This article was originally published in a journal published by Elsevier, and the attached copy is provided by Elsevier for the author’s benefit and for the benefit of the author’s institution, for non-commercial research and educational use including without limitation use in instruction at your institution, sending it to specific colleagues that you know, and providing a copy to your institution’s administrator. All other uses, reproduction and distribution, including without limitation commercial reprints, selling or licensing copies or access, or posting on open internet sites, your personal or institution’s website or repository, are prohibited. For exceptions, permission may be sought for such use through Elsevier’s permissions site at: http://www.elsevier.com/locate/permissionusematerial

Accident Analysis and Prevention 39 (2007) 843–852

py

Effects of major-road vehicle speed and driver age and gender on left-turn gap acceptance

co

Xuedong Yan a,∗ , Essam Radwan a,1 , Dahai Guo b,2 a

Center for Advanced Transportation Systems Simulation, Department of Civil & Environmental Engineering, University of Central Florida, Orlando, FL 32816-2450, United States b Computer Science Program, College of Business, Florida Gulf Coast University, 10501 FGCU Boulevard South, Fort Myers, FL 33965-6565, United States

al

Received 28 August 2006; received in revised form 1 November 2006; accepted 12 December 2006

Abstract

pe

rs

on

Because the driver’s gap-acceptance maneuver is a complex and risky driving behavior, it is a highly concerned topic for traffic safety and operation. Previous studies have mainly focused on the driver’s gap acceptance decision itself but did not pay attention to the maneuver process and driving behaviors. Using a driving simulator experiment for left-turn gap acceptance at a stop-controlled intersection, this study evaluated the effects of major traffic speed and driver age and gender on gap acceptance behaviors. The experiment results illustrate relationships among drivers’ left-turn gap decision, driver’s acceleration rate, steering action, and the influence of the gap-acceptance maneuver on the vehicles in the major traffic stream. The experiment results identified an association between high crash risk and high traffic speed at stop-controlled intersections. The older drivers, especially older female drivers, displayed a conservative driving attitude as a compensation for reduced driving ability, but also showed to be the most vulnerable group for the relatively complex driving maneuvers. © 2007 Elsevier Ltd. All rights reserved. Keywords: Driving simulator; Gap acceptance; Driving behavior; Stop-controlled intersection; Driver age difference; Driver gender difference

1. Introduction

∗

Au

th o

r's

Two-way stop-controlled (TWSC) intersections are the most prevalent intersection type in the United States (Gattis and Low, 1999). At stop-controlled intersections, drivers on the minor road need to make use of proper gaps among the traffic to cross or merge into the major road. Because the driver’s gapacceptance maneuver is a complex and risky driving behavior, it is a highly concerned topic for traffic safety and operation. Retting et al. (2003) reported that nearly 700,000 motor vehicle crashes occurred annually at stop signs in US and approximately one-third of these crashes involved injuries in US. In Minnesota, there were 34,175 reported crashes on rural twolane roads between 2000 and 2002. Over 32% (11,069) of these crashes were intersection related, with 22% of the fatal rural Corresponding author. Tel.: +1 407 823 5810; fax: +1 407 823 4676. E-mail addresses: [email protected] (X. Yan), [email protected] (E. Radwan), [email protected] (D. Guo). 1 Tel.: +1 407 823 2945; fax: +1 407 823 4676. 2 Tel.: +1 239 590 7583; fax: +1 239 590 7330. 0001-4575/$ – see front matter © 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.aap.2006.12.006

accidents occurring at stop-controlled intersections (Laberge et al., 2006). Previous research has identified gap acceptance problems as a significant contributor to the stop-controlled intersection accidents and indicated that incorrect gap acceptance might cause around 30% of left-turn accidents (Chovan et al., 1994). As an essential driving performance at stop-controlled intersections, the gap acceptance decision has been used as an important measurement to analyze and predict traffic conflicts and accident rates at intersections (Spek et al., 2006; Alexander et al., 2002). Furthermore, in the current AASHTO Manual (2001), the gap acceptance methodology is applied to determine intersection sight distances, which is based on previous research works (Harwood et al., 2000). In the Highway Capacity Manual (HCM, 2000), the critical gap accepted by drivers is a key parameter to calculate the minor traffic capacity. However, both important highway design manuals ignored the effects of important factors such as driver age and gender and major-road vehicle speed on the gap acceptance, which has generally been paid attention to by researchers in traffic safety aspects.

844

X. Yan et al. / Accident Analysis and Prevention 39 (2007) 843–852

same sex. Through both simulator and field measures, Staplin (1995) indicated that older drivers show relative insensitivity to vehicle approach speed in left-turn maneuvers across the major road traffic when compared with younger drivers. This may increase the risk of accidents if there is a lone speeder in the traffic scheme. Abdel-Aty et al. (1999) pointed out that at intersections, elderly drivers are over-represented in right and left turns as well as angle collisions. Scialfa et al. (1991) concluded that older drivers generally over-estimate the speed of vehicles traveling at low speeds, while under-estimating the speed of those traveling much faster, which could explain the over-involvement of elderly drivers in accidents at junctions.

1.2. Age and gender differences in gap acceptances

co

al

Although numerous previous studies have been done for the gap-acceptance research, all of them only focused on the gap acceptance decision itself but did not pay attention to the maneuver process and related driving behavior. Using a driving simulator experiment for left-turn gap acceptance, this study evaluated the effects of major traffic speed and driver age and gender on gap acceptance behaviors, such as minimum accepted gap, driver’s acceleration rate, steering action, and influence of the gap-acceptance maneuver on the vehicles in the major traffic stream. This research aimed at analyzing the interactive effect between gap-acceptance drivers and vehicles on the major road, and identifying the driving behavior differences associated with major traffic speed and driver age and gender.

Au

th o

r's

pe

rs

It was reported that the process that driving performance becomes progressively poorer with age accelerates somewhere in the fifth decade of life (National Highway Traffic Safety Administration, 1993). Older drivers have problems to adequately detect, perceive and accurately judge the safety of a gap (Laberge et al., 2006). Therefore, older drivers may experience greater difficulties at nonsignalized intersections as the result of diminished visual capabilities, such as depth and motion perception. Prior results indicated that judgments about whether a potential collision would occur were less accurate for older drivers (40–64 years) compared with younger drivers (18–29 years) (DeLucia et al., 2003). Lyles and Staplin (1991) examined police-reported accidents in Michigan and Pennsylvania and found that when these were ordered according to olderdriver involvement rate, turning left across on-coming traffic and crossing or turning into a traffic stream were found to be the most dangerous maneuvers. Moreover, it was indicated that older drivers are over-represented in severe injury crashes at intersections due in part to increases in frailty and functional disabilities that occur with age (Oxley et al., 2006). Lerner et al. (1995) concluded that older drivers are particularly overrepresented in multiple-vehicle, intersection-related accidents, which could be interpreted to reflect slow detection of and reaction to other vehicles, or slow decision time for intersection-related maneuvers. Lerner et al. (1995) collected gap acceptance data as a function of driver age for left turn, right turn, and through movements at stop-controlled intersections. The findings indicated that younger drivers (20–40 years) accept shorter gaps than older drivers (over 65 years). Alexander et al. (2002) found that females require a larger median gap than males, irrespective of age, and older subjects (65–79 years) require a larger gap than younger subjects (under 60 years) of the

1.3. Objective of the study

on

In the AASHTO and HCM manuals, it is assumed that the critical gap accepted by drivers does not vary with major-road vehicle speed, which is based on the previous analysis by Kyte et al. (1996). However, in the simulator experiment for turning left from the major road into the minor road (Alexander et al., 2002), the velocity of the on-coming traffic was the variable that had the greatest effect on the median accepted gap size. This result corresponds to those of other previous studies involving gap acceptance, which had shown that drivers accept a smaller gap at higher approach velocities (Darzentas et al., 1980; Staplin, 1995). Another field study indicated that for a given time gap, the probability that a driver accepts the gap increases as the speed of the opposing vehicle increases, which implies that the distance is the primary determinant of gap acceptance (Davis and Swenson, 2004). Based on the speed dependency of gap acceptance decision, Spek et al. (2006) developed a theoretical crash prediction model and found that the probability that a crossing vehicle collides with the major stream vehicle can be expected to increase when the major traffic speed increases.

py

1.1. Effect of traffic speed on gap acceptances

2. Methodology 2.1. Participants The experiment was a 3 (age) × 2 (gender) × 2 (major traffic speed) within-subject repeated measures design. Sixty-three paid participants in three age groups, ranging in age from 20 to 83 years, were recruited for this research. Table 1 lists the descriptive statistics of subjects by age and gender. The young group ranged from 20 to 30 years old, the middle-age group ranged from 31 to 55 years old, and the old group ranged from 56 to 83 years old. Every participant held a valid Florida’s driver’s license with at least 2 years of driving experience. The experiTable 1 Descriptive statistics of subjects by age and gender Age

Gender

Mean

Standard deviation

N

Young (20–30)

Female Male

24.00 23.57

2.42 2.85

14 14

Middle (31–55)

Female Male

39.10 33.64

9.27 3.29

10 11

Old (56–83)

Female Male

69.75 73.30

8.66 8.86

4 10

Total

Female Male

35.93 40.94

16.93 21.83

28 35

X. Yan et al. / Accident Analysis and Prevention 39 (2007) 843–852

ment lasted for about 20 min in total and the participants were compensated 20 dollars for their participation.

845

of the experiment, the driver first drives the simulator for 400 m along the minor road (stage 1). When the simulator approaches the intersection, all vehicles on the major road begin to enter the scene. Then, according to the voice reminder and indication sign (stop sign and left-turn sign), the driver stops at the intersection and waits for the appropriate gap in the traffic on the major road to make his/her left turn maneuver (stage 2). When the simulated vehicle enters the major road and its speed is caught up with the major road traffic speed (stage 3), the experiment is completed. All subjects were tested in two scenarios, 40.2 km/h (25 mph) major road speed (scenario A) and 88.5 km/h (55 mph) major road speed (scenario B), which is shown in Fig. 1.

2.2. Apparatus/equipment

co

py

This study used the UCF (University of Central Florida) driving simulator as a tool for data collection. The driving simulator is an I-Sim Mark-II system with a motion base capable of operation with 6 degrees of freedom. It includes five channels (one forward, two side views and two rear view mirrors) of image generation, an audio and vibration system, and steering wheel feedback. The simulated environment is projected at 180◦ of field view and at a resolution of 1280 × 1024 pixels. The driving simulation system is composed of the following components:

2.4. Traffic design on the major road

on

al

A challenge in designing this experiment was how to make the drivers perform their minimum left-turn gap-acceptance maneuver in the same way accomplished in the real world. For obtaining that goal, the oncoming traffic on the major road from the right was composed of two classes of intermingled gaps to make the traffic appear random. One gap classification had very small gaps (less than 3 s) that were unlikely to be accepted by the participants. The other class consisted of increasing gaps in which the subsequent gap was 1 s larger than the previous one. This design assured that the selected gap would be close to the driver’s minimum acceptable gap and the major road traffic looked realistic. This concept is illustrated in Fig. 2. The uniformly increasing gaps ranged in duration from 1 to 16 s, a large enough variation to accommodate all drivers. A previous gap acceptance study suggested that there is no meaningful information regarding driver gap acceptance behavior when the accepted gap size is above 12 s (Gattis and Low, 1999).

rs

• Simulator Cab: Saturn Sedan, automatic transmission, air conditioning, the left back mirror and the back mirror inside the cab. • Simview: The software provides the graphical display based on the computation. • Motion base: It provides motion when driving. • Scenario Editor: The software helps researchers to edit a tested traffic scenario. • APIs for reading real-time data: APIs (Application Programmer Interface) can read the real-time data from Simview. The sampling frequency is 60 Hz. 2.3. Description of traffic scenarios

Au

th o

r's

pe

In the gap acceptance experiments, a long straight undivided two-lane highway was used as the major road. Its length is around 3000 m with a lane width of 3.66 m. A driving scenario consists of three stages as shown in Fig. 1. During the course

Fig. 1. Traffic scenario design for left-turn gap acceptance.

Fig. 2. Traffic design on the major road.

846

X. Yan et al. / Accident Analysis and Prevention 39 (2007) 843–852

Whenever a car on the major road approached the intersection, it was automatically changed from “record vehicle” to the “normal vehicle”. The “record vehicles” move along a fixed path with the unchangeable speed from the start point to the end point, whereas the “normal vehicles” intelligently move along the given route. While the normal vehicles follow a designed speed, they can decelerate, accelerate, and pass the slow impeding car according to the traffic situation. Therefore, if the simulator is turning left into the major road very slowly, the coming car on the major road will decrease its speed to avoid a crash. Unless some participants drive the simulator too aggressively or disobey the rationale, there will be no crashes.

The orders of scenarios were presented as A–B or B–A randomly for subject so as to eliminate the time order effect. For security and liability reasons, each subject was escorted to the simulator cabin to commence the experiment and he/she was allowed at least 2 min to rest before running the next scenario. 2.6. Dependent measures

co

py

In the simulator experiment, the measured parameters related to gap acceptance maneuver include minimum time gap accepted by left turners (GAP), average steering angle velocity during the left turn (SAV), average acceleration of the simulator during the left-turn maneuver (ACC), minimum separation between the minor-road vehicle and the following major-road vehicle (MCD), speed reduction rate of the following vehicle on the major road (SRR), and the deceleration rate of the following vehicle (DEC). These dependent measures are explained as follows: • GAP (in s): the time headway between two vehicles on the major road into which a left-turn driver may choose to turn. • SAV (in rad/s): total rotation angles are divided by the total time during which the simulator steer turn left and turn back when subjects complete the left turn maneuver. • ACC (in m/s2 ): average acceleration of simulator vehicle during the period of making left turn maneuver. • MCD (in m): minimum clearance distance between the simulator and the following major-road vehicle before the left-turn maneuver is completed. • SRR: the speed reduction rate of the following vehicle on the major road to accommodate the turning vehicle.

on

pe

rs

Upon arrival, the subjects were asked to fill out and sign an informed consent form (per IRB). The subjects were then advised to drive and behave as they normally would and to adhere to traffic laws as in real life situations. The subjects were also notified that they could quit the experiment at any time in case of motion sickness or any kind of discomfort. Prior to the formal experiment, drivers were trained for at least 5 min to familiarize with the driving simulator operation. During the course of the practice, subjects exercised selected maneuvers including straight driving, acceleration, deceleration, left/right turn, and other basic driving behaviors. After completing the familiarity course, the formal experiments began during which all subjects faced the same set of two driving scenarios, denoted A for the lower major traffic speed (40.2 km/h) and B for the higher major traffic speed (88.5 km/h).

al

2.5. Experiment procedure

Table 2 Descriptive statistical results for dependent variables Parameter

Speed 40.2 km/h (N = 63)

Middle (N = 42) Old (N = 28)

Gender Female (N = 56) Male (N = 70)

Total (N = 126)

MCD (m)

SRR

DEC (m/s2 )

7.44 2.08

3.06 0.79

1.82 0.61

54.16 17.75

0.02 0.05

0.11 0.21

5.82 1.46

1.89 0.68

24.53 13.38

0.24 0.13

1.20 1.25

2.99 0.82

Mean S.D.

6.29 1.79

3.08 0.69

2.04 0.60

37.82 18.68

0.12 0.14

0.58 0.81

Mean S.D.

6.20 1.50

3.16 0.82

1.86 0.61

34.71 18.58

0.12 0.14

0.85 1.51

Mean S.D.

7.94 2.38

2.71 0.92

1.47 0.63

49.35 27.96

0.16 0.17

0.52 0.47

Mean S.D.

6.93 2.02

3.03 0.80

1.81 0.66

40.88 22.90

0.14 0.17

0.65 0.86

Mean S.D.

6.38 1.90

3.02 0.81

1.89 0.64

38.12 20.58

0.12 0.12

0.66 1.18

Mean S.D.

6.63 1.97

3.03 0.80

1.85 0.65

39.34 21.60

0.13 0.14

0.65 1.05

Mean S.D. Mean S.D.

Au

Age Young (N = 56)

ACC (m/s2 )

th o

88.5 km/h (N = 63)

SAV (rad/s)

GAP (s)

r's

Factors

X. Yan et al. / Accident Analysis and Prevention 39 (2007) 843–852

• DEC (in m/s2 ): the deceleration rate used by the following vehicle on the major road to accommodate the turning vehicle.

Table 5 Mean values of GAP for speed × gender × age

The basic statistical descriptions of the experiment results are listed in Table 2. In the subsequent statistical analyses, an ANOVA was used to investigate differences between factors (see Table 3). The hypothesis testing in the following analysis was based on a 0.05 significance level.

Young

Middle

Old

40.2 km/h Female Male

7.56 6.35

6.97 6.60

10.99 8.76

88.5 km/h Female Male

6.00 5.26

5.63 5.61

7.11 6.23

is lowest and its speed reduction is greatest. An example for this analysis is illustrated in Fig. 3. When time unit is equal to 114.45, the clearance distance is minimum and equal to 19.53 m, and the speed of the following car is 63.46 km/h, which is close to that (63.14 km/h) of the simulator vehicle. It can be concluded that the minimum clearance distance and the maximum speed reduction always happen at the same experiment sampling time. After minimum clearance distance occurs, the left turn maneuver can be considered to be successfully completed by the subject. The smaller MCD and the larger SRR and DEC indicate that the left-turn vehicles have more effects on the major road traffic.

on

al

co

3.1. Correlation analyses for dependent variables

Category

py

3. Experiment results

3.2. Minimum gap acceptance

Table 5 lists mean values of GAP for each speed and for each combination of gender × age. The ANOVA results show the significant effects of age (p < 0.001), gender (p = 0.004), speed (p < 0.001), and two-way interaction between age and speed (p = 0.035) on driver’s gap acceptance (see Table 3).

pe

rs

Table 4 shows the correlation analysis among the dependent variables. Clearly, the GAP is correlated with all the other parameters. Intuitively, if drivers accept smaller gaps, they tend to make left-turn maneuvers with a faster steering velocity and higher acceleration rate, and result in shorter minimum clearance distances to the following vehicles, and larger percentage of speed reduction and higher deceleration rates of the following vehicles. Furthermore, it was found that MCD, SRR, and DEC are highly correlated with each other. When the left-turn vehicle enters onto the major road at a slow speed, the following car on the major road decreases its speed to avoid a crash. As the following car on the major road is approaching the turning-into vehicle whose speed keeps increasing, the distance between the turning-into vehicle and the following major-road vehicle gets smaller, until the minimum clearance distance between the two vehicles occurs. At that same time, the following car’s speed

847

Table 3 Analysis of variance table for dependent measures Source

d.f.

F-ratio

*

GAP

SAV

ACC

1 −.280** −.302** .591** −.218* −.298**

−.280**

−.302**

14.798** 37.044** 1.468 1.314 3.462* .204 2.560

ACC

MCD

2.601 9.587** .159 1.557 1.003 .018 .171 .382

SRR 6.690*

3.660 9.749** 111.678** .337 .009 1.824 .063 219.173

4.447* 155.082** 4.053* 2.012 .812 1.194 .009

DEC .024 1.325 36.781** .138 .019 .545 .302 .837

Significant at the 0.01 level. Significant at the 0.05 level.

Au

**

.112 3.587* .752 2.943 1.129 .388 .029 .621

1 2 1 2 1 2 2 114

8.514**

th o

Gender Age Speed Gender × age Gender × speed Age × speed Gender × age × speed Mean square error

SAV

r's

GAP

Table 4 Correlation among dependent variables

GAP SAV ACC MCD SRR DEC ** *

1 .425** −.074 −.224* −.081

Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level (2-tailed).

.425** 1 .324** −.455** −.234**

MCD .591** −.074 .324** 1 −.557** −.448**

SRR

DEC

−.218*

−.298** −.081 −.234** −.448** .593** 1

−.224*

−.455** −.557** 1 .593**

X. Yan et al. / Accident Analysis and Prevention 39 (2007) 843–852

co

py

848

Fig. 4. Minimum GAP vs. driver’s actual age by gender.

r's

pe

rs

on

al

accepted by male drivers tend to gradually increase as the driver age increases while those for female drivers show a U-shape pattern, which indicates that middle age groups accept smaller gaps and the younger and older groups have relatively larger gaps. Furthermore, it shows that the gender difference in GAP is not significant for drivers from 35 to 55 years; however, for drivers younger than 35 years or older than 55 years, females appear to accept larger gaps than males. The results show that the oncoming vehicle speed on the major road plays an important role in the gap size accepted by drivers. Drivers would accept smaller gaps (M = 5.82 s, S.D. = 1.46 s) for the higher major-road traffic speed scenario compared to those (M = 7.44 s, S.D. = 2.08 s) for the lower major-road traffic speed scenario. It implies that drivers rely on both space and time information to perform gap-acceptance judgments, which is corresponding to the conclusion of prior studies (Alexander et al., 2002; Davis and Swenson, 2004). Furthermore, the interaction effect between speed and driver age shows that for the lower major-road traffic speed scenario, there is a strong gap acceptance difference in driver age [F(2, 60) = 34.756, p < 0.001], and the old drivers apparently accept larger gaps (M = 9.40 s, S.D. = 2.24 s) compared to the younger drivers (M = 6.96 s, S.D. = 1.82 s) and the middle-age drivers (M = 6.77 s, S.D. = 1.49 s); however, for the higher major-road traffic speed scenario, the minimum gaps accepted by old drivers (M = 6.48 s, S.D. = 1.48 s) are not significantly larger than the younger drivers (M = 5.63 s, S.D. = 1.51 s) and the middle-age drivers (M = 5.62 s, S.D. = 1.31 s) [F(2, 60) = 1.926, p = 0.155]. The interaction effect between speed and driver age is illustrated in Fig. 5.

th o

Fig. 3. Relationship between clearance distance and speeds of the left-turning vehicle (simulator) and the following car.

Au

Old drivers tend to accept larger gaps (M = 7.94 s, S.D. = 2.38 s), compared to younger drivers (M = 6.29 s, S.D. = 1.79 s) and middle-age drivers (M = 6.20 s, S.D. = 1.50 s). There is no significant difference between young drivers and middle-age drivers [t(96) = 0.283, p = 0.778]. For the gender factor, it appears that male drivers accept smaller gaps (M = 6.38 s, S.D. = 1.90 s) than female drivers (M = 6.93 s, S.D. = 2.02 s). The findings suggest that older drivers and female drivers are more conservative than the other groups. Although the interaction effect between age and gender was not found to be significant by ANOVA, a plot of the minimum gap versus driver’s actual age shows different trends of gap acceptance between males and females (see Fig. 4). The comparison of the 2ndorder polynomial regression lines shows that the minimum gaps

3.3. Steering angle velocity during left turning In the output file of driving simulator experiment, steering control data display the angle of the selected simulator’s steering wheel. According to the ANOVA results, only age (p = 0.031) has a significant effect on SAV. During the process of turning left, old drivers (M = 2.71 rad/s, S.D. = 0.92 rad/s) tend to turn the simulator steering wheel more slowly than younger drivers (M = 3.08 rad/s, S.D. = 0.69 rad/s) [t(82) = 2.116, p = 0.037]

849

al

co

py

X. Yan et al. / Accident Analysis and Prevention 39 (2007) 843–852

Fig. 5. Interaction effect between age and speed on gap acceptance.

MCD (M = 49.35 m, S.D. = 27.96 m), compared to those for young drivers (M = 37.82 m, S.D. = 18.68 m) and middle-age drivers (M = 34.71 m, S.D. = 18.58 m). Another interesting finding shows that drivers have much smaller MCD for the higher major-road traffic speed (88.5 km/h) compared to that for the lower major-road speed (40.2 km/h), as shown in Fig. 6. Furthermore, the ANOVA results show that speed (p < 0.001), age (p = 0.014), gender (p = 0.011), and two-way interaction between age and gender (p = 0.020) have significant effects on the speed reductions of the following vehicles on the majorroad (SRR) (see Table 3). When drivers turn into the higher speed major-road traffic, they cause the following vehicle to reduce speed by a higher rate (M = 0.24, S.D. = 0.13), compared to turning into the lower speed traffic (M = 0.02, S.D. = 0.05). As aforementioned before, the speed reduction rate of the major-road vehicle is highly correlated to its deceleration rate. The ANOVA shows that the deceleration rate (M = 1.20 m/s2 , S.D. = 1.25 m/s2 ) of the following vehicle in the higher majorroad traffic speed scenario is also much higher than that (M = 0.11 m/s2 , S.D. = 0.21 m/s2 ) in the lower major-road traffic speed scenario [F(1, 124) = 45.863, p < 0.001]. The results imply a general trend that left-turn vehicles have more effects on higher speed major-road traffic and drivers take more risks for gap acceptance maneuvers at intersections with higher design speeds. It was also found that the left turns of older drivers contribute to more SRR (M = 0.16, S.D. = 0.17), compared to those for young drivers (M = 0.12, S.D. = 0.14) and middle-age drivers (M = 0.12, S.D. = 0.14); and female drivers result in a higher SRR (M = 0.14, S.D. = 0.17) than male drivers (M = 0.12, S.D. = 0.12). Furthermore, the interaction effect between age and gender illustrates that the old female drivers cause the highest SRR (see Fig. 7) but other age-gender groups show similar SRR. Therefore, older female drivers have a main contribution to gen-

pe

rs

and middle-age drivers (M = 3.16 rad/s, S.D. = 0.82 rad/s) [t(68) = 2.152, p = 0.035]. Between young and middle-age groups, there is no statistical difference in SAV [t(96) = 0.488, p = 0.627]. This finding indicates the trend of decreasing driving ability for older drivers. Experiment results show that the mean of SAV is 3.06 and 2.99 rad/s for scenarios A and B, respectively, and there is no statistical difference in the effect of the major road design speed [F(1, 124) = 0.299, p = 0.585].

on

Fig. 6. Speed effect on minimum clearance distance.

3.4. Left-turning acceleration

Au

th o

r's

The ANOVA result shows that only driver age (p < 0.001) has significant effect on driver’s acceleration rate during the period of the left turn at the intersection. Older drivers are more likely to use smaller acceleration rates (M = 1.47 m/s2 , S.D. = 0.63 m/s2 ) to achieve the major road traffic speed compared to the younger drivers (M = 2.04, S.D. = 0.60) and middle-age drivers (M = 1.86 m/s2 , S.D. = 0.61 m/s2 ). Between young and middle-age groups, there is no statistical difference in ACC [t(96) = 1.473, p = 0.144]. Generally, the larger gaps that drivers accepted, the smaller accelerations that drivers used to turn into the major road. Therefore, the driver age effect on acceleration rate is corresponding to the trend that older drivers accept larger gaps, and reflect older drivers’ conservative driving attitude. 3.5. Minimum clearance distance and speed reduction rate and deceleration rate of the following vehicle on the major road Based on the ANOVA, it was found that both age (p < 0.001) and speed (p < 0.001) are significantly associated with the minimum clear distance (MCD) between the left-turning simulator and the following vehicle. Older drivers have larger

X. Yan et al. / Accident Analysis and Prevention 39 (2007) 843–852

pening at stop-controlled intersections. Also, it is suggested that a further analysis of the quantitative relationship between intersection operation speed and gap-acceptance crash rate based on crash databases be conducted. The experiment results showed that there is an apparent age difference in the left-turn driving behavior, as well as gender difference. Compared to younger drivers and middle-age drivers, the older subjects tend to select larger gaps to make left turns, turn the steering wheel more slowly, and keep a further clearance distance from the following car. The conservative driving attitude of older drivers would compensate driving ability decrease. In the previous research, there is also evidence that elderly drivers perceive potential risk situations at intersections and change their driving habits to avoid them. It was demonstrated that drivers 55 years and older, particularly those who had a previous history of crashes, avoided driving in rain, during rush hour and making left turns across traffic (Ball et al., 1998). However, when making left turns in the higher major-road traffic speed scenario, the older subjects did not show age difference in gap acceptance. The results confirmed the conclusion that older drivers rely primarily or exclusively on perceived distance to perform gap-acceptance judgments, reflecting a reduced ability to integrate time based on speed and distance information with increasing age (Staplin et al., 1993). Spek et al. (2006) indicated that the effect that drivers tend to accept smaller time gaps as the approach speed increases appears to be stronger for older drivers than for younger drivers, suggesting that the older driver is more prone to collide with speeding vehicles. For the gender factor, the results showed that male drivers accept smaller gaps than female drivers, which reflect a general conservative driving attitude of female drivers. By assessing both general perceptual-motor performance and safety concerns, it was found that male drivers consistently overestimate their perceptual-motor skills, whereas safety skills are more prominent among female drivers (Lajunen et al., 1998; Lajunen and Summala, 1995). However, the result also showed that the older female drivers contribute to the most significant influence in the major road vehicle speed. The finding implied that older females could be the most vulnerable group for the relatively complex driving maneuvers because of more reductions in their driving ability. It is worthwhile to mention that the older female group in this study only includes four observations, which led to larger variations of the related experiment results. Nevertheless, the observed trend in this study implies that a further safety concern is needed for this high risk driver group. Driving simulators provide a safe and controlled virtual environment to test high risk driving behaviors. The convenient traffic design in the driving simulator allowed researchers to observe the minimum gap selected by subjects, while it is very difficult in a field study since drivers’ gap acceptances are influenced by the traffic volume distribution on the major road. According to the experience of this simulator experiment, the scenario design method can be used to analyze other gapacceptance behaviors at priority intersections, such as right turn from the minor road and crossing maneuver from the minor road or major road. Furthermore, the driving simulator enabled the traffic parameters, such as SAV, ACC, MCD, SRR, and DEC

al

co

py

850

rs

der and age difference in the following vehicle speed reduction rate.

on

Fig. 7. Interaction effect between age and gender on speed reduction rate.

4. Discussions

Au

th o

r's

pe

The experiment results illustrated that the major road traffic speed contributes to an interactive effect between gapacceptance drivers and vehicles on the major road. First, drivers are more likely to accept smaller gaps in the higher major-road traffic speed scenario than the lower major-road traffic speed scenario. It implied that drivers show relatively less sensitivity to vehicle approach speed but more sensitivity to on-coming vehicle’s distance or position. The perceptual task of drivers for gap acceptance requires integrating speed and distance information of a longitudinally coming-through vehicle moving in depth without significant change in visual direction. Olson (1996) suggests that the speed of an approaching vehicle may not easily be perceived until it is very near. This finding corresponds to the previous study results of gap acceptances for left-crossing maneuver (Cooper et al., 1977; Hancock et al., 1991; Alexander et al., 2002). On the other hand, due to accepting smaller gaps for the higher traffic speed, drivers have shorter separation from the following vehicle, result in more speed reduction and deceleration rate of the following vehicles to accommodate the turning vehicle, and therefore contribute to more effects on major road traffic operation. These findings implied a connection between high crash risk and high traffic speed at stop-controlled intersections. A previous study reported that stop-controlled intersections where high-speed, high-volume roads are intersected by lowerspeed, lower-volume roads controlled by a stop sign are a major problem for traffic safety, particularly in rural areas (Laberge et al., 2006). Therefore, effective speed control countermeasures to reduce speeding behaviors may lead to decrease in crashes hap-

X. Yan et al. / Accident Analysis and Prevention 39 (2007) 843–852

directly to practice in absolute terms. However, the observed trends and patterns associated with the gap-acceptance behaviors should be valid and contribute to the further research or practical work. 5. Conclusions

al

co

py

In summary, this driving simulator experiment illustrated relationships among drivers’ left-turn gap decision, corresponding driving behaviors, and consequent influence on the vehicles in the major traffic stream. The results indicated that major road traffic speed and driver age and gender have significant effects on the gap-acceptance maneuver. The findings identified an association between high crash risk and high traffic speed at stop-controlled intersections. The older drivers displayed a conservative driving attitude as a compensation for the reducing driving ability. However, due to the aging effect on cognitive, perceptual, and motor abilities, they may still be the most vulnerable group for the relatively complex driving maneuvers, especially for older female drivers.

on

Acknowledgments

The authors would like to acknowledge the Florida Department of Transportation (FDOT) for its sponsorship of this research project. Appreciation is also extended to CATSS staff and students for providing assistance with the execution of the simulation of the experiment. The recommendations of this study are those of the authors and they do not represent views of FDOT.

Au

th o

r's

pe

rs

related to drivers’ gap decisions, to be efficiently collected. These parameters were generally omitted by previous gap acceptance studies. In this study, it was found that older drivers tended to accept larger gaps to compensate the reducing driving ability, but they still showed more influence on the major road traffic than younger and middle-age drivers. Thus, the results implied that the gap acceptance accident risk may not be accurately predicted only based on the probability that drivers accept small gaps. However, it should be mentioned that for simulator experiments, there are some general limitations and validation considerations. A problem that occurs frequently with all driving simulators is simulator sickness. Based on the authors’ experiences in previous driving simulator experiments, the simulator sickness is particularly correlated with turning maneuvers and braking behaviors. Without a real safety risk as field test, sickness in a simulator experiment may significantly harm drivers’ gap acceptance decisions because drivers would try to complete the experiment as soon as possible to reduce the discomfort level. Therefore, this study attempted to avoid the sickness effect on the experiment results through encouraging participants with sickness to quit the experiment (they still obtained payment for their involvements even if not completing the experiment). Another purpose of this policy is to protect older participants since the oldest subject is 83 years old. In this study, the older female subjects were most sensitive to simulator sickness and only 4 of 10 participants fully completed the experiments without suffering any sickness. Previous driving simulator validity studies related driving behaviors are measured using two types of validity: absolute validity (when the numerical values between the two systems are the same) and relative validity (when differences found between experimental conditions are in the same direction, and have a similar or identical magnitude on both systems) (Blaauw, 1982). For the absolute validity, the conclusions of the simulator validity varied in different types of driving simulators and test scenarios. However, from the perspective of the relative validity, most previous studies concluded that driving simulators can reflect a similar trend of the driving performance in driving simulators to that in the real world. Lee (2002) compared driving performance of older adult drivers (60–90 years) between a PC-based simulator and subjects’ own cars. He found a covariance (r2 = 0.66) between the two measures and concluded that simulator usage was a safer and more economical method than the on-road testing to assess the driving performance of older adult drivers. For the speed and lane position control studies, Godley et al. (2002) and T¨ornros’s (1998) reported that participants generally drove faster in the instrumented car than the simulator, whereas relative validity was achieved for both speed and lateral position. More recently, a speed validation study focused on the effectiveness of temporary traffic signs on highways (Bella, 2005). The results showed that differences between the speeds observed in the real situation and those measured with the simulator were not statistically significant and validated the driving simulator in absolute terms. In this study, since the authors did not conduct a field validation for the simulator data, caution should be taken in transferring these findings

851

References American Association of State Highway and Transportation Officials (AASHTO), 2001. A Policy on Geometric Design of Highways and Streets Washington, DC. Abdel-Aty, M., Chen, C., Radwan, E., 1999. Using conditional probability to find the driver age effect in crashes, J. Transp. Eng., 1999, ASCE, vol. 125, No. 6. Alexander, J., Barham, P., Black, I., 2002. Factors influencing the probability of an incident at a junction: results from an interactive driving simulator. Accid. Anal. Prev. 34, 779–792. Ball, K., Owsley, C., Stalvey, B., Roenker, D.L., Sloane, M.E., Graves, M., 1998. Driving avoidance and functional impairment in older drivers. Accid. Anal. Prev. 30, 313–322. Bella, F., 2005. Validation of a driving simulator for work zone design. Transp. Res. Rec. 1937, 136–144. Blaauw, G.J., 1982. Driving experience and task demands in simulator and instrumented car: a validation study. Hum. Factors 24, 473–486, 2004. Chovan, J.D., Tijerian, L., Everson, J.H., Pierowicz, J.A., Hendricks, D.L., 1994. Examination of intersection left turn across path crashes and IVHS countermeasures (DOT HS 808 154). National Highway Traffic Safety Administration. Cooper, D.F., Storr, P.A., Wennell, J., 1977. Traffic studies at T-junctions: the effect of speed on gap acceptance and conflict rate. Traffic Eng. Control 18, 110–112. Darzentas, M.R., McDowell, C., Cooper, D., 1980. Minimum acceptable gaps and conflict involvement in a simple crossing maneuver. Traffic Eng. Control 21 (2), 58–61. Davis, G.A., Swenson, T., 2004. Field study of gap acceptance by left-turning drivers. Transp. Res. Rec. 1899, 71–75.

X. Yan et al. / Accident Analysis and Prevention 39 (2007) 843–852

al

co

py

Transport Research (CAITR 2002) University of New South Wales, Sydney, Australia, December 4–6 (2002). Lerner, N.D., Huey, R.W., McGee, H.W., Sullivan, A., 1995. Older Driver Perception-Reaction Time for Intersection Sight Distance and Object Detection: Report No. FHWA-RD-93-168. Federal Highway Administration: Washington, DC. Lyles, R.W., Staplin, L., 1991. Age differences in motion perception and specific traffic maneuver problems. Transp. Res. Rec. 1325, 23–33. National Highway Traffic Safety Administration, 1993. Addressing the Safety Issues Related to Younger and Older Drivers. DOT HS 807 957 NTS-31. US Department of Transportation. http://www.nhtsa.dot.gov/ people/injury/olddrive/pub/yorept.pdf. Access date, August 2006. Olson, P., 1996. Forensic Aspects of Driver Perception and Response. Lawyers & Judges Publishing Co., Tuscon, AZ. Oxley, J., Fildes, B., Corben, B., Langford, J., 2006. Intersection design for older drivers. Transp. Res. Part F: Traff. Psychol. Behav. 9, 335–346. Retting, R.A., Weinstein, H.B., Solomon, M.G., 2003. Analysis of motor-vehicle crashes at stop signs in four US cities. J. Saf. Res. 34, 485–489. Scialfa, C.T., Gucy, L.T., Leibowitz, H.W., Garvey, P.M., Tyrrell, R.A., 1991. Age differences in estimating vehicle velocity. Psychol. Aging 6, 60–66. Spek, A.C.E., Wieringa, P.A., Janssen, W.H., 2006. Intersection approach speed and accident probability. Transp. Res. Part F: Traffic Psychol. Behav. 9, 155–171. Staplin, L., 1995. Simulator and field measures of driver age differences in left-turn gap judgments. Transp. Res. Rec. 1485, 49–55. Staplin, L., Lococo, K., Sim, J., 1993. Traffic maneuver problems of older drivers. Report No. FHWA-RD-92-092. McLean, VA: Federal Highway Administration. T¨ornros, J., 1998. Driving behaviour in a real and a simulated road tunnel—a validation study. Accid. Anal. Prev. 30, 497–503.

Au

th o

r's

pe

rs

DeLucia, P.R., Bleckley, M., Meyer, L.E., Bush, J.M., 2003. Judgments about collision in younger and older drivers. Transp. Res. Part F: Traffic Psychol. Behav. 6, 63–80. Gattis, J.L., Low, S.T., 1999. GAP acceptance at atypical stop-controlled intersections. J. Transp. Eng. 125, 201–207. Godley, S.T., Triggs, J.T., Fildes, B.N., 2002. Driving simulator validation for speed research. Accid. Anal. Prev. 34, 589–600. Hancock, P.A., Caird, J.K., Shekhar, S., Vercruyssen, M., 1991. Factors Influencing Drivers’ Left Turn Decisions Proceedings of the Human Factors Society 35th Annual Meeting. Harwood, D., Mason, J., Brydia, R., 2000. Sight distance for stop-controlled intersections based on gap acceptance. Transp. Res. Rec. 1701, 32–41. Highway Capacity Manual (HCM), 2000. TRB, National Research Council, Washington, DC. Kyte, M., Tian, Z., Mir, Z., Hameedmansoor, Z., Kittelson, W., Vandehey, M., Robinson, B., Brilon, W., Bondzio L., Wu, N., Troutbeck, R., 1996. Capacity and Level of Service at Unsignalized Intersections, Final Report: Volume 1-Two-Way Stop-Controlled Intersections. National Cooperative Highway Research Program, pp. 3–46. Laberge, J.C., Creaser, J.I., Rakauskas, M.E., Ward, N.J., 2006. Design of an intersection decision support (IDS) interface to reduce crashes at rural stopcontrolled intersections. Transp. Res. Part C: Emerg. Technol. 14, 39–56. Lajunen, T., Summala, H., 1995. Driver experience, personality, and skill and safety motive dimensions in drivers’ self-assessments. Personal. Individual Differ. 19, 307–318. Lajunen, T., Corry, A., Summala, H., Hartley, L., 1998. Cross-cultural differences in drivers’ self-assessments of their perceptual-motor and safety skills: Australians and Finns. Personal. Individual Differ. 24, 539–550. Lee, H., 2002. The validity of driving simulator to measure on-road driving performance of older drivers, The 24th Conference of Australian Institutes of

on

852