedestrian fatalities occurred on road sides and 22% occurred while crossing. One third of pedestrian deaths were to people under the age of 16 years. Mabunda, Swart & Seedat (2008) analysed data on 7433 pedestrian fatalities in four South African cities between 2001 and Over half of these deaths (56.7%) were to pedestrians between ages 20 and 44 years; there were 3.3 male pedestrian deaths for every female pedestrian death, and over half (58%) of the 4004 cases tested were positive for alcohol. Three categories of pedestrian fatalities were identified: male pedestrian deaths with high levels of alcohol concentrations, female and elderly pedestrian deaths occurring between 6am and midday, and children, adolescents and young adult pedestrian deaths during weekday afternoons and evenings. Deficiencies related to road infrastructure and hazards have also been identified as major contributors to pedestrian fatalities. In Bangladesh, an irap infrastructure assessment showed that most major highways were 2-star or less suggesting that these present substantial hazards to pedestrians (Hoque, Hossain, Islam & Rahman, 2013). Socio-economic factors that contribute to pedestrian crashes in Western countries appear to be different from those for developing countries. Recently, Chimba, Emaasit, Cherry and Pannell (2014) investigated the relationship between demographic and socioeconomic factors (based on census data) and pedestrian crashes in Tennessee mapped using geospatial information. Crash data for 5360 pedestrian crashes was drawn from the Tennessee Department of Transportation and Department of Safety databases ( ). Positive correlations between pedestrian crash occurrence and low vehicle ownership, high poverty levels, high minority populations, residential and fringe areas, presence of a rolling terrain, high speed limits, high number of lanes and presence of a school zone. Also in the USA, Cottrill and Thakuriah (2010) undertook an exploratory analysis of pedestrian crashes in areas with high low-income or minority populations in the Chicago metropolitan area. They found that pedestrian crashes in such areas are related to issues of exposure, crime rates, transit availability, and general population demographic characteristics such as income and presence of children. Cultural factors Cultural factors have been investigated in relation to their influence over pedestrian behaviour. In Israel, pedestrian behaviour in orthodox versus non-orthodox regions, with orthodox citizens having a stronger belief in the supremacy of religious laws over state laws than other citizens (Rosenbloom, Nemrodov & Barkan, 2004; Rosenbloom, Shahar & Perlman, 2008; Rosenbloom, 2009) has been compared. Rosenbloom et al (2004) observed 1,047 pedestrians at two busy urban intersections, one in a secular area and one in an ultra-orthodox area. Pedestrians in the orthodox environment were found to commit violations Austroads 2016 page 90

101 three times more frequently than those in the secular environment. Results also showed that males committed significantly more violations than females, and there was a negative correlation between age and frequency of violations. Rosenbloom et al (2008) went on to further investigate this apparent difference. A total of 995 pedestrians were observed (461 in an ultra-orthodox city and 534 in a neighbouring secular city). As in the previous study, ultra-orthodox pedestrians committed more violations than secular pedestrians did, and there were more road violations in the ultra-orthodox location than there were in the secular location. Consistent with previous research, males crossed on red much more than females did, regardless of religiosity. The role of cultural factors and attitudes for pedestrian behaviour was investigated in a Turkish sample by Nordfjaern and Simsekoglu (2013). A questionnaire was conducted among pedestrians in urban environments in Izmir (n=239) and Istanbul (n=50) (reflecting different socio-economic neighbourhoods). The questionnaire included questions about demographic factors (including religious affiliation), the number of hours per day spent as a pedestrian in traffic and the number of accidents in which they had been injured in the past 5 years. The Pedestrian Behaviour Scale and Turkish versions of a collectivism and individualism and uncertainty avoidance instruments were also used. When looking at pedestrian safety attitudes, cultural factors explained approximately 60% of the variance. Results also showed that vertical collectivism (the interrelation between people but where respect for authority and differences are emphasised) was related to lower levels of self-reported pedestrian risk taking behaviour. Horizontal collectivism (interpersonal interdependence is stressed, with a lower emphasis on hierarchy and authority) was related to greater levels of pedestrian risk taking behaviour. These findings are especially relevant when considering conformity towards regulations set by authorities and also the influence of social conformity on risky crossing. Other factors Other factors that influence pedestrian behaviour which have been investigated include the effect of carrying baggage, sleep restriction, trip length and trip purpose. The effect of carrying a backpack was explored by Schwebel, Pitts and Stavrinos (2009). A total of 96 college students took part in 20 street crossings using a virtual pedestrian environment. They wore a backpack for half of the crossings. Results showed that participants walked more slowly, left less safe time to spare after crossing the virtual street, and experienced more frequent hits or close calls with traffic when crossing while carrying the backpack. They also missed fewer safe opportunities to cross while carrying the backpack. Although wearing a backpack represents only facet of pedestrian behaviour, the authors noted that it was interesting that carrying a heavy burden didn t affect how quickly participants entered the street or the size of the traffic gaps selected for when to cross. The effect of sleep restriction on adolescent pedestrian behaviour was investigated by Davis, Avis and Schwebel (2013). In a within-subjects design, 55 adolescents (aged years) took part in a virtual pedestrian environment both after sleep restriction (measured using actigraphy) and with adequate sleep (a week apart). When sleep restricted, participants generally behaved in more risky ways. They took more time to initiate pedestrian crossings, crossed with less time before contact with vehicles, experienced more virtual hits or close calls, and looked left and right more often compared with when adequately rested. Findings were consistent after controlling for age, gender, ethnicity, and average total sleep duration before the two conditions. In an intercept survey of 367 respondents, Kothuri, Clifton and Monsere (2014) explored pedestrian attitudes and perceptions regarding safety, delay and determinants of crossing decisions at signalized intersections. Questions centred on current trip length, frequency of crossing at the intersections of interest and trip purpose, and influence of safety perceptions and compliance with traffic laws on decision to cross. Results showed that the presence of groups, positive perceptions of safety at an intersection and work trip purpose all reduced the likelihood of responding that safety was a factor in crossing. Recreational trips and moderate trip lengths increased the likelihood that compliance was indicated as a factor in crossing decisions. Younger adults, users of public transportation and people making shorter trips were less satisfied with delay and waiting time. In contrast, people who had positive perceptions of safety and were frequent users of the intersections were more likely to be satisfied with delay and waiting time. Younger respondents perceived greater safety while crossing compared to older respondents, and more women compared to men strongly agreed that their crossing decisions were based on safety considerations. Austroads 2016 page 91

102 Ibrahim, Day, Hirshon and El-Setouhy (2012) conducted a self-report survey among 1,324 university students aged years in Cairo, Egypt. Twenty-two percent of respondents had suffered a pedestrian injury during the last 6 months. Positive associations were found between road risk-perception, road behaviours and attitude towards traffic injury prevention. Using observations and interviews, Hatfield and Job (2004) explored pedestrian and driver perceptions of priority rules at pedestrian crossings in three NSW locations (Glebe, Chatswood and Goulburn). While most of the observed pedestrians started crossing on a pedestrian Walk signal, a sizeable proportion of the sample started crossing on a Don t Walk signal, with variation seen across the three locations for some behaviours. When asked about key situations, right of way was generally recognised, but some drivers reported an intention to take right of way nonetheless. Quite a high proportion of the sample reported that they would start crossing on a flashing Don t Walk signal when cars are a long way away (72.5%), when the signal has just started flashing (73.5%), when there are no cars in sight (79%) and when it is safe to do so (85.2%). There was recognition that both pedestrians and drivers can contribute to problems on the roads. Austroads 2016 page 92

103 Appendix B Intercept Interview Schedule Would you like to participate? Yes (continue) No (thank and terminate) For University ethical reasons, we can only talk to people between years. For statistical purposes, what is your age? Intersection A / B When approached were they using technology: Yes (phone) Yes (music) No Gender: M / F Go on if YES and between years: [VERBAL SCRIPT HERE] Do you have any questions? Here is the information sheet for you to keep. It basically repeats what I have just said. Our contact details are on this. Please feel free to read over this before we start, if you d like. Is it all right now to ask you the questions? I would like to note your answers while we talk. Will that be all right with you? YES / NO We can offer you a coffee/drink voucher at the end as our way of saying thank you. QUESTIONS DEMOG: These are some questions about you for statistical purposes. These will not be used to identify you in any way 1. [record without asking] (circle) Male Female 2. What do you regard as your employment/study status [CARD A] Main purpose of this walking trip [CARD B] How often do you use public transport [CARD C] About how much time do you spend each week walking for transport (including getting between where you work and public or private transport) (minutes, hours/minutes) 6. About how many times a week do you have to cross a road while you are walking? Austroads 2016 page 93

104 7. I am going to read some activities to you. Using this scale [point to CARD D] I will ask you to tell me how often in the last week have you done the activity. The first one is, how often have you used your device to initiate a text while walking? [repeat for the others] (circle) Walking Crossing More than once a day Once a day Several times a week Once a week or less Never More than once a day Once a day Several times a week Once a week or less Initiate a text Monitor text messages Respond to a text Never Initiate a call Answer a call Initiate internet search or interaction Monitor internet Respond to internet Use an audio only device with earphones/buds in ear listening to music/radio Have you ever had any near misses, that is, with a vehicle (including bicycles), while walking or crossing the road when you were distracted? If yes, please explain briefly: If over 18: [By walking we mean on the footpath or road for 50m or more] 9. a. Many people tell us they walk after they have consumed alcohol. We are very interested in your own experiences and views about this. By consuming alcohol we mean, having at least 2 standard drinks in the previous hour immediately prior to walking. Thinking about the past 3 months, about how often have you walked after consuming at least two standard drinks in the previous hour? [CARD E] 1. Daily times a week 3. Once a week 4. Once a month 5. Less often 6. Never Thank and give voucher b. On any of those occasions did you experience any near misses with a vehicle (including bicycles)? Y / N If yes, how many times? Austroads 2016 page 94

105 c. 9C. Thinking back over those occasions, in the last 3 months, when you did walk after having at least 2 standard drinks, about how many times were there in total? d. 9D. How many of those times do you think you were affected by the alcohol you drank? e. 9E. On any of those occasions where you walked after consuming alcohol, did you sustain an injury? Y / N If yes, how many times? f. 9F. How serious was the most serious injury? [CARD F] 1. Trivial 2. Minor-no medical treatment needed 3. Required medical attention 4. Required Emergency Department visit 5. Hospitalised Austroads 2016 page 95

106 Appendix C Summary of Results for Intercept Interview Responses to Drink Walking Questions For the summary below, all participants (n = 211) were asked about drink walking. However, only those who indicated that they had drink walked on at least one occasion in the previous 3 months (n = 116) were asked subsequent questions, resulting in smaller n for later questions. How often have you walked after consuming at least two standard drinks? (n = 210) Never % (n) Daily % (n) 1-5 times per week % (n) Once per month % (n) Less than once per month % (n) 42.7 (94).9 (2) 19.6 (43) 22.3 (49) 10.0 (22) How many times during the last 3 months did you walk after having at least 2 standard drinks? (n = 116) How many of those times did you think you were affected by the alcohol you drank? (n = 116) On any of those occasions did you experience near misses? (n = 116) On any of those occasions did you sustain an injury? (n = 116) 1-3 occasions % (n) 4-12 occasions % (n) more than 12 occasions % (n) 39.7 (46) 37.1 (43) 23.3 (27) None % (n) 1-3 % (n) 4-12 % (n) more than 12 occasions % (n) 31.0 (36) 27.6 (32) 28.4 (33) 7.1 (15) No Yes 97.4 (113) 2.6 (3) No Yes 95.7 (111) 4.3 (5) Austroads 2016 page 96

107 Austroads 2016 page 97