Accident Analysis and Prevention 37 (2005) 10561065

Gap acceptance and risk-taking by young and mature drivers,

both sober and alcohol-intoxicated, in a simulated driving task
Stefanie Leung

, Graham Starmer
Department of Pharmacology, The University of Sydney, Bosch Building D05, Sydney, NSW 2006, Australia
Received 24 January 2005; received in revised form 19 May 2005; accepted 2 June 2005
Abstract
A single-blind randomized study was conducted on young (1821 years, n =16) and mature (2535 years, n =16) drivers to assess how
age, combined with a modest dose of alcohol (0.7 g/kg for males and 0.6 g/kg for females), inuenced performance on a driving simulator.
The driving tasks included detecting the presence of a vehicle on the horizon as quickly as possible, estimating the point on the road that an
approaching vehicle would have passed by the participants vehicle (time-to-collision) and overtaking another vehicle against a steady stream
of oncoming trafc. The results of the vehicle detection task showed that detection times were signicantly slower with maturity, alcohol
consumption and lower approaching vehicle speeds (50 kph), particularly on curved sections of road. Approaching vehicle speed was also
found to signicantly inuence time-to-collision (TTC) judgments, such that faster approach speeds led to less underestimated (and therefore
riskier) judgments of TTC than slower speeds. In the overtaking task, mature participants demonstrated impaired discrimination skills with
varying approaching vehicle speeds, while young participants recorded signicantly slower speeds while overtaking a vehicle, thus increasing
the time that they spent in the opposing lane. In conclusion, young and mature drivers demonstrated pivotal differences in behavior in this
study. Young drivers showed a greater tendency to engage in risky driving, while experienced drivers appeared to be more susceptible to
perceptual inuences. Overall, alcohol consumption impaired a drivers ability to divide attention, but had little effect on decision-making
processes.
2005 Elsevier Ltd. All rights reserved.
Keywords: Risk-taking; Time-to-collision; Overtaking; Young drivers; Mature drivers; Speed; Alcohol
1. Introduction
A large number of countermeasures have been adopted in
Australia in an attempt to reduce the high incidence of road
accidents. These have included a reduced suburban speed
limit (from 60 to 50 kph), more stringent criteria for provi-
sional license holders, speed cameras, random breath testing
and educational campaigns directed towards the dangers of
drink-driving and speeding. Over the last two decades these
measures have dramatically reduced the road toll, although
this reduction is now at plateau. That is, speeding and alco-
hol still remain the two most frequent contributors to fatal
road accidents. In addition, the apparent over-representation
of the young driver population (aged 1720 years) in fatal

Corresponding author. Tel.: +61 413488682.

E-mail address: leungs@med.usyd.edu.au (S. Leung).
on-road collisions is of increasing concern. In 1998, young
drivers (aged 1720 years) comprised approximately 5.4%of
the licensed population in New South Wales but were repre-
sented in approximately 15.5%of all fatal overtaking crashes
for that year (Australian Transport Safety Bureau, 2004).
1.1. Hazard detection
A study by Quimby (1987) found that drivers who took
longer to detect and respond to potential hazards had more
accidents. These drivers were more prone to react impul-
sively and without attention to safety. In addition, Deery and
Love (1996) found that, compared with experienced drivers,
novice drivers detected hazards less quickly and less ef-
ciently. This was attributed to the fact that novice drivers
tended to look at the road and lane markings close to the
hood of their vehicle, unlike experienced drivers who looked
0001-4575/$ see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.aap.2005.06.004
S. Leung, G. Starmer / Accident Analysis and Prevention 37 (2005) 10561065 1057
to the horizon and monitored their lateral position via their
peripheral vision (Mourant and Rockwell, 1970). Further-
more, less experienced drivers adopted a narrower horizontal
scanning pattern, especially while negotiating curves (Cohen
and Studach, 1977; Shinar et al., 1977). As a result, they
were less equipped to, and consequently slower at, detecting
road hazards ahead. This has been shown to occur despite the
fact that young drivers have faster reaction times than mature
drivers in non-driving settings (Summala, 1987). It is there-
fore possible that, when their vehicles are in motion, young
(less experienced) drivers will detect an approaching vehicle
more slowly than mature (more experienced) drivers.
Alcohol intoxication has also been associated with sig-
nicant deterioration in arousal and attention (Roehrs et
al., 1992). Thus, alcohol-affected drivers, and in particular
alcohol-affected young drivers, would be expected to detect
approaching vehicles most slowly. Furthermore, considering
that the perception of risk is reduced after alcohol (Deery
and Love, 1996a,b) and that this effect is more prominent
in novice drivers (Deery, 1999), it is possible that alcohol-
affected drivers and, in particular, alcohol-affected young
drivers, might be more likely to misjudge the margins of
safety afforded by gaps in trafc.
1.2. Time estimations
In performing an overtaking maneuver, drivers must rst
detect an oncoming vehicle and then estimate when it will
pass them (time-to-collision and time-to-contact). This criti-
cal sub-task of driving has been considerably researched and
has typically manifested as an underestimate of time (Schiff
and Detwiler, 1979; Schiff et al., 1992a,b; Caird and Han-
cock, 1994; DeLucia et al., 2003). That is, observers predict
that an approaching vehicle will arrive at their position sooner
than is actually the case. Various factors have been shown to
inuence such judgments of time.
Approach speed has been shown to signicantly inuence
arrival time judgments in both video-based and in-vehicle
experiments (Cavallo and Laurent, 1988; Sidaway et al.,
1996). It was found that, as the approaching vehicles speed
increased, so too did the accuracy of time estimations. In
addition, older drivers (aged 6583 years) demonstrated a
tendency to underestimate time-to-contact to a greater extent
than younger drivers (aged 2045 years). DeLucia et al.
(2003) further found that judgments about whether a poten-
tial collision would occur were less accurate for older drivers
(aged 4064 years) compared with younger drivers (aged
1829 years). These outcomes beg the question of whether
the over-representation of young drivers in overtaking acci-
dents is due to risk-taking, a misperception of arrival time,
idle attention, or a combination of some or all of these fac-
tors. Furthermore, given that young drivers have a higher
rate of accidents involving alcohol (Triggs and Smith, 1996),
determination of the effects of alcohol consumption on haz-
ard detection and the judgment of acceptable gaps in trafc
is desirable. To date, however, research into the interaction
of age and alcohol on specic faculties critical for accurate
perceptions of speed while driving has been neglected.
The purpose of this study was to investigate the effects of
a moderate dose of alcohol on a drivers perception of speed,
hazards and risk acceptance. Focus was directed towards the
experience of the driver in order to address the question of
the extent to which driving is a learned behavior. That is,
increased exposure to the task results in greater prociency.
It was hypothesized that relative inexperience would be man-
ifest as riskier driving and poorer decision-making. Samples
from the provisional license population (1721 years) and
the full license population were used to test this hypothesis.
It was also expected that at a blood alcohol concentration
(BAC) of 0.08 g/100 ml, implicated in a large proportion of
fatal crashes involving alcohol in New South Wales (Roads
and Trafc Authority of NSW, 2004), drivers would show
marked impairment of their perceptual skills. In particular,
this would be more obvious in the inexperienced driver group.
This premise is supported by the ndings of Compton et al.
(2002), which afrmed that many driving-related skills were
degraded at BACs as low as 0.03 g/100 ml, and that a BAC
of 0.08 g/100 ml was associated with an almost three-fold
increase in crash-risk. Thus, a target BAC of 0.08 g/100 ml,
exceeding the Australian legal per se limit of 0.05 g/100 ml,
was adopted for this study in an effort to elicit a marked effect
on driving performance.
2. Method
2.1. Experimental design
Using a single-blind randomized mixed-model design,
participants (young and mature) attended two experimen-
tal sessions on the driving simulator; an alcohol condition
(target BAC=0.08 g/100 ml) and a placebo condition (target
BAC=0.00 g/100 ml). Participants were administered a bev-
erage at the commencement of each experimental session.
For the alcohol treatment, alcohol (0.7 g/kg) was adminis-
tered orally as a 1:3 mixture of vodka (37.5% alcohol by
volume) with orange juice, with the dose adjusted for gen-
der according to the Widmark factor (Widmark, 1932). That
is, a 70 kg man would be required to consume 165 ml vodka
to achieve the desired BAC of 0.08 g/100 ml. The placebo
beverage comprised of an equivalent amount of orange juice
with 5 ml of vodka oated on the surface to provide olfactory
masking (Starmer and Bird, 1984). Participants were allowed
20 min to consume the beverage at a constant rate.
2.2. Participants
Sixteen (eight male and eight female) young volunteers
(aged 1821 years; mean =20 years; S.D. =0.9 years) and
16 (10 male and 6 female) mature volunteers (aged 2535
years; mean =28 years; S.D. =2.7 years) were recruited for
the study through advertisement in The University of Sydney
1058 S. Leung, G. Starmer / Accident Analysis and Prevention 37 (2005) 10561065
Table 1
Young and mature driver characteristics
Driver characteristics Young drivers (1821 years) Mature drivers (2535 years) p-value
Mean S.D. Mean S.D.
Years of driving experience 2.9 0.96 9.9 3.38 <0.001
a,
Previous crash history (number crashes) 0.3 0.60 1.0 0.89 0.016
Alcohol consumption (number standard drinks/week) 13.2 9.87 8.8 6.04 0.141
a
Equal variances not assumed.
on-line and street press. The above age ranges were chosen
in an effort to mimic, as closely as possible, those adopted
by the Roads and Trafc Authority of New South Wales
in reporting annual crash statistics. First-time consumers of
alcohol were not accepted, and volunteers were required to
have held a current drivers license for at least 1 year in order
to be eligible for participation. Self-reports of driving experi-
ence, crash history and average weekly alcohol consumption
were collated for the young and mature driver groups. Sig-
nicant differences were observed in the mean number of
years driving (t
(df=30)
=7.96, p <0.001) and mean number of
previous crashes (t
(df=30)
=2.55, p <0.05), reecting a rela-
tionship between crash risk and exposure to driving, while
average weekly alcohol consumption appeared comparable
between the two groups (Table 1). Participants were told to
abstain from consuming alcohol for at least 24 h prior to test-
ing and to avoid eating a meal within 2 h of testing. They
were also instructed not to drive on either of the testing days.
Participants were paid for their involvement in the study and,
as incentive to drive as safely as possible, were offered a
monetary bonus for maintaining a clear experimental driving
record across both sessions. Ethics approval for the study was
obtained through The University of Sydney Human Ethics
Committee.
2.3. Driving simulator
This study used the STISIM Drive (Version 1.1.0.10)
driving simulator located at The University of Sydney. The
simulator consisted of a full car cabin tted with a steering
wheel, dashboard speedometer, indicators, horn, accelerator
and brake pedals. Computer-generated visual images were
projected onto three screens resulting in a 135

eld of view.
The driving simulator incorporated fully interactive driving
(with immediate visual and auditory feedback) and provided
a exible experimental environment, whereby trafc scenes
could be constructed and manipulated.
2.4. Experimental tasks
The study aimed to explore the involvement of haz-
ard detection, speed perception, time perception, distance
perception and risk perception in performing overtaking
maneuvers. This was monitored via performance in a 15 min
driving simulation, commenced 60 min after participants
began drinking, which encompassed the following tasks.
2.4.1. Detection time
Participants were requiredtopress the hornbutton(located
on the steering wheel) as soon as they detected an approach-
ing vehicle on the horizon. These vehicles were programmed
to enter the opposing lane by appearing 500 m away from
behind buildings, or 800 m away in an area dense in trees.
The approaching vehicle speeds were randomized between
50 and 90 kph, while the participants speed was limited to
90 kph. In addition, at the point that the approaching vehicles
appeared, roadway curvature was counterbalanced between
a straight stretch of road, a left curve and a right curve. This
task was designed to measure the level of hazard awareness of
the participants at the different experimental times and under
different experimental conditions.
2.4.2. Time-to-collision estimations
Approaching vehicles were programmed to appear in view
for 9 s then to disappear when they were 3 s away from the
participants vehicle. The approaching vehicle speeds were
randomized between 50 and 90 kph, while the participants
speed was limited to 90 kph. Participants were required to
estimate when the hood of the approaching vehicle would
have reached alignment with the hood of their own vehi-
cle (i.e. when the vehicles would have passed each other),
had the approaching vehicle not disappeared. At this point,
participants were instructed to press the horn button located
on the steering wheel. All time-to-collision estimations were
performed on a straight stretch of road. This task provided a
measure of the participants perception of time, and relative
speed and distance.
2.4.3. Overtaking maneuver
Participants were required to overtake a vehicle being fol-
lowed (lead vehicle), against a succession of approaching
vehicles, as soon as they deemed it to be safe. The speed of
the vehicle being followed was kept constant at 80 kph, while
successions of approaching vehicles all traveled at the same
speed (either 50 or 90 kph). The time between these succes-
sive approaching vehicles was programmed to become pro-
gressively larger; increasing from 9 to 17 s when the vehicles
were approaching at 50 kph; and from7 to 15 s when the vehi-
cles were approaching at 90 kph. All overtaking maneuvers
were performed on a straight stretch of road with a sign-
posted speed limit of 100 kph for participants. In addition, all
vehicles (boththe vehicle beingfollowedandthe approaching
vehicles) appeared 500 m ahead of the participants vehicle.
S. Leung, G. Starmer / Accident Analysis and Prevention 37 (2005) 10561065 1059
This task provided a measure of the participants perception
of and willingness to take risks in the light of variable margins
of safety.
2.5. Blood alcohol analysis
An Alco Sensor III (St. Louis, MO) breath analysis device
was used to measure blood alcohol levels prior to the sim-
ulation test drive. The instrument employs a blood:breath
factor of 2100:1, and readings were corrected to the true mean
blood:breath factor of 2285:1 (Dubowski, 1974). Previous
studies have shown a correlation in excess of 97% between
BACs determinedfrombreathanddirectlyfromvenous blood
(Slemeyer, 1986).
2.6. Data analysis
Data were analyzed by a mixed model ANOVA with
planned orthogonal contrasts. Each of the three tasks (detec-
tion time, time-to-collision and overtaking) was analyzed
separately using participant age (young versus mature), par-
ticipant gender (male versus female), alcohol consump-
tion (placebo versus alcohol), approaching vehicle speed
(50 kph versus 90 kph) and roadway curvature (straight ver-
sus curved) as the independent variables. The dependent
variables for each task were:
(a) Detection task time to detect an approaching vehicle.
(b) Time-to-collision task time-to-collision of the partic-
ipants vehicle with the approaching vehicle when the
horn button was depressed.
(c) Overtaking task headway distance between the partici-
pants vehicle and the vehicle being overtaken at the com-
mencement of the overtaking maneuver and the mean
speed of travel of the participants vehicle while engaged
in the overtaking maneuver.
This study primarily aimed to test the null hypotheses that
alcohol consumption, driver age and driver gender would not
selectively affect risk-taking and perceptual performance on
the driving simulator. The Type I error rate was set at =5%.
3. Results
3.1. Blood alcohol concentration
In the alcohol condition, young participants attained a
mean peak BAC of 0.059 g/100 ml (S.D. =0.013 g/100 ml) at
the time of testing (60 min after drinking was commenced),
while mature participants attained a mean peak BAC of
0.064 g/100 ml (S.D. =0.010 g/100 ml) also at 60 min. There
was no signicant difference between the mean peak BACs
of young and mature participants but there was a signi-
cant difference between both of the mean peak BACs and
the target BAC of 0.08 g/100 ml (t
(df=15)
=6.127, p <0.05
Fig. 1. Mean (S.E.) detection times for young and mature participants,
both when traveling on a straight and a curved section of road.
and t
(df=15)
=6.536, p <0.05). In the placebo condition, par-
ticipants recorded a zero BAC at all times.
3.2. Detection time
A signicant main effect of roadway curvature was found
on response times to an approaching vehicle (F
(1,24)
=302.64,
p <0.001), whereby detection times were signicantly slower
on curved roads than on straight roads (4.01 s versus 0.93 s)
for all drivers.
Age was signicant in a drivers ability to detect oncoming
vehicles as an interaction with straight and curved sections
of road (F
(1,24)
=8.92, p <0.01). As Fig. 1 shows, aver-
aged across the BAC conditions, mature participants were
slower than young participants to detect the presence of
approaching vehicles when on a curved road (4.38 s versus
3.65 s). In contrast, mature participants had marginally faster
mean detection times when on a straight road (0.77 s versus
1.10 s).
Alcohol consumption also signicantly (F
(1,31)
=4.94,
p <0.05) interacted with detection times (Fig. 2). That is,
averaged across age, responses appeared not to be inuenced
by alcohol when participants were traveling on straight roads
Fig. 2. Mean (S.E.) detection times in the alcohol and placebo conditions,
when traveling on a straight and a curved section of road.
1060 S. Leung, G. Starmer / Accident Analysis and Prevention 37 (2005) 10561065
Fig. 3. Mean (S.E.) detection times of vehicles traveling at variable speeds,
while on straight and curved sections of roadway.
(0.94 s versus 0.93 s), but when negotiating a curve, detection
times were slower in the alcohol-affected condition than in
the placebo state (4.31 s versus 3.72 s).
Fig. 3 illustrates a similar signicant interaction
(F
(1,24)
=42.38, p <0.001) whereby all participants were
slower in detecting the presence of approaching vehicles
on curved roadways when those vehicles were traveling at
50 kph compared to when they were traveling at 90 kph
(4.80 s versus 3.24 s). Incontrast, onstraight roadways, detec-
tion times were comparable for both of the approaching
vehicle speeds (1.04 s versus 0.82 s). No signicant differ-
ences were observed between male and female detection time
responses.
3.3. Time-to-collision
The time-to-collision (TTC) between the participants
vehicle and the approaching vehicle, when it disappeared
from view, was programmed at 3 s. On average, for all test
variables, TTC was always underestimated.
While no signicant effect of BAC was found on
mean TTC estimations, a signicant effect (F
(1,24)
=44.97,
p <0.001) was observed with approaching vehicle speed. As
shown in Fig. 4, TTC was underestimated to a lesser extent
when the approaching vehicle was traveling at 90 kph than
when it was traveling at 50 kph (2.72 s versus 2.29 s). In
addition, male drivers underestimated TTC to a signicantly
(F
(1,28)
=4.36, p <0.05) lesser extent than female drivers
(2.26 s versus 2.66 s), thus affording themselves a smaller
margin of safety (Fig. 5).
3.4. Overtaking task
No signicant effect of age or BAC was observed with
the gap-acceptance results in the overtaking task. It was
interesting to observe, however, that participants perceived
substantially greater risk in overtaking in the face of a vehicle
approaching at 50 kph than one at 90 kph. That is, although
the time between successive approaching vehicles was kept
Fig. 4. Mean (S.E.) time-to-collision estimations when the approaching
vehicle was traveling at 50 and 90 km/h.
constant regardless of the speed of the approaching vehicle,
participants chose toovertake withinlarger vehicle gaps when
the approaching vehicles were traveling at 50 kph than they
were traveling at 90 kph (15 s versus 10 s).
In considering the headway distances that participants
chose to adopt immediately before commencing the overtak-
ing maneuver, it was interesting to nd a signicant age inter-
action (F
(1,20)
=13.15, p <0.005) with the speed of approach-
ing vehicles (Fig. 6). That is, while headway distances did not
vary between the two approaching vehicle speeds for young
drivers (14.18 m versus 14.15 m), mature drivers maintained
a greater distance from the vehicle being overtaken when
the oncoming trafc was approaching at 90 kph compared
to at 50 kph (16.05 m versus 12.59 m). Furthermore, female
drivers created signicantly (F
(1,20)
=6.66, p <0.05) larger
headway distances (16.48 m versus 12.00 m) between their
vehicle and the vehicle being overtaken than the male drivers
(Fig. 7).
The participants mean speed of travel was also measured
during the time they were in the opposing lane (while execut-
ing an overtaking maneuver). The area in which overtaking
maneuvers were conducted had a sign-posted speed limit of
100 kph. Analysis of the mean speed results revealed that
on average, across all variables, all participants exceeded
Fig. 5. Mean (S.E.) time-to-collision estimations of the male and female
drivers.
S. Leung, G. Starmer / Accident Analysis and Prevention 37 (2005) 10561065 1061
Fig. 6. Mean (S.E.) headway distances of young and mature drivers against
vehicles approaching at 50 and 90 km/h.
Fig. 7. Mean (S.E.) headway distances of male and female drivers.
the sign-posted speed limit of 100 kph. A signicant over-
all effect of age was found with mean speed (F
(1,28)
=6.24,
p <0.05). That is, as Fig. 8 shows, while executing an
overtaking maneuver, mature drivers tended to speed to
a greater extent than young drivers (106.05 kph versus
102.86 kph).
Fig. 8. Mean (S.E.) speed of young and mature participants while in the
opposing lane and engaged in an overtaking maneuver.
4. Discussion
The ability to detect and estimate the relative speed of
other vehicles, and consequently judge TTC, is critical for the
execution of a number of common trafc maneuvers, such as
intersection crossing, merging and overtaking. Acute aware-
ness and accurate judgments of time are therefore imperative
for effective and safe driving, as even split seconds can differ-
entiate an achievable maneuver from an imminent accident.
It has been shown in this study as well as in others, how-
ever, that estimates of TTC are generally very inaccurate and
are usually manifest as underestimates of time (Schiff and
Oldak, 1990; Caird and Hancock, 1994; Hoffmann and Mor-
timer, 1994). Such underestimation, while being regarded at
face value as inaccuracy, also translates to added safety. That
is, by judging the available time before a collision as being
less than the actual time, drivers afford themselves a greater
margin of safety in which to complete an intended maneu-
ver, subsequently minimizing the risk involved. The decision
to engage in risky driving, however, is not always conscious
and is often inadvertently the result of a distorted perception
of risk, timing and speed (Finn and Bragg, 1986; Summala,
1987).
4.1. Detection time
On straight roads, mature drivers recorded marginally
faster vehicle detection times than young drivers. In consid-
ering the participants speed of travel of 90 kph during the
task, the detection time difference (0.33 s) can be translated
to a distance of approximately 6.4 m for a vehicle approach-
ing at 50 kph or 8.3 m for a vehicle approaching at 90 kph.
Therefore, irrespective of the speed of the approaching vehi-
cle, young drivers deprived themselves of a valuable margin
of safety. This nding can be attributed to the variable visual
strategies adopted by drivers of different ages and experi-
ences. Shinar (1978) demonstrated that on a straight expanse
of road, experienced drivers directed their line of sight to the
horizon in anticipation of occurrences on the road ahead and
relied heavily on their peripheral vision to maintain roadway
position (Mourant and Rockwell, 1970). Novice drivers, on
the other hand, focused their attention on the road and lane
markings close to the hood of their vehicle, and were con-
sequently less equipped to notice or deal with unexpected
events (Mourant and Rockwell, 1970). This suggested that
the dynamics of their own vehicle and, in particular their
lateral lane position, rather than consideration of the wider
aspects of the driving scenario, are the focus for young drivers
on straight roads (Mourant and Rockwell, 1970).
Curve negotiation, as opposed to cruising on straight
roads, is a more complex task. Greater demand is placed
on the driver in having to maintain an appropriate heading
angle with respect to the roadway curvature while concur-
rently adopting an efcient visual search pattern and retaining
proper control of the vehicle (Wooldridge et al., 2000). As
a result of the increased challenge presented to drivers on
1062 S. Leung, G. Starmer / Accident Analysis and Prevention 37 (2005) 10561065
curved roads, peripheral visual performance is decreased
(Miura, 1990). Consequently, vehicles would be expected to
be detected later on curved roads than on straight roads, as
was the case in this experiment. In addition, mature drivers
were slower (0.73 s) than young drivers to detect the pres-
ence of an approaching vehicle on curved sections of road,
perhaps simply reecting an age-associated increase in reac-
tion time (where reaction time =detection time +movement
time in this experiment) (Mullin, 2000; Ratcliff et al., 2001),
rather than being a consequence of any driving experience-
related differences in visual search and tracking strategies on
curves. That is, given that a drivers attention is occupied by
navigation to a greater extent on curved roads than on straight
roads, all drivers regardless of experience are compelled to
adopt similar, proximate (following the road geometry) and
frequent scanning patterns (Cohen and Studach, 1977; Shinar
et al., 1977).
In everyday trafc, drivers are often confronted with two
or more tasks concurrently, and are required to execute these
tasks in an efcient and safe manner. In order to do so, how-
ever, it is essential that a driver possess the abilitytoefciently
divide attention. The act of general driving, for example, has
been described as a skilled divided attention task (Moskowitz,
1973), whereby drivers scan the surroundings while simul-
taneously maintaining vehicle control. Many studies have
investigated the ability of a driver to divide attention under
the inuence of alcohol, and have shown progressive deterio-
ration of performance with increasing BAC (Buikhuisen and
Jongman, 1972; Allen et al., 1975; Gustafson, 1986; Koel-
ega, 1995; Carscadden, 2002). It was postulated that alcohol
had the effect of reducing the spare attentional capacity of
the driver (Allen et al., 1975), thus rendering the task too
demanding. The concept that impairment of performance is
conditional upon task demands, even after moderate doses of
alcohol (Moskowitz et al., 1985), bears relevance to the out-
comes of this experiment. That is, given the relative complex-
ity of curve negotiation in comparison to driving on straight
roads, it follows that the attentional demands are greater on
curves and the ability to divide attention is depleted after the
consumption of alcohol. Moreover, it has also been shown
that the usable eld of view shrinks at the point of xation
(perceptual narrowing) with an increase in demand (Crundall
et al., 1998). Thus, after alcohol consumption, drivers are rel-
egated to using a narrowand misrepresented eld of viewand
deteriorated visual functions to process information about
objects on the road. It is, therefore, not surprising that detec-
tion times after the consumption of alcohol were signicantly
worse on curves and not on straight sections of road.
Vehicles traveling at 90 kph were detected more rapidly
than vehicles traveling at 50 kph, but only on curved roads.
This can be explained in terms of the visual angle from the
drivers line of sight (focus of expansion) to an object mov-
ing through the drivers visual eld, which increases as the
driver continues forward motion toward the focus of expan-
sion. Notably, the visual angle increases to a greater extent
the further the object is from the drivers central line of
sight. Thus, peripheral vision (utilized when monitoring a
vehicle approaching on a curve) produces greater changes
to the visual angle (referred to as angular velocity) than
central vision (Salvatore, 1967). Furthermore, angular veloc-
ity has been documented to increase proportionally with
the observers speed (Triggs, 1986). Thus, when a vehicle
approaches at a fast speed on a curve, there is increased con-
trast in the drivers retinal image (Snowden et al., 1998).
Changes in the environment, such as the appearance of an
approaching vehicle, are therefore perceived to occur sud-
denly and, hence, more noticeably. On the other hand, a
slowly traveling vehicle is able to deceive the driver by blend-
ing in with the environment. In comparison, there is little
identiable change in the visual angle of an approaching vehi-
cle on a straight road, such that there would be no signicant
discrimination between slower and faster approaching vehi-
cle speeds on those roads.
4.2. Time-to-collision
It was hypothesized that alcohol would negatively affect
TTC by further limiting a drivers spare attentional capacity
and consequently causing the driver to concentrate dispro-
portionately on handling the vehicle. However, in this study,
a signicant effect of alcohol on TTC estimations was not
observed. This result concurs with the ndings of Kearney
and Guppy (1988), who examined the effects of alcohol on a
drivers perception of his/her own vehicle speed. The lack of
statistical signicance in that study was attributed to the rel-
ative ease of the task, such that the intoxicated participants
retained adequate spare attentional capacity to make speed
judgments. It is possible, therefore, that in the present study,
alcohol did not exert an identiable effect on time judgments
because the BACs were not sufciently high (did not reach
target of 0.08 g/100 ml) to elicit impairment. That is, in this
study, the drivers spare attentional capacity may not have
been depleted by the presence of the alcohol.
Earlier studies on TTC estimation have reported signif-
icant differences in accuracy when the observers approach
speed was a variable (Cavallo and Laurent, 1988; Sidaway
et al., 1996). These experiments found that the faster the
observer approached a stationary object, the more accurate
the estimates of TTC. In the present study, both vehicles
were put in motion but the speed of the approaching vehicle
rather than that of the observer was manipulated. Neverthe-
less, a similar trend to the one above was observed whereby
vehicles approaching at fast speeds (90 kph) yielded more
accurate (less underestimated) approximations of TTC than
did vehicles approaching at medium speeds (50 kph). Simi-
lar to the inuence of approaching vehicle speed on detection
times, the effect of different approach speeds on TTCestima-
tions can be explained by the concept of contrast sensitivity
(Blakemore and Snowden, 1999). For example, in consider-
ing an approaching vehicle traveling at a faster speed, the rate
of change of background environmental cues and the angu-
lar velocity of the approaching vehicle are both increased
S. Leung, G. Starmer / Accident Analysis and Prevention 37 (2005) 10561065 1063
(Salvatore, 1967). This allows the viewer a more accurate
perception of the approaching vehicles speed, resulting in
more accurate (less underestimated) times. Such perceptual
effects might make the chance of a gap acceptance collision
more likely to occur on fast roads than on slow roads, as
well as more likely in built-up urban environments than in
sparse rural environments, due to increased contrast in the
surroundings and thus more accurate (less cautious) estima-
tions of TTC.
Gender differences were also apparent in this study.
Females, across all variables, tended to make more conser-
vative estimates of TTC than males. Earlier studies produced
similar results and attributed them primarily to risk-taking
differences between the two genders (McLeod and Ross,
1983; Schiff and Oldak, 1990; Schiff et al., 1992a,b). That is,
males, compared with females, possess an exaggerated sense
of their own driving competency and perceive less risk in a
variety of dangerous driving behaviors. The most pronounced
differences occur in ratings of relative driving skill, driver
aggression, sensation-seeking behavior and in their general
risk acceptance (DeJoy, 1992; Turner and McClure, 2003).
Schiff and Oldak (1990), however, further postulated that in
women a variable ability with spatio-temporal skills (whether
poorer or more attuned), or an inherent predisposition to err
on the side of safety rather than tolerate a dangerous variable
error could contribute to their relatively cautious judgments.
They considered that risk-acceptance alone could not account
for the observed gender differences as, in tasks where risk-to-
self was not a factor, women continued to elicit conservative
responses. Additional research into the specic aspects of
driving that are susceptible to gender differences was there-
fore considered to be desirable.
4.3. Overtaking task
Mature drivers adopted smaller headway distances when
a vehicle was approaching at 50 kph rather than at 90 kph,
but younger drivers did not. This result suggests that young
and mature participants perceived different degrees of risk
in overtaking in the face of a slow vehicle (50 kph) and a
fast vehicle (90 kph). The actual risk involved, however, was
kept constant because the gap between successive approach-
ing vehicles was calculated on the basis of time rather than on
distance. Thus, theoretically, drivers should have approached
all overtaking maneuvers in the same manner, regardless of
the speed of the approaching vehicle. It would seem, how-
ever, that mature drivers were less anxious when overtaking
in the face of faster vehicles, presumably because these vehi-
cles appeared further away. This nding is supported by
other studies, whereby younger drivers judged time more
accurately than mature drivers (Schiff et al., 1992a,b; Han-
cock and Manser, 1997). It would thus appear that, unlike
their younger counterparts, mature participants made over-
taking decisions based largely on the relative distance of
the approaching vehicle, rather than on relative speed and
time.
At the point where the maneuver was initiated, males
allowed for smaller safety margins between their vehicle
and the next approaching vehicle, and smaller safety mar-
gins between their vehicle and the vehicle being followed.
In addition, males decided to overtake between smaller gaps
in trafc than females. In contrast, however, females spent
more time in the opposing lane than males. While this would
normally equate to risky conduct, the fact that female drivers
overtook between larger gaps in safety serves to offset the
result. It can thus be inferred, fromthe results of this task and
other tasks, that increased driver aggression, a distorted per-
ception of risk, and an overcondence in driving ability all
contribute to the relatively riskier behavior demonstrated by
males when overtaking. These results are in accord with the
ndings of a study that examined the driving records of a large
population of drivers (Rajalin et al., 1997), in which close-
following male drivers showed involvement in 2.3 times more
trafc offences than control drivers. Furthermore, a survey
of the drivers revealed that the motivation underlying close-
following on roads stemmed primarily from an increased
desire and urgency to overtake. That is, male drivers in this
study could have consciously traded-off safety for a shorter
travel time.
When overtaking in the opposing lane, mature participants
adopted faster speeds than young participants. It was inter-
esting to note, however, that both young and mature drivers
found it necessary to overtake at speeds in excess of the
100 kph sign-posted speed limit for the area, although they
were following a vehicle traveling at only 80 kph. The signif-
icance of this nding has implications for risk-involvement.
That is, although the experienced drivers tended to hurry
the maneuver by speeding excessively, they also demon-
strated more caution by making it a priority to return to their
own lane as quickly as possible. In contrast, the inexperi-
enced drivers spent relatively more time in the opposing lane
and this could be interpreted as riskier behavior. These results
reinforce the belief that younger and less experienced drivers,
whether consciously or not, take more risks when driving
(Ulleberg, 2001; Turner andMcClure, 2003). That is, younger
and less experienced drivers tend to overestimate their driv-
ing skills and underestimate potential trafc hazards on the
road (Summala, 1987; Trankle et al., 1990; Deery, 1999).
5. Conclusion
It is evident from the results of this study that there
are differences in the driver behavior of young and mature
individuals. In most cases young drivers demonstrated an
increased tendency to engage in risky tactics when compared
to mature drivers. However, it is still unclear whether the
increased risk-taking exhibited by young drivers is a result
of a lack of driving skills, poor awareness of a potentially
hazardous situation, or a combination of both factors. Mature
drivers demonstrated decits in their ability to detect the pres-
ence of potential hazards on curved sections of road, showed
1064 S. Leung, G. Starmer / Accident Analysis and Prevention 37 (2005) 10561065
relative inaccuracy in judgments of arrival-time, and tended
to speed while overtaking. Further research into the behav-
iors and attitudes of young, middle-aged and elderly drivers
in high and low-risk situations is necessary to verify the
appropriate focus for education. Experimental scenarios in
which participants received immediate feedback about their
performance would allow the driver to recognize a poten-
tially dangerous situation and, at the same time, instill in the
driver an awareness of their capabilities and limitations on the
road.
References
Allen, R.W., Jex, H.R., McRuer, D.T., DiMarco, R.J., 1975. Alcohol
effects on driving behavior and performance in a car simulator. IEEE
Trans. Syst. Man Cybernet. 5 (5), 498505.
Australian Transport Safety Bureau, 2004. Overtaking crash statistical
summary 1998, Canberra.
Blakemore, M.R., Snowden, R.J., 1999. The effect of contrast upon per-
ceived speed: a general phenomenon? Perception 28 (1), 3348.
Buikhuisen, W., Jongman, R.W., 1972. Trafc perception under the inu-
ence of alcohol. Quart. J. Stud. Alcohol 33 (3-A), 800806.
Caird, J.K., Hancock, P.A., 1994. The perception of arrival time for differ-
ent oncoming vehicles at an intersection. Ecol. Psychol. 6 (2), 83109.
Carscadden, J.L., 2002. Alcohol effects on visual attention: the impact of
information processing. Dissertation Abstracts International: Section
B: The Sciences and Engineering 62 (12-B).
Cavallo, V., Laurent, M., 1988. Visual information and skill level in time-
to-collision estimation. Perception 17 (5), 623632.
Cohen, A.S., Studach, H., 1977. Eye movements while driving cars around
curves. Percept. Motor Skills 44 (3, Pt 1), 683689.
Compton, R.P., Blomberg, R.D., Moskowitz, H., Burns, M., Peck, R.C.,
Fiorentino, D., 2002. Crash risk of alcohol impaired driving. In: Pro-
ceedings of the 16th International Conference on Alcohol Drugs and
Trafc Safety, Montreal.
Crundall, D.E., Underwood, G., Chapman, P.R., 1998. How much do
novice drivers see? The effects of demand on visual search strate-
gies in novice and experienced drivers. In: Underwood, G. (Ed.), Eye
Guidance in Reading and Scene Perception, pp. 395417.
Deery, H.A., 1999. Hazard and risk perception among young novice
drivers. J. Safety Res. 30 (4), 225236.
Deery, H.A., Love, A.W., 1996a. The Driving Expectancy Questionnaire:
development, psychometric assessment and predictive utility among
young drink-drivers. J. Stud. Alcohol 57 (2), 193202.
Deery, H.A., Love, A.W., 1996b. The effect of a moderate dose of alco-
hol on the trafc hazard perception prole of young drink-drivers.
Addiction 91 (6), 815827.
DeJoy, D.M., 1992. An examination of gender differences in trafc acci-
dent risk perception. Accident Anal. Prevent. 24 (3), 237246.
DeLucia, P.R., Bleckley, M., Meyer, L.E., Bush, J.M., 2003. Judgments
about collision in younger and older drivers. Trans. Res. Part F: Trafc
Psychol. Behav. 6 (1), 6380.
Dubowski, K.M., 1974. Biological aspects of breath-alcohol analysis.
Clin. Chem. 20, 294299.
Finn, P., Bragg, B.W., 1986. Perception of the risk of an accident by
young and older drivers. Accident Anal. Prevent. 18 (4), 28998.
Gustafson, R., 1986. Alcohol and vigilance performance: effect of small
doses of alcohol on simple visual reaction time. Percept. Motor Skills
62 (3), 951955.
Hancock, P.A., Manser, M.P., 1997. Time-to-contact: more than tau alone.
Ecol. Psychol. 9 (4), 265297.
Hoffmann, E.R., Mortimer, R.G., 1994. Drivers estimates of time to
collision. Accid Anal. Prevent. 26 (4), 511520.
Kearney, S.A., Guppy, A., 1988. The effects of alcohol on speed per-
ception in a closed-course driving situation. J. Stud. Alcohol 49 (4),
340345.
Koelega, H.S., 1995. Alcohol and vigilance performance a review. Psy-
chopharmacology 118 (3), 233249.
McLeod, R.W., Ross, H.E., 1983. Optic-ow and cognitive factors in
time-to-collision estimates. Perception 12 (4), 417423.
Miura, T., 1990. Active function of eye movement and useful eld of view
in a realistic setting. In: Groner, R., dYdewalle, G. (Eds.), From Eye
to Mind: Information Acquisition in Perception, Search, and Reading.
Studies in Visual Information Processing, pp. 119127.
Moskowitz, H., 1973. Laboratory studies of the effects of alcohol on
some variables related to driving. J. Safety Res. 5 (3), 185199.
Moskowitz, H., Burns, M.M., Williams, A.F., 1985. Skills performance
at low blood alcohol levels. J. Stud. Alcohol 46 (6), 4825.
Mourant, R.R., Rockwell, T.H., 1970. Mapping eye-movement patterns
to the visual scene in driving: an exploratory study. Human Factors
12 (1), 8187.
Mullin, J.P., 2000. Reaction time in normal aging: an examination of the
complexity hypothesis of cognitive slowing. Dissertation Abstracts
International: Section B: The Sciences and Engineering 60 (11-B),
5805, US: Univ Microlms International.
Quimby, A.R., 1987. Perceptual abilities of accident-involved drivers. J.
Safety Res. 18 (1), 45.
Rajalin, S., Hassel, S.-O., Summala, H., 1997. Close-following drivers
on two-lane highways. Accident Anal. Prevent. 29 (6), 723
729.
Ratcliff, R., Thapar, A., McKoon, G., 2001. The effects of aging on
reaction time in a signal detection task. Psychol. Aging 16 (2), 323
341.
Roads and Trafc Authority of NSW, 2004. Road Trafc Accidents in
New South Wales 2003. Roads and Trafc Authority Road Safety
Strategy Branch, New South Wales.
Roehrs, T., Zwyghuizen-Doorenbos, A., Knox, M., Moskowitz, H., Roth,
T., 1992. Sedating effects of ethanol and time of drinking. Alcoholism:
Clin. Exp. Res. 16 (3), 5537.
Salvatore, S., 1967. Vehicle speed estimation from visual stimuli. Public
Roads, 128131.
Schiff, W., Detwiler, M.L., 1979. Information used in judging impending
collision. Perception 8 (6), 647658.
Schiff, W., Oldak, R., 1990. Accuracy of judging time to arrival: effects
of modality, trajectory, and gender. J. Exp. Psychol.: Human Percept.
Perform. 16 (2), 303316.
Schiff, W., Oldak, R., Shah, V., 1992a. Aging persons estimates of vehic-
ular motion. Psychol. Aging 7 (4), 518525.
Schiff, W., Oldak, R., Shah, V., 1992b. Aging Persons estimates of vehic-
ular motion. Psychol. Aging 7 (4), 518525.
Shinar, D., 1978. Psychology on the Road: The Human Factors in Trafc
Safety. Wiley, New York.
Shinar, D., McDowell, E.D., Rockwell, T.H., 1977. Eye movements in
curve negotiation. Human Factors 19 (1), 6371.
Sidaway, B., Fairweather, M., Sekiya, H., McNitt-Gray, J., 1996. Time-
to-collision estimation in a simulated driving task. Human Factors 38
(1), 101113.
Slemeyer, A., 1986. The Alcomat: a new mobile evidential breath tester.
Alcohol Drugs Traff. Safety, 12451255.
Snowden, R.J., Stimpson, N., Ruddle, R.A., 1998. Speed perception fogs
up as visibility drops. Nature 392, 450.
Starmer, G.A., Bird, K.D., 1984. Investigating drug ethanol interactions.
Brit. J. Clin. Pharmacol. 18 (1), 27S35S.
Summala, H., 1987. Young driver accidents: risk taking or failure of
skills? Alcohol Drugs Driving 3 (34), 7991.
Trankle, U., Gelau, C., Metker, T., 1990. Risk perception and age-specic
accidents of young drivers. Accident Anal. Prevent. 22 (2), 119125.
Triggs, T.J., 1986. Speed estimation. In: Peters, G.A., Peters, B.J. (Eds.),
Automotive Engineering and Litigation. Garland Press, New York, pp.
95124.
S. Leung, G. Starmer / Accident Analysis and Prevention 37 (2005) 10561065 1065
Triggs, T.J., Smith, K.B., 1996. Young driver research program: digest
of reports and principle ndings of the research (CR 164). Federal
Ofce of Road Safety, Canberra.
Turner, C., McClure, R., 2003. Age and gender differences in risk-taking
behaviour as an explanation for high incidence of motor vehicle
crashes as a driver in young males. Injury Contr. Safety Promot. 10
(3), 123130.
Ulleberg, P., 2001. Personality subtypes of young drivers. Relationship to
risk-taking preferences, accident involvement, and response to a trafc
safety campaign. Transport. Res. Part F: Trafc Psychol. Behav. 4 (4),
279297.
Widmark, E.M.P., 1932. Die theoretischen Grundlagen und die prak-
tische Verwendbarkeit der gerichtlich-medinizischen Alkoholbestim-
mung. Urban und Schwarzenberg, Berlin.
Wooldridge, M.D., Fitzpatrick, K., Koppa, R., Bauer, K., 2000. Effects of
horizontal curvature on driver visual demand. Geometric Des. Effects
Trafc Oper. 2000 (1737), 7177.