Can simulator-based training improve street-crossing safety for elderly pedestrians?

Transportation Research Part F 15 (2012) 206–218

Contents lists available at SciVerse ScienceDirect

Transportation Research Part F journal homepage: www.elsevier.com/locate/trf

Can simulator-based training improve street-crossing safety for elderly pedestrians? Aurélie Dommes ⇑, Viola Cavallo IFSTTAR French Institute of Science and Technology for Transport, Development and Networks, Laboratory of Driver Psychology, Versailles, France

a r t i c l e

i n f o

Article history: Received 14 June 2011 Received in revised form 29 November 2011 Accepted 17 December 2011

Keywords: Pedestrian Street-crossing Ageing Safety Training Simulator

a b s t r a c t Older adults are known to be over-involved in pedestrian fatalities. Past research has shown that many of them make unsafe decisions when vehicles are approaching at a high speed and also miss many crossing opportunities when vehicle speed is low. They seem to have trouble taking the speed of approaching cars into account and predominantly base their crossing decisions on vehicle distance. A randomized controlled design was employed here to evaluate the effectiveness of combined behavioural and educational interventions on crossing decisions. Forty seniors aged 60 years and older were randomly assigned to an intervention group receiving a simulator-based street-crossing training programme (n = 20), or a control group (unrelated internet-use training course, n = 20). Baseline data and post-intervention data (immediately after, and 6 months after training) were collected, and included street-crossing decisions and behaviours on a simulated street-crossing task. Although the groups did not differ significantly from each other on the baseline measures, the results showed significant group differences immediately after training: interventiongroup participants crossed more rapidly, adopted larger safety margins, and had fewer tight fits than participants in the control group. However, 6 months after training, significant group differences were no longer observed: improvements in street crossing decisions and behaviours were apparent for both intervention and control groups. These results indicate a clear shift of the decision criterion towards more safety for all participants over time. However, the ability to take the oncoming car’s speed into account did not improve: on both post-intervention tests, both groups of participants still made more unsafe decisions when the car was approaching at a high speed and missed more crossing opportunities at a low speed. This finding may reflect age-related perceptual and cognitive difficulties that cannot be remedied by a behavioural or educational training method. Implications for further research and opportunities for enhanced training interventions are discussed. Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction Developing countermeasures for improving the safety of street crossing among older adults is becoming a crucial problem. Walking is essential to the mobility of elderly road users, not only for carrying out daily living tasks, but also for social interaction and physical exercise. However, pedestrians, particularly senior pedestrians, are vulnerable road users and at high risk of death or serious injury. International crash statistics indicate that elderly pedestrians are an extremely vulnerable road-user group (NHTSA, 2001; OECD, 2001). In France, more than half of all pedestrians killed on the road (51%) are over 65 years old, whereas this age group represents less than 15% of the population (ONISR, 2006). ⇑ Corresponding author. Address: IFSTTAR-LPC, 25 allée des Marronniers, Satory, 78000 Versailles, France. Tel.: +33 (0)1 30 84 39 43; fax: +33 (0)1 30 84 40 01. E-mail address: [email protected] (A. Dommes). 1369-8478/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.trf.2011.12.004

A. Dommes, V. Cavallo / Transportation Research Part F 15 (2012) 206–218

207

Previous research (Holland & Hill, 2010; Lobjois & Cavallo, 2007, 2009; Oxley, Fildes, Ihsen, Charlton, & Day, 1997; Oxley, Ihsen, Fildes, Charlton, & Day, 2005) has identified several age-related changes in road-crossing behaviour that may contribute to this over-representation in pedestrian accidents, such as slowing of decision-making, decreased walking speed, and difficulty in selecting safe gaps and adopting sufficient safety margins. These problems appear to be particularly marked in complex environments such as on two-way roads (cf. Holland & Hill, 2010; Oxley et al., 1997) and/or when the approaching vehicle’s speed was high (cf. Lobjois & Cavallo, 2007, 2009; Oxley et al., 2005). Recent research (Dommes & Cavallo, 2011; Lobjois & Cavallo, 2007, 2009) also revealed that while younger pedestrians chose constant time gaps, older people accepted shorter and shorter time gaps as speed increased. Instead of basing their decision on time gap, as young pedestrians appear to do, seniors preferentially seem to use more simple heuristics based primarily on the distance of the approaching vehicle. For a given available time gap, the distance of the approaching vehicle is actually greater at high than at low speeds. Considering the greater vehicle distance, older people more often decide that it is safe to cross, walk more slowly, and adopt shorter safety margins when the speed of the approaching vehicle is high than when it is low. In this situation, the use of a distance-based heuristic is very dangerous because the time available for crossing is overestimated. As a corollary, shorter car distances (which are also associated with lower approach speeds) lead older pedestrians to decide not to cross and to miss safe crossing opportunities. Older adults’ difficulty taking the speed of approaching cars into account and their tendency to predominantly base their crossing decisions on vehicle distance could be linked to their poorer performance not only in estimating the time-to-arrival of approaching objects (Andersen & Enriquez, 2006) or approaching cars (Schiff, Oldak, & Shah, 1992; Scialfa, Guzy, Leibowitz, Garvey, & Tyrell, 1991), but also in detecting collisions with an obstacle (DeLucia, Bleckley, Meyer, & Bush, 2003). The decline in motion perception, especially the ability to discriminate different speeds and to perceive slow angular movements (Snowden & Kavanagh, 2006) may be one of the reasons why seniors disregard speed information. This sensitivity decline is particularly handicapping whenever a car is approaching at a fast speed, since the vehicle is far away and generates a very slow visual movement (Lobjois & Cavallo, 2007). While high speed increases the odds of a collision for older pedestrians, who have difficulty taking into account an oncoming car’s movement in the allotted time, high speed is also a risk factor that increases the consequences of an injury for this vulnerable population. In addition the issue of countermeasures centred on speed limits and road design in urban environments, a question that arises is whether training could be an effective means of helping older pedestrians make safer crossing decisions. The promotion of safe behaviours amongst older road users has generally been studied in the field of driving. Several studies have focused on the potential for training among older drivers (for a review, see Korner-Bitensky, Kua, von Zweck, & Van Benthem, 2009), and have assessed the effectiveness of different kinds of intervention (i.e., educational programmes, see e.g., Owsley, McGwin, Phillips, McNeal, & Stalvey, 2004; physical activity, see e.g., Ostrow, Shaffron, & McPherson, 1992; cognitive training, see e.g., Roenker, Cissel, Ball, Wadley, & Edwards, 2003; and mixed intervention, see e.g., McCoy, Tarawneh, Bishu, Ashman, & Foster, 1993) on crash rates or driving skills. Educational interventions appear to increase selfregulation and avoidance of challenging situations among older drivers, but not to reduce collisions (Owsley et al., 2004). In contrast, training programmes targeting specific abilities (e.g. physical, cognitive) have been shown to improve driving skills in older adults (Ostrow et al., 1992; Roenker et al., 2003), although they did not assess whether these improved driving skills resulted in lower crash rates. To our knowledge, the only interventions targeting older pedestrians that have been developed to date are knowledgebased, providing seniors with information to cope with hazards, through video campaigns, booklets, etc. Greater and greater numbers of local and governmental initiatives are now being implemented, but most of them have not been evaluated or published. The few existing information programmes were shown to improve knowledge but there was no evidence of a safety benefit (see Dunbar, Holland, & Maylor, 2004). Many more interventions were however aimed at child pedestrians (e.g. Thomson et al., 2005). To enhance child pedestrian skill learning and reduce street-crossing crashes, educational programmes have been used in classroom teaching (e.g., Cross & Pitkethly, 1988) or given by parents (e.g., Limbourg & Gerber, 1981). While these prevention strategies allow children to acquire knowledge of traffic at a conceptual level, many studies have reported limited success: knowledge-enhancement approaches have failed, or have done little to improve children’s know-how on roads, which requires practice in traffic situations (Duperrex, Bunn, & Roberts, 2002; Rothengatter, 1981; Zeedyk, Wallace, Carcary, Jones, & Larter, 2001). In contrast, other methods have address pedestrian skills directly by training children in real traffic environments (e.g., Young & Lee, 1987) or in computer-simulated ones (e.g., Oxley, Congiu, Whelan, D’Elio & Charlton, 2008; Thomson et al., 2005; Tolmie et al., 2005). The results indicate significant improvement which, in most cases, is long-lasting. The method developed by Thomson et al. (2005) appears to be particularly effective: it tackles more than conceptual and strategic issues involved in safe street crossing, since it includes behavioural training and proposes practice exercises in a computer-simulated traffic environment. In their study, feedback was provided in two ways. The first kind of feedback was delivered by the training software and informed the children about the safety of their decisions (e.g. a simulated collision in case of a risky decision). The second kind involved discussions between children (i.e. peer discussion, Tolmie et al., 2005) or with a trainer (e.g. mothers of the children, Thomson et al., 2005). Simulators and virtual reality techniques have proven to be powerful training devices in preventing child pedestrian injury (Schwebel, Gaines, & Severson, 2008), rehabilitating the brain-damaged (e.g. Weiss, Naveh, & Katz, 2003) and teaching basic driving skills (e.g., De Winter, Wieringa, Kuipers, Mulder, & Mulder, 2007; Kappé & Emmerik, 2005). Simulators provide feedback, allow for graduated levels of task difficulty, and make it possible to adapt the training to each individual’s

208

A. Dommes, V. Cavallo / Transportation Research Part F 15 (2012) 206–218

abilities, in such a way that effective and personalised skill learning or relearning can be achieved (Weiss et al., 2003). The use of simulators as training devices for older drivers has been recently tested by Roenker et al. (2003) and Romoser and Fisher (2009). The simulator-based intervention proposed by Roenker et al. (2003) involved two educational sessions about prevention behaviours and general road rules, with collective discussions and demonstrations on the simulator. The simulator-trained group was found to improve temporarily on the specific driving manoeuvres presented in the educational training sessions. These improvements were not maintained 18 months later. In contrast, Romoser and Fisher (2009) used a more active and immersive kind of simulator-based training, through repeated individualised practice and personalised feedback. This method was shown to be an effective means for immediately increasing older drivers’ performance and skills, especially visual scanning at intersections. Further research is needed to evaluate whether such improvements are longlasting. Surprisingly, little research has evaluated whether a programme combining behavioural and educational interventions in a safe and simulated traffic environment can improve the older adults’ decisions to cross. The present study was aimed at contributing new knowledge to this issue. The aim was to assess the effectiveness of a training method specifically designed to rehabilitate the behavioural and decisional components of street-crossing by providing repeated practice on a simulator, personalised feedback, and educational discussions.

2. Method 2.1. Participants A total of 40 seniors participated in the study and met the inclusion/exclusion criteria. The criteria were as follows: (1) being aged 60 years or over; (2) being fluent in French; (3) being retired; and (4) no suffering from cognitive impairment, as defined by a Mini-Mental State Examination (Folstein, Folstein, & McHugh, 1975) score of 27 or higher. They all underwent a medical examination (which included sight and hearing acuity and current medication) to also ensure study eligibility (i.e., absence of major cardiac, neurological, and visual disorders or diseases). Moreover, participants took a battery of functional tests designed to assess visual, attentional and motor abilities. The UFOV test (Ball & Owsley, 1993) and measures of walking speed were included for instance.1 The 40 participants were paid for their participation and signed an informed consent form before beginning the experiment. The institutional ethics committee approved the study. The 40 participants selected for the study were between the ages of 61 and 83 years (M = 72.2; SD = 5.29). They were divided into two age groups: 21 younger-old participants (12 women, 9 men) ranging in age from 61 to 71 (M = 68.1, SD = 2.7) and 19 older-old participants (11 women, 8 men) ranging in age from 72 to 83 (M = 76.7, SD = 3.5). Based on their age and sex, the 40 participants were randomly assigned to either the intervention or control group. Twenty seniors were assigned to the intervention group (receiving the street-crossing training programme): 11 women and 9 men ranging in age from 65 to 83 years (M = 73.05 years, SD = 4.44; 10 participants from the younger-old age group and 10 from the older-old age group). Twenty other seniors comprised the control group (12 women and 8 men ranging in age from 61 to 82; M = 71.35 years; SD = 5.78; 11 participants from the younger-old age group and 9 from the older-old age group). Instead of the street-crossing training, the control group participants were given an unrelated internet-use training course. The two groups did not differ significantly from each other on age (F(1, 38) = 1.03, p = .32), MMSE scores (F(1, 38) = 2.62, p = .12), walking speed (F(1, 38) = 0.81, p = .38) or UFOV measures (F(1, 38) = 0.83, p = .37). The sex ratio was similar (60% women in the control group vs 55% in the intervention group; p = .75). 2.2. Procedure Both groups participated in four sessions. First, they completed a 1-h pre-test session to assess their baseline street-crossing behaviours using a simulated road-crossing environment. Approximately one week later, participants in the intervention group completed two 1.5-h sessions of street-crossing training, and participants in the control group completed two 1.5-h sessions of internet-use training. The training sessions were separated by approximately 1 week. One week after the training, participants took a 1-h post-test to assess their street-crossing behaviours (immediate post-test). Six months later, their street-crossing behaviours were retested (6-month post-test). 2.3. Experimental setup A street-crossing simulation device was used to evaluate street-crossing behaviours in the three test sessions as well as to train the intervention group. The interactive street-crossing simulator not only allowed for safe street crossing and a perfect control over the characteristics of the traffic, but also produced a high immersion; pedestrians walked an actual distance of four meters every time they decided to cross in the virtual traffic.

1

Details of the outcomes of these assessments and their relationship to road-crossing safety are reported elsewhere (Dommes & Cavallo, 2011).

A. Dommes, V. Cavallo / Transportation Research Part F 15 (2012) 206–218

209

The device was based on the INRETS simulation tools (Espié, 1999) adapted to the street-crossing situation (Cavallo, Lobjois, & Vienne, 2006). The device included a portion of experimental road (4.2-m wide, materialised on the ground), an imagegeneration system, three-screen projection (2.70  1.90 m), a 3D sound-rendition system, and a recording system. The setup provided the participant with a horizontal visual field between 90° (at the starting point) and 140° (in the middle). The vertical visual field was 40°. The images (30 Hz refresh rate) were calculated and projected at the participant’s eye height. Scenes were updated interactively the participant’s motion (indicating a crossing decision) which was measured by a movement-tracking system using a cable attached to her/his waist. The visual scenes represented a one-way street 4.20 m wide sidewalk-to-sidewalk. Traffic consisted of a motorcycle followed by two identical cars moving at a constant speed from left to right (with respect to the participant standing on the sidewalk). 2.4. Pre- and post-test assessments During the pre- and post-test sessions, participants performed a simulated street-crossing task. They were tested individually. On each trial, participants were asked to stand at the edge of the sidewalk, facing the experimental road, to look left at the visual scene, paying attention to the approaching vehicles, and decide whether or not to cross between the two cars. If they thought there was not enough time, they were to remain standing still on the sidewalk. If they thought it was safe to cross (without needing to run), they had to walk (at any pace) over to the sidewalk on the other side of the street. The participant’s crossing decisions and motion were recorded on each trial. At the beginning of each scene, the motorcycle was 1.5 s away from the pedestrian and the first car was 1 s away from the motorcycle. The time gap between the two cars (1 –7 s, in 1-s increments) and the vehicle speed (30, 40, 50, 60, and 70 km/h) were varied. The number of trial repetitions per time gap differed according to their probability of being considered acceptable for crossing, as it has previously been shown that the shortest gaps are always refused and the longest gaps are always accepted (e.g., Lobjois & Cavallo, 2007). For this reason, the 1- and 7-s time gaps were presented once only, the 2- and 6-s time gaps twice, and the 3-, 4-, and 5-s time gaps, three times. This combination of 15 trials and 5 speeds resulted in a total of 75 trials. The 75 trials were randomized and divided into 2 blocks, with a break for the participants between the blocks. All trials were initiated by the experimenter. The complete experimental pre- and post-test sessions lasted approximately 30– 45 min each. In the pre-test session, the experimenter first presented the basic principles of the street-crossing simulator. Then the participant performed a maximum of 18 practice trials (at vehicle speeds of 30, 50, and 70 km/h and time gaps of 1, 4, and 7 s). The practice trials were stopped when the participant was comfortable and fully understood the task. 2.5. Street-crossing training The training method promoted individual sensory-motor practice on the simulator. Implicit and explicit behavioural feedback on street-crossing safety was provided. As speed of the approaching car was shown to be ignored by senior pedestrians (e.g., Lobjois & Cavallo, 2007), the training chose to also address the strategies older adults brought into play when deciding to cross. Seniors were encouraged to assess the approaching car’s speed before crossing, rather than considering its distance only. They were also encouraged to observe the approaching traffic as they crossed and to increase their pace if the car was arriving faster than expected. In sum, by means of repeated practice and a better understanding of the task constraints, the training programme was aimed at improving the safety of older pedestrians and helping them better take the approaching car’s speed into account. The first training session began with a discussion (approximately 30 min) about what information should be taken into account in order to cross the street safely and what safety-conscious behaviours should be adopted. The importance of speed information and the consequences of neglecting it were also explained. The participant was then asked to complete three street-crossing training modules. The presentation order of the three modules was counterbalanced over the two training sessions. The modules presented the same visual scenes as those used in the pre- and post-test sessions. The tasks were also identical. Vehicle speed (Module A: 30 vs 50 km/h; Module B: 40 vs 60 km/h; Module C: 50 vs 70 km/h) and time gap (1–7 s) were varied, making for a total of 42 randomly presented trials per module. Each module was repeated immediately, so the participant completed each module twice in a row. In all, 252 training trials were proposed. The experimenter gave the participant two kinds of feedback. The first concerned the safety margin, which was computed online for each trial and indicated to the participant (cf. data analysis section). If the participant’s safety margin was above 1.5 s (criterion set on the basis of Simpson, Johnston, & Richardson, 2003), the crossing was scored as safe. The experimenter then initiated the next trial. If the safety margin was below 1.5 s, the experimenter informed the participant and they talked together about what made this behaviour risky (time gap too short, initiation time too long, high speed of approaching car, etc.). If the participant appeared to fail to understand, the trial was repeated. The second kind of feedback pertained to the median accepted time gap. For each speed of the approaching cars, the participant’s median accepted time gap was computed via a logistic function (see data analysis section) and presented in a chart. At the end of each training module, the

210

A. Dommes, V. Cavallo / Transportation Research Part F 15 (2012) 206–218

experimenter and the participant examined the effect of the approaching car’s speed on her/his median accepted time gap. If the participant’s decisions exhibited a speed effect (i.e., the median accepted time gap decreased as speed increased), the experimenter stressed the importance of paying better attention to the speed of the approaching car before deciding whether or not to cross the street. 2.6. Internet training Participants in the control group were given computer and internet-use training so that they could experience the same amount of social contact and have the same number of experimental sessions as the intervention group. The control group received an introduction to the computer and instruction in how to use and access websites. Participants practiced searching for information individually by doing exercises on topics that interested them (e.g., painting, health, etc.). 2.7. Data analysis Trials were scored as to whether the participant did or did not accept the available gap for crossing the street. The participant’s motion (distance travelled over time) was also recorded whenever s/he decided to cross. Eight key measures were derived from these data, as follows:  Median accepted time gap (s). For each speed, the median time gap between vehicles accepted by the participant was computed using a logistic regression analysis on the row data, i.e. the percentage of crossings accepted by the participant at each time gap. The following logistic function was used to determine the transition point between the decision not to cross and the decision to cross (argument a of the function): F(v) = 1/(1 + e ((v a)/b)); where v is the time gap and b the slope of the logistic curve at point a.  Initiation time (s). Initiation Time (IT) was computed for each participant and each accepted crossing from the pedestrian’s position on the experimental road and the respective position of the first car on the virtual road. IT was equal to the time between when the rear end of the first car passed in front of the participant and when s/he started to walk; IT was negative if the participant started to cross before the rear end of the first car passed her/his crossing line.  Crossing time (s). Crossing Time (CT) was equal to the time taken from when the participant started to walk and when s/he completed the crossing and crossed the curb on the opposite sidewalk.  Safety margin (s). Safety Margin (SM) was computed for each participant and each accepted crossing from the pedestrian’s position on the experimental road and the respective position of the second car on the virtual road. SM was measured as the time between when the participant reached the opposite sidewalk and when the front end of the second car reached the crossing line; SM was negative if the participant was still on the road when the front of the second car passed the crossing line.  Collisions (%). A crossing was scored as a collision when the participant was virtually hit by the approaching car: s/he was on the path of the approaching car (width of the front end of the car) when it passed the crossing line. This variable was expressed as percentage of the total number of crossings accepted by the participant.  Unsafe decisions (%). A crossing was scored as an unsafe decision when SM was negative: the participant was not hit but was still on the road when the second car passed the crossing line (s/he did not reach the opposite sidewalk yet, but was not on the path of the approaching car). This variable was expressed as percentage of the total number of crossings accepted by the participant.  Tight fits (%). A tight fit occurred when SM was between 0 and 1.5 s (the participant had reached the opposite sidewalk when the approaching car passed the crossing line). The number of tight fits was divided by the total number of crossings accepted by the participant.  Missed opportunities (%). A missed opportunity was counted when the participant decided not to cross, even though based on her/his mean CT and IT, there would have been enough time to cross safely (i.e., SM would have been above 1.5 s). Missed opportunities were expressed as a percentage of the total number of crossings refused by the participant. A multivariate analyses of variance (MANOVA) was implemented with group (intervention, control) as a betweengroup factor, and with testing point (pre-test, immediate post-test, and 6-month post-test) and speed of the approaching car as within-group factors. Because of a moderately high degree of correlation between the dependent variables, MANOVA was a more appropriate analysis than independent ANOVAs (Tabachnick & Fidell, 1989). The Wilks’ Lambda multivariate statistic (K) was used. Wilks’ lambda ranges from 0 to 1; the lower it is the more the given effect contributes to the model. The significance level was set at .05. Effects found to be significant in the MANOVA were evaluated against each of the dependent measures, using univariate analyses. The computation of relative effect size (g2p ), the measure of observed power (x2), and post-hoc comparisons (Tukey’s post-hoc tests) completed the analyses.

A. Dommes, V. Cavallo / Transportation Research Part F 15 (2012) 206–218

211

Table 1 Statistical results. Effects

Multivariate tests

Univariate tests

Group

Nonsignificant

–

Testing point

K = .14

Median gap: F(2,76) = 9.7, p < .01, g2p = .20, x2 = .98

F(14,25) = 11, p<.01, g2p = .86, x2 = 1

Initiation time: F(2,76) = 6.1, p < .01, g2p = .14, x2 = .87 Crossing time: F(2,76) = 28.1, p < .01, g2p = .43, x2 = 1 Safety margin: F(2,76) = 68.8, p < .01, g2p = .64, x2 = 1 Unsafe decisions: F(2,76) = 43.8, p < .01, g2p = .54, x2 = 1 Tight fits: F(2,76) = 31.5, p < .01, g2p = .45, x2 = 1 Missed opportunities: F(2,76) = 10.6, p < .01, g2p = .22, x2 = .99

Speed

K = .01

Median gap: F(4,152) = 185, p < .01, g2p = .83, x2 = 1

F(28,11) = 57.9, p<.01, g2p = .99, x2 = .1

Initiation time: F(4,152) = 497.5, p < .01, g2p = .93, x2 = 1 Crossing time: F(4,152) = 14.6, p < .01, g2p = .28, x2 = 1 Safety margin: F(4,152) = 421.1, p < .01, g2p = .92, x2 = 1 Unsafe decisions: F(4,152) = 55, p < .01, g2p = .59, x2 = 1 Tight fits: F(4,152) = 118, p < .01, g2p = .76, x2 = 1 Missed opportunities: F(4,152) = 60.7, p < .01, g2p = .62, x2 = 1

Group  Testing point

Crossing time: F(2,76) = 4.2, p < .05, g2p = .10, x2 = .72

K = .39 2 p

2

F(14,25) = 2.7, p<.05, g = .60, x = .93

Safety margin: F(2,76) = 16.8, p < .01, g2p = .31, x2 = 1 Tight fits: F(2,76) = 9.2, p < .01, g2p = .20, x2 = .97 Missed opportunities: F(2,76) = 4.9, p < .01, g2p = .12, x2 = .79

Group  Speed

Nonsignificant

–

Testing point  Speed

Could not be calculateda

Median gap: F(8,304) = 2.6, p<.05, g2p = .06, x2 = .91 Safety margin: F(8,304) = 2.5, p<.05, g2p = .06, x2 = .91 Unsafe decisions: F(8,304) = 15.8, p<.01, g2p = .29, x2 = 1 Tight fits: F(8,304) = 5.9, p<.01, g2p = .14, x2 = 1 Missed opportunities: F(8,304) = 8, p<.01, g2p = .17, x2 = 1

Group  Testing point  Speed

Could not be calculatedb

Unsafe decisions: F(8,304) = 2.5, p<.05, g2p = . 06, x2 = .91 Missed opportunities: F(8,304) = 3.4, p<.05, g2p = .08, x2 = .98

Note: non significant results are not reported in this Table. a The error degrees of freedom were too low considering the number of dependent variables (7), modalities of the repeated factors (3 testing points  5 speeds) and subjects (40). b The error degrees of freedom were too low considering the number of dependent variables (7), modalities of the repeated factors (2 groups  3 testing points  5 speeds) and subjects (20).

3. Results The MANOVA included seven of the eight measures of the street-crossing behaviour, i.e. median accepted time gap, initiation time, crossing time, safety margin, unsafe decisions, tight fits, and missed opportunities. Very few collisions occurred (10 of the 9000 total trials), therefore these data were excluded from the statistical analyses. Statistical results of the analyses are presented in Table 1. 3.1. Main differences between groups, testing points and speeds of the approaching cars The results yielded a nonsignificant multivariate main effect of group. The multivariate main effect of testing point was significant (see Table 1). Univariate analyses indicated that all dependent measures contributed to the multivariate main effect (see Table 1). Post-hoc analyses revealed significant changes between the pre-test and the immediate post-test that were long lasting (p’s < .05): no significant difference appeared between the immediate and 6-month post-tests (with the exception of initiation time). The multivariate main effect of speed was significant. Table 1 shows that all seven dependent measures contributed to the multivariate main effect. Post-hoc analyses revealed that when the speed of the approaching vehicle increased, participants accepted smaller time gaps, started walking later, crossed more slowly, and adopted smaller safety margins (p’s < .05). Tables 2 and 3 present the means and standard deviations of these four behavioural measures. Participants made also more unsafe decisions (see Fig. 2) and more tight fits, and missed fewer opportunities to cross as the car’s speed increased (p’s < .05).

212

A. Dommes, V. Cavallo / Transportation Research Part F 15 (2012) 206–218

Table 2 Mean (and standard deviation) of median accepted time gaps (in seconds). Group

Test

Intervention

Pre-test

3.9

Immediate post-test

(0.9)

6-Month post-test

Control

Pre-test

3.5

Immediate post-test

(0.9)

6-Month post-test

Speed (km/h) 30 4.6 (0.9)

40 4 (0.9)

50 3.6 (0.7)

60 3.3 (0.7)

70 2.9 (0.5)

Total

4.6 (1) 5.2 (0.6) 4.9 (0.5)

4 (0.8) 4.6 (0.8) 4.4 (0.6)

3.5 (0.7) 4.1 (0.3) 3.7 (0.4)

3.2 (0.6) 3.6 (0.6) 3.6 (0.5)

2.9 (0.5) 3.1 (0.3) 3.1 (0.4)

3.6 (0.9) 4.1 (0.9) 3.9 (0.9)

4.2 (1.1) 4.5 (0.9) 4.3 (1.1)

3.5 (0.9) 3.7 (1) 4 (1.1)

3.3 (0.6) 3.4 (0.6) 3.5 (0.9)

3 (0.7) 3.2 (0.6) 3.1 (0.8)

2.8 (0.5) 2.8 (0.5) 2.9 (0.9)

3.4 (0.9) 3.5 (0.9) 3.6 (1.1)

Table 3 Mean (and standard deviation) of initiation time (IT), crossing time (CT), and safety margin (SM), in seconds. Group

Test

Speed (km/h) 30

IT Intervention

Pre-test Post-test 6 Months

Control

Pre-test Post-test 6 Months

CT Intervention

Pre-test Post-test 6 Months

Control

Pre-test Post-test 6 Months

SM Intervention

Pre-test Post-test 6 Months

Control

Pre-test Post-test 6 Months

40

50

60

70

Total

1.5 (0.4) 1.6 (0.4) 1.4 (0.4)

1.2 (0.4) 1.3 (0.4) 1.2 (0.4)

1 (0.4) 1.2 (0.4) 1 (0.3)

0.9 (0.4) 1.1 (0.4) 0.9 (0.4)

0.8 (0.4) 1 (0.4) 0.8 (0.4)

1.1 (0.4) 1.2 (0.4) 1.1 (0.4)

1.6 (0.4) 1.6 (0.5) 1.5 (0.4)

1.3 (0.4) 1.4 (0.5) 1.2 (0.4)

1.1 (0.4) 1.2 (0.5) 1.1 (0.4)

1 (0.4) 1 (0.4) 0.9 (0.4)

0.9 (0.4) 0.9 (0.5) 0.8 (0.4)

1.2 (0.4) 1.2 (0.5) 1.1 (0.4)

4.7 (0.4) 4.2 (0.4) 4.3 (0.3)

4.6 (0.4) 4.2 (0.3) 4.3 (0.3)

4.7 (0.4) 4.3 (0.3) 4.3 (0.3)

4.7 (0.4) 4.3 (0.4) 4.4 (0.3)

4.7 (0.4) 4.3 (0.4) 4.4 (0.3)

4.7 (0.4) 4.3 (0.4) 4.3 (0.3)

4.6 (0.5) 4.4 (0.4) 4.2 (0.4)

4.5 (0.4) 4.4 (0.5) 4.2 (0.4)

4.6 (0.4) 4.4 (0.4) 4.3 (0.4)

4.6 (0.4) 4.5 (0.4) 4.3 (0.4)

4.7 (0.4) 4.5 (0.4) 4.3 (0.4)

4.6 (0.4) 4.4 (0.4) 4.3 (0.4)

2.4 (0.5) 3.4 (0.5) 3 (0.5)

1.8 (0.4) 2.7 (0.6) 2.3 (0.5)

1.4 (0.4) 2.3 (0.4) 1.9 (0.4)

1.1 (0.4) 1.9 (0.5) 1.6 (0.4)

0.8 (0.3) 1.6 (0.4) 1.3 (0.4)

1.5 (0.7) 2.4 (0.8) 2 (0.7)

2.4 (0.6) 2.8 (0.6) 2.8 (0.8)

1.8 (0.4) 2.2 (0.5) 2.3 (0.8)

1.5 (0.5) 1.8 (0.4) 1.9 (0.6)

1.2 (0.4) 1.4 (0.4) 1.6 (0.5)

1 (0.4) 1.2 (0.4) 1.3 (0.5)

1.6 (0.7) 1.9 (0.7) 2 (0.8)

A. Dommes, V. Cavallo / Transportation Research Part F 15 (2012) 206–218

213

3.2. Interaction effects The multivariate interaction between group and testing point was significant (see Table 1). Univariate analyses indicated that four of the seven dependent measures contributed to the multivariate main effect (see Table 1). Whereas the groups did

Fig. 1. (a) Mean safety margin (SM); (b) mean percentage of tight fits; and (c) mean percentage of missed opportunities by group and testing point. Vertical bars represent standard deviations.

214

A. Dommes, V. Cavallo / Transportation Research Part F 15 (2012) 206–218

not differ significantly from each other on baseline, post-hoc analyses revealed significant group differences on the immediate post-test, where the intervention group adopted larger safety margins, made fewer tight fits, and missed more crossing opportunities than the control group did (p’s < .05). See Fig. 1, which illustrates the mean safety margin, percentage of tight fits, and percentage of missed opportunities in each group and for each testing point. Contrary to the control group, which exhibited no significant changes, the intervention group exhibited significant crossing-time and tight-fit decreases between the pre-test and the immediate post-test sessions, as well as a significant safety-margin increase. Surprisingly, no significant group differences were observed on the 6-month post-test: the control group showed significant improvements that matched those exhibited by the intervention group. In contrast, some of the intervention group’s improvements on the immediate post-test declined on the 6-month post-test session (safety margin decreased significantly, and tight fits rose significantly between the immediate and 6-month post-tests, p’s < .05). But importantly, performance was better on the 6-month follow-up than on the baseline measure for both groups: all participants showed significant crossing-time and tight-fit decreases, as well as a significant safety-margin increase. While missed opportunities remained relatively stable over time in the control group, participants in the intervention group missed more crossing opportunities on the immediate post-test than on the pre-test, but this increase did not last until the 6-month follow-up (see Fig. 1). The multivariate interaction between group and speed was not significant. The multivariate interaction between testing point and speed could not be calculated (see Table 1). Univariate analyses yielded a significant interaction effect for five of the seven dependent variables (see Table 1). Post-hoc analyses showed that improvements and changes across testing points differed significantly according to the speed of the approaching car. The multivariate interaction between group, testing point and speed could also not be calculated (see Table 1). Univariate analyses indicated a significant triple interaction in the measures of unsafe decisions and missed opportunities (see Table 1). Both groups made significantly more unsafe decisions at 70 than at 30 km/h (p < .001) on baseline (pre-test), however, speed-related differences were no longer observed in the intervention group on the immediate post-test: they made as many unsafe decisions at 30 as at 70 km/h (see Fig. 2). In contrast, the control group still made significantly more unsafe decisions at 70 than at 30 km/h (p < .01) on the immediate post-test. However, the reduction in unsafe decisions at the highest speed in the intervention group did not last: on the 6-month post-test, both groups made significantly more unsafe decisions at 70 km/h than at 30 km/h (p < .05). Regarding missed opportunities, post-hoc tests showed a significant difference between groups on the immediate post-test only when the vehicle was approaching at 30 km/h (p < .01), where the intervention group missed more crossing opportunities than the control group.

Fig. 2. Mean percentage of unsafe decisions by group, speed, and testing point.

A. Dommes, V. Cavallo / Transportation Research Part F 15 (2012) 206–218

215

3.3. Regression toward the mean? A factor that could cast doubt on the internal validity of pre-test post-test designs is regression toward the mean (Dimitrov & Rumrill, 2003), i.e. ‘‘a statistical phenomenon that can make natural variation in repeated data look like real change’’ (Barnett, van der Pols, & Dobson, 2004, p. 215). To control for this effect, separate Ancovas on the change scores between testing sessions (short- and long-term changes) were carried out for each of the seven dependent variables, with the pre-test data (mean baseline scores) as covariates. The results of the Ancovas confirmed the effects uncovered by the multivariate and univariate analyses presented above. Thus, the changes observed in the present experiment do not seem to be linked to the statistical phenomenon of regression toward the mean. 4. Discussion This study was undertaken in order to assess the effectiveness of combined behavioural and educational interventions on improving crossing decisions amongst older adults. The results showed an immediate benefit of the training programme. The participants who benefitted from the intervention exhibited a marked improvement in the overall safety of their streetcrossing decisions and behaviours: they crossed more rapidly, adopted larger safety margins and had fewer tight fits than participants in the control group. These behavioural improvements were accompanied, however, by an increase in the number of missed opportunities. In contrast to the rapid progress made by the participants in the intervention group, participants in the control group showed no significant behavioural changes on the immediate post-test. Two possible explanations can be proposed. The first is that participants in the intervention group may have taken advantage of the explicit behavioural feedback and the educational information given to them during the training programme. The second is that they may have gained greater awareness of street-crossing dangers through task repetition alone: they performed 252 more crossing trials than participants in the control group. With task repetition, they may have made better use of the implicit continuous visual feedback available in the interactive street-crossing task: the simulator provided a perception–action coupling that enabled participants to adapt their actions to their visual perceptions. Although a study by Lobjois and Cavallo (2009) showed that elderly participants did not take advantage of the adjustment possibilities offered by an interactive simulation, the results of the present study are compatible with the idea that the training programme promoted better use of visual feedback. The fact of having attracted the participants’ attention to the availability of this information may have helped them better adjust their actions to what they perceived and make safer and safer decisions as they repeated the task. Furthermore, many more trials were carried out in the present training experiment. These findings raise a number of important questions. The training programme combined behavioural and educational interventions, and it is important to determine the respective roles of each of these components in the immediate improvements made by the intervention group. For example, was repeated practice alone effective in bringing about improvements in road-crossing behaviour? Repeated practice may involve both greater awareness of street-crossing dangers, and higher exposure to implicit visual feedback. Also, did explicit behavioural feedback (e.g., safety margins) account for the rapid improvements observed in the intervention group? Another question is whether educational intervention by itself can improve elderly pedestrians’ safety, or whether a combination of the three components is necessary (i.e., repeated practice, behavioural feedback, and educational intervention). Further studies are required to separate the respective effects of the three components of the programme tested here. While the immediate effects were encouraging, contrary to what we expected, significant differences between groups were no longer observed 6 months after training. Participants in the control group showed surprising improvements that were similar to those exhibited by the participants in the intervention group. Between the baseline and the long-term follow-up, participants in the control group crossed the experimental street faster, adopter larger safety margins and had fewer tight fits. More importantly, while participants in the control group showed significant improvements in the final post-test, participants in the intervention group exhibited slight declines in crossing performance between the two post-test sessions (i.e. safety margins decreased, and tight fits rose significantly). Consequently, both groups adopted just about as much as safe street-crossing decisions and behaviours on the 6-month post-test, suggesting that the training programme had no long-term effect. The improvements of the control group could be ascribed to enhanced awareness of street-crossing dangers acquired implicitly over task repetition. With repeated practice, participants in the control group could have benefitted from the implicit visual feedback, i.e. the perception–action coupling offered by the simulator. Indeed, both groups carried out the same crossing task at three testing sessions, which added up to 225 crossing trials per participant. The main group difference was that participants in the intervention group performed additional crossing trials while receiving both safety education and explicit behavioural feedback. These two specific components of the training programme seemed to act as boosters, thereby triggering an immediate behavioural modification. As no significant group differences appeared on the long-term post-test, that is, the third time participants performed the test, individual sensory-motor repeated practice on the simulator seemed to be effective alone in bringing about improvements in road-crossing safety. This finding gives an element of answer to the questions raised by the immediate improvements observed in the intervention group just after training. One of the most significant findings of this study is related to vehicle approach speed. While our results indicate an overall improvement in street-crossing behaviour throughout the experiment (between pre- and post-tests), participants of both

216

A. Dommes, V. Cavallo / Transportation Research Part F 15 (2012) 206–218

groups still failed to make better use of the approaching vehicle’s speed in their decisions and actions. A robust effect of speed was observed in both the intervention and control groups on pre- and post-test sessions: participants accepted shorter gaps, started to cross later, took more time to cross, and adopted smaller safety margins as speed increased. Fewer missed opportunities were found and more unsafe decisions and tight fits were observed at high speeds than at low speeds. The lower impact of speed on unsafe decisions in the intervention group immediately after training was accompanied by a sharp rise in missed opportunities at low speeds, which means that the training did not have any effect of improving processing of the oncoming vehicle’s speed. The lack of effectiveness of training in the use of speed suggests that age-related perceptual and cognitive impairments cannot be addressed by a simulator-based behavioural method. Indeed, a recent study by Dommes and Cavallo (2011) highlighted the role of processing speed and visual attention abilities (assessed via the Useful Field of View, UFOV test, see Ball, Beard, Roenker, Miller, & Griggs, 1988; Ball, Roenker, & Bruni, 1990) in the way pedestrians took or did not take information about vehicle speed into account in their decisions. In earlier studies (e.g., Lobjois & Cavallo, 2007), elderly pedestrians were shown to mainly use heuristics based on the distance of the approaching vehicle, rather than grounding their decisions on time gap, as younger adults did. The behaviour-based training method does not seem to have modified the perceptual strategies of elderly pedestrians or to have helped them better take into account the oncoming vehicle’s speed in their street-crossing decisions. Repeated practice on the simulator did not either. What we found in both groups, instead, was a shift of the decision criterion in the direction of more conservative judgments: the ‘‘critical distance’’ deemed unsafe for crossing seem to be raised. This change in criterion resulted in a substantial decrease in unsafe decisions at high speeds. In this respect, the simulator-based street-crossing practice led to an undeniable gain in safety. This kind of training method recently showed its effectiveness when comparing older pedestrians trained on a simulator with young individuals (Dommes, Cavallo, Vienne, & Aillerie, 2012): seniors improved their behaviour considerably to the point that their overall safety-related indicators were similar to those of younger individuals, even 6 months after training. Another promising methodology for improving elderly gap judgments about crossing the street was also tested recently by Hunt, Harper, and Lie (2011). Their training programme was aimed at teaching people to judge vehicle speed more accurately and at determining whether better speed estimations made participants use speed more effectively in their gap-acceptance decisions. Three techniques were tested and the most promising one required participants to judge whether the oncoming car was travelling at an atypical or normal speed (without feedback on response accuracy), and then to estimate the speed of the car. The results were encouraging. However, the experiment had no follow-up phase, the task did not include an actual crossing, and the extent to which speed influenced gap-acceptance decisions was not examined. Furthermore, no control group was involved. Future studies should address these questions before deciding if this method can successfully promote long-term improvements in seniors’ street-crossing decisions and behaviours. Although the present experiment provides some answers by examining whether elderly pedestrians can be sustainably trained using a simulator-based behavioural method, future studies are needed to confirm our interpretation. The number of participants was low and the age range high. Given the increasing number of seniors in our ageing societies, future studies will be able to test several age groups within the older-adult cohort (younger-old vs older-old), and find out to what extent men and women with similar/different mobility patterns can be trained. Holland and Hill (2010) recently highlighted these last factors as predictors of unsafe crossing decisions. Maybe these factors influence also the way seniors improve or not with training. The complexity of the traffic environment is also an issue to address in future training studies, because older pedestrians are known to be more vulnerable on two-way roads than in one-way traffic situations. 5. Conclusions The present study suggests that repeated practice on a street-crossing simulator is able to shift the decision criterion towards more conservative judgments, but fails to help older pedestrians make better use of the approaching vehicle’s speed in their decisions and actions. On the basis of current knowledge, several ways for improving elderly pedestrians safety can be proposed. The first one is speed management, which involves both measures targeting road infrastructures and introducing appropriate speed limits for areas where there is high pedestrian activity. Increasing the number of speed ramps and narrow streets, as well as pedestrian-only zones and 30-km/h zones would considerably improve senior safety. Given the slow walking speed of seniors, car-free islands should be set up in the middle of two-way roads so that the elderly can cross in two stages. This would not only reduce their risk of collision by decreasing the time spent in the street, but would also lighten the cognitive load of the street-crossing task. As a supplement to the above safety regulations and ergonomic measures, cognitive training programmes can be a promising and innovative measure worth exploring further. As in Roenker et al.’s (2003) programme for elderly drivers, speed of processing and visual attention training (i.e. UFOV training) could be a successful method for specifically improving the way older pedestrians process information, and more particularly, for helping them better take the oncoming car’s speed into account. The effectiveness of information campaigns should also be considered and objectively evaluated. Providing education about functional declines and information about the role of self-awareness (see Holland & Rabbitt, 1992) and self-identity

A. Dommes, V. Cavallo / Transportation Research Part F 15 (2012) 206–218

217

(see Holland, Hill, & Cooke, 2009) in crossing behaviour could be an effective means for helping older pedestrians adopt selfregulatory practices (for example, avoiding high-speed roads). Acknowledgments This research was supported by grants from the MAIF Foundation. The authors are grateful to the older participants for their participation, interest, and cooperation. We would also like to thank Dr. Elsa Rautou for the medical examinations, Fatma Boustelitane and Denis Courtot for their help in running the experiment, Roland Donat (IFSTTAR – Laboratory of New Technologies) for his assistance with the data processing, Fabrice Vienne, Isabelle Aillerie, and Stéphane Caro (IFSTTAR – Road Operations, Perception, Simulators and Simulations Research Unit) for their help in designing and setting up the experimental device, and Claude Perrot and Jean-Louis Mondet (Laboratory of Driver Psychology) for their technical assistance. The authors are also grateful to Vivian Waltz for editing and translating this article into English. We extend special thanks to Jennifer Oxley (MUARC, Malaysia) for her helpful comments. References Andersen, G. J., & Enriquez, A. (2006). Aging and the detection of observer and moving object collisions. Psychology and Aging, 21, 74–85. Ball, K. K., Beard, B., Roenker, D. L., Miller, R., & Griggs, D. (1988). Age and visual search: Expanding the useful field of view. Journal of the Optical Society A, 5, 2210–2219. Ball, K. K., & Owsley, C. (1993). The useful field of view test: A new technique for evaluating aged-related declines in visual functions. Journal of the American Optometric Association, 64, 71–79. Ball, K. K., Roenker, D. L., & Bruni, J. (1990). Developmental changes in attention and visual search throughout adulthood. In J. Enns (Ed.), The development of attention: Research and theory (pp. 489–508). Amsterdam: Elsevier Science. Barnett, A. G., van der Pols, J. C., & Dobson, A. J. (2004). Regression to the mean: What it is and how to deal with it. International Journal of Epidemiology, 34, 215–220. Cavallo, V., Lobjois, R., & Vienne, F. (2006). The interest of an interactive road crossing simulation for the study of adaptive road crossing behavior. In Proceedings of the first driving simulation conference Asia-Pacific, Tsukuba, Japan. Cross, R. T., & Pitkethly, A. (1988). Speed, education and children as pedestrians: A cognitive change approach to a potentially dangerous naïve concept. International Journal of Science Education, 10(5), 531–540. De Winter, J. C. F., Wieringa, P. A., Kuipers, J., Mulder, J. A., & Mulder, M. (2007). Violations and errors during simulation-based driver training. Ergonomics, 50(1), 138–158. DeLucia, P. R., Bleckley, M. K., Meyer, L. E., & Bush, J. M. (2003). Judgments about collision in younger and older drivers. Transportation Research Part F, 6, 63–80. Dimitrov, D. M., & Rumrill, P. D. Jr., (2003). Pretest–postest designs and measurement of change. Work, 20, 159–165. Dommes, A., & Cavallo, V. (2011). The role of perceptual, cognitive, and motor abilities in street-crossing decisions of young and older pedestrians. Ophthalmic and Physiological Optics, 31(3), 292–301. 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, 42–47. Dunbar, G., Holland, C. A., & Maylor, E. A. (2004). Older pedestrians: A critical review of the literature. Road Safety Research Report No. 37. DfT, London. Duperrex, O., Bunn, F., & Roberts, I. (2002). Safety education of pedestrians for injury prevention: A systematic review of randomised controlled trials. British Medical Journal, 324, 1129–1131. Espié, S. (1999). Vehicle-driven simulator versus traffic-driven simulator: The INRETS approach. In Proceedings of the 1st driving simulation conference Europe, Paris, France (pp. 367–376). Folstein, M., Folstein, S., & McHugh, P. (1975). Mini-mental state. A practical method for grading the cognitive state of patients for the clinician. Journal of Psychiatric Research, 12(3), 189–198. Holland, C. A., & Hill, R. (2010). Gender differences in factors predicting unsafe crossing decisions in adult pedestrians across the lifespan: A simulation study. Accident Analysis and Prevention, 42(4), 1097–1106. Holland, C. A., Hill, R., & Cooke, R. (2009). Understanding the role of self-identity in habitual risky behaviours: Pedestrian road-crossing decisions across the lifespan. Health Education Research, 24(4), 674–685. Holland, C. A., & Rabbitt, P. M. A. (1992). People’s awareness of their age-related sensory and cognitive deficits and the implications for road safety. Applied Cognitive Psychology, 6, 217–231. Hunt, M., Harper, D. N., & Lie, C. (2011). Mind the gap: Training road users to use speed and distance when making gap-acceptance decisions. Accident Analysis and Prevention, 43, 2015–2023. Kappé, B., & Emmerik, M. L. (2005). The use of driving simulators for initial driver training and testing. Report No 75151.01. The Hague: TNO Defence, Security and Safety. Korner-Bitensky, N., Kua, A., von Zweck, C., & Van Benthem, K. (2009). Older driver retraining: An updated systematic review of evidence of effectiveness. Journal of Safety Research, 40, 105–111. Limbourg, M., & Gerber, D. A. (1981). A parent training program for the road safety education of preschool children. Accident Analysis and Prevention, 13(3), 255–267. 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(5), 934–943. Lobjois, R., & Cavallo, V. (2009). The effects of aging on street-crossing behavior: From estimation to actual crossing. Accident Analysis and Prevention, 41(2), 259–267. McCoy, P. T., Tarawneh, M. S., Bishu, R. R., Ashman, R. D., & Foster, B. G. (1993). Evaluation of countermeasures for improving driving performance of older drivers. Transportation Research Record, 1405, 72–80. National Highway Traffic Safety Administration (NHTSA) (2001). Traffic Safety Facts 2000: Pedestrians (Report No. DOT-HS-809311). US Department of Transport, Washington, DC. Organisation for Economic Co-operation Development (OECD) (2001). Aging and transport. Mobility needs and safety issues. Paris: OECD Publications. Observatoire National Interministériel de Sécurité Routière (ONISR) (2006). Grand thèmes de la sécurité routière en France: Piétons. La Documentation Française, Paris. Ostrow, A. C., Shaffron, P., & McPherson, K. (1992). The effects of a joint range-of-motion physical fitness training program on the automobile driving skills of older adults. Journal of Safety Research, 23, 207–219. Owsley, C., McGwin, G., Jr., Phillips, J. M., McNeal, S. F., & Stalvey, B. T. (2004). Impact of an educational program on the safety of high-risk, visually impaired, older drivers. American Journal of Preventive Medicine, 26, 222–229.

218

A. Dommes, V. Cavallo / Transportation Research Part F 15 (2012) 206–218

Oxley, J., Congiu, M., Whelan, M., D’Elio, M., & Charlton, J. (2008). Teaching young children to cross roads safely. Annual Proceedings – Association for the Advancement of Automotive Medicine, 52, 215–223. Oxley, J., Fildes, B., Ihsen, E., Charlton, J., & Day, R. (1997). Differences in traffic judgments between young and old adults pedestrians. Accident Analysis and Prevention, 29(6), 839–847. Oxley, J., Ihsen, E., Fildes, B., Charlton, J., & Day, R. (2005). Crossing roads safely: An experimental study of age differences in gap selection by pedestrians. Accident Analysis and Prevention, 37(5), 962–971. Roenker, D. L., Cissel, G. M., Ball, K. K., Wadley, V. G., & Edwards, J. D. (2003). Speed-of-processing and driving simulator training result in improved driving performance. Human Factors, 45(2), 218–233. Romoser, M. R. E., & Fisher, D. L. (2009). The effect of active versus passive training strategies on improving older drivers’ scanning in intersections. Human Factors, 51(5), 652–668. Rothengatter, J. A. (1981). Traffic safety education for young children. Lisse: Swets & Reitlinger. Schiff, W., Oldak, R., & Shah, V. (1992). Aging person’s estimates of vehicular motion. Psychology and Aging, 7, 518–525. Schwebel, D. C., Gaines, J., & Severson, J. (2008). Validation of virtual reality as a tool to understand and prevent child pedestrian injury. Accident Analysis and Prevention, 40(4), 1394–1400. Scialfa, C. T., Guzy, L. T., Leibowitz, H. W., Garvey, P. M., & Tyrell, M. A. (1991). Age differences in estimating vehicle velocity. Psychology and Aging, 6, 60–66. Simpson, G., Johnston, L., & Richardson, M. (2003). An investigation of road crossing in a virtual environment. Accident Analysis and Prevention, 35(5), 787–796. Snowden, R. J., & Kavanagh, E. (2006). Motion perception in the aging visual system: Minimum motion, motion coherence, and speed discrimination thresholds. Perception, 35, 9–24. Tabachnick, B. G., & Fidell, L. S. (1989). Using multivariate statistics (2nd ed.). New York: Harper Collins. Thomson, J. A., Tolmie, A. K., Foot, H. C., Sarvary, P., Whelan, K. M., & Morrison, S. (2005). Influence of virtual reality training on the roadside crossing judgments of child pedestrians. Journal of Experimental Psychology: Applied, 11(3), 175–186. Tolmie, A., Thomson, J. A., Foot, H. C., Whelan, K., Morrison, S., & McLaren, B. (2005). The effects of adult guidance and peer discussion on the development of children’s representations: Evidence from the training of pedestrian skills. British Journal of Psychology, 96(2), 181–204. Weiss, P. L., Naveh, Y., & Katz, N. (2003). Design and testing of a virtual environment to train stroke patients with unilateral spatial neglect to cross a street safely. Occupational Therapy International, 10(1), 39–55. Young, D. S., & Lee, D. N. (1987). Training children in road crossing skills using a roadside simulation. Accident Analysis and Prevention, 19(5), 327–341. Zeedyk, M. S., Wallace, L., Carcary, B., Jones, K., & Larter, K. (2001). Children and road safety: Increasing knowledge does not improve behaviour. British Journal of Educational Psychology, 71(4), 573–594.