1

Mistakes or Deliberate Violations? A Study into the Origins of Rule Breaking at 1

Pedestrian Train Crossings 2

3

James Freeman and Andry Rakotonirainy 4

Centre for Accident Research and Road Safety – Queensland (CARRS-Q) 5

Queensland University of Technology 6

Abstract 7

Train pedestrian collisions are the most likely to result in severe injuries and fatalities 8

when compared to other types of rail crossing accidents. However, there is currently scant 9

research that has examined the origins of pedestrians’ rule breaking at level crossings. As a 10

result, this study examined the origins of pedestrians’ rule breaking behaviour at crossings, 11

with particular emphasis directed towards examining the factors associated with making 12

errors versus deliberation violations. A total of 636 individuals volunteered to participate in 13

the study and completed either an online or paper version of the questionnaire. Quantitative 14

analysis of the data revealed that knowledge regarding crossing rules was high, although up 15

to 18% of level crossing users were either unsure or did not know (in some circumstances) 16

when it was legal to cross at a level crossing. Furthermore, 156 participants (24.52%) 17

reported having intentionally violated the rules at level crossings and 3.46% (n = 22) of the 18

sample had previously made a mistake at a crossing. In regards to rule violators, males 19

(particularly minors) were more likely to report breaking rules, and the most frequent 20

occurrence was after the train had passed rather than before it arrives. Regression analysis 21

revealed that males who frequently use pedestrian crossings and report higher sensation 22

seeking traits are most likely to break the rules. This research provides evidence that 23

pedestrians are more likely to deliberately violate rules (rather than make errors) at crossings 24

and it illuminates high risk groups. This paper will further outline the study findings in 25

regards to the development of countermeasures as well as provide direction for future 26

research efforts in this area. 27

2

28

Keywords: Trains, pedestrian level crossings, deliberation violations, mistakes. 29 30

Address for Correspondence: James Freeman. Centre for Accident Research and Road Safety, 31

Institute of Health and Biological, Innovation, K Block, 130 Victoria Rd Kelvin Grove 32

Campus, Kelvin Grove 4059 Queensland University of Technology, Australia. Ph: +61 7 33

31384677; Fax: +61 7 3138 0111; Email: je.freeman@qut.edu.au 34

35

1. Introduction 36

Collisions between trains and pedestrians are the most likely to result in rail-related 37

fatalities (Federal Railroad Administration 2006; Nelson 2008; Sochon, 2008), as only one 38

third of pedestrians survive these collisions (Lobb, Harre & Terry, 2003). This statistic is also 39

much higher than train motor vehicle collisions (Australian Transport Safety Bureau 2004). 40

Within Australia, there were 392 rail accidents (involving pedestrians) that resulted in 41

fatalities between January 2001 and December 2010, excluding suicides (Australian Transport 42

Safety Bureau 2011). An earlier study reported that pedestrians (excluding suicides) 43

comprise 66% of all fatalities at crossings (Australian Transport Safety Bureau 2004). 44

Furthermore, in regards to serious injuries, researchers examined data between 2003 and 2007 45

and reported 51 people on average were hospitalised each year due to train-pedestrian 46

collisions in Australia (Henley & Harrison 2009). Given the significant personal, social and 47

emotional burden of such collisions, understanding the origins of such events is of primary 48

importance. 49

50

1.1 Errors versus Deliberate Violations 51

One of the central unanswered questions relating to collisions between trains and 52

pedestrians is whether they result from pedestrians making errors or deliberating violating 53

crossing rules. In regards to this question, the largest body of research to date has focused on 54

3

train vehicle collisions. This research has generally indicated that drivers are likely to 55

inadvertently engage in risky behaviours as a result of not detecting crossings, failing to 56

notice approaching trains and misjudging the risk of approaching trains (Australian Transport 57

Safety Bureau 2002; CRC 2010; Wallace 2008). These may broadly be conceptualised as 58

human error, and may result from various cognitive factors including inattention, distraction, 59

poor knowledge, misjudgment and limited sight distance (Freeman et al., 2013). In fact, 60

failure to detect warning signals (or understand their meaning) has previously been calculated 61

to account for nearly half of all fatal level crossing crashes in Australia (ATSB Statistical Unit 62

2002). Additionally, poor knowledge of level crossing procedures and/or road rules has also 63

gained attention within the field, as researchers have proposed that road users have poor 64

knowledge of train speeds as well as the ability to slow quickly (if needed) to avoid an 65

accident (Richards & Heathington 1990). Researchers have hypothesised that pedestrians 66

may also have poor knowledge regarding the penalties associated with breaching crossing 67

rules (Wallace 2008), although little research has been conducted into this area. One of the 68

few studies to focus on pedestrians’ knowledge of penalties revealed that almost half of a 69

sampled group did not believe or were unaware that it was illegal to cross when a train was 70

approaching (Lloyd’s Register Rail 2007). The same Victorian study reported that 18% of the 71

sample reported unintentionally being caught on train tracks when a train was approaching 72

(Lloyd’s Register Rail 2007). The most common reason reported for this outcome was not 73

being aware of the train approaching or a second train approaching. 74

75

The issue of the presence (and awareness) of a second train shortly after the first has 76

passed is receiving increasing focus within the literature. An American study calculated that 77

18% of pedestrian accidents were related to the presence of a second train (Federal Railroad 78

Administration 2008; Illinois Commerce Commission 2005). The above mentioned Victorian 79

4

study reported that 16% of the sample would at least sometimes cross (and violate warning 80

signals) after one train had passed (Lloyd’s Register Rail 2007). The authors also examined 81

Victorian coroners’ reports of pedestrian level crossing fatalities which revealed that the 82

presence of a second train was often mentioned (Lloyd’s Register Rail 2007). Furthermore, 83

poor knowledge (and/or under estimation) of train speeds has also been proposed to be a 84

contributing factor (CRC 2010). For example, an earlier review of 18 pedestrian fatalities 85

between 2002 and 2004 revealed that misjudgment was a primary factor in 8 cases, and a 86

further 6 pedestrians were under the influence of alcohol or drugs (Australian Transport 87

Safety Bureau 2004). 88

89

While it has been hypothesised that road users are more likely to make errors 90

(compared to violations) at crossings, the reverse has been proposed for pedestrians (CRC 91

2010; Freeman et al., 2013). For example, pedestrians ignoring warning signs has been 92

reported as a contributor in some United States studies (Federal Railroad Administration 93

2008; Illinois Commerce Commission 2005). One of the few Australian studies to focus on 94

the issue reported that 31% of a sample of pedestrians admitted crossing the tracks even when 95

they knew a train was approaching (Lloyd’s Register Rail 2007). The most common reason 96

reported for such behaviour was being in a hurry. Another study that examined both vehicle 97

drivers and pedestrians reported that pedestrians are more likely to speed up and cross the 98

tracks when a train is approaching (Beanland, 2013). A review of rail literature 99

commissioned by the Coopeartive Research Centre (CRC) for Rail Innovation (2010) found 100

that a small group of studies reported that the origins of pedestrians’ behaviour (involving 101

violating traffic or level crossing rules) was due to maximizing convenience (Daff & 102

Cramphorn, 2006; Federal Railroads Administration 2008; Lobb 2006; Lobb, Harre & 103

Sudendorf 2001). It has yet to be confirmed whether pedestrians make calculated risks, 104

5

although it has been hypothesised that some weigh the perceived safety of a route against the 105

time and effort that would be required to use it (Lobb, 2006). Additionally, researchers have 106

suggested that risky crossing behaviour may be reinforced if individuals consistently engage 107

in such behaviour and avoid the negative consequences (CRC 2010; Davey et al. 2008). This 108

phenomenon is otherwise known as “punishment avoidance.” Preliminary research has also 109

found that violations are more likely to occur in the presence of pedestrians (Khattak & Luo 110

2011). 111

112

1.2 High Risk Groups and Factors 113

An analysis of the individual characteristics associated with fatalities and serious 114

injuries at level crossings reveals some groups are at a disproportionate risk of being struck by 115

a train. These groups are: (a) males, (b) school children and other young persons, (c) people 116

with disabilities as well as (d) older pedestrians (Lloyd’s Register Rail 2007). However, 117

comprehensive information about why these groups remain at-risk has yet to be obtained 118

(CRC 2010). Nevertheless, what is known is that males are consistently over-represented in 119

both vehicle and pedestrian incidents at level crossings (CRC 2010; Henley & Harrison 120

2009). For example, previous Australian research calculated that 84% of train-pedestrian 121

fatalities involved males (Australian Transport Safety Bureau 2004), which is also reflected in 122

train-vehicle collisions e.g., 80% (Australian Transport Safety Bureau 2002). A European 123

study also reported males are over represented in train-pedestrian fatalities (Silla & Luoma 124

2012). 125

Secondly, school children have also been identified to be at an increased risk at level 126

crossings (Khattak & Luno 2011; Lobb et al. 2003; Spicer 2008). A video surveillance study 127

of crossing behaviours in Nebraska revealed that children approximately 8 years of age 128

accounted for a disproportionate amount of gate-violations (25%) compared to older 129

6

pedestrians and cyclists (Khattak & Luno 2011). An earlier New Zealand study reported that 130

50% of train-pedestrian fatalities and 40% of injuries involved persons aged between 10 and 131

19 (New Zealand Health Information Service, 1999, cited in Lobb et al. 2003). However, and 132

similar to above, there is limited research into the origins of why school children are at an 133

increased risk (CRC 2010). Nevertheless, these might include poor scanning behaviours, 134

underdeveloped cognitive and risk perception abilities as well as impulsiveness (CRC 2010). 135

In regards to scanning behaviours, one of the only studies in the area reported pedestrians 136

aged between 12 and 17 years had the poorest scanning behaviours at crossings (McPherson 137

& Daff 2005). This is consistent with what is known within the broader developmental 138

neuroscience field that indicates adolescences are most likely to make risky decisions (Albert 139

& Steinberg 2011). More specifically, underdeveloped perceptual/cognitive skills (Congiu et 140

al. 2008) divided attention combined with poor scanning (Dunbar, Lewis & Hill 2001), and 141

limited crossing experience (Connelly et al. 1998) have all been proposed to contribute to 142

younger persons increased risk. Additionally, intentional risk taking has also been 143

hypothesized to be a significant contributing factor (Lloyd’s Register Rail 2007), although 144

consistent with broader knowledge in the field, it remains unclear why this group are at an 145

increased risk of being struck by a train (Freeman et al., 2013). 146

147

In addition to males and younger groups, people with physical disabilities (or mobility 148

aids) and older pedestrians are at an increased risk of being struck by a train. This may stem 149

from a number of reasons, including: (a) getting mobility aids stuck in train tracks (b) being 150

vulnerable to uneven or poorly maintained surfaces and (c) slower crossing speeds (CRC 151

2010; McPherson & Daff 2004). However similar to above, little is currently known 152

regarding the exact risks for this group’s increased vulnerability at pedestrian crossings. 153

154

7

Taken together a number of outstanding questions remain regarding the origins of 155

pedestrians’ behaviours at level crossings, which has clear implications for the development 156

of effective evidence-based interventions. Due to this lack of research, the present study aims 157

to examine the etiology of pedestrian’s crossing behaviour (in a large sample) through 158

quantitative analysis. The three main areas of focus are examining: (a) whether pedestrians 159

are more likely to make deliberate violations versus errors at crossings, (b) whether these 160

factors are consistent across age groups and (c) what other factors increase the likelihood of 161

breaking crossing rules. 162

163

2. Method 164

2.1 Participants 165

A total of 636 individuals volunteered to participate in the study (in the Brisbane 166

region) comprising of 119 minors (<18 years), 460 younger adults (18 – 59 years) and 57 167

older adults (60+ years). The mean age of the sample was 34.94 (range 10 to 82) years. 168

Males constituted 56.6% (n = 360) of the sample while 43.4% (n = 276) were females. For 169

the younger adults, 56.50% (n = 260) had obtained a tertiary degree, and another 19.34% (n = 170

89) had completed year 12. For the older adults, 58.9% (n = 33) had obtained a tertiary 171

degree, 12.5% (n = 7) had completed year 12. The most frequent reported income brackets 172

for the younger adult group was $52,000-$83,999 p.a. (27.9%, n = 121) followed by the 173

>$84,000 p.a. bracket (22.9%, n = 99). For the older adult group, the most frequent income 174

bracket was $13,000-$31,199 p.a. (31.4%, n = 16) with 19.6% (n = 10) participants earning 175

below this level and 31.4% (n = 16) earning above this level. In regards to detection, 10 176

participants had previously been fined for breaking the rules at a pedestrian level crossing. 177

178

2.2. Materials and Procedure 179

8

A questionnaire was developed (and piloted with 20 participants) which contained 180

questions on demographics and level crossing usage, knowledge and experiences with level 181

crossings, frequency of making deliberate violations and mistakes as well as reasons for 182

making violations and errors. The Brief Sensation Seeking Scale (BSSS) developed by 183

Hoyle, Stephenson, Palmgreen, Lorch & Donhew (2002) was also utilised to measure the 184

extent of sensation seeking propensities, particularly among the younger group. 185

186

2.3 Procedure 187

The minors were recruited from two high schools in the Brisbane metropolitan area 188

situated in close proximity to a level crossing. Two procedures were utilised. The 189

questionnaire was completed via pen and paper at schools. Participants were provided with a 190

movie voucher. The younger adult and older adult groups were recruited via a radio interview 191

promoting the study, newspaper and online advertisements, and disseminating flyers at level 192

crossings as well as snowballing techniques. All adult participants completed an online 193

questionnaire and were given a $20 Coles-Myer voucher. Both versions of the questionnaire 194

were identical, with some irrelevant demographic questions omitted from the minors’ version 195

of the questionnaire (e.g. current employment status, yearly income and highest level of 196

education.) 197

198

3. Results 199

3.1 Level Crossing Usage 200

In terms of usage, the minor group were proportionately the most frequent users of 201

level crossings with 10.92% (n = 13) using crossings daily and 23.52% (n = 28) using 202

crossings weekly. In contrast, for the younger adult group, 5.43% (n = 25) reported using 203

crossings daily and 10.00% (n = 46) used crossings weekly or less often. For the older adult 204

9

group, no participants used level crossings on a daily basis, but 42.10% (n = 24) used the 205

crossings weekly or monthly. The usage frequency for the different age groups is shown in 206

Table 1. 207

208

209

Table 1. Frequency of Level Crossing Usage 210

Age Group Daily Weekly Monthly Less than

Monthly

Minors 10.92% (n = 13) 26.89% (n = 32) 21.01% (n = 25) 41.17% (n = 49)

Younger Adults 5.65% (n = 26) 15.43% (n = 71) 17.83% (n = 82) 61.08% (n = 281)

Older Adults 0.00% (n = 0) 19.30% (n = 11) 22.81% (n = 13) 57.90% (n = 33)

211

3.2 Knowledge and Experience 212

A series of questions focused on participants’ knowledge and experiences at 213

pedestrian crossings. Participants responded “yes”, “no” or “unsure” as to whether they 214

believed it was legal to cross a level crossing in the presence of: (a) a bell sounding, (b) lights 215

flashing and (c) when the automatic gates were closed. In the current context, a bell may 216

sound (and/or lights flash) before automatic gates close at active sites in the sampled area. 217

Overall, rule knowledge was high, with the lowest percentage of correct answers being for 218

crossing when the bell is sounding (combined percentage 82.28%). For all three rules the 219

age group with the greatest percentage of participants answering correctly were the older 220

adults. For the individual groups, younger participants were least likely to report accurate 221

knowledge regarding lights flashing (75.63%) and crossing after the gates are closed 222

(75.63%). However, and from a different perspective, the results indicate that between 10% 223

to 18% of level crossing users (in the current sample) were either unsure or did not know (in 224

some circumstances) when it was legal to cross a level crossing. The percentages of correct 225

answers overall and for individual age groups are shown in Table 2. None of the differences 226

between age groups were statistically significant. 227

10

Table 2. Percentage of Participants Answering Crossing Rule Questions Correctly 228

Age Group Bell Sounding Light Flashing Gate Closed

Minors 76.47% (n = 91) 75.63% (n = 90) 75.63% (n = 90)

Younger Adults 83.47% (n = 384) 87.61% (n = 403) 88.48% (n = 407)

Older Adults 92.90% (n = 52) 93.0% (n = 53) 93.0% (n = 53)

Overall 82.28% (n = 527) 85.85% (n = 546) 86.47% (n = 550)

229

3.3 Deliberate Violations 230

A central aim of the study was to investigate the frequency of making deliberate 231

violations at pedestrian crossings versus making mistakes. As a result, participants were 232

asked if they had ever crossed at a railway pedestrian crossing when the lights were flashing, 233

the gate was closed, deliberately ignored warning signal, made a mistake at a level crossing 234

or been fined for breaking the rules at a level crossing. Responses were in a yes or no 235

format. For any of the intentional violation questions (e.g., lights flashing, gates closed or 236

deliberately ignored a signal), 156 participants (24.52%) reported having intentionally 237

violated the rules at level crossings in at least one of these ways. Participants were provided 238

with an opportunity to explain the reasons for their behaviour, with the overwhelming 239

majority indicating they were “in a rush” or “running late.” Males were significantly more 240

likely to break the rules compared to females (111 versus 45), X2 (2, N = 636) = 18.32, p = 241

.000. The age group with the highest incidence of intentional rule breaking was minors 242

(35.3%, n = 42). However, for the younger adult group, 23.0% (n = 106) and 14.0% (n = 8) 243

of the older adult group had also intentionally broken the rules. Percentage and frequency 244

scores for the intentional violations can be seen in Table 3. A chi-square analysis conducted 245

between the intentional violation rates across the ages determined there was a statistically 246

significant difference, X2 (2, N = 636) = 11.38, p = .003. A corresponding question regarding 247

crossing when the lights were flashing (measured on a 7 point likert scale) revealed that the 248

11

participants were statistically more likely to violate the rules after the train has passed (M = 249

2.32) versus before the train has passed (M = 1.71) [ t (636) = -10.65, p < .001]. 250

251

3.4 Errors 252

In contrast, errors were not as common an occurrence. More specifically, only 3.46% 253

(n = 22) of participants reported this experience. Similar to the violation data, there was 254

proportional variation in the incidence of mistakes between the age groups with the highest 255

percentage of minors making mistakes (5.0%, n = 6) followed by older adults (5.2%, n = 3) 256

and lastly younger adults (2.8%, n = 13). In regards to crossing after the train had passed, 17 257

participants (2.67%) reported crossing after a train had passed and not realised a second train 258

was approaching. Due to the small numbers involved, statistical analyses could not be 259

undertaken to determine if these differences were statistically significant or expected by 260

chance. However, participants who made deliberate violations were also more likely to make 261

errors at level crossings X2 (2, N = 636) = 11.09, p = .002. The prevalence of deliberate 262

violations and mistakes made between the three age groups are shown below, in Table 3. 263

264

Additionally, intentions to break crossing rules again in the next six months were 265

measured on a 7 point likert scale (1 = likely to 7 = unlikely). While the mean of 6.63 (SD = 266

1.16) indicates the sample were generally not likely to violate crossing rules, a closer 267

examination of the data revealed that males were statistically more likely than females t (636) 268

= -5.07, p < .000 and minors were more likely than the combined group of adults t (636) = -269

4.52, p < .000. Consistent with previous research that has indicated past behaviour is a good 270

predictor of future behaviour (Freeman & Watson, 2006), those who admitted breaking 271

crossing rules in the past, were more likely to indicate a willingness to engage in similar 272

behaviour in the future (M = 5.91 versus M = 6.86; t (636) = -9.55, p < .000). 273

12

274

Table 3. Incidence of Deliberate Violations and Errors between the Age Groups 275

Age Group Deliberate Violations Errors

Minors 35.3% (n = 42) 5.0% (n = 6)

Younger Adults 23.0% (n = 106) 2.80% (n = 13)

Older Adults 14.0% (n = 8) 5.20% (n = 3)

Overall 24.52% (n = 156) 3.46% (n = 22)

276

3.5 Sensation Seeking 277

To determine if reported violation behaviours were related to sensation seeking 278

(particularly among the younger cohort), the Brief Sensation Seeking Scale (BSSS) (Hoyle, 279

Stephenson, Palmgreen, Lorch & Donhew, 2002) was utilised. For the current study, 280

sensation seeking was considered to be “the need for varied, novel, and complex sensations 281

and the willingness to take physical and social risks for the sake of such experiences” 282

(Zuckerman, 1979, p.10). The BSSS is an 8-item scale which evaluates an individual’s 283

propensity to seek these sensations and experiences. Consistent with previous research 284

(Martins et al., 2008), minors recorded the highest sensation seeking scores, followed by 285

younger adults and older adults. The mean scores for minors was also comparable with 286

previous BSSS-based research (Hoyle et al., 2002). Not surprisingly, minors reported 287

significantly higher BSSS scores compared to adults (M = 3.05 versus M = 2.62, correlation r 288

= -.20**), although it may be suggested the difference is minimal in practical terms, t (636) = 289

5.29, p < .000. Furthermore, a similar finding was identified between BSSS scores and those 290

who did violate the rules (M = 2.96) and those who did not (M = 2.60), t (636) = 5.01, p < 291

.000. 292

293

To determine the predictors of violating crossing rules, a logistic regression was 294

performed to investigate the predictive role of age, gender, frequency of crossing usage, 295

crossing rule knowledge and sensation seeking to the outcome variable of breaking the rules. 296

13

Given the relatively small proportion of the sample who acknowledged breaking the rules, 297

those who admitted a deliberate violation or an error were combined to form the dependent 298

variable. Frequency of crossing usage was recoded into: (a) daily/almost daily, (b) weekly, 299

(c) monthly, (d) less than monthly. Crossing rule knowledge was divided into those who 300

answered all three questions correctly versus those who did not (82.1% versus 17.9%). Table 301

4 displays the regression coefficients, Wald statistics, odds ratios (OR), and 95% confidence 302

intervals. 303

304

The overall model was a significant predictor of the outcome variable (χ2(8) = 60.331, 305

p = .000). Taken together, 13.6% of the estimated total variance was accounted for in the 306

model and 77.4% of the sample was correctly classified. In regards to significant predictors, 307

three items were found to be significant predictors of crashes: gender (Wald = 7.938, p < .05); 308

Usage frequency (Monthly Wald = 19.259, p < .001 & Daily Wald = 5.286, p < .001) and 309

sensation seeking (Wald = 11.315 , p < .001). As a result, being male, using crossings more 310

frequently and reporting higher sensation seeking propensities were significant predictors of 311

breaking crossing rules. An additional regression analysis that only included deliberate 312

violators (and not errors) revealed the same predictive variables. 313

314

315

316

317

318

Table 4. Predictors of Breaking Crossing Rules 319

320

95% Confidence

14

interval for OR

Independent variables

B

S.E.

Wald

OR

Lower

Upper

Age -.003 .006 .195 .997 .985 1.010

Gender -.595 .221 7.938* .551 .364 .834

Usage Frequency (<monthly) 21.85

Usage Frequency (monthly) 1.180 .269 19.259** 3.253 1.921 5.510

Usage Frequency (weekly) .47 .304 2.384 1.600 .881 2.904

Usage Frequency (daily) .839 .365 5.286* 2.315 1.132 4.734

Knowledge .407 .240 2.876 1.503 .939 2.407

Sensation seeking

Model Chi-Square 60.331** (df = 7)

-.456 .136 11.315* .634 .486 .827

321

Note: *p < .05, **p < .001; OR = Odds Ratio; 322

4. Discussion 323

Currently, a number of outstanding questions remain regarding the origins of 324

pedestrians’ behaviours at level crossings as research has traditionally focused on motorists’ 325

behaviour. The present research aimed to examine: (a) whether pedestrians are more likely to 326

make deliberate violations versus errors at crossings, (b) whether these factors are consistent 327

across age groups and (c) what factors increase the likelihood of breaking crossing rules. 328

Firstly, participants generally exhibited a high level of level crossing rule knowledge in 329

regards to the presence of: (a) a bell sounding, (b) lights flashing and (c) automatic gates 330

closing. In regards to the latter, it is currently possible for pedestrians to push open closed 331

gates at active crossings in the Brisbane region, although a trial of “locking gates” is soon to 332

be implemented. Regardless of the wider implementation of this initiative, it is encouraging 333

that the sample generally recognised the inappropriateness of attempting to open gates that are 334

15

closed. Within the wider rail safety arena, research has demonstrated that the effective 335

positioning and use of barriers can reduce trespassing onto railway tracks (Silla, 2009). 336

However, from a different perspective, it may be considered somewhat surprising that up to 337

18% of the sample were either unsure or did not know (in some circumstances) when it was 338

legal to cross at a level crossing. This is consistent with one of the few studies in the area that 339

revealed almost half of an Australian-sampled group did not believe or were unaware that it 340

was illegal to cross when a train was approaching (Lloyd’s Register Rail 2007). This has 341

implications for the development of awareness/education schemes, particularly given the 342

tremendous amount of research that has demonstrated injury prevention education is an 343

important and effective tool to reduce injuries and fatalities (Crandall, Zarzaur & Tinkoff, 344

2013). However, it is noted that preliminary education-based campaigns have not proven 345

extremely effective (Lobb et al., 2001; Lobb et al., 2003; Shonfeld & Musumeci, 2003), 346

although this may be due to other deleterious effects on safety discussed below e.g., sensation 347

seeking propensities. 348

349

Secondly, and in regards to a central question of this program of research, participants 350

were more likely to report deliberately breaking the rules compared to making errors (24.52% 351

versus 3.46%). That is, almost a quarter of the sample admitted previously breaking crossing 352

rules at one time. This study is one of the first to confirm the hypothesis that pedestrians at 353

level crossings are more likely to ignore warning signals rather than make mistakes (CRC, 354

2010; Freeman et al., 2013). A corresponding preliminary analysis revealed that the reason 355

for this behaviour was most closely associated with being “in a rush” or “running late”, 356

which supports one of the only studies to examine this issue (Lloyd’s Register Rail 2007). 357

Further research is now required to determine whether pedestrians make calculated risks 358

before crossing or if such behaviour is more impulsive, as well as what interventions would 359

16

be most effective in reducing the likelihood of engaging in such behaviours e.g., barriers, 360

enforcement or education-based signs. 361

362

Thirdly, further analysis revealed that those most likely to engage in such behaviours 363

were males, particularly, younger males. More specifically, a comparative age analysis also 364

revealed that minors (e.g., under 18) were statistically more likely to report intentionally 365

violating the rules compared to younger and older adults. This finding again supports the 366

assertion that males, including school children, constitute a significant high risk pedestrian 367

group (Lloyd’s Register Rail 2007), which is evidenced in their overrepresentation in 368

recorded incidents at level crossings (Australian Transport Safety Bureau 2002; CRC 2010; 369

Henley & Harrison 2009). In regards to school children, this group has been proposed to be 370

at an increased risk for a number of reasons, including poor scanning behaviours, 371

underdeveloped cognitive and risk perception abilities as well as impulsiveness (CRC 2010). 372

While a corresponding examination of scanning and cognitive abilities was not possible in the 373

current study, a complementary analysis of sensation seeking traits revealed this factor to be 374

predictive of crossing violations, as was as frequency of crossing usage. In regards to the 375

former, developmental neuroscience has consistently indicated that adolescence are more 376

likely to make risky decision (Albert & Steinberg 2011), and/or take physical and social risks 377

for the sake of experience (Zuckerman, 1979a). In fact, sensation seeking has been shown to 378

be a predictor of a wide array of problem behaviours (Zuckerman, 1994), and now, risk taking 379

at pedestrian level crossing may yet prove to fall into this category. Sensation seeking has 380

also been shown to be higher in males (Hoyle et al., 2002), which may suggest why a 381

disproportionate percentage of males reported rule violation in the current study. Research 382

has indicated that individuals high in sensation seeking are drawn to high risk activities 383

(Hoyle et al., 2000) as this group underestimate the risks associated with such behaviours 384

17

(Hoyle et al., 2002). Furthermore, risk estimates have been hypothesized to be negatively 385

associated with novel activities (Zuckerman, 1979b), although this may not necessarily be the 386

case for the current context given that more frequent crossing usage was predictive of 387

breaking the rules. Rather, high sensation seekers may be more likely (than low sensation 388

seekers) to both try (and repeat) a wide array of risky activities (Hoyle et al., 2002). If further 389

research proves this to be the case, than a challenge remains as to the development of 390

effective countermeasures that can negate (or prohibit) the expression of high sensation 391

seekers’ behaviours, particularly among minors. Future applied research may yet demonstrate 392

this can be achieved through incapacitation and/or deterrent based interventions that either 393

prohibit or deter offending behaviours. 394

395

A final theme to emerge from the current study was that participants who reported 396

violating crossing rules were more likely to use crossings regularly. This is in contrast to an 397

earlier hypothesis that younger persons’ increased risk is attributable to limited crossing 398

experience (Connelly et al. 1998). From a different perspective, the current finding is 399

consistent with sensation seeking research which has indicated that high sensation seekers are 400

more likely to engage in risky behaviours, and then afterwards, diminish the level of risk 401

associated with the behaviour (Hoyle et al., 2002). Yet another alternative perspective 402

emerges from the deterrence-based literature, which has illuminated the powerful effects of 403

“punishment avoidance” (Freeman & Watson, 2006; Paternoster & Piquero, 1995). More 404

specifically, an increasing body of literature has demonstrated that committing an offence and 405

avoiding detection (and subsequent punishment) is a strong predictor of repeating the 406

behaviour in the future (Freeman & Watson, 2006). In the current circumstances, and given 407

the low probability of actually be apprehended for a pedestrian crossing offence, violators 408

may be prone to repeat the behaviour out of habit and/or due to reduced perceived threat of 409

18

any associated consequences. Although this hypothesis will need to be tested through in 410

future studies. Further research is also required to determine whether pedestrians make 411

calculated risks (Lobb, 2006), or if such behaviour is impulsive and/or instinctual. In regards 412

to calculated risks, it is noteworthy that the current sample was more likely to violate the rules 413

after (rather than before) a train has passed. This is the first study to examine this issue and 414

the finding warrants further research into the risks associated with being hit by a second train. 415

Preliminary research into risks associated with a second train passing at a crossing has 416

revealed an increase in accidents (Federal Railroad Administration 2008; Illinois Commerce 417

Commission 2005; Lloyd’s Register Rail, 2007). 418

419

The study’s limitations should be borne in mind when interpreting the results. Firstly, 420

participants were not randomly selected, and as a result, questions remain regarding the 421

representativeness of the sample. Secondly, the study only contains self-reported data, and as 422

such, there may be some discrepancy between stated behaviour and actual events, particularly 423

when responses focus on aberrant behaviours. Thirdly, a large proportion of the sample did 424

not regularly use pedestrian crossings e.g., less than monthly. 425

426

5. Conclusions 427

Despite such limitations, the study findings indicate that pedestrians are more likely to 428

deliberately violate crossing rules (compared to making errors) and a sizeable proportion of 429

the sample reported this behaviour e.g., a quarter. The results provide support for a renewed 430

focus on understanding the origins of pedestrian level crossing behaviour as well as 431

developing effective tailored countermeasures. Such a focus can only assist in the pursuit to 432

reduce the burden associated with train-pedestrian collisions. 433

434

19

Acknowledgements: The authors are grateful to the CRC for Rail Innovation (established 435

and supported under the Australian Government's Cooperative Research Centres program) for 436

the funding of this research. Project No. R2.120: Understanding Pedestrian Behaviour at 437

Level Crossings 438

439

20

440

REFERENCES 441

Albert, D & Steinberg, L., 2011, Judgement and decision making in adolescence, Journal of 442

Research on Adolescence 21 (1), 211-224. 443

444

Australian Transport Safety Bureau Transport Safety Statistical Unit 2002, Level crossing 445

incidents: fatal crashes at level crossing, In paper presented at the 7th International 446

Symposium on Railroad-highway Grade Crossing Research and Safety, Melbourne, Victoria, 447

Australia. 448

449

Australian Transport Safety Bureau 2004, Level crossing accident fatalities, Available from 450

http://www/atsb.gov.au/publications/2004/lev_crossfat.aspx 451

452

Australian Transport Safety Bureau 2011, Australian Rail Safety Occurrence Data,1 January 453

2001 to 31 December 2011, ARSB, Canberra. 454

455

Beanland, V., Lenne, M., Salmon, P., & Stanton, N. (2013). A self-report study of factors 456

influencing decision-making at rail level crossings: comparing car drivers, motorcyclists, 457

cyclists and pedestrians. Paper presented at the Australasian Road Safety Research, 458

Policing, Education Conference, Brisbane, Queensland. 459

460

Congiu, M, Whelan, M, Oxley, J, Charlton, J, D’Elia, A & Muir, C 2008, Child pedestrians: 461

Factors associated with ability to cross roads safely and development of a training package 462

(report no. 283), Monash University Accident Research Centre, Victoria, Australia. 463

464

Connelly, ML, Congalen, HM, Parsonon, BS & Isler, RB 1998, Child pedestrians’ crossing 465

gap thresholds, Accident Analysis and Prevention, vol. 30, no. 4, pp. 443-453. 466

467

Cooperative Research Centre (CRC) for Rail Innovation 2010, An Investigation of Risk 468

Takers at Railway Level Crossings, Project # R2.114, CRC. 469

470

Crandall, M., Zarzaur, B., & Tinkoff, G. (2013). American association for the surgery of 471

trauma prevention committee topical review: national trauma data bank, geographic 472

21

information systems, and teaching injury prevention. The American Journal of Surgery, 473

206(5), 709-73. 474

475

Daff, M & Cramphorn, B 2006, Signal design for pedestrians as if they were thinking adults, 476

Paper presented at the 7th International Conference on Walking and Liveable Communities, 477

Melbourne. 478

479

Davey, J, Wallace, A, Stenson, N & Freeman, J 2008, The experiences and perceptions of 480

heavy vehicle drivers and train drivers of dangers at railway level crossings, Accident 481

Analysis and Prevention, vol. 40, pp. 1217-1222. 482

483

Dunbar, G, Lewis, V & Hill, R 2001, Children’s attentional skills and road behaviour, Journal 484

of Experimental Psychology: Applied, vol. 7, no. 3, pp. 227-234. 485

486

Federal Railroad Administration 2008, A compilation of pedestrian safety devices in use at 487

grade crossings, Federal Railroad Administration, Washington D. C. 488

489

Federal Railroad Administration 2006, Railroad safety statistics: 2005 annual report, Federal 490

Railroad Administration, Washington D. C. 491

492

Freeman, J., Rakotonirainy, A., Stefanova, T., McMaster, M., 2013. Understanding pedestrian 493

behaviour at railway level crossings : is there a need for more research? Road and Transport 494

Research Journal 22(3), 29-39. 495

496

Freeman, J., & Watson, B. (2006). An application of stafford and warr's reconceptualization 497

of deterrence to a group of recidivist drink drivers. Accident Analysis and Prevention, 38(3), 498

462-71. 499

500

Henley, G & Harrison, JE 2009, Serious injury due to transport accidents involving a railway 501

train, Australia, 2002-03 to 2006-07, Australian Institute of Health and Welfare, Canberra 502

503

Hoyle, R., H., Fejfar, M.C., & Miller, J.D. (2000). Personality and sexual risk-taking: a 504

quantitative review. Journal of Personality, 68, 1203-1231. 505

506

22

Hoyle, R.H., Stephenson, M.T., Palmgreen, P., Lorch, E.P., Donohew, R.L. (2002). 507

Reliability and validity of a brief measure of sensation seeking. Personality and Individual 508

Differences, 32, 401 – 414. 509

510

Illinios Commerce Commission 2005, Pedestrian safety at rail grade crossings in north-511

eastern Illinois, Illinois Commerce Commission, Illinois. 512

513

Khattak, A & Luo, Z 2011, Pedestrian and bicyclist violations at highway-rail grade 514

crossings, Transportation Research Board, vol. 2250, pp. 76-82. 515

516

Lobb, B 2006, Trespassing on the tracks: a review of railway pedestrian safety research. 517

Journal of Safety Research, vol. 37, pp. 359-365. 518

519

Lobb, B, Harre, N & Suddendorf, T 2001, ‘An evaluation of a suburban railway pedestrian 520

crossing safety programme’, Accident Analysis and Prevention, vol. 33, pp. 157–65. 521

522

Lobb, B, Harre, N & Terry, N 2003, ‘An evaluation of four types of railway pedestrian 523

crossing safety interventions’, Accident Analysis and Prevention, vol. 35, no. 4, pp. 487–94. 524

525

Llyod’s Rail Register 2007, Study of pedestrian behaviour at public railway crossings. 526

Victoria, Australia: Public Transport Safety Victoria. 527

528

McPherson, CD & Daff, M 2004, Innovations in disability access at pedestrian rail crossing, 529

Paper presented at the 8th International level Crossing Symposium, Sheffield, United 530

Kingdom. 531

532

McPherson, CD & Daff, M 2005, Pedestrian behaviour and the design of accessible rail 533

crossings, Paper presented at the 28th Australasian Transport research Forum, Sydney, 534

Australia. 535

536

23

Martins, S., Storr, C., Alexandre, P., & Chilcoat, H. (2008). Adolescent ecstasy and other 537

drug use in the National Survey of Parents and Youth: the role of sensation-seeking, parental 538

monitoring and peer’s drug use. Addictive Behaviors, 33(7), 919-933. 539

540

Nelson, A 2008, Applying a partnership approach to level crossing risk: a strategic 541

opportunity, Paper presented at the 10th World Level Crossing Symposium: Safety Tresspass 542

and Prevention, Paris, France. 543

544

New Zealand Health Information Service 1999, Published accident statistics, Ministry of 545

Health, Wellington, New Zealand, Cited in Lobb, B, Harre, N & Terry, N 2003, An 546

evaluation of four types of railway pedestrian crossing safety interventions, Accident Analysis 547

and Prevention, vol. 35, no. 4, pp. 487-494. 548

549

Paternoster, R., & Piquero, A. (1995). Reconceptualizing deterrence: an empirical test of 550

personal and vicarious experiences. Journal of Research in Crime and Delinquency, 32(2), 551

251-286. 552

553

Richards, SH & Heathington, KW 1990, Assessment of warning time needs at railroad-554

highway grade crossings with active control devices and associated traffic laws, 555

Transportation Research Record, vol. 1160, pp. 52-59. 556

557

Schonfeld, C & Musumeci, I 2003, Pedestrian Travel: Getting Queenslanders Walking Safely, 558

Centre for Accident Research and Road Safety-Queensland: Brisbane. 559

560

Silla, A 2009, The effect of different countermeasures on railway trespassing, Young 561

Researchers Seminar, Torino, Italy 3-5 June. 562

563

Silla, A & Luoma, J 2009, ‘Trespassing on Finnish railways: identification of problem sites 564

and characteristics of trespassing behaviour’, European Transport Research Review, vol. 1, 565

no. 1, pp. 47–53. 566

24

567

Sochon, P 2008, Innovative and cooperative leadership to improve safety at Australian level 568

crossings, Paper presented at the 10th World Level Crossing Symposium: Safety and Trespass 569

Prevention, Paris, France. 570

571

Spicer, T 2008, Bentleigh, Victoria- Australia – a railway pedestrian crossing case study, 572

Paper presented at the 10th World Level Crossing Symposium: Safety and Trespass 573

Prevention, Paris, France. 574

575

576

Wallace, A 2008, Motorists behaviour at railway level crossings: the present context in 577

Australia, Unpublished PhD thesis, Queensland University of Technology. 578

579

Zuckerman, M. (1979a). Sensation seeking: beyond the optimal level of arousal. Hillsdale, 580

NJ: Erlbaum. 581

Zuckerman, M. (1979b). Sensation seeking and risk-taking behavior. In C.E. Izard, Emotions 582

in Personality and Psychopathology (pp.163-197). New York: Plenum. 583

584

Zuckerman, M. (1994). Behavioral Expression and Biosocial Bases of Sensation Seeking. 585

New York: Cambridge University Press. 586