A simulator study of factors influencing drivers’ behavior at traffic lights

A simulator study of factors influencing drivers’ behavior at traffic lights

Transportation Research Part F 37 (2016) 107–118 Contents lists available at ScienceDirect Transportation Research Part F journal homepage: www.else...

499KB Sizes 0 Downloads 7 Views

Transportation Research Part F 37 (2016) 107–118

Contents lists available at ScienceDirect

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

A simulator study of factors influencing drivers’ behavior at traffic lights Blazej Palat ⇑, Patricia Delhomme Ifsttar, Planning Mobilities Environment Department, Mobility and Behavior Psychology Lab, 25 Allée des Marronniers, F-78000 Versailles-Satory, France

a r t i c l e

i n f o

Article history: Received 9 June 2014 Received in revised form 17 November 2015 Accepted 30 November 2015

Keywords: Driving Time pressure Social context Traffic lights

a b s t r a c t Drivers’ reactions to changing traffic lights have an impact on safety at intersections. We examined the influence of transient factors – more specifically time pressure and social context, both conducive to traffic-light violation – on behavior behind the wheel when a traffic light changes. We carried out an experiment on a driving simulator. The participants were 94 car drivers (53 males) with a mean age of 21.7 years. They drove under time pressure vs. no time pressure. At several intersections the participants were alone (no other drivers present), whereas at several other intersections they were behind a line-up of vehicles, the last of which ran the yellow light (other drivers present). As expected, time pressure and social context (presence of other drivers) increased participants’ risky behaviors while approaching, and going through traffic lights, as well as undesirable rapid accelerations when the signal changes to green. The effect of time pressure on yellow-light running was not mediated by approach speed, which showed that participants in a hurry were likely to run lights intentionally. The results are interpreted in view of proposing effective measures for reducing yellow-light running and rapid accelerations at traffic lights. Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction The behavior of drivers as they approach and go through intersections with traffic lights is important for several reasons. Obviously, stop/go decisions and hard breaking at yellow and red signals remain a major concern due to their consequences on road safety (for frequent types of crashes, see Green, 2003; Jørgensen, 1988). However, reacting to a signal changing from red to green while stopped at a red light and accelerating after starting up so as to reach one’s target speed are also important aspects of driving behavior. In heavy traffic, the first driver in a line-up of vehicles waiting at a red light should start up shortly after the light turns from red to green to allow as many cars as possible to go through the intersection during the green-light phase. But before starting up, the driver should check to see whether all pedestrians have stepped off the crosswalk and whether the intersection is clear of red-light runners. Factors likely to influence drivers’ behavior at intersections with traffic lights can be deduced from general models of the driving task like Summala’s (1997) ‘‘Driver Task Cube” (see also Summala, 2007). According to this model, available time (Hollnagel, 2002) is a crucial factor of almost all facets of the driving task. Time control is achieved by maintaining a targeted speed. When drivers believe they are short of time, they may want to accelerate, which in turn reduces time-to-collision margins (Godthelp, Milgram, & Blaauw, 1984). In other words, the higher the speed, the faster the driver has to react to

⇑ Corresponding author. Tel.: +33 6 88 98 01 54; fax: +33 1 30 84 40 01. E-mail addresses: [email protected] (B. Palat), [email protected] (P. Delhomme). http://dx.doi.org/10.1016/j.trf.2015.11.009 1369-8478/Ó 2016 Elsevier Ltd. All rights reserved.

108

B. Palat, P. Delhomme / Transportation Research Part F 37 (2016) 107–118

different aspects of the road environment in order to keep a proper trajectory and avoid collisions, and the higher the risk that they might not react on time to an impending danger. Hence, the objective risk of a crash increases as speed increases. Also, the higher the speed, the stronger the probability that if the light turns yellow, drivers will find themselves at a distance from the light (dilemma zone: Gazis, Herman, & Maradudin, 1960; Liu, Herman, & Gazis, 1996) that prevents them both from stopping before the intersection without slamming on the brakes or continuing at a constant speed so as to get through the intersection before the light turns red. In this kind of situation, accelerating to avoid red-light running seems to be the most suitable behavior, but the higher the speed, the lower the acceleration capacity of the vehicle. It is therefore not surprising that high-speed driving was found to be conducive to yellow- and red-light running (Elmitiny, Yan, Radwan, Russo, & Nashar, 2010; Konecˇni, Ebbesen, & Konecˇni, 1976). Stopping at a yellow or red traffic light is different from stopping before a fixed or moving obstacle insofar as going through a yellow or red light does not necessarily imply a collision. Thus, as a means of time control when there is little available time, drivers may attempt to go through the yellow light even though they are far enough away to stop safely before the intersection or they may even deliberately run the red light (Porter & Berry, 2001). They may adopt these risky behaviors either because they consider the risk acceptable (as predicted by Wilde’s ‘‘Risk Homeostasis Theory”, 1982) or because they do not perceive any risk at all as long as they do not get a ticket, get in a crash, or nearly miss getting into that kind of situation (Fuller, 1984; Näätänen & Summala, 1976). They may also see no risk because in most situations, running the yellow or the red light does not induce feelings of difficulty or loss of control over the vehicle (Fuller, McHugh, & Pender, 2008). Moreover, unlike time pressure, risk perception is not a predictive factor of the intention to run a yellow light (Palat & Delhomme, 2012), whereas it influences the choice of speed at which drivers approach an intersection when they have the right of way (Saad, Delhomme, & Van Elslande, 1990; Saad, Van Elslande, Delhomme, Lepesant, & Gaujé, 1989). If a driver in a hurry stops at a yellow or red light and waits until the light turns green again, he/she might want to start up as quickly as possible after the light changes and accelerate rapidly in order to reach his/her targeted speed as fast as possible. Time control is one of the primary motives underlying the choice of a speed and the decision to run a yellow- and/or red light. There may also be several additional, situation-specific or affective motives acting independently or in interaction with time control. For some people, high speed provides pleasurable sensations (Rothengatter, 1988), and driving in general can foster a ‘‘flow” experience (Csíkszentmihályi, 1989). Hence, unwanted stops at traffic lights that change from green to yellow when the driver is approaching an intersection can cause annoyance (Palat & Delhomme, 2012) because they interrupt or hinder these positive feelings (Lupton, 2002). Other possible motives underlying the choice of speed (Connolly & Åberg, 1993) and the decision to abide by traffic lights (Elmitiny et al., 2010; Palat & Delhomme, 2012) are related to interactions with other car drivers. These interactions are regulated by legal norms of on-road behavior. If the traffic laws had not existed, a road user’s behavior would have been totally unpredictable for others. Their interactions would have generated far more risk of crash and traffic in general would have been less fluent. Of course, knowledge of the traffic laws is indispensable, but driving also demands skill and experience that can be developed by imitating (Bandura, 1977; Scott-Parker, Watson, King, & Hyde, 2012) the way other drivers behave in different situations behind the wheel. Under uncertainty, imitating others or conforming to them is a way of choosing the most appropriate behavior. However, it could also merely be a way of avoiding social rejection or sanctions in cases where the driver believes he/she knows what the most appropriate behavior in a given situation is yet still conforms to others who adopt a different behavior (Deutsch & Gerard, 1955). These two primary reasons for conforming serve a more general goal which consists of maintaining a positive self-concept (Cialdini & Trost, 1998). Last but not least, conforming or imitating may sometimes be an automatic reaction that is not entirely intentional (Heyes, 2011). Automatic imitation can play a role when the driver has to respond rapidly to a given stimulus such as a traffic light changing from green to yellow. There are a number of driving situations in which drivers are particularly likely to conform to others. Sometimes drivers try to break the rules if they perceive the benefits of such a decision (for example, time gain) and a low level of risk. Unruly behavior in this type of situation will be observed frequently and considered normal by the majority of drivers. In France, for instance yellow-light running is illegal (Article R412-31 of the French traffic laws), but remains quite a common behavior (Moget-Monseur & Biecheler-Fretel, 1985; TNS Sofres, AXA, AXA Prévention, 2011). This kind of behavior can therefore be regarded as an unofficial norm specific to the driving culture of this country (Özkan & Lajunen, 2011). In such a contradictory normative context, it is likely that people guide their behavior by conforming to others rather than to the official, explicitly formulated legal norm (Sigelman & Sigelman, 1976; Taubman-Ben-Ari & Katz-Ben-Ami, 2012). Hence, if a driver goes through several intersections with traffic lights during a trip and sees other drivers who stop at the yellow light, he/she will probably not try to run the yellow light at subsequent intersections (if the opportunity arose) for fear of disapproval from other road users. However, a driver who witnesses yellow-light running during such a trip will be likely to think about the potential benefits of this unruly behavior for the drivers who did break the law, so he/she may try to run the yellow light at subsequent intersections. Actually, comparing oneself to others helps in evaluating one’s performance behind the wheel. In a traffic jam, the line-up of vehicles waiting at a red light is so long that only those who are close to the light will be able to cross the intersection during the next green-light phase. During this phase, one driver in the moving line will have to make a stop/go decision at the onset of the yellow light. At that moment, other drivers in the line become targets of social comparison (Festinger, 1954; for recent advances in social comparison theory, see Suls, Martin, & Wheeler, 2002) insofar as they are involved in the same driving situation. If the driver runs the yellow light, he/she will belong to the group of drivers who were able to get through the intersection. According to past research, heavy traffic at intersections seems to be conducive to red-light

B. Palat, P. Delhomme / Transportation Research Part F 37 (2016) 107–118

109

running (Porter & England, 2000; Shinar, 1998). However, if a driver decides to stop while the vehicles in front of him/her manage to go, he/she will feel outdone by them, which for some is a rather frustrating driving situation (Gulian, Matthews, Glendon, & Davies, 1989). Such an upward social comparison to those drivers who got through the intersection could be a source of motivation to improve one’s situation (Shelley, Lobel, & Lobel, 1989) by starting back up as rapidly as possible when the light turns from red to green, and/or accelerating sharply right afterwards in order to catch up with those drivers, or to show off one’s driving skills (Yinon & Levian, 1995). To the best of our knowledge, the influence of heavy traffic on the prevalence of this kind of behavior at a red-to-green signal change has not yet been tested empirically. The aims of this study were to use a driving simulator to test for the effect of time pressure and a social context conducive to yellow-light running, on drivers’ behavior when traffic lights change. Regarding the social context, our intention was especially to examine whether other drivers’ behavior can not only have an immediate effect, but also a deferred one. More specifically, we assumed that the behavior of a driver who is behind a line of vehicles might be influenced by the social context at two moments when the traffic light change: (a) when the driver in front of him/her runs the yellow light, and (b) when, after having stopped, he/she is starting back up after the signal change from red to green. We also wanted to analyze the relationship between time pressure, the choice of a speed as the primary means of time control behind the wheel, and yellow- and/or red-light running. In particular, we tried to find out whether drivers run yellow and/or red lights under time pressure because they generally drive faster in such circumstances and therefore are often caught in the indecision zone (i.e. the choice of speed is a mediator between time pressure and yellow- and/or red-light running), or because they deliberately take risks by running traffic signals. Our secondary aim was to determine whether drivers’ behavior at traffic lights is not affected much by variations in the level of perceived risk involved in traffic-light violations. We set forth four hypotheses: 1. Time pressure will have a significant effect on drivers’ behavior. Drivers in a hurry (a) will approach traffic lights at higher speeds, (b) be more likely to brake abruptly, or (c) less likely to stop at a yellow light as compared to drivers who are not under time pressure. In cases where drivers in a hurry stop at a yellow light, (d) they will react sooner to the red-to-green signal change and (e) will accelerate faster afterward in order to reach their targeted speed, than will drivers who are not in a hurry. 2. There will be an effect of social context conducive to yellow-light running, regardless of whether or not the driver is aware of being influenced. At a traffic light where they are about to be outdone by other drivers, and when the unofficial norm of going through yellow lights is made salient by the other drivers’ behavior, (a) drivers will be more likely to brake abruptly, or (b) less likely to stop at the yellow light, than when they approach and go through the intersection in the absence of others. If a driver stops at the yellow light while about to be outdone by other drivers, and when the unofficial norm of going through yellow lights is made salient by the others’ behavior, (c) the driver will react sooner to the red-togreen signal change, and (d) accelerate more rapidly afterward in order to reach his/her targeted speed, than if he/she had stopped at the light in the absence of others. 3. Approach speed will not mediate the relationship between time pressure and yellow- or red-light running. The most important reason why drivers in a hurry are less likely to stop at a yellow light is that they deliberately run the light rather than that they drive at high speeds and are therefore often caught in the indecision zone. 4. Risk perception will affect (a) approach speed, but neither (b) braking nor (c) yellow-light running. 2. Method 2.1. Participants The participants were 94 car drivers (53 males) averaging 21.7 years of age (r = 1.86, range: 18–25). They had had their car driver’s license for 2.6 years on average (r = 1.8, range: 1 month–6 years 7 months), and had driven a car for an average of 24,113 (Mdn = 13,750, range: 300–200,000) kilometers since obtaining their license. Almost a half of the participants (39) had been involved in at least a minor collision during the three years prior to the study, and 24 had already been ticketed for various driving violations. Four participants (1 male) dropped out because of simulator sickness. They were therefore not included in the sample. 2.2. Materials 2.2.1. Driving simulator We used the driving simulator belonging to the Mobility and Behavior Psychology Lab (Ifsttar-LPC). The equipment is composed of ten parallelepiped-shaped panels and visual channels (2.44 m  1.83 m) as well as an instrumented vehicle (Peugeot 308, Fig. 1). Seven of these panels are equipped with a classic video projector (F22 Projection Design) while the other three have a Titan stereoscopic video projector (Digital Projection, 3D). The instrumented vehicle is positioned in the center of five panels with a ‘‘triptych” facing the driver. Various driving parameters (e.g., speed, acceleration, braking, wheel movements) are recorded at a frequency of 30 Hz in accordance with the virtual traffic situation to which the driver is exposed. In the scenarios programmed for our study, there was no possibility of collision. Instead of colliding with an object, the driver would ‘‘pass through” it.

110

B. Palat, P. Delhomme / Transportation Research Part F 37 (2016) 107–118

Fig. 1. Driving simulator of the Mobility and Behavior Psychology Lab (Ifsttar-LPC).

2.2.2. Familiarizing trip The sole aim of this trip was to familiarize the participants with driving in a simulator and to check to see whether they suffered from simulator sickness. The trip consisted of an itinerary approximately 5.8 km long in an urban environment during which the drivers went through six intersections with traffic lights. Speed limits of 50 km/h were visible at regular intervals along the itinerary. Traffic-light cycles had variable durations, except one that turned yellow according to a triggering rule described in Section 2.2.3.2, and one that was red as the participants approached it and turned green when they were 60 m from it. For a section of the itinerary, traffic flow on the opposite side of the road (i.e., in the reverse direction to the participants’ direction of travel) was generated in order to enhance realism. 2.2.3. Experimental trip This trip consisted of an itinerary on the simulator approximately 12.6 km long, in an urban environment, during which the drivers went through 12 intersections with traffic lights (going straight ahead, without turning). Speed limits of 50 km/h were visible at regular intervals along the itinerary. Half of these lights changed from green to yellow as the participant was approaching the intersection. The remainder of the situations at traffic lights, along with some traffic generated in the vicinity of the intersections in the opposite direction to the participants’ direction of travel, were intended to enhance realism. All 12 traffic lights (whether they changed or not from green to yellow) were set up along the itinerary in such a way that participants were never in the same driving situation at two or more intersections in a row. However, it should be noted that during this itinerary, the participants came across yellow lights at a higher frequency than generally. 2.2.3.1. Time pressure. Right before the experimental trip, near a half (n = 48) of the participants were explicitly told that one of the experiment’s goals was to evaluate their time-saving abilities, so they were asked to drive as if they were rushing to make it for an important meeting. Time pressure could not be operationalized as a within-subject variable because it would necessitate more than one experimental trip, and therefore a higher risk of simulator sickness. 2.2.3.2. Traffic lights changing from green to yellow. Among the 12 traffic lights, 6 turned from green to yellow when the participants were 4.3 s from the light if they continued to travel at a constant speed. This triggering rule was fixed, given that a vehicle traveling at 50 km/h needs 4.3 s to cover a distance of 59.73 m. The indecision zone began at 59.94 m before a yellow light, which lasts for 3 s (see Gazis et al., 1960). This is the yellow-light duration set for urban zones in France (Ministère de l’Ecologie, de l’Energie, du Développement Durable et de la Mer, 2012). Therefore, a vehicle approaching a light at a speed above 50 km/h was necessarily within the indecision zone. The visibility in the intersections where the lights changed from green to yellow as the participant approached was good. 2.2.3.3. Social context conducive to yellow-light running. At three of the six traffic lights that turned from green to yellow based on the triggering rule described above, participants were alone at the intersection (neither followed nor preceded by any other vehicles). These were situations where there was no social context conducive to yellow-light running, and the unofficial norm of going through the yellow light was not salient. At the other three traffic lights, participants stopped at the end of a line-up of 21 vehicles waiting at a red light. The length of the queue and the dynamic parameters of the situation which followed were set up so that the participants found themselves driving at a certain speed, and at a certain distance from the light while it was changing from green to yellow. The vehicles were separated from each other by 1 m so that the line-up measured 123.75 m. After 5 s, the cars in the line took off at an acceleration rate of 1.5 m/s2. Hence, the vehicles (in front of the participant) needed to cover 64.31 m to reach a speed of 50 km/h. Participants who had reached 50 km/h as they traveled behind the line of vehicles found themselves just before the indecision zone when the light turned from green to yellow. If the participants were following closely behind the

B. Palat, P. Delhomme / Transportation Research Part F 37 (2016) 107–118

111

vehicle in front of them (which all participants did), the light turned from green to yellow while the vehicle ahead was still in front of the light. In this case, the vehicle ahead ran the yellow light. The presence of vehicles that managed to get through the intersection before the participant, and especially the behavior of the last vehicle that ran the yellow light, created a social context conducive to yellow-light running. 2.2.3.4. Other situations at traffic lights. These included two situations where the participants were alone as they approached a green traffic light that did not change to yellow, three situations almost exactly the same as the situations where the implicit norm was salient except that the light did not change to yellow as the participants approached, and one situation where the participants were alone as they approached a red light that changed to green after 5 s (provided the participant stopped, which all participants did). 2.3. Behavioral measures These were the participants’ mean travel speed, and the following measures of behavior at six intersections where the light turned yellow as the participants approached: the participants’ instantaneous speed just before the green-to-yellow change (approach speed), the time elapsed before pressing the brake pedal, and the traffic-light phase (green, yellow, or red) as the participant passed the light. In cases where the participant stopped, reaction time to the red-to-green signal change, and the time needed to reach one’s mean travel speed after starting back up (as a measure of acceleration) were also recorded. 2.4. Questionnaires The purpose of the pre-experimental questionnaire was to record the participants’ driving characteristics, whereas the purpose of the post-experimental one was to assess their perceptions of risks related to yellow- and red-light running. 2.4.1. Pre-experimental questionnaire Participants reported the number of kilometers they had driven since obtaining their driver’s license, whether they were involved in collisions or ticketed during the three years prior to the study, their age, the number of years since obtaining their license, their sex, and filled in the French translation (Delhomme, 2002) of a driving sensation seeking scale (Taubman, Mikulincer, & Iram, 1996 cited in Yagil, 2001). According to several studies (Rosenbloom & Wolf, 2002a, 2002b), sensation seeking could be an important predictor of the decision to run a yellow light. 2.4.2. Post-experimental questionnaire There were eight measures of risk perception (a = 0.72) created ad hoc to fit the aims of our study. The participants rated the risk of a crash if they went straight through, turned right, or turned left at the intersection after running the yellow (3 items) or red (3 items) light, and the risk of getting a ticket for those violations (2 items), on 5-point Likert scales ranging from 1 (low risk) to 5 (high risk). All the risk-perception measures were aggregated so as to obtain a single measure of yellow- and red-light running risk. 2.5. Procedure The study was approved by Ifsttar’s ethics committee. Announcements were put up on two large university campuses in the suburbs of Paris to recruit participants for a study aimed at describing car drivers’ behavior in urban areas. Additional participants were recruited by sending an e-mail to individuals who had signed up on an external database to volunteer for experiments in psychology and neuroscience. Participants were paid 30 € for their time. The instructions for the participants were presented from a dictaphone by the experimenter. Experimental sessions were individual, lasted for about 1 h, and comprised four stages, as follows: 1. Welcoming stage. After arriving at the session, the participants received a brief description of the study and information regarding their rights. They then filled in free consent and payment forms. None of the drivers refused to participate at this stage. 2. Pre-experimental questionnaire. After filling in the questionnaire, the participants were assigned to one of two timepressure conditions so that there would be a similar sex ratio (see Table 1 for more details), a similar level of driving experience (no significant difference of the mean reported number of kilometers driven between no-pressure and in-a-hurry conditions: M = 23574.6, M = 24629.46, t = 0.159, p < .87), a similar number of drivers who had past experiences of risky situations (no significant difference of the mean sum of self-reported numbers of collisions and tickets between no-pressure and in-a-hurry conditions tested with a Poisson generalized linear model: M = 1.02, M = 0.9, LR v2 = 0.22, p < .64), and a similar mean level of drivers’ sensation seeking in each time-pressure condition (no significant difference of the mean score on the sensation seeking scale between no-pressure and in-a-hurry conditions: M = 2.75, M = 2.96, t = 1.21, p < .23).

112

B. Palat, P. Delhomme / Transportation Research Part F 37 (2016) 107–118 Table 1 Number of male and female participants in the experimental conditions of time pressure. There is no significant difference of sex ratio between the two conditions: Pearson v2 = 0.0007, p < .98.

nmale nfemale

No time pressure

In a hurry

26 20

27 21

3. Familiarizing trip. The participants took a familiarizing trip on the simulator, accompanied by the experimenter sitting in the passenger’s seat next to them. The experimenter could answer if the participant had any questions concerning the simulator. If a participant suffered from nausea, dizziness, or headache, the session was immediately stopped. The participant was debriefed, and then taken to a place where he/she could rest until feeling well enough to leave. The familiarizing trip was followed by a short break. None of the participants had any collisions with moving or fixed objects during the familiarizing trip. 4. Experimental trip. Before taking the wheel, the participants who were supposed to drive under time pressure were told to drive as if they were in a hurry. The experimenter was no longer present in the car during the experimental trip. He still could monitor their behavior on the road and in the vehicle, thanks to cameras installed inside the instrumented vehicle. He could also communicate with them from the control room if needed.1 None of the participants had any collisions with moving or fixed objects during the experimental trip. 5. Post-experimental questionnaire. After filling in the questionnaire, the participants were thanked, and debriefed. 2.6. Design The experiment used a mixed subject design with one independent within-subject variable – social context (operationalized by different situations at traffic lights), which had two experimental conditions, social context conducive to running the light vs. no such context – and one between-subject variable – time pressure, which had two experimental conditions, in a hurry vs. no time pressure. 3. Results The perception of risks related to yellow- and red-light running was first analyzed with descriptive statistics. Hypotheses pertaining to a given behavior were then tested using linear-mixed-effect, and generalized linear-mixed-effect models. The latter was employed to assess the impact of factors conducive to yellow- or red-light running, which was a binary response variable. Wald chi-square statistics are given for fixed effects of statistically significant independent categorical variables, and standardized regression coefficients for independent continuous ones. The effect of time pressure, risk perception, and their interaction on approach speed was tested only with respect to the situations were the participants approached and went alone through the intersection (i.e., no social context conducive to yellow-light running). In the situations where the social context was conducive to yellow-light running, the participants’ approach speed was highly constrained since they were following other vehicles. We examined the effects of time pressure, social context, and risk perception, and their firstand second-order interactions on time to press the brake pedal at the onset of the yellow light, and on yellow- and red-light running as response variables. In addition, approach speed and its first-, second-, and third-order interactions with time pressure, social context, and risk perception were included in the models of braking and yellow- or red-light running so as to control for the impact of approach speed on these dependent variables in traffic-light situations where the driver approaches and goes through the traffic light alone. We also examined the effect of time pressure, social context, and their interaction on reaction time to the red-to-green signal change, and the measure of acceleration upon starting back up in cases where the participant stopped at a yellow light. Finally, mean travel speed and its first- and second-order interactions with time pressure and social context were controlled for in the model where acceleration was the response variable. Table 2 summarizes the effects of time pressure and social context (as predicted in Hypotheses 1 and 2). 3.1. Risk perception analysis In general, the participants considered traffic light violations to be quite risky (M = 3.74, r = 0.62). The risk of a crash was deemed moderate when the driver went straight through (M = 3.15, r = 1.28) or turned right (M = 2.48, r = 1.17), but quite high if he/she turned left (M = 3.8, r = 1.16) at the intersection after running a yellow light. The risk of being ticketed for running the yellow light was rated as moderate (M = 2.73, r = 1.25). The crash risk was estimated as high if the driver went straight through (M = 4.7, r = 0.72), turned right (M = 4.35, r = 0.98), or turned left (M = 4.69, r = 0.76) at the intersection after running a red light. The risk of being ticketed for running a red light was considered high (M = 4, r = 1.14). 1

One participant reported feeling sick and hence dropped out at the beginning of the experimental trip.

113

B. Palat, P. Delhomme / Transportation Research Part F 37 (2016) 107–118 Table 2 Effects of time pressure and social context on drivers’ behavior on the simulator.

Time pressure

Behavioral measure

Experimental condition

v2

M

r

min

max

Approach speed

No pressure In a hurry No pressure In a hurry No pressure In a hurry No pressure In a hurry

8.58**

47.49 54.94 0.77 0.82 1.87 1.33 6.72 5.32 No. of lights run 30 68

10.45 13.9 0.35 0.39 1.02 0.99 2.36 1.85

25.35 29.37 0 0 0 0 2.58 2.37

99.05 116.46 2.12 3.27 7.42 8.95 16.5 17.13

0.76 0.83 1.58 1.66 6.37 5.71 No. of lights run 35 63

0.39 0.34 1 1.08 2.42 1.97

0 0 0 0 2.37 2.63

3.27 2.06 7.7 8.95 17.13 15.37

Braking Starting up on green light

Reaction time Acceleration

Social context

3.97*

Braking

Alone With others Alone With others Alone With others

14.23***

Reaction time

Running a traffic light

***

12.39***

No pressure In a hurry

Acceleration

*

16.34***

Running a traffic light

Starting up on green light

**

0.14

Alone With others

0.76 29.59***

12.13***

p < .05. p < .01. p < .001.

3.2. Testing hypotheses regarding behavioral measures The numbers of the hypotheses being tested are given in parentheses. 3.2.1. Approach speed (1a, 4a) Time pressure (v2 = 8.58, p < .004) and risk perception (b⁄ = 2.38, p < .02) significantly predicted participants’ approach speed. The participants driving in a hurry approached traffic lights at higher speeds (M = 54.94, r = 13.9, min = 29.37, max = 116.46) than those who had no time pressure (M = 47.49, r = 10.45, min = 25.35, max = 99.05), which is in line with Hypothesis 1a. Consistent with Hypothesis 4a, risk perception affected the speed at which the participants approached traffic lights: as reflected by the negative value of the regression coefficient, the lower the level of perceived risk, the higher the speed. 3.2.2. Braking (1b, 2a, 4b) Approach speed (b⁄ = 0.03, p < .007) and social context (v2 = 14.23, p < .001) significantly predicted the time elapsed before the participants started to brake. As illustrated by the positive value of the standardized regression coefficient, the faster the participants drove, the later they pressed the brake pedal to stop at the traffic light. Contrary to Hypothesis 1b, time pressure did not affect braking behavior (v2 = 0.14, p < .71). In line with Hypothesis 2a, the participants braked later when they were potentially influenced by others (M = 0.83, r = 0.34, min = 0, max = 2.06) than when they were approaching the intersection alone (M = 0.76, r = 0.39, min = 0, max = 3.27). In accordance with Hypothesis 4b, the perception of risk related to yellow- and/or red-light running did not affect braking behavior (b⁄ = 0.06, p < .44). 3.2.3. Yellow- or red-light running (1c, 2b, 3, 4c) Time pressure (v2 = 3.97, p < .05) and social context (v2 = 12.13, p < .001) significantly predicted the occurrence of yellowor red-light running. In line with Hypotheses 1c, more yellow lights and red lights were run by the participants who were in a hurry (68 lights), than by those who were not under time pressure (30 lights). In general, the participants ran more yellow or red lights when they could be influenced by others (63 lights) than when they were approaching and going through the intersection alone (35 lights), which is consistent with Hypothesis 2b. In line with Hypothesis 3, the main effect of approach speed was not significant either (v2 = 1.61, p < .2), hence, the possibility that approach speed mediates the relationship between time pressure and yellow- or red-light running can be ruled out (Baron & Kenny, 1986). However, the timepressure by approach-speed interaction was significant (v2 = 4.5, p < .04). An additional analysis showed that approach speed was predictive of yellow- or red-light running for the participants with no time pressure (b⁄ = 0.31, p < .004). The faster these participants drove, the greater the probability that they would not stop at the yellow light. However, approach speed was not predictive of traffic-light violation for the participants who were in a hurry (b⁄ = 0.04, p < .7, see Fig. 2). In accordance with Hypothesis 4c, risk perception was not predictive of yellow- or red-light running (b⁄ = 0.47, p < .66).

114

B. Palat, P. Delhomme / Transportation Research Part F 37 (2016) 107–118

Fig. 2. Approach speed of participants who had no time pressure, and of participants who were in a hurry when they stopped or did not stop at the yellow light. Slopes of the relationship between approach speed and yellow-light running are indicated by point lines.

3.2.4. Reaction time to a signal change from red to green (1d, 2c) Time pressure significantly predicted reaction time to a red-to-green signal change (v2 = 16.34, p < .001). As stated in Hypothesis 1d, the participants who were in a hurry took less time to start back up after the signal change (M = 1.33, r = 0.99, min = 0, max = 8.95) than did those who had no time pressure (M = 1.87, r = 1.02, min = 0, max = 7.42). Reaction time did not depend on the social context (v2 = 0.76, p < .38), a result that failed to support Hypothesis 2c. 3.2.5. Accelerating when starting back up (1e, 2d) Time pressure (v2 = 12.39, p < .001) and social context (v2 = 29.59, p < .001) significantly predicted the time that the participants needed to reach their mean travel speed after starting back up when the light turned green. In line with Hypothesis 1e, the participants who were in a hurry took less time (M = 5.32, r = 1.85, min = 2.37, max = 17.13) to reach their mean travel speed when starting up (they accelerated more rapidly) than did those who had no time pressure (M = 6.72, r = 2.36, min = 2.58, max = 16.5). As stated in Hypothesis 2d, in situations where the participants stopped at the yellow light and could potentially be influenced by other drivers, they took less time to reach their mean travel speed when the light turned green (M = 5.71, r = 1.97, min = 2.63, max = 15.37) than in the situations where they stopped before the light in the absence of others (M = 6.37, r = 2.42, min = 2.37, max = 17.13). 4. Discussion We carried out an experimental study aimed at measuring transient factors (time pressure, and social context conducive to yellow-light running) likely to influence drivers’ behavior when traffic lights change. We set forth several hypotheses pertaining to drivers’ behavior when the light changes from green to yellow, or from red to green. Most of our hypotheses were supported by the results of the experiment. All except one behavioral measure exhibited an effect of time pressure. The participants who were asked to drive as if they were in a hurry approached traffic lights at higher speeds, and stopped at the yellow light less often than did the participants who were not under time pressure. Those participants in the in-a-hurry condition who stopped at yellow lights reacted more rapidly to the signal change from red to green, and accelerated at a faster rate afterward than did the participants in the no-time-pressure condition. However, time pressure did not significantly affect the braking behavior. One possible explanation for this could be that time-control decisions are usually semi-automatic (Allen, Lunenfeld, & Alexander, 1971; Hollnagel, Nåbo, & Lau, 2003; Michon, 1985; van der Molen & Botticher, 1988) and therefore do not require

B. Palat, P. Delhomme / Transportation Research Part F 37 (2016) 107–118

115

a lot of time for decision making. At the onset of a yellow light, then, drivers would decide very rapidly to stop or go through. If they decide to stop, the braking reaction would not be significantly delayed by hesitation. Still, the decision to stop vs. run the yellow light is likely to be intentional and must comply with time constraints. One could argue that drivers in a hurry do not run yellow lights because they intend to, but because they are driving at higher speeds and are therefore more likely to be caught in the indecision zone when the light turns yellow. Our analysis showed that the effect of time pressure, as a factor conducive to yellow-light running, was not mediated by the speed at which the driver was approaching the traffic light when it changed from green to yellow. In line with the results of a self-report study aimed at motivations to run a red light (Porter & Berry, 2001), we provided further evidence that traffic-light running can often result from a conscious decision related to time control. In our study, an overly high approach speed predicted yellow-light running for participants who were not under time pressure, but not for those who were in a hurry. This result suggests that the latter took such risky decisions regardless of their approach speed. Social context as a second factor influencing drivers’ behavior at traffic lights is more transient in nature. All the expected effects of social context were observed except on one behavioral measure. The participants started to brake later, and stopped at the yellow light less often, in situations where the social context was conducive to yellow-light running compared to situations where an influencing social context was absent. Moreover, even when the participants stopped at the yellow light in a situation where they could be influenced by other drivers, they accelerated more rapidly when the signal turned green than when they were the only car approaching the yellow light. However, contrary to what happened in the in-a-hurry and no-time-pressure conditions, reaction time to the red-to-green signal change did not differ according to the social-interaction circumstances in which the participants were approaching and going through the intersection. We believe that this was because time-saving demands enhanced reactivity in almost every driving situation encountered during a trip, whereas catching up with other drivers is a goal pertaining to a specific situation. Unless the others are driving at a very high speed, accelerating and maintaining an increased speed for a short period of time is sufficient for catching up with them without an extra attentional effort aimed at reacting rapidly to the signal change from red to green. As expected, all but one (approach speed) of the behavioral measures taken when the participants were approaching traffic lights were not significantly affected by risk perceptions related to yellow- or red-light running. According to the threat avoidance framework (Fuller, 1984), when a driver approaches an intersection, he/she perceives an increased level of risk linked to possible interactions with drivers on the other streets and adjusts his/her speed accordingly in order to better anticipate these interactions. In this case, the decrease in speed is related to the level of subjective risk (Delhomme & Meyer, 1998; Saad et al., 1990; Saad et al., 1989). If the intersection is equipped with traffic lights, and the light is green as the driver approaches (as in this study), the driver adjusts his/her speed according to the perceived risk related to yellowor red-light running. Indeed, the less a driver decreases his/her speed, the greater the probability that he/she will not be able to stop if the light turns yellow. However, it must be stressed that the presence of other road users (whether moving or stationary) on the other streets was not operationalized in our experiment. In our study, approach speed was also related to braking behavior at the onset of the yellow light. Interestingly, the faster the participants drove, the later they put on the brakes to stop at the traffic light. This result goes against what could have been expected if braking before a traffic light were perceived by drivers in the same way as braking before a fixed or moving obstacle. We know that the higher the speed, the narrower the time-to-collision margin, which should make drivers react more rapidly to a signal change. It is therefore not surprising that the participants who approached the intersection at a high speed were the ones who perceived less risk related to yellow- or red-light running. Low risk perception could be the reason why they also adopted less cautious braking behavior. There are several limitations to the conclusions drawn in our study. First of all, the perception of time pressure and its consequences are relatively complex psychological phenomena. Among other factors, they depend on the social (Xu, Li, & Jiang, 2014) and situational (Naveteur, Cœugnet, Charron, Dorn, & Anceaux, 2013) context. Personality and motivational factors may also play a role. For instance, an individual may be more motivated to avoid being late to an important professional appointment if he/she is particularly dedicated to his/her work, prone to stress, or has previous negative experiences of being late to such appointments. Also, different driving goals imply different degrees of motivation to save time (Cœugnet, Naveteur, Antoine, & Anceaux, 2013; Schmidt-Daffy, 2012). A driver might be more likely, for example, to take risks when taking someone suffering from a heart attack to the hospital, than when hurrying to get to a shopping mall before it closes. The determinants of perceived time pressure, and their effects on drivers’ behavior at traffic lights, are difficult to study by means of observation or on a driving simulator. These research methods make it possible to directly record various driving parameters, but provide no information about the specific personal context of a trip by car. Future research on factors influencing the perception of time pressure and its effect on drivers’ behavior will probably have to use self-report methods, while taking care to tackle the biases inherent in this approach (Lajunen & Summala, 2003). A second important point concerns the perception of risk while approaching and going through intersections with a traffic light. It is important to distinguish subjective risk from risk related feelings (Fuller, 2005). Risk feelings emerge as a psychophysiological response (fear) to situations perceived as dangerous, whereas subjective-risk assessment is a judgment of the probability that a threat and the associated risk feelings will emerge in a given driving situation. The latter is a cognitively elaborate process that demands time and attentional resources. It follows that it would be difficult for a driver to estimate the subjective level of risk related to every decision made behind the wheel unless he/she has enough time and available attentional resources (e.g., when driving on a straight, empty highway). Such probability judgments can be derived, however, from the frequency of occurrence (Estes, 1976) of risk feelings in a given driving situation (e.g., overtaking, obstacle

116

B. Palat, P. Delhomme / Transportation Research Part F 37 (2016) 107–118

avoidance, etc.), defined at some level of abstraction (e.g., overtaking a long vehicle on a two-lane road). The probability of undesirable consequences (a crash, a ticket) in different driving situations can also be learned during driver’s training courses, from other individuals, and/or via risk communication campaigns. Representations of driving situations can contain many contextual elements (e.g., weather conditions, time of day, etc.), but they cannot include the exact configuration of vehicles, infrastructure, and dynamically changing time-to-collision margins relating the driver to the road environment. Reduced time-to-collision margins remain the most important factor triggering risk feelings. Our research examined the influence of subjective risk related to yellow- and red-light running, i.e., the perceived probability that running a yellow or red light will have undesirable consequences. However, we did not examine whether risk feelings (e.g., feelings related to the anticipation of interactions with traffic on the other streets, or the car in front) play a role in determining drivers’ behavior at traffic lights. This kind of feeling (fear) could have been assessed with physiological measures such as heart rate or skin conductance. We chose not to take such measures because it could inconvenience the participants. Besides, driving a simulator would probably not elicit the same level of physiological response as real driving. The behavior of our participants on the simulator should nevertheless be very similar to their real driving behavior, since they were explicitly asked to drive as if they were driving the car they use the most often. During everyday trips by car, drivers can often be distracted in their otherwise highly automated task. Less automated reflexes such as, for instance, directing attention to important elements of road environment could sometimes be strongly affected by concurrent activities (FHWA Safety Program mentions eating, talking on mobile phone, and manipulating devices inside the car cabin as distractions responsible for red-light running). For this reason, in real driving conditions drivers who run red lights deliberately have to be distinguished from those who do not stop because they fail to see the traffic light (Green, 2003). However, in a driving scenario which was used in the present study such distractions were absent. Moreover, participants were given the necessary time required to get used to driving on the simulator so that they should not be distracted by the characteristics of an unfamiliar vehicle. We therefore assumed that it would be very improbable in these conditions that a participant fails to see a traffic light. In contrary, it is quite probable that a driver who approaches a yellow traffic light does not see another vehicle approaching the crossroads on the perpendicular road (for which the light is still red) until that vehicle gets very close to the crossroads. If neither the driver for whom the light is yellow, nor the one for whom it is red then stops, a dangerous lateral collision will be very likely. Now, let us imagine that the driver who approaches the yellow light sees the vehicle approaching on the perpendicular road sufficiently early to react on time. It is then less probable that the driver would attempt to run the yellow light. It would also be less probable that he would attempt to run the yellow light if he sees a stationary vehicle waiting at the red light on the perpendicular road by fear that the stationary vehicle could start moving into the crossroads and collide with him. Therefore from the standpoint of the risk of lateral collisions, situations where the driver approaching a yellow light does not see any vehicles (neither moving nor stationary) on the perpendicular road merit more attention. This is the primary reason why in our study the presence of vehicles on the roads perpendicular to the participants’ direction of travel was not operationalized. Moreover, from a purely technical standpoint, measuring the influence of vehicles on the perpendicular roads would demand a more complex and lengthy driving scenario, and therefore a greater risk of simulator sickness. Besides, when too many factors at a time are examined in one study, they can interact in a way difficult to interpret for the researcher who is then incapable to arrive at valid conclusions. Hence, we believe that the presence of other road users on perpendicular roads as a factor influencing drivers’ decision making at traffic lights should be thoroughly examined in future research by means of experimental designs conceived specially for this purpose. Last but not least, one could argue that the effects of social influence were due to an increased workload in the dual-task situation where the drivers were simultaneously following a vehicle in front of them and watching the traffic light. Even though we are convinced that driving is so highly automated that both these tasks could be regulated very efficiently, perhaps more attention should be paid to measuring workload in order to account for its effects on decision making at traffic lights in the future research. As regards the practical implications of our study, it seems that stressing the risks related to yellow- and red-light running or slamming on the brakes at a traffic light during driver’s training or communication campaigns might not be effective in reducing this kind of dangerous behavior. On the other hand, time pressure and social context of driving are powerful factors influencing drivers’ decision-making at traffic lights, irrespective of subjective risk. Increasing subjective risk might nevertheless be effective in reducing yellow- and red-light running indirectly. When drivers are not under time pressure, they often run a yellow or red light if they are approaching the intersection at a high speed. An increased level of subjective risk could make drivers slow down before the intersection, and therefore reduce the probability that they run the yellow light. The same effect could be achieved by further reducing speed limits before intersections with traffic lights. Under time pressure, however, drivers are still likely to be highly motivated to run a yellow or red light. In such cases, the only way to override the weight of time gain may be to show drivers that there will inevitably be negative consequences of this illegal behavior. This could be achieved if red-light runners were always ticketed. For this, the policy of increasing the number of red-light cameras as a means of law enforcement seems reasonable. However, before this policy becomes effective, drivers’ perceived risk of being ticketed has to be very high. In other words, many cameras have to be installed to convince drivers that running a red light always results in a ticket. By contrast, non-coercive measures could be used to tackle the impact of social context on drivers’ behavior at traffic lights. Social comparisons increase competition between individuals (McClintock & McNeel, 1966), which is one of the factors conducive to risky driving behaviors (Vingilis et al., 2013). Emphasizing that driving is a cooperative rather than

B. Palat, P. Delhomme / Transportation Research Part F 37 (2016) 107–118

117

competitive activity (Delhomme & Meyer, 1998) should foster relaxed and safety-oriented attitudes among drivers. The perception of the driving activity may be affected via driver’s training, highway message boards, or communication campaigns. More cooperation-oriented drivers are likely to be less sensitive to situations where others are about to go through a yellow light when they themselves have to stop and wait. This would make them more likely to stop at the yellow light instead of running it to follow the other cars. Then, when the light turns green, instead of accelerating rapidly to catch up with drivers who managed to get through the previous light, or to impress the drivers behind them, they may be more careful, but still start back up in time in order to enable as many vehicles as possible to pass through the intersection during the green-light phase. References Allen, T. M., Lunenfeld, H., & Alexander, G. J. (1971). Driver information needs. Highway Research Record, 366, 102–115. Bandura, A. (1977). Social learning theory. New York, NY: General Learning Press. Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173–1182. Cialdini, R. B., & Trost, M. R. (1998). Social influence social norms conformity and compliance. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.). The handbook of social psychology: Special fields and applications (4th ed.) (Vol. 2, pp. 151–192). Boston, MA: McGraw-Hill. Cœugnet, S., Naveteur, J., Antoine, P., & Anceaux, F. (2013). Time pressure and driving: Work, emotions and risks. Transportation Research Part F, 20, 39–51. Connolly, T., & Åberg, L. (1993). Some contagion models of speeding. Accident Analysis and Prevention, 25(1), 57–66. Csíkszentmihályi, M. (1989). Optimal experience in work and leisure. Journal of Personality and Social Psychology, 56(5), 815–822. Delhomme, P. (2002). Croyances des jeunes automobilistes en matière de vitesse [Rapport final. Convention DSCR-INRETS N°00/010/T- étude n°7]. Arcueil: INRETS. Delhomme, P., & Meyer, T. (1998). Control motivation and young drivers’ decision making. Ergonomics, 41(3), 373–393. Deutsch, M., & Gerard, H. B. (1955). A study of normative and informational social influences upon individual judgment. The Journal of Abnormal and Social Psychology, 51(3), 629–636. Elmitiny, N., Yan, X., Radwan, E., Russo, C., & Nashar, D. (2010). Classification analysis of driver’s stop/go decision and red-light running violation. Accident Analysis and Prevention, 42, 101–111. Estes, W. K. (1976). The cognitive side of probability learning. Psychological Review, 83(1), 37–64. Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7(2), 117–140. FHWA Safety Program. Jul. 19, 2011. Fuller, R. (1984). A conceptualization of driving as threat avoidance. Ergonomics, 27(11), 1139–1155. Fuller, R. (2005). Towards a general theory of driver behaviour. Accident Analysis and Prevention, 37, 461–472. Fuller, R., McHugh, C., & Pender, S. (2008). Task difficulty and risk in the determination of driver behaviour [la difficulté de la tâche et le risque dans le comportement des conducteurs]. Revue Européenne de Psychologie Appliquée, 58, 13–21. Gazis, D., Herman, R., & Maradudin, A. (1960). The problem of the amber signal light in the traffic flow. Operations Research, 8(1), 112–132. Godthelp, H., Milgram, P., & Blaauw, G. J. (1984). The development of a time-related measure to describe driving strategy. Human Factors, 26(3), 257–268. Green, F. (2003). Red-light running (Research report no. APR 356). Vermont South, Australia: ARRB Transport Research Ltd. Gulian, E., Matthews, G., Glendon, A. I., & Davies, D. R. (1989). Dimensions of driver stress. Ergonomics, 32, 719–726. Heyes, C. (2011). Automatic imitation. Psychological Bulletin, 137(3), 463–483. Hollnagel, E. (2002). Time and time again. Theoretical Issues in Ergonomics Science, 3(2), 143–158. Hollnagel, E., Nåbo, A., & Lau, I. (2003). A systemic model for driver-in-control. In Proceedings of the 2nd international driving symposium on human factors in driver assessment, training, and vehicle design. Park City, UT: Public Policy Center, University of Iowa. Jørgensen, N. (1988). Risky behaviour at traffic signals: A traffic engineer’s view. Ergonomics, 31(4), 657–661. Konecˇni, V. J., Ebbesen, E. B., & Konecˇni, D. K. (1976). Decision process and risk taking in traffic: Driver response to the onset of yellow light. Journal of Applied Psychology, 61(3), 359–367. Lajunen, T., & Summala, H. (2003). Can we trust self-reports of driving? Effects of impression management on driver behaviour questionnaire responses. Transportation Research Part F, 6, 97–107. Liu, C., Herman, R., & Gazis, D. C. (1996). A review of the yellow interval dilemma. Transportation Research Part A, 30(5), 333–348. Lupton, D. (2002). Road rage: Drivers’ understandings and experiences. Journal of Sociology, 38(3), 275–290. McClintock, C. G., & McNeel, S. P. (1966). Reward and score feedback as determinants of cooperative and competitive game behavior. Journal of Personality and Social Psychology, 4(6), 606–613. Michon, J. A. (1985). A critical view of driver behavior models: What do we know, what should we do? In L. Evans & R. C. Schwing (Eds.), Human behavior and traffic safety (pp. 485–520). New York, NY: Plenum Press. Ministère de l’Ecologie, de l’Energie, du Développement Durable et de la Mer (2012). Instructioninterministérielle sur la signalisation routière. Retrieved September 25, 2013. Moget-Monseur, M., & Biecheler-Fretel, M. B. (1985). Le comportement de base du conducteur. Un essai de conceptualisation du système de normes légales et sociales de l’usager de la route (Cahiers d’études Onser no. 64). Arcueil, France: Organisme National de Sécurité Routière. Näätänen, R., & Summala, H. (1976). Road-user behaviour and traffic accidents. Amsterdam, North Holland: Elsevier. Naveteur, J., Cœugnet, S., Charron, C., Dorn, L., & Anceaux, F. (2013). Impatience and time pressure: Subjective reactions of drivers in situations forcing them to stop their car in the road. Transportation Research Part F, 18, 58–71. Özkan, T., & Lajunen, T. (2011). Person and environment: Traffic culture. In B. E. Porter (Ed.), Handbook of traffic psychology (1st ed., pp. 179–192). London, UK: Academic Press, Elsevier. Palat, B., & Delhomme, P. (2012). What factors can predict why drivers go through yellow traffic lights? An approach based on an extended theory of planned behavior. Safety Science, 50(3), 408–417. Porter, B. E., & Berry, T. D. (2001). A nationwide survey of self-reported red light running: Measuring prevalence, predictors, and perceived consequences. Accident Analysis and Prevention, 33(6), 735–741. Porter, B., & England, K. (2000). Predicting red-light running behavior: A traffic safety study in three urban settings. Journal of Safety Research, 31(1), 1–8. Rosenbloom, T., & Wolf, Y. (2002a). Signal detection in conditions of everyday life traffic dilemmas. Accident Analysis and Prevention, 34, 763–772. Rosenbloom, T., & Wolf, Y. (2002b). Sensation seeking and detection of risky road signals: A developmental perspective. Accident Analysis and Prevention, 34, 569–580. Rothengatter, T. (1988). Risk and the absence of pleasure: A motivational approach to modelling road user behaviour. Ergonomics, 31(4), 559–607. Saad, F., Delhomme, P., & Van Elslande, P. (1990). Driver’s speed regulation when negotiating intersections. In M. Koshi (Ed.), Transportation and traffic theory. Proceedings of the eleventh international symposium on transportation and traffic theory (pp. 193–212). Yokohama, Japan: Elsevier. Saad, F., Van Elslande, P., Delhomme, P., Lepesant, C., & Gaujé, T. (1989). Analyse des comportements en situation de conduite critique: Le franchissement d’intersections. Convention DSCR-INRETS. Contrat N°8741050 T 00 222 75 01, 50P. (+ XXVP.). Arcueil, France: Institut National de Recherche sur les Transports et leurs Sécurité.

118

B. Palat, P. Delhomme / Transportation Research Part F 37 (2016) 107–118

Schmidt-Daffy, M. (2012). Velocity versus safety: Impact of goal conflict and task difficulty on drivers’ behaviour, feelings of anxiety, and electrodermal responses. Transportation Research Part F, 15(3), 319–332. Scott-Parker, B., Watson, B., King, M. J., & Hyde, M. K. (2012). ‘‘They’re lunatics on the road”: Exploring the normative influences of parents, friends, and police on young novices’ risky driving decisions. Safety Science, 50(9), 1917–1928. Shelley, E., Lobel, T., & Lobel, M. (1989). Social comparison activity under threat: Downward evaluation and upward contacts. Psychological Review, 96(4), 569–575. Shinar, D. (1998). Aggressive driving: The contribution of the drivers and the situation. Transportation Research Part F, 1, 137–160. Sigelman, C. K., & Sigelman, L. (1976). Authority and conformity: Violation of a traffic regulation. The Journal of Social Psychology, 100, 35–43. Suls, J., Martin, R., & Wheeler, L. (2002). Social comparison: Why, with whom, and with what effect? Current Directions in Psychological Science, 11(4), 159–163. Summala, H. (2007). Towards understanding motivational and emotional factors in driver behaviour: Comfort through satisficing. In P. Cacciabue (Ed.), Modelling driver behaviour in automotive environments: Critical issues in driver interactions with intelligent transport systems (pp. 189–207). London, UK: Springer-Verlag. Summala, H. (1997). Hierarchical model of behavioural adaptation and traffic accidents. In T. Rothengatter & E. C. Vaya (Eds.), Traffic and transport psychology: Theory and application (pp. 41–52). Oxford, UK: Elsevier Science. Taubman, O., Mikulincer, M., & Iram, A. (1996). The cognitive, motivational and emotional system of driving. Research report. Israel: Department of Casualties and Road Safety of the Israeli Army. Taubman-Ben-Ari, O., & Katz-Ben-Ami, L. (2012). The contribution of family climate for road safety and social environment to the reported driving behavior of young drivers. Accident Analysis and Prevention, 47, 1–10. TNS Sofres, AXA, AXA Prévention (2011). Baromètre axa prévention du comportement des français au volant – Vague 7 [A survey on risky driving behaviors]. Retrieved April 22, 2011. van der Molen, H., & Botticher, A. (1988). A hierarchical model for traffic participants. Ergonomics, 36(5), 557–567. Vingilis, E., Seeley, J., Wiesenthal, D. L., Wickens, C. M., Fischer, P., & Mann, R. E. (2013). Street racing video games and risk-taking driving: An internet survey of automobile enthusiasts. Accident Analysis and Prevention, 50, 1–7. Wilde, G. J. S. (1982). The theory of risk homeostasis: Implications for safety and health. Risk Analysis, 2, 209–225. Xu, Y., Li, Y., & Jiang, L. (2014). The effects of situational factors and impulsiveness on drivers’ intentions to violate traffic rules: Difference of driving experience. Accident Analysis and Prevention, 62, 54–62. Yagil, D. (2001). Reasoned action and irrational motives: A prediction of drivers’ intention to violate traffic laws. Journal of Applied Social Psychology, 31, 720–740. Yinon, Y., & Levian, E. (1995). Presence of other drivers as a determinant of traffic violations. The Journal of Social Psychology, 135(3), 299–304.